1 paper · 1 filter
Haowen Pan, Pengyang Shao, Yunshan Ma +4
Long-term memory enables LLM agents to learn from past interactions but exposes memory formation to adversarial influence. Existing studies primarily organize memory poisoning by a…