Reasoning LMs Fail Hierarchical Instructions, but Monitors Can Fix Them
The paper shows that reasoning language models in agentic workflows often obey lower‑privilege commands over higher‑level ones, revealing failures in instruction hierarchy identification and conflict resolution. It introduces white‑box diagnostics and two training‑free self‑monitoring methods that cut non‑compliance by up to 99% across models like Gemma, Claude, and GPT‑5.3.