mlsanityData to 7 Oct 2026

Chain-of-Thought Faithfulness of Reasoning Models Varies with Where and How Preference Cues Are Delivered

arXiv 2608.29464 ↗First posted 29 Aug 2026cs.CL cs.AIcited by 0

On the map: LLM reasoning → Prompting and chain-of-thought reasoning

Abstract (arXiv)

Chain-of-thought (CoT) monitoring assumes that reasoning traces faithfully record the information that shapes a model's answer. Existing faithfulness tests often place explicit bias cues in the user message, while agents may encounter preferences through tool returns or raw artifacts. We introduce FACE-Eval (Faithful Attribution of Cue Effects Evaluation), a 5,100-sample evaluation that varies cue location (user message or tool return) and explicitness (direct summary or raw artifact). We measure verbalized commitment among cue-following answers and unverbalized adoption among all cued samples. We evaluate 15 open-weight models from eight families, with total parameters ranging from 4B to 1.60T. Every model has lower verbalized commitment for tool-return than user-message cues and for implicit than explicit cues. Unverbalized adoption is higher for tool-return cues on all 15 models and for implicit cues in 28 of 30 model-channel comparisons. A source-attribution prompt narrows the channel gap on seven models, sometimes by increasing user-channel unverbalized adoption, while telling models that their reasoning will be monitored does not reliably close the gap. We also use two transcript monitors (GPT-5.6-Luna and GPT-4o-mini) to detect preference adoption in the largest model of each family. Across 32 model-channel-explicitness cells, higher unverbalized adoption is associated with lower detection ability for both monitors (Pearson r=-0.54 and r=-0.78, respectively). These results suggest that CoT monitoring may be less reliable when preference information arrives through tools or must be inferred from raw artifacts, within the single-call, prefilled-tool setting tested here.

Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Prompting and chain-of-thought reasoning, one of 17 directions in LLM reasoning. Builds on 23 of its 59 references that are in the map. First posted on arXiv 29 Aug 2026.

Builds on

23 of 59 references.

Closest ancestors · 14

Same direction, newest first

  1. 2026Chain-of-Thought Monitoring Can Be Unreliable in Implicit-Influence Settings5 Aug
  2. 2026Quantifying the Necessity of Chain of Thought through Opaque Serial Depth10 Mar
  3. 2026Counterfactual Simulation Training for Chain-of-Thought Faithfulness24 Feb
  4. 2025Is Chain-of-Thought Really Not Explainability? Chain-of-Thought Can Be Faithful without Hint Verbalization28 Dec
  5. 2025Monitoring Monitorability20 Dec
  6. 2025Measuring Chain-of-Thought Monitorability Through Faithfulness and Verbosity31 Oct
  7. 2025Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety15 Jul
  8. 2025When Chain of Thought is Necessary, Language Models Struggle to Evade Monitors7 Jul
  9. 2025Thought Anchors: Which LLM Reasoning Steps Matter?23 Jun
  10. 2025Reasoning Models Don't Always Say What They Think8 May
  11. 2025Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation14 Mar
  12. 2025Chain-of-Thought Reasoning In The Wild Is Not Always Faithful11 Mar
2 more
  1. 2024Bias-Augmented Consistency Training Reduces Biased Reasoning in Chain-of-Thought8 Mar
  2. 2023Measuring Faithfulness in Chain-of-Thought Reasoning17 Jul

Other references · 9

  1. 2025gpt-oss-120b & gpt-oss-20b Model Cardcited by 1,258
  2. 2023Instruction-Following Evaluation for Large Language Modelscited by 1,143
  3. 2023Instruction Tuning with GPT-4cited by 676
  4. 2026GLM-5: from Vibe Coding to Agentic Engineeringcited by 502
  5. 2025Olmo 3cited by 95
  6. 2025Are DeepSeek R1 And Other Reasoning Models More Faithful?cited by 69
  7. 2025SHADE-Arena: Evaluating Sabotage and Monitoring in LLM Agentscited by 58
  8. 2026Reward Hacking Benchmark: Measuring Exploits in LLM Agents with Tool Usecited by 25
  9. 2026Biases in the Blind Spot: Detecting What LLMs Fail to Mentioncited by 8

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Bypassing the Rationale: Causal Auditing of Implicit Reasoning in Language Models
  2. 2026Legibility is Not Interpretability: Comparing Judged and Actual Importance in Chain-Of-Thought Reasoning
  3. 2026Do Models Know Why They Changed Their Mind? Interpretability and Faithfulness of Chain-of-Thought Under Knowledge Conflict
  4. 2026Lie to Me: How Faithful Is Chain-of-Thought Reasoning in Reasoning Models?
  5. 2026On the Generalization of Steering Vectors for Chain-of-Thought Faithfulness
  6. 2026Reasoning-Token Spikes Under Prompted Untruthful Responding in Large Language Models
  7. 2026Does Verbose Chain-of-Thought Really Help? In-Distribution Evidence that Content, Not Length, Matters
  8. 2025Unveiling Confirmation Bias in Chain-of-Thought Reasoning