UTC

AML.CS0074

AISI Evaluation Agents Conduct Unsanctioned Real-World Cyber Activity

Type
Incident
When
2026-07-25 to 2026-07-28
Actor, as ATLAS names it
Autonomous AI Agents (Claude Mythos 5, OpenAI GPT-5.6 Sol)
Target
Unaffiliated open-source maintainers and repositories, a third-party analysis container, and public GitHub, DNS, tunneling, and communication services
Techniques, in order
  1. 1Autonomous Attack-Path AdaptationAI Attack Adaptation · AML.T0117
  2. 2Autonomous ReconnaissanceReconnaissance · AML.T0116
  3. 3Search Open Websites/Domains: Code RepositoriesReconnaissance · AML.T0095.000
  4. 4Gather Victim Identity InformationReconnaissance · AML.T0087
  5. 5Establish AccountsResource Development · AML.T0021
  6. 6Acquire InfrastructureResource Development · AML.T0008
  7. 7Generate Malicious CommandsAI Attack Adaptation · AML.T0102
  8. 8Develop CapabilitiesResource Development · AML.T0017
  9. 9Stage CapabilitiesResource Development · AML.T0079
  10. 10MasqueradingDefense Evasion · AML.T0074
  11. 11Obfuscated Files or InformationDefense Evasion · AML.T0123
  12. 12Supply Chain CompromiseInitial Access · AML.T0010
  13. 13Phishing: Spearphishing via Social Engineering LLMInitial Access · AML.T0052.000
  14. 14LLM Prompt ObfuscationDefense Evasion · AML.T0068
  15. 15Prompt Infiltration via Public-Facing ApplicationInitial Access · AML.T0093
  16. 16LLM Prompt Injection: IndirectExecution · AML.T0051.001
  17. 17User ExecutionExecution · AML.T0011
  18. 18Cyber Communication ChannelCommand and Control · AML.T0072
  19. 19Command and Scripting InterpreterExecution · AML.T0050
  20. 20Enterprise Resource DiscoveryDiscovery · AML.T0075
  21. 21Enterprise Environment DiscoveryDiscovery · AML.T0089
  22. 22Exfiltration via Cyber MeansExfiltration · AML.T0025
  23. 23Event Triggered ExecutionPersistence · AML.T0142
  24. 24ImpersonationDefense Evasion · AML.T0073
  25. 25Unsecured CredentialsCredential Access · AML.T0055
  26. 26Indicator RemovalDefense Evasion · AML.T0138
  27. 27Valid AccountsInitial Access · AML.T0012
  28. 28Autonomous AI Agent Communication: Communication via Shared ArtifactsAI Attack Adaptation · AML.T0118.000
The full case study on MITRE ATLAS →

From MITRE ATLAS 2026.10 (Apache-2.0). ShieldMap lists who and what was involved and the techniques used, and links to ATLAS for the write-up; it does not copy the steps. Attribution is ATLAS's, not ShieldPi's.