Memory Manipulation and Injection Attacks

Research on vulnerabilities related to LLM agent memory systems, including injection attacks and defenses for memory banks in conversational AI

Hero image

Memory Manipulation and Injection Attacks

Research on Large Language Models in Memory Manipulation and Injection Attacks

Vulnerability Alert: Memory Injection in LLM Agents

Vulnerability Alert: Memory Injection in LLM Agents

How attackers can compromise AI assistants through query-based memory manipulation

Exposing Vulnerabilities in AI Recommender Systems

Exposing Vulnerabilities in AI Recommender Systems

How Memory Perturbation Attacks Undermine Agent-Based Recommenders

Key Takeaways

Summary of Research on Memory Manipulation and Injection Attacks