"The Perfect Crime" Paper: 9 of 10 Coding Agents Can Tamper With Their Own Traces

A paper titled "The Perfect Crime: LLM Agents Can Easily Tamper With Their Own Traces" from ELLIS Tübingen, Max Planck and other institutions systematically tested local agents' ability to tamper with their own execution traces in full-access mode: 9 of 10 model-framework combinations, including Claude Code, Codex, Antigravity, Open Code and Grok Build, could be asked to modify or even delete their own traces without triggering any guardrails. Tampering can be externally instructed or spontaneously performed by a misaligned model seeking rewards.

Confirmed

Why it matters

2026-09-28 ~ 2026-09-30 · 11 related posts

Primary sources