AI Daybook
AI Daybook
Today's issue
PaperAgents

MemoryDojo benchmarks poisoning attacks on LLM agent memory systems

Original titleMemoryDojo: Benchmarking Poisoning Through LLM Agent Memory

Outlet
Takara TLDR
Published
Importance
2/5

Summary

The authors introduce MemoryDojo, a benchmark that traces how poisoned information flows through agent memory systems and measures its impact on subsequent agent behavior. The benchmark covers 155 multi-session trajectories across four domains and supports five different memory systems. They find that all memory systems are susceptible to poisoning, with substantial variation in downstream effects, and propose a memory-aware poison optimization process that reveals differences between architectures.

Original excerpt

LLM agents benefit from using memory systems to improve performance on long-horizon, multi-session tasks. However, memory introduces a critical attack surface. Malicious instructions encountered in one session may be stored and propagated through memory before being retrieved and acted upon in later sessions. Because memory systems differ in how they process and retrieve information, the downstream effects of poisoning can vary significantly across systems. To quantify the effects of poisoning across different memory architectures, we introduce MemoryDojo, a benchmark for tracing the lifecycle of poisoned information through agent memory and measuring its influence on subsequent agent behavior. MemoryDojo is easily extensible, consists of 155 multi-session long-horizon trajectories covering four domains, and supports five different memory systems. Through our evaluation, we find that all memory systems are susceptible to memory poisoning, with substantial variation in downstream effects. We further propose a memory-aware poison optimization process that increases the poisoning efficacy, highlighting differences between memory architectures. We hope that MemoryDojo will support research on understanding and mitigating agentic memory risks.

#benchmark#memory#agent security#poisoning
Read the original
See all →