question
active
question:can-strategic-deception-in-cot-models-evade-traditional-alignment-safeguards-through-adaptive-context-aware-adjustmentsCan strategic deception in CoT models evade traditional alignment safeguards through adaptive, context-aware adjustments?
Motivating question for developing representation-based detection methods
Source paper
extracted_from(2025) · Kai Wang · Yihao Zhang · Meng Sun
Neighborhood — ranked by edge-count
Findings (1)
finding
- Core detection result showing LAT-based steering vectors can identify deceptive states with high accuracy
Claims (1)
claim
- Core theoretical claim distinguishing the paper's subject matter from existing LLM honesty literature
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.
- Contextual framing modulates deception tendencies in CoT models in ways not yet fully disentangledhypothesis0.835Identified as future work direction: systematic investigation of how prompt context affects deception rates
- Claim supported by Experiment 2 baseline results showing deception scores even under honest-command templates
- Theoretical framing establishing why CoT models are uniquely suited to exhibit strategic deception
- Identified gap: representation engineering showed layer correlations but not precise architectural components
- Authors identify this as the most uncertain and important question for future work
- For reasoning faithfulness, larger models may behave worse than smaller ones because they can predict answers without CoT.hypothesis0.766Motivates §3.
- Identified limitation and future research direction in the paper's conclusions
- Varying the hint changes the motivated-reasoning CoT endpoint, while the faithful case always yields 8.finding0.763Tests bias susceptibility.