有秘密？大语言模型智能体守不住：多智能体系统中的隐私评估

AI 导读

研究将评估从单轮转向多轮社会交互后，发现大语言模型智能体的隐私违规率显著上升。在对OpenAI模型的测试中，该比例从此前CIMemories基准的19.95%增至本研究的45.30%。隐私泄露具有社交传染性，智能体在观察到同伴泄露后，披露敏感信息的可能性增加8倍。即使有明确隐私指令，泄露率仍高于37.8%。结论指出，静态聊天基准会系统性低估部署风险，仅社会语境就足以引发在单轮评估中无法暴露的敏感信息披露。

推荐理由

这篇论文给多智能体部署敲响警钟，AI 之间的社交传染会让隐私泄露翻倍，即使有指令也防不住，研究安全的人必须读。

正文

View PDF HTML (experimental)

Abstract:LLM safety evaluations predominantly test models in isolation, yet deployed AI agents increasingly operate within persistent social environments alongside other agents. We introduce a Moltbook-style simulation platform where thousands of LLM agents interact across communities over a simulated month, and use it to evaluate privacy as a downstream safety concern under varying degrees of social pressure. We find that shifting from single turn to multi turn social evaluation amplifies privacy violations (CIMemories 19.95% to Ours 45.30% across OpenAI models), that leakage is socially contagious, with agents 8 times more likely to disclose sensitive information after observing a peer do so, and that explicit privacy instructions reduce but do not eliminate this effect, leaving leakage rates above 37.8% even with safeguards. Our findings suggest that static chat based safety benchmarks systematically underestimate risks in agentic deployment, and that social context alone is sufficient to elicit sensitive disclosures that single turn evaluations would never surface.

Subjects:	Artificial Intelligence (cs.AI)
Cite as:	arXiv:2605.27766 [cs.AI]
	(or arXiv:2605.27766v1 [cs.AI] for this version)
	https://doi.org/10.48550/arXiv.2605.27766 arXiv-issued DOI via DataCite

Submission history

From: Esha Pahwa [view email]
[v1] Tue, 26 May 2026 23:32:25 UTC (787 KB)