KEMBAR78
UQ_Cybercrime_Professionalism_Lecture_2024_07.pdf
Cyber Crime Professionalism
Jon Oliver
Adjunct Professor at Uni Qld
My Background
Academic background in Machine Learning
Building antispam filters 2002 -> 2008
Cyber security 2008 to 2024
Academia Industry
Sweet Spot
Technical Capability
1999
Melissa
Virus
2002
Spamming
2005
Botnets
2007
Webthreats
2010 Zeus:
Banking
Trojan
2017 Fileless
/ Living off
the Land
2011
Blackhole:
Exploit Kits
2014
CryptoLocker:
Ransomware
2016: Cridex:
Breach based
Ransomware
2017
Coinminer
2019
Ransomware
as a Service
(RaaS)
2021 RaaS
double
extortion
Business Model
Cybercrime over 25 years…
Cyber Criminal Professionalism
• Often requires no technical innovation
• Excellent execution
• Measuring every aspect of a cyber attack
• Presenting this data in a UI
• Optimizing each step
• Perfecting their business model
Preventing Cyber Crime
Technical Knowledge
+
Understanding the business model
(study underground forums,
read reports of criminals being arrested, …)
1. Measuring
BlackHole Exploit Kit (BHEK)
Ref: http://media.blackhat.com/bh-us-12/Briefings/Jones/BH_US_12_Jones_State_Web_Exploits_Slides.pdf
compromised.com/fHNDZW/index.html
Malicious
web server
The Threat: Exploit Kit Spam
BHEK UI
Advertising / Release Notes
BlackHole exploit Kit 2.0
Рады приветствовать вам совершенно новую версию связки эксплойтов. За более чем 2 года
существования нашего проекта, старый движок связки изрядно заездили и затаскали, АВ компании
стали очень быстро распознавать по неким критериям что это BlackHole и помечать ее как malware. В
новой версии мы переписали все с нуля, причем с нуля переписана не только часть с выдачей
эксплойтов, но и админ панель.
Из новшеств по выдаче:
1. Мы максимально защитили эксплойты от автоскачивания их АВ компаниями, теперь генерируется
динамический УРЛ, который действителен в течении нескольких секунд, нужных лишь для одного
заражения потенциально заходящего на ссылку человека.
2. Теперь так же максимально защищен и ваш ехе, АВ компания не сможет его просто так скачать, что
позволит держать ваши ехе максимально долго в чистоте.
3. JAR и PDF файлы выдаются только тем версиям плагинов, которые уязвимы, если плагин не уязвим,
сплойт не выдается, и не пачкается лишний раз.
BlackHole exploit Kit 2.0
Are pleased to welcome you to a brand new version of the bundle of exploits. For more than
2 years of existence of our project, the old engine arrival and ligaments badly worn, AV
companies have become very quick to recognize that this kind of criteria BlackHole and flag
it as malware. In the new version we have rewritten from scratch, and re-written from
scratch is not only part of the issuance of exploits, but also the admin panel.
Of the innovations on the issue:
1. Implemented maximum protection from Automatic systems for downloading exploits,
used by AV companies: generate a dynamic URL, which is valid for a few seconds, you need
only to one victim at a time.
2. Now, Your executable also protected from multiple downloads, AV company can not just
download it, which will keep your exe as long as clean.
3. JAR and PDF exploits show only for detected vulnerable versions of plug-ins if the plug is
not vulnerable,exploits not issued, and not get in detection loop.
…
JRE <=
1.6.0_27
1.5.0_31
JRE <=
1.7.0_02
1.6.0_30
1.5.0_33
JRE <=
1.7.0_03
1.6.0_32
1.5.0_32
JRE <=
1.7.0_06
JRE <=
1.7.0_07
JRE <=
1.7.0_10
JRE <=
1.7.0_15
Flash <=
11.5.502.1
46
CVE-2011-3544
gogonaggogonaExp.class
CVE-2012-0507
GondadExx.Ohno.class
CVE-2012-1723
Gond1723.Gondattack.class
CVE-2012-4681
cve2012xxxx.Gondvv.class
CVE-2012-5076
gond20125076.Gondqq.class
CVE-2013-0422
Xml20130422.XML20130422
.class
CVE-2013-1493
yQCI.class
CVE-2013-0634
xxxx.swf
MSIE <= 7
CVE-2012-1889
javascript
CVE-2012-4969
javascript
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
Yes
No
No
No
No
No
No
No
No
No
Dispatcher
SWF Vulnerability
MSIE Vulnerability
Infection!
Pain to crawl
Anti-crawling and serving
vulnerable documents by
client environment
Serve blank
page
Client IP in
black list?
Yes
No
Landing
Page
Cybercriminal Firewall Rules
iptables -A INPUT -s aa.119.145.16/28 -j DROP #Kaspersky Labs
iptables -A INPUT -s aa.119.145.16/28 -j DROP #Kaspersky Labs
…
iptables -A INPUT -s bb.188.199.136/29 -j DROP #SYMANTECGMBH
iptables -A INPUT -s bb.227.131.216/29 -j DROP #Symantec Ltd
…
iptables -A INPUT -s xx.110.201.32/29 -j DROP #TREND MICRO DEUTSCHLAND GMBH
iptables -A INPUT -s yy.93.166.128/28 -j DROP #Trend Micro UK Limited
iptables -A INPUT -s zz.23.206.192/29 -j DROP #TRENDMICROITALYS.R.L.
…
Cybercriminal Firewall Rules
/dev/null
Security
Companies
2. Evading Security Solutions
Scan4U
Scan4U Competitors
AVDetect
VirusCheckMate.
https://documents.trendmicro.com/assets/white_papers/wp-the-rise-and-fall-of-scan4you.pdf
Professional Operations:
TorrentLocker 2014-2017
Ref: http://www.trendmicro.com.au/cloud-content/au/pdfs/security-intelligence/white-papers/australian_web_threat_landscape_-v7.pdf
2
Evading Spam Filters
Technique TorrentLocker Outbreaks
Honeypots Send email to legitimate email accounts
IP Reputation Use rented servers
Header analysis / phishing detection Don’t spoof
Use email authentication
Antispam Content Send plausible content
Phishing lists Compromise many legitimate web-
servers
Misc Low volume. Optimize timing
TorrentLocker – A typical spam run
• state-nsw-gov.com, state1-nsw-gov.com, state2-nsw-gov.com, state3-nsw-
gov.com, state4-nsw-gov.com sent a spam outbreak
• Between 1am and 9am
• Constituted approx 0.13% of email sent to Australian email addresses
• Appeared to be sent to a carefully culled email address list
• < 1% were to invalid email addresses
• Used Email Authentication (DKIM and SPF)
TorrentLocker – using Email Authentication
;; ANSWER SECTION:
info-service-osr.org. 3593 IN TXT
"v=DMARC1; p=reject; rua=mailto:admin@info-service-osr.org"
info-service-osr.org. 3593 IN TXT
"v=spf1 ip4:194.58.1.1/16 ip4:193.124.1.1/16 ip4:151.248.1.1/16 a
mx ~all"
TorrentLocker – compromised webservers
96 on a single outbreak
hxxp://ahaliapublicschool.org/files/awmGz1ZvAL4zyy.php
hxxp://alokitogrambangla.org/jscripts/P8sMnGioHtzHwwlvzqklwqGuUeh7Z54.php
hxxp://alternativ-medizin.org/cms2/media/WHw0iGEgoEdkWEn.php
...
hxxp://ttcchurch.org/uploadify/ND4wLfkrAizuISqskln2AX2WJ.php
hxxp://turismodegalicia.org/uploadify/dh4F61g7aAupmqaiEDKgNZcY.php
hxxp://www.bkmlaw.org/8XNjrWBqEIFXWA45N1eoCrBCjpGx.php
TorrentLocker – using Captcha
Cyber Business Models
Value of Compromised Machine
Monetization Approach Return
Spam $
Banking Trojan $$
Stealing Intellectual Property $$
Ransomware (Opportunistic – spam based) $$$
Coinmining $ to $$$
Ransomware (Breach based) $$$$
Double Extortion Ransomware $$$$$
https://www.techtarget.com/searchsecurity/definition/double-extortion-ransomware
Security: Before / After Ransomware
Criteria Pre Ransomware Post Ransomware
Detection urgency Within a few hours was
often adequate
At time of infection
Errors to avoid Critical False Positives Missed detections
Impact of missed
detection
Often small
Use some resources:
network / power / …
Files encrypted + more
Blackmail / extortion
ML Strength
Comment 1
Industry was too slow in adopting
Machine Learning (ML) for malware
https://www.cpomagazine.com/cyber-security/as-ukraine-war-rages-conti-ransomware-gang-throws-support-behind-russian-government/
2021:
4640 papers on
ML for
detecting
ransomware
2022:
17,100
papers on
ML for
detecting
ransomware
So why are ransomware
attacks still happening?
ML evasion?
Something else?
x
Source: “Intriguing properties of Neural Networks”, Szegedy et al., Feb 2014, https://arxiv.org/pdf/1312.6199.pdf
Tricking ML
Speaker
Insect
Dog
Ostrich
x
Not Quite…
Breach-based/ Fileless Ransomware
Ref: https://blogs.vmware.com/cloud/2020/08/27/investigating-stopping-fileless-attacks-carbon-black-cloud/
https://community.carbonblack.com/t5/Knowledge-Base/Cb-Defense-Understanding-quot-repmgr-quot-or-quot-Cb-Defense/ta-p/47542
Ransomware by RDP Abuse: Phase 1
Step 1
RDP
server
Brute force:
Johnsmith abc123
Maryjones qw3rty
…
Bot
Step 2
Found User and
Password
admin l3tm31n!
Bot
Bot
Step 3
Bad actor sells RDP credentials
on Underground forum
Bad
Actor
Phase 1
(cont.)
Sells
RDP
credentials
Ransomware by RDP Abuse: Phase 2
Ransomware (RW) Actor purchases credentials on underground forum
Bad Actor
RW Actor
© 2020 Trend Micro Inc.
41
A growing category of cyber-crime consists of breaking into corporate networks and doing
nothing else – except selling that illicit access to others for about $7,000 a go, says infosec
biz Digital Shadows.
The firm described what it said was a "notable increase" in the number of stolen-creds-for-
sale postings, with the average price for a working access method being $7,100 and
comprising around 17 per cent of listings seen by Digital Shadows. This price increases to
$9,800 for remote desktop protocol (RDP) access, echoing research from ESET showing a
700 per cent increase in the number of RDP access attempts during 2020.
Ransomware by RDP Abuse: Phase 3
RDP
server
1. Ransomware (RW) actor logs in with stolen credentials
2. Turns off Security / Makes multiple attempts / Privilege Escalation / …
3. Infects with ransomware
OR
1. Ransomware (RW) actor logs in with stolen credentials
2. Fileless ransomware
RW Actor
https://www.vmware.com/content/dam/digitalmarketing/vmware/en/pdf/docs/vmwcb-solution-brief-vmware-carbon-black-next-gen-av.pdf
Ransomware and ML
Machine Learning does not
work if someone turns it off…
Many ransomware groups
have adopted this approach
https://www.microsoft.com/security/blog/2020/04/28/ransomware-groups-continue-to-target-healthcare-critical-services-heres-how-to-reduce-risk/
Advantages of Breach Based Ransomware
• Can target victims who may be in a situation
where they feel compelled to pay ransom
• Can increase the amount of money you
demand ($$$$)
Cyber criminal professionalism
Perfect the Business Model
Perfect the Business Model
Optimising the infection process
Many elements of a cyber attack are available for rent
Pay $$$ for premium services (the latest vulnerability, freshly compromised credentials)
Customer service
Release notes
Tracking every component of the infection process

UQ_Cybercrime_Professionalism_Lecture_2024_07.pdf

  • 1.
    Cyber Crime Professionalism JonOliver Adjunct Professor at Uni Qld
  • 2.
    My Background Academic backgroundin Machine Learning Building antispam filters 2002 -> 2008 Cyber security 2008 to 2024 Academia Industry Sweet Spot
  • 3.
    Technical Capability 1999 Melissa Virus 2002 Spamming 2005 Botnets 2007 Webthreats 2010 Zeus: Banking Trojan 2017Fileless / Living off the Land 2011 Blackhole: Exploit Kits 2014 CryptoLocker: Ransomware 2016: Cridex: Breach based Ransomware 2017 Coinminer 2019 Ransomware as a Service (RaaS) 2021 RaaS double extortion Business Model Cybercrime over 25 years…
  • 4.
    Cyber Criminal Professionalism •Often requires no technical innovation • Excellent execution • Measuring every aspect of a cyber attack • Presenting this data in a UI • Optimizing each step • Perfecting their business model
  • 5.
    Preventing Cyber Crime TechnicalKnowledge + Understanding the business model (study underground forums, read reports of criminals being arrested, …)
  • 6.
    1. Measuring BlackHole ExploitKit (BHEK) Ref: http://media.blackhat.com/bh-us-12/Briefings/Jones/BH_US_12_Jones_State_Web_Exploits_Slides.pdf
  • 7.
  • 8.
  • 9.
    Advertising / ReleaseNotes BlackHole exploit Kit 2.0 Рады приветствовать вам совершенно новую версию связки эксплойтов. За более чем 2 года существования нашего проекта, старый движок связки изрядно заездили и затаскали, АВ компании стали очень быстро распознавать по неким критериям что это BlackHole и помечать ее как malware. В новой версии мы переписали все с нуля, причем с нуля переписана не только часть с выдачей эксплойтов, но и админ панель. Из новшеств по выдаче: 1. Мы максимально защитили эксплойты от автоскачивания их АВ компаниями, теперь генерируется динамический УРЛ, который действителен в течении нескольких секунд, нужных лишь для одного заражения потенциально заходящего на ссылку человека. 2. Теперь так же максимально защищен и ваш ехе, АВ компания не сможет его просто так скачать, что позволит держать ваши ехе максимально долго в чистоте. 3. JAR и PDF файлы выдаются только тем версиям плагинов, которые уязвимы, если плагин не уязвим, сплойт не выдается, и не пачкается лишний раз.
  • 10.
    BlackHole exploit Kit2.0 Are pleased to welcome you to a brand new version of the bundle of exploits. For more than 2 years of existence of our project, the old engine arrival and ligaments badly worn, AV companies have become very quick to recognize that this kind of criteria BlackHole and flag it as malware. In the new version we have rewritten from scratch, and re-written from scratch is not only part of the issuance of exploits, but also the admin panel. Of the innovations on the issue: 1. Implemented maximum protection from Automatic systems for downloading exploits, used by AV companies: generate a dynamic URL, which is valid for a few seconds, you need only to one victim at a time. 2. Now, Your executable also protected from multiple downloads, AV company can not just download it, which will keep your exe as long as clean. 3. JAR and PDF exploits show only for detected vulnerable versions of plug-ins if the plug is not vulnerable,exploits not issued, and not get in detection loop. …
  • 11.
    JRE <= 1.6.0_27 1.5.0_31 JRE <= 1.7.0_02 1.6.0_30 1.5.0_33 JRE<= 1.7.0_03 1.6.0_32 1.5.0_32 JRE <= 1.7.0_06 JRE <= 1.7.0_07 JRE <= 1.7.0_10 JRE <= 1.7.0_15 Flash <= 11.5.502.1 46 CVE-2011-3544 gogonaggogonaExp.class CVE-2012-0507 GondadExx.Ohno.class CVE-2012-1723 Gond1723.Gondattack.class CVE-2012-4681 cve2012xxxx.Gondvv.class CVE-2012-5076 gond20125076.Gondqq.class CVE-2013-0422 Xml20130422.XML20130422 .class CVE-2013-1493 yQCI.class CVE-2013-0634 xxxx.swf MSIE <= 7 CVE-2012-1889 javascript CVE-2012-4969 javascript Yes Yes Yes Yes Yes Yes Yes Yes Yes No No No No No No No No No Dispatcher SWF Vulnerability MSIE Vulnerability Infection! Pain to crawl Anti-crawling and serving vulnerable documents by client environment Serve blank page Client IP in black list? Yes No Landing Page
  • 12.
    Cybercriminal Firewall Rules iptables-A INPUT -s aa.119.145.16/28 -j DROP #Kaspersky Labs iptables -A INPUT -s aa.119.145.16/28 -j DROP #Kaspersky Labs … iptables -A INPUT -s bb.188.199.136/29 -j DROP #SYMANTECGMBH iptables -A INPUT -s bb.227.131.216/29 -j DROP #Symantec Ltd … iptables -A INPUT -s xx.110.201.32/29 -j DROP #TREND MICRO DEUTSCHLAND GMBH iptables -A INPUT -s yy.93.166.128/28 -j DROP #Trend Micro UK Limited iptables -A INPUT -s zz.23.206.192/29 -j DROP #TRENDMICROITALYS.R.L. …
  • 13.
  • 14.
  • 15.
  • 16.
  • 17.
    Professional Operations: TorrentLocker 2014-2017 Ref:http://www.trendmicro.com.au/cloud-content/au/pdfs/security-intelligence/white-papers/australian_web_threat_landscape_-v7.pdf
  • 18.
  • 19.
    Evading Spam Filters TechniqueTorrentLocker Outbreaks Honeypots Send email to legitimate email accounts IP Reputation Use rented servers Header analysis / phishing detection Don’t spoof Use email authentication Antispam Content Send plausible content Phishing lists Compromise many legitimate web- servers Misc Low volume. Optimize timing
  • 20.
    TorrentLocker – Atypical spam run • state-nsw-gov.com, state1-nsw-gov.com, state2-nsw-gov.com, state3-nsw- gov.com, state4-nsw-gov.com sent a spam outbreak • Between 1am and 9am • Constituted approx 0.13% of email sent to Australian email addresses • Appeared to be sent to a carefully culled email address list • < 1% were to invalid email addresses • Used Email Authentication (DKIM and SPF)
  • 21.
    TorrentLocker – usingEmail Authentication ;; ANSWER SECTION: info-service-osr.org. 3593 IN TXT "v=DMARC1; p=reject; rua=mailto:admin@info-service-osr.org" info-service-osr.org. 3593 IN TXT "v=spf1 ip4:194.58.1.1/16 ip4:193.124.1.1/16 ip4:151.248.1.1/16 a mx ~all"
  • 22.
    TorrentLocker – compromisedwebservers 96 on a single outbreak hxxp://ahaliapublicschool.org/files/awmGz1ZvAL4zyy.php hxxp://alokitogrambangla.org/jscripts/P8sMnGioHtzHwwlvzqklwqGuUeh7Z54.php hxxp://alternativ-medizin.org/cms2/media/WHw0iGEgoEdkWEn.php ... hxxp://ttcchurch.org/uploadify/ND4wLfkrAizuISqskln2AX2WJ.php hxxp://turismodegalicia.org/uploadify/dh4F61g7aAupmqaiEDKgNZcY.php hxxp://www.bkmlaw.org/8XNjrWBqEIFXWA45N1eoCrBCjpGx.php
  • 23.
  • 24.
  • 25.
    Value of CompromisedMachine Monetization Approach Return Spam $ Banking Trojan $$ Stealing Intellectual Property $$ Ransomware (Opportunistic – spam based) $$$ Coinmining $ to $$$ Ransomware (Breach based) $$$$ Double Extortion Ransomware $$$$$
  • 26.
  • 27.
    Security: Before /After Ransomware Criteria Pre Ransomware Post Ransomware Detection urgency Within a few hours was often adequate At time of infection Errors to avoid Critical False Positives Missed detections Impact of missed detection Often small Use some resources: network / power / … Files encrypted + more Blackmail / extortion ML Strength
  • 28.
    Comment 1 Industry wastoo slow in adopting Machine Learning (ML) for malware
  • 30.
  • 31.
    2021: 4640 papers on MLfor detecting ransomware
  • 32.
  • 33.
    So why areransomware attacks still happening? ML evasion? Something else?
  • 34.
    x Source: “Intriguing propertiesof Neural Networks”, Szegedy et al., Feb 2014, https://arxiv.org/pdf/1312.6199.pdf Tricking ML Speaker Insect Dog Ostrich
  • 35.
  • 36.
    Not Quite… Breach-based/ FilelessRansomware Ref: https://blogs.vmware.com/cloud/2020/08/27/investigating-stopping-fileless-attacks-carbon-black-cloud/ https://community.carbonblack.com/t5/Knowledge-Base/Cb-Defense-Understanding-quot-repmgr-quot-or-quot-Cb-Defense/ta-p/47542
  • 37.
    Ransomware by RDPAbuse: Phase 1 Step 1 RDP server Brute force: Johnsmith abc123 Maryjones qw3rty … Bot Step 2 Found User and Password admin l3tm31n! Bot Bot Step 3 Bad actor sells RDP credentials on Underground forum Bad Actor
  • 38.
  • 39.
    Ransomware by RDPAbuse: Phase 2 Ransomware (RW) Actor purchases credentials on underground forum Bad Actor RW Actor
  • 40.
    © 2020 TrendMicro Inc. 41 A growing category of cyber-crime consists of breaking into corporate networks and doing nothing else – except selling that illicit access to others for about $7,000 a go, says infosec biz Digital Shadows. The firm described what it said was a "notable increase" in the number of stolen-creds-for- sale postings, with the average price for a working access method being $7,100 and comprising around 17 per cent of listings seen by Digital Shadows. This price increases to $9,800 for remote desktop protocol (RDP) access, echoing research from ESET showing a 700 per cent increase in the number of RDP access attempts during 2020.
  • 41.
    Ransomware by RDPAbuse: Phase 3 RDP server 1. Ransomware (RW) actor logs in with stolen credentials 2. Turns off Security / Makes multiple attempts / Privilege Escalation / … 3. Infects with ransomware OR 1. Ransomware (RW) actor logs in with stolen credentials 2. Fileless ransomware RW Actor
  • 42.
  • 43.
    Ransomware and ML MachineLearning does not work if someone turns it off… Many ransomware groups have adopted this approach
  • 44.
  • 45.
    Advantages of BreachBased Ransomware • Can target victims who may be in a situation where they feel compelled to pay ransom • Can increase the amount of money you demand ($$$$)
  • 46.
    Cyber criminal professionalism Perfectthe Business Model Perfect the Business Model Optimising the infection process Many elements of a cyber attack are available for rent Pay $$$ for premium services (the latest vulnerability, freshly compromised credentials) Customer service Release notes Tracking every component of the infection process