APT Meets GPT: Targeted Operations with Untamed LLMs
Over the course of three months, Volexity observed UTA0388 using various themes and fictional identities across dozens of spear phishing campaigns. As time passed, Volexity observed UTA0388 broaden their targeting and send emails in a variety of different languages, including English, Chinese, Japanese, French, and German. In most cases, the initial email sent by UTA0388 contained a link to phishing content hosted on a cloud-based service that would lead to malware.
AI Analysis
Technical Summary
The UTA0388 threat actor has been observed conducting targeted spear phishing campaigns over a three-month period, utilizing large language models (LLMs) to generate convincing and diverse phishing emails. These emails employ various fictional identities and themes, sent in multiple languages including English, Chinese, Japanese, French, and German, to broaden their reach and evade detection. The initial phishing emails typically contain links to cloud-hosted phishing sites that serve malware payloads. The malware uses PowerShell scripts and achieves persistence through scheduled tasks and hijacking execution flows. Command and control communications leverage advanced techniques such as WebSocket-based Govershell C2 channels, enabling stealthy and resilient remote control. The malware delivery often involves archive files (ZIP, RAR) to bypass security controls. The threat actor employs a wide range of MITRE ATT&CK techniques including T1071 (Application Layer Protocol), T1102 (Web Service), T1123 (Audio Capture), T1134 (Access Token Manipulation), T1036 (Masquerading), T1105 (Ingress Tool Transfer), T1059 (Command and Scripting Interpreter), T1574 (Hijack Execution Flow), T1053 (Scheduled Task), and T1566 (Phishing). Although no known exploits are currently in the wild, the campaign demonstrates a sophisticated blend of social engineering and technical exploitation, leveraging LLMs to enhance phishing effectiveness and evade traditional detection methods. The use of multiple languages and cloud infrastructure indicates a broad targeting strategy, likely aiming at high-value organizations globally, including Europe.
Potential Impact
European organizations face significant risks from this threat due to the use of multi-lingual spear phishing campaigns that can bypass traditional email security filters and deceive users across different countries. The malware's use of PowerShell and persistence mechanisms can lead to long-term footholds within networks, enabling data exfiltration, espionage, or disruption. The WebSocket-based Govershell C2 channel allows stealthy communication that may evade network monitoring tools. Organizations relying heavily on cloud services are particularly vulnerable since the phishing content is hosted on cloud platforms, complicating blocking efforts. The broad language support increases the likelihood of successful compromise across diverse European workforces. Potential impacts include loss of sensitive data, intellectual property theft, operational disruption, and reputational damage. The medium severity rating reflects the threat's sophistication and potential for significant harm if not mitigated effectively.
Mitigation Recommendations
1. Implement advanced email filtering solutions capable of detecting and blocking multi-lingual spear phishing attempts, including heuristic and AI-driven analysis to identify LLM-generated content. 2. Conduct targeted user awareness training in multiple languages (English, French, German, etc.) focusing on recognizing phishing tactics and suspicious links. 3. Monitor and restrict PowerShell usage on endpoints, enforcing execution policies and logging all PowerShell activity for anomaly detection. 4. Deploy endpoint detection and response (EDR) tools with capabilities to detect persistence techniques such as scheduled tasks and hijacked execution flows. 5. Monitor network traffic for unusual WebSocket connections, especially those communicating with cloud-hosted domains, and implement strict egress filtering. 6. Regularly audit and restrict use of archive files (ZIP, RAR) in email attachments and downloads, scanning them thoroughly for malicious content. 7. Maintain up-to-date threat intelligence feeds and integrate indicators of compromise (IOCs) related to UTA0388 into security monitoring systems. 8. Enforce multi-factor authentication (MFA) and least privilege principles to limit the impact of compromised credentials. 9. Conduct regular phishing simulation exercises tailored to the languages and themes used by this threat actor to improve user resilience. 10. Collaborate with cloud service providers to quickly identify and take down phishing sites hosted on their platforms.
Affected Countries
Germany, France, United Kingdom, Netherlands, Belgium, Switzerland, Luxembourg
Indicators of Compromise
- url: https://aesthetic-donut-1af43s2.netlify.app/file/rar
- url: https://aesthetic-donut-1af43s2.netlify.app/index/file/A_Introduction_Docs_v00546823.rar
- ip: 185.144.28.68
- hash: cf8d7017d025dd7ad65a946e33ed23d8
- hash: 7351db2af3139c2b8eed820c9938060b18ba4a99
- hash: 0414217624404930137ec8f6a26aebd8a3605fe089dbfb9f5aaaa37a9e2bad2e
- hash: 126c3d21a1dae94df2b7a7d0b2f0213eeeec3557c21717e02ffaed690c4b1dbd
- hash: 2ffe1e4f4df34e1aca3b8a8e93eee34bfc4b7876cedd1a0b6ca5d63d89a26301
- hash: 4c041c7c0d5216422d5d22164f83762be1e70f39fb8a791d758a816cdf3779a9
- hash: 53af82811514992241e232e5c04e5258e506f9bc2361b5a5b718b4e4b5690040
- hash: 7d7d75e4d524e32fc471ef2d36fd6f7972c05674a9f2bac909a07dfd3e19dd18
- hash: 88782d26f05d82acd084861d6a4b9397d5738e951c722ec5afed8d0f6b07f95e
- hash: 998e314a8babf6db11145687be18dc3b8652a3dd4b36c115778b7ca5f240aae4
- hash: a5ee55a78d420dbba6dec0b87ffd7ad6252628fd4130ed4b1531ede960706d2d
- hash: ad5718f6810714bc6527cc86d71d34d8c556fe48706d18b5d14f0261eb27d942
- hash: fbade9d8a040ed643b68e25e19cba9562d2bd3c51d38693fe4be72e01da39861
- ip: 104.194.152.137
- ip: 104.194.152.152
- ip: 31.192.234.22
- ip: 45.141.139.222
- ip: 74.119.193.175
- ip: 80.85.154.48
- ip: 80.85.156.234
- ip: 80.85.157.117
- ip: 82.118.16.173
- url: http://80.85.154.48:443
- url: http://80.85.157.117:443
- url: http://82.118.16.173:443
- url: http://api.twmoc.info/ws
- url: http://onedrive.azure-app.store/ws
- url: http://outlook.windows-app.store/ws
- url: https://app-site-association.cdn-apple.info:443/updates.rss
- domain: azure-app.store
- domain: cdn-apple.info
- domain: doccloude.info
- domain: sliddeshare.online
- domain: twmoc.info
- domain: windows-app.store
- domain: api.twmoc.info
- domain: app-site-association.cdn-apple.info
- domain: onedrive.azure-app.store
- domain: outlook.windows-app.store
- domain: www.twmoc.info
APT Meets GPT: Targeted Operations with Untamed LLMs
Description
Over the course of three months, Volexity observed UTA0388 using various themes and fictional identities across dozens of spear phishing campaigns. As time passed, Volexity observed UTA0388 broaden their targeting and send emails in a variety of different languages, including English, Chinese, Japanese, French, and German. In most cases, the initial email sent by UTA0388 contained a link to phishing content hosted on a cloud-based service that would lead to malware.
AI-Powered Analysis
Technical Analysis
The UTA0388 threat actor has been observed conducting targeted spear phishing campaigns over a three-month period, utilizing large language models (LLMs) to generate convincing and diverse phishing emails. These emails employ various fictional identities and themes, sent in multiple languages including English, Chinese, Japanese, French, and German, to broaden their reach and evade detection. The initial phishing emails typically contain links to cloud-hosted phishing sites that serve malware payloads. The malware uses PowerShell scripts and achieves persistence through scheduled tasks and hijacking execution flows. Command and control communications leverage advanced techniques such as WebSocket-based Govershell C2 channels, enabling stealthy and resilient remote control. The malware delivery often involves archive files (ZIP, RAR) to bypass security controls. The threat actor employs a wide range of MITRE ATT&CK techniques including T1071 (Application Layer Protocol), T1102 (Web Service), T1123 (Audio Capture), T1134 (Access Token Manipulation), T1036 (Masquerading), T1105 (Ingress Tool Transfer), T1059 (Command and Scripting Interpreter), T1574 (Hijack Execution Flow), T1053 (Scheduled Task), and T1566 (Phishing). Although no known exploits are currently in the wild, the campaign demonstrates a sophisticated blend of social engineering and technical exploitation, leveraging LLMs to enhance phishing effectiveness and evade traditional detection methods. The use of multiple languages and cloud infrastructure indicates a broad targeting strategy, likely aiming at high-value organizations globally, including Europe.
Potential Impact
European organizations face significant risks from this threat due to the use of multi-lingual spear phishing campaigns that can bypass traditional email security filters and deceive users across different countries. The malware's use of PowerShell and persistence mechanisms can lead to long-term footholds within networks, enabling data exfiltration, espionage, or disruption. The WebSocket-based Govershell C2 channel allows stealthy communication that may evade network monitoring tools. Organizations relying heavily on cloud services are particularly vulnerable since the phishing content is hosted on cloud platforms, complicating blocking efforts. The broad language support increases the likelihood of successful compromise across diverse European workforces. Potential impacts include loss of sensitive data, intellectual property theft, operational disruption, and reputational damage. The medium severity rating reflects the threat's sophistication and potential for significant harm if not mitigated effectively.
Mitigation Recommendations
1. Implement advanced email filtering solutions capable of detecting and blocking multi-lingual spear phishing attempts, including heuristic and AI-driven analysis to identify LLM-generated content. 2. Conduct targeted user awareness training in multiple languages (English, French, German, etc.) focusing on recognizing phishing tactics and suspicious links. 3. Monitor and restrict PowerShell usage on endpoints, enforcing execution policies and logging all PowerShell activity for anomaly detection. 4. Deploy endpoint detection and response (EDR) tools with capabilities to detect persistence techniques such as scheduled tasks and hijacked execution flows. 5. Monitor network traffic for unusual WebSocket connections, especially those communicating with cloud-hosted domains, and implement strict egress filtering. 6. Regularly audit and restrict use of archive files (ZIP, RAR) in email attachments and downloads, scanning them thoroughly for malicious content. 7. Maintain up-to-date threat intelligence feeds and integrate indicators of compromise (IOCs) related to UTA0388 into security monitoring systems. 8. Enforce multi-factor authentication (MFA) and least privilege principles to limit the impact of compromised credentials. 9. Conduct regular phishing simulation exercises tailored to the languages and themes used by this threat actor to improve user resilience. 10. Collaborate with cloud service providers to quickly identify and take down phishing sites hosted on their platforms.
Affected Countries
For access to advanced analysis and higher rate limits, contact root@offseq.com
Technical Details
- Author
- AlienVault
- Tlp
- white
- References
- ["https://www.volexity.com/blog/2025/10/08/apt-meets-gpt-targeted-operations-with-untamed-llms/"]
- Adversary
- UTA0388
- Pulse Id
- 68e68c8d506e04cc0474a83b
- Threat Score
- null
Indicators of Compromise
Url
Value | Description | Copy |
---|---|---|
urlhttps://aesthetic-donut-1af43s2.netlify.app/file/rar | — | |
urlhttps://aesthetic-donut-1af43s2.netlify.app/index/file/A_Introduction_Docs_v00546823.rar | — | |
urlhttp://80.85.154.48:443 | — | |
urlhttp://80.85.157.117:443 | — | |
urlhttp://82.118.16.173:443 | — | |
urlhttp://api.twmoc.info/ws | — | |
urlhttp://onedrive.azure-app.store/ws | — | |
urlhttp://outlook.windows-app.store/ws | — | |
urlhttps://app-site-association.cdn-apple.info:443/updates.rss | — |
Ip
Value | Description | Copy |
---|---|---|
ip185.144.28.68 | CC=RU ASN=AS44493 chelyabinsk-signal llc | |
ip104.194.152.137 | CC=US ASN=AS53667 frantech solutions | |
ip104.194.152.152 | CC=US ASN=AS53667 frantech solutions | |
ip31.192.234.22 | CC=RU ASN=AS44493 chelyabinsk-signal llc | |
ip45.141.139.222 | CC=KR ASN=AS138195 moack.co.ltd | |
ip74.119.193.175 | CC=HK ASN=AS135330 adcdata.com | |
ip80.85.154.48 | CC=RU ASN=AS44493 chelyabinsk-signal llc | |
ip80.85.156.234 | CC=RU ASN=AS44493 chelyabinsk-signal llc | |
ip80.85.157.117 | CC=RU ASN=AS44493 chelyabinsk-signal llc | |
ip82.118.16.173 | CC=UA ASN=AS15626 itl llc |
Hash
Value | Description | Copy |
---|---|---|
hashcf8d7017d025dd7ad65a946e33ed23d8 | MD5 of fbade9d8a040ed643b68e25e19cba9562d2bd3c51d38693fe4be72e01da39861 | |
hash7351db2af3139c2b8eed820c9938060b18ba4a99 | SHA1 of fbade9d8a040ed643b68e25e19cba9562d2bd3c51d38693fe4be72e01da39861 | |
hash0414217624404930137ec8f6a26aebd8a3605fe089dbfb9f5aaaa37a9e2bad2e | — | |
hash126c3d21a1dae94df2b7a7d0b2f0213eeeec3557c21717e02ffaed690c4b1dbd | — | |
hash2ffe1e4f4df34e1aca3b8a8e93eee34bfc4b7876cedd1a0b6ca5d63d89a26301 | — | |
hash4c041c7c0d5216422d5d22164f83762be1e70f39fb8a791d758a816cdf3779a9 | — | |
hash53af82811514992241e232e5c04e5258e506f9bc2361b5a5b718b4e4b5690040 | — | |
hash7d7d75e4d524e32fc471ef2d36fd6f7972c05674a9f2bac909a07dfd3e19dd18 | — | |
hash88782d26f05d82acd084861d6a4b9397d5738e951c722ec5afed8d0f6b07f95e | — | |
hash998e314a8babf6db11145687be18dc3b8652a3dd4b36c115778b7ca5f240aae4 | — | |
hasha5ee55a78d420dbba6dec0b87ffd7ad6252628fd4130ed4b1531ede960706d2d | — | |
hashad5718f6810714bc6527cc86d71d34d8c556fe48706d18b5d14f0261eb27d942 | — | |
hashfbade9d8a040ed643b68e25e19cba9562d2bd3c51d38693fe4be72e01da39861 | — |
Domain
Value | Description | Copy |
---|---|---|
domainazure-app.store | — | |
domaincdn-apple.info | — | |
domaindoccloude.info | — | |
domainsliddeshare.online | — | |
domaintwmoc.info | — | |
domainwindows-app.store | — | |
domainapi.twmoc.info | — | |
domainapp-site-association.cdn-apple.info | — | |
domainonedrive.azure-app.store | — | |
domainoutlook.windows-app.store | — | |
domainwww.twmoc.info | — |
Threat ID: 68e68dec47cdb70919db7f0b
Added to database: 10/8/2025, 4:14:36 PM
Last enriched: 10/8/2025, 4:29:47 PM
Last updated: 10/9/2025, 11:15:26 AM
Views: 13
Community Reviews
0 reviewsCrowdsource mitigation strategies, share intel context, and vote on the most helpful responses. Sign in to add your voice and help keep defenders ahead.
Want to contribute mitigation steps or threat intel context? Sign in or create an account to join the community discussion.
Related Threats
From Phishing to Malware: AI Becomes Russia's New Cyber Weapon in War on Ukraine
MediumFake Teams Installers Dropping Oyster Backdoor (aka Broomstick) in New Malvertising Scam
MediumVampire Bot Malware Sinks Fangs Into Job Hunters
MediumXWorm 6.0 Returns with 35+ Plugins and Enhanced Data Theft Capabilities
MediumBatShadow Group Uses New Go-Based 'Vampire Bot' Malware to Hunt Job Seekers
MediumActions
Updates to AI analysis are available only with a Pro account. Contact root@offseq.com for access.
External Links
Need enhanced features?
Contact root@offseq.com for Pro access with improved analysis and higher rate limits.