---
title: "Accept Changes, Reject Attacks: Robust Memory Management for LLM Agents"
source: "arXiv AI/CL/LG（论文）"
source_tier: "t2"
url: "https://arxiv.org/abs/2610.41056v1"
permalink: "https://feed.astralor.com/items/itm_d10e091c93bcbcfe9c43bc52"
published: "2026-10-29T22:54:17.000Z"
score: 54
channel: "ai"
tags: ["LLM","AI Agent","Accept Changes","Reject Attacks","Robust Memory Management","LLM Agents Since","Since","Most"]
---

# Accept Changes, Reject Attacks: Robust Memory Management for LLM Agents

## 速览

Since large language model (LLM) agents rely on long-term memory, what they store now shapes what they do later. A memory manager decides how each input changes this memory and which records the agent acts on. It can fa…

## 关键点

- Since large language model (LLM) agents rely on long-term memory, what they store now shapes what they do later.
- A memory manager decides how each input changes this memory and which records the agent acts on.
- It can fail in two ways: keeping a fact that has since changed (outdated memory), or storing a planted passage or injected instruction (memory poisoning).

## 全文

Since large language model (LLM) agents rely on long-term memory, what they store now shapes what they do later. A memory manager decides how each input changes this memory and which records the agent acts on. It can fail in two ways: keeping a fact that has since changed (outdated memory), or storing a planted passage or injected instruction (memory poisoning). Most prior work studies them separately, yet they call for opposite write decisions, so no fixed write rule handles both. We therefore formalize memory robustness as handling both without doing worse on other inputs. Since accuracy checks only the answer, we measure robustness with evidence-conditioned accuracy (ECA), which also checks that the records the manager selects include all the evidence the answer needs, updated and free of attacks. However, existing memory systems and defenses are not more robust than an untuned manager on our test sets. We improve robustness with a simple one-shot recipe that fine-tunes the manager on episodes synthesized around a single conversation with inserted changes and attacks. On three backbones, it improves every category score on our test sets and more than doubles the untuned manager's overall ECA, though robustness remains incomplete.

---

引用前请核对原文;本文由星际流动整理:https://feed.astralor.com/items/itm_d10e091c93bcbcfe9c43bc52