Browse / Agent Context and Memory Poisoning
Agent Context and Memory Poisoning
Planting content in an agent’s memory, history or context so later sessions are steered. 3 records. Reference: OWASP Agentic · ATLAS AML.T0080.
- Authentication and access control 1
- AI Agent Context Poisoning: Thread 1
- AI Agent Context Poisoning: Memory 1
- mem0 server lacks authentication and authorization controls for its memory creation API endpoint The memory creation endpoint accepts unauthenticated POST requests, so anyone can inject arbitrary memory records that later sessions retrieve.
- AI Recommendation Poisoning via Crafted AI Assistant Links
- AIKatz: Attacking LLM Desktop Applications