claim
active
claim:cot-monitoring-may-be-more-effective-for-threats-requiring-extensive-reasoning-or-environment-interactionCoT monitoring may be more effective for threats requiring extensive reasoning or environment interaction
Subsets amenable to monitoring.
Source paper
extracted_from(2025) · Yanda Chen · Joe Benton · Ansh Radhakrishnan · Jonathan Uesato +11
Neighborhood — ranked by edge-count
Papers (1)
paper
Claims (2)
claim
- CoT monitoring is promising for noticing unintended behaviors but not sufficient to rule them outextendsCentral conclusion.
- Risk to tool-use monitoring.
Related by similarity (8)
cosine ≥ 0.65 · no typed edgeEntities in the same semantic neighborhood but without a typed relation to this one — candidates for new edges or unrecognized duplicates.
- Lower faithfulness on GPQA.
- Implication for test-time monitoring.
- Interpretation of early answering/mistakes results.
- Contrasts with Baker et al.
- Monitoring CoT to detect misaligned intentions; paper tests its reliability.
- Baker et al.'s CoTs verbalize hacks more because hacks involve exploration/planning and coding has better tool-use faithfulnesshypothesis0.789Explanation for differing results.
- From decoupling of AOC and accuracy gain.
- Additional test-time computation from longer context is responsible for CoT's performance boost.hypothesis0.779Tested via filler tokens.