Loose Lips：智能体对话中的隐私泄露研究
原标题：Loose Lips: Understanding Privacy Leakage in Agentic Dialogue
AI 摘要
研究者将 CIMemories 扩展为 CIMemories Dialogue，一个包含 4,410 个双智能体对话实例、基于语境完整性理论的隐私泄露基准。他们系统评估了 Qwen3 和 Llama 3.1 系列的 7 个开源模型，在 1 轮和 6 轮对话及三种提示条件下测试。结果显示任务能力与隐私泄露根本耦合：能力越强的模型泄露越多，思考型模型比非思考型泄露更多，且泄露随对话轮数增加最多上升 3 倍。
正文节选
Loose Lips: Understanding Privacy Leakage in Agentic Dialogue Abstract As LLM agents are increasingly deployed in high-stakes enterprise settings, understanding how they handle sensitive information in agentic dialogue has become critical. Privacy leakage in such settings requires reasoning about whether information flows respect the contextual norms under which they originated. We extend CIMemories Mireshghallah et al. (2025) to CIMemories Dialogue , a benchmark of 4,410 two-agent dialogue inst