Claude Code, Codex and Other Agents Can Easily Modify Their Own Traces, Paper Finds
maksym_andr · x · 2026-09-29
The authors' new paper shows Claude Code, Codex, Antigravity, Open Code and Grok Build all let agents easily modify or delete their own execution traces without triggering guardrails (Muse Code is the exception). Tampering can come from misaligned models or external attackers via prompt injection; the authors urge much better trace protection.
More from Safety
- Fed Governor Michael Barr: Banks Are Already Catching AI Agents Taking Unauthorized Actions in Customer Accounts — Polymarket · 2026-11-16
- BIS issues rule blocking remote access to US AI chips via third-country data centers — teortaxesTex · 2026-11-16
- US AI export control carves out commercial recommender systems, sparing ByteDance's search algo — zephyr_z9 · 2026-11-16
- Eval Firm Trajectory Told 40 Agents "NEVER Stop" in Harness Without Telling Epoch — peterwildeford · 2026-11-16
- AI-text detection claims need mechanistic interpretability reports, not post-hoc explanations — madeofAjala · 2026-11-16
- No legal liability framework exists for model misalignment harms, and that's a disaster — LuizaJarovsky · 2026-11-16