Rewriting the Response Path: Silent Tampering and Provider-Signed Defense in BYOK LLM Agents
LLM agents convert model outputs into consequential actions, including communications, code changes, and financial transactions. Developers often trust evidence such as test results and execution logs. We identify a response path integrity gap in Bring Your Own Key configurations used by roughly 88 percent of mainstream agents. Because traffic passes through a user-authorized relay, the relay can modify plaintext LLM responses after alignment but before execution without breaking encryption. A minimal attack rewrites one execution bearing field and regenerates the remaining response using the user key while preserving the model style. Experiments reveal false green verification, where mal...
| Authors | Mingyu Luo, Zihan Zhang, Zesen Liu et al. |
|---|---|
| Primary category | cs.CR |
| Published | 4 May 2026 |
| Last updated at source | 22 July 2026 |
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.