arXiv cs.AI· Zhishang Xiang, Zerui Chen, Yunbo Tang, Zhimin Wei, Ruqin Ning, Yujie Lin, Qinggang Zhang, Jinsong Su·· 2 天前AI 评分27
MemSyco-Bench：评测 Agent 记忆引发的谄媚问题
MemSyco-Bench: Benchmarking Sycophancy in Agent Memory
AI 导读
研究者提出 MemSyco-Bench，一个评测 Agent 系统中记忆诱发谄媚行为的基准，覆盖五项任务：判断 Agent 能否拒绝将记忆当作事实证据、遵守记忆的适用范围、解决记忆与客观证据的冲突、跟踪记忆更新，以及用有效记忆做个性化。现有记忆基准只关注记忆是否正确存储、检索或更新，忽略了检索到的记忆如何影响下游推理与决策。相关资源已向社区开放。
正文 · 原文
Abstract:Memory has emerged as a cornerstone of modern LLM-based agents, supporting their evolution from single-turn assistants to long-term collaborators. However, memory is not always beneficial: retrieved memories often induce a critical issue of sycophancy, causing agents to over-align with the user at the cost of factual accuracy or objective reasoning. Despite this emerging risk, existing memory benchmarks primarily evaluate whether memories are correctly stored, retrieved, or updated, while overlooking how retrieved memories influence downstream reasoning and decision-making. To bridge this gap, we propose MemSyco-Bench, a comprehensive benchmark for evaluating memory-induced sycophancy in agent systems. MemSyco-Bench measures when memory should influence a decision and how valid memory should be used. Specifically, it covers five tasks that assess whether agents can reject memory as factual evidence, respect its applicable scope, resolve conflicts between memory and objective evidence, track memory updates, and use valid memory for personalization. All related resources are collected for the community at this https URL.
| Subjects: | Information Retrieval (cs.IR); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2607.01071 [cs.IR] |
| (or arXiv:2607.01071v3 [cs.IR] for this version) | |
| https://doi.org/10.48550/arXiv.2607.01071 arXiv-issued DOI via DataCite |
Submission history
From: Qinggang Zhang [view email]
[v1]
Wed, 1 Jul 2026 15:30:33 UTC (10,562 KB)
[v2]
Thu, 2 Jul 2026 15:40:30 UTC (10,475 KB)
[v3]
Thu, 15 Oct 2026 15:04:45 UTC (10,340 KB)
来源：arXiv cs.AI · arxiv.org