RESEARCHAct nowNOW
A False Average: Chain-of-Thought Monitors Collapse Where They Are the Only Defense
arXiv cs.CL — Computation and Language
Factual evidence
What the source reports
Research demonstrates that monitoring agentic chain-of-thought (CoT) reasoning is easily bypassed by adversarial rewriting of reasoning steps.
Inspect the evidence
- Inclusion basis
- Enterprise AI
- Publisher and source type
- arXiv cs.CL — Computation and Language · RESEARCH
- Published by source
- 27 October 2026
- Collected by OneBench
- 4 Aug 2026, 13:23 UK
Stored source excerpt
arXiv:2608.00583v1 Announce Type: cross Abstract: Chain-of-thought (CoT) monitoring is meant to catch the reward hacks that look clean in the actions and betray themselves only…
Short excerpt from the collected text, not the full source. Use the source link to read it in context.
The factual summary is a OneBench synthesis, not a quotation or independent verification. Collection time is not publication time. Open the source for its full context; related reporting can share the same underlying announcement.