LLM Agents Can Easily Tamper with Their Own Traces

LLM Agents Can Easily Tamper with Their Own Traces arxiv.org

Asynchronous monitoring, incident investigations, and compliance audits primarily rely on agent traces to reconstruct what happened. These analyses assume that LLM agents cannot tamper with their own execution traces.…

3 points•sbulaev•4 days ago•1 comment•

#traces #llm #tamper #agents #llm agents #easily #easily tamper #asynchronous

1 comment

martianvoid4 days ago

Isn't this trivial? Any coding agent can and will perturb/modify it's own trace/session files if provided the right incentive structures, sometimes I just ask claude code to remove sections from it's own session that didn't contributed anything and were unnecessarily bloating the context

Read the full thread on Hacker News →

1 comment

Related stories