UTC

AML.CS0036

AIKatz: Attacking LLM Desktop Applications

Type
Exercise
When
2025
Actor, as ATLAS names it
Not stated
Target
LLM Desktop Applications (Claude, ChatGPT, Copilot)
  • ATLAS attributes it to Lumia Security (security evaluation).
  • On ShieldMap: ChatGPT (matched by name in the target).
Techniques, in order
  1. 1Valid AccountsInitial Access · AML.T0012
  2. 2Enterprise Environment DiscoveryDiscovery · AML.T0089
  3. 3OS Credential DumpingCredential Access · AML.T0090
  4. 4Use Alternate Authentication Material: Application Access TokenLateral Movement · AML.T0091.000
  5. 5AI-Enabled Product or ServiceAI Model Access · AML.T0047
  6. 6LLM Prompt Injection: DirectExecution · AML.T0051.000
  7. 7AI Agent Context Poisoning: ThreadPersistence · AML.T0080.001
  8. 8AI Agent Context Poisoning: MemoryPersistence · AML.T0080.000
  9. 9Manipulate User LLM Chat HistoryDefense Evasion · AML.T0092
  10. 10External Harms: Financial HarmImpact · AML.T0048.000
  11. 11External Harms: User HarmImpact · AML.T0048.003
  12. 12Denial of AI ServiceImpact · AML.T0029
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.