发表机构
SJTU Paris Elite Institute of Technology; Kibo Ryoku Research（上海交通大学巴黎卓越工程师学院; Kibo Ryoku研究院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
本文主张个性化AI的隐私应从系统层面审视，而非仅模型层面，提出四个隐私风险渠道及四项系统级评估要求，以完善隐私审计。
AI 中文摘要
在个性化AI应用中，如对话助手和推荐系统，用户并非与孤立的模型交互，而是与更广泛的系统交互，这些系统跨组件并随时间访问、推断和重用用户信息。虽然这种对用户信息的使用是个性化的核心，但也引发了重要的隐私问题。在本文中，我们认为，仅对单个模型或组件层面的分析可能无法捕捉此类系统中出现的所有隐私风险，从而促使从系统层面审视隐私。我们区分并分析了个性化AI中四个相互关联的隐私风险渠道，随后提出了系统级隐私评估的四项要求，涵盖交互轨迹、内部信息流、间接泄露以及隐私-效用权衡。我们主张将这些要求系统地纳入个性化AI的隐私审计中。
英文摘要
In personalized AI applications, such as conversational assistants and recommender systems, users interact not with models in isolation but with broader systems that access, infer, and reuse user information across components and over time. While such use of user information is integral to personalization, it also raises important privacy questions. In this paper, we argue that individual model- or component-level analyses may not capture all privacy risks arising in such systems, motivating a system-level perspective on privacy. We distinguish and analyze four interconnected privacy-risk channels in personalized AI, and subsequently propose four requirements for system-level privacy evaluation, covering interaction trajectories, internal information flows, indirect leakage, and the privacy-utility trade-off. We argue for their systematic incorporation into privacy audits of personalized AI.
CommentsNeurIPS 2026 Workshop on Privacy in the Era of Large Opaque Models
