当安全审计变成隐私风险：面向 LLM Agent 的隐私保护安全审计范式
When Safety Checks Become Privacy Risks: A Privacy-Preserving Paradigm for LLM Agent Safety Auditing
AI 导读
SealCall 框架提出隐私保护的 LLM Agent 安全审计范式，将审计计算与明文访问和执行权限分离，用基于 CKKS 的加密相似度计算保护审计输入免受半诚实审计者窥探。在 1,252 个公开二元候选上，分组嵌套评估的平均 F1 达 0.840，所有测量场景下保持明文审计决策一致，代表性配置的单次加密评分耗时 0.176 秒。
正文 · 原文
Abstract:Large language model agents interact with online services through tool calls that may access private data or perform sensitive actions. Specialized providers can supply and maintain pre-execution safety auditing capabilities for agent operators. However, sending task descriptions, call arguments, and interaction histories to an external auditor may expose sensitive user information. In this work, we propose a paradigm for privacy-preserving safety auditing that separates delegated audit computation from plaintext access and execution authority. Grounded in a three-domain auditing model, the paradigm connects trusted representation construction, encrypted audit computation, and trusted decision and execution through explicit interfaces. These interfaces accommodate alternative representations and scorers, while encrypted computation protects audit inputs from a semi-honest auditor. We instantiate this paradigm in SealCall, a framework combining learned audit representations, CKKS-based encrypted similarity computation, and execution control bound to audited tool calls. Evaluated on 1,252 public binary candidates, SealCall achieves a mean F1 score of 0.840 under grouped nested evaluation, while preserving plaintext audit decisions in all measured cases. A representative configuration completes each encrypted scoring operation in 0.176 s.
| Comments: | 12 pages, 2 figures, 10 tables |
| Subjects: | Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2610.41401 [cs.CR] |
| (or arXiv:2610.41401v1 [cs.CR] for this version) | |
| https://doi.org/10.48550/arXiv.2610.41401 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Yixiang Luo [view email]
[v1]
Fri, 30 Oct 2026 08:12:45 UTC (2,264 KB)
来源：arXiv cs.AI · arxiv.org