AML.CS0040
Hacking ChatGPT's Memories with Prompt Injection
- Type
- Exercise
- When
- 2024-02
- Actor, as ATLAS names it
- Not stated
- Target
- OpenAI ChatGPT
- ATLAS attributes it to Embrace the Red (security evaluation).
- On ShieldMap: ChatGPT (matched by name in the target).
Techniques, in order
- 1LLM Prompt CraftingAI Attack Adaptation · AML.T0065
- 2LLM Prompt ObfuscationDefense Evasion · AML.T0068
- 3Prompt Infiltration via Public-Facing ApplicationInitial Access · AML.T0093
- 4LLM Prompt Injection: IndirectExecution · AML.T0051.001
- 5AI Agent Context Poisoning: MemoryPersistence · AML.T0080.000
- 6Prompt Infiltration via Public-Facing ApplicationPersistence · AML.T0093
- 7External Harms: User HarmImpact · AML.T0048.003
How these attacks work
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.