English(EN) MemAudit: Post-hoc Auditing of Poisoned Agent Memory via Causal Attribution and Structural Anomaly Detection

MemAudit框架审计中毒的LLM代理内存

作者 PulseAugur 编辑部 · [2 个来源] · 2026-05-22 15:03

研究人员开发了MemAudit，一个旨在识别和审计大型语言模型代理内存中恶意数据的新框架。该事后审计系统解决了对抗性用户可以将有害记录注入代理内存，从而可能操纵其行为的安全漏洞。MemAudit利用因果归因和结构异常检测来精确定位导致不良输出的特定内存，在测试场景中显著降低了攻击成功率。 AI

影响提供了一种通过审计其内存存储来检测和缓解LLM代理安全风险的方法。

排序理由该集群包含一篇详细介绍LLM代理内存审计新框架的学术论文。

AI 生成摘要 · Google Gemini · 来自 2 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Research; 该集群包含一篇详细介绍LLM代理内存审计新框架的学术论文。
Source corroboration: 2 independent sources; Multiple independent publishers reporting the same story raises confidence that it's real and newsworthy.
Topics: paper, safety; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: 95 days old; Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

arXiv cs.AI TIER_1 English(EN) · Zhewen Tan, Yilun Yao, Huiyan Jin, Wenhan Yu, Guoan Wang, Mengyuan Fan, liang lu, Feng Liu, Xiangzheng Zhang, Duohe Ma, Tong Yang, Lin Sun · 2026-05-25 04:00

MemAudit：通过因果归因和结构异常检测对中毒代理记忆进行事后审计

arXiv:2605.23723v1 Announce Type: new Abstract: Large language model agents increasingly rely on persistent memory to store past interactions, retrieve relevant demonstrations, and improve long-horizon task execution. However, this memory mechanism also creates a practical securi…
arXiv cs.AI TIER_1 English(EN) · Lin Sun · 2026-05-22 15:03

MemAudit：通过因果归因和结构异常检测对中毒代理记忆进行事后审计

Large language model agents increasingly rely on persistent memory to store past interactions, retrieve relevant demonstrations, and improve long-horizon task execution. However, this memory mechanism also creates a practical security vulnerability: an adversarial user may inject…