arXiv cs.AI· Prakhar Ganesh, Yash More, Marco Romanelli, Ferdinando Fioretto, Golnoosh Farnadi·· 13 小时前AI 评分22
通用 LLM 时代，上下文隐私正被忽视
Privacy in Context Is Being Overlooked in the Age of General-Purpose LLMs
AI 导读
论文指出，提示词隐私研究正趋向设计统一框架，这忽视了隐私本质上是上下文相关的，并可能带来虚假的安全感。作者认为即便未来工作也需处理提示词隐私多项理想属性间的根本张力，建议转向应用特定部署、上下文特定评估与更好的隐私素养。
正文 · 原文
Abstract:As large language models (LLMs) continue to grow in scope and capability, privacy concerns are on the rise. Since users increasingly interact with a single general-purpose system across many distinct contexts, there is a natural tendency to pursue a one-size-fits-all solution to privacy. Unfortunately, this overlooks a crucial insight from decades of privacy research: privacy is inherently contextual. In this paper, we argue that existing work in prompt privacy is drifting towards attempts to design a unified framework, which is highly misleading and can foster a false sense of security. Furthermore, we argue that even future work in this direction needs to wrestle with fundamental tensions across various desirable properties of prompt privacy. Thus, such continued attempts towards a unified framework need to be reconsidered. We recommend three critical paths forward: application-specific deployment of LLMs, context-specific evaluation of privacy frameworks, and better privacy literacy.
| Comments: | Accepted at Personalization in the Era of Large Foundation Models Workshop @ AAAI 2026 |
| Subjects: | Machine Learning (cs.LG); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.29925 [cs.LG] |
| (or arXiv:2610.29925v1 [cs.LG] for this version) | |
| https://doi.org/10.48550/arXiv.2610.29925 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Prakhar Ganesh [view email]
[v1]
Wed, 21 Oct 2026 16:01:42 UTC (605 KB)
来源：arXiv cs.AI · arxiv.org