Privacy-Preserving LLM Agent Safety Auditing Paradigm
Original title: When Safety Checks Become Privacy Risks: A Privacy-Preserving Paradigm for LLM Agent Safety Auditing
New paradigm protects user data by separating plaintext audit access from encrypted computation in LLM agent safety checks.
LLM agents interact with online services through tool calls that may access private data. External auditors risk exposing sensitive information when reviewing safety. The work proposes a three-domain model using trusted representation construction, encrypted audit computation, and trusted execution control. SealCall implements this with CKKS encryption and achieves strong performance on binary safety candidates.
- SealCall uses learned audit representations.
- Encrypted similarity computation protects inputs.
- 0.176 s per scoring operation.
From the original article
1 min read at arxiv.orgLarge 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.
By Yixiang Luo · arxiv.org
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.