Hidden Thoughts Are Not Secret: Reasoning Trace Exposure in LLMs
On the map: LLM reasoning → Prompting and chain-of-thought reasoning
Abstract (arXiv)
Reasoning traces have become a valuable form of learning signals for improving and transferring the capabilities of large language models. In particular, detailed traces can help distill reasoning behavior from stronger teacher models into weaker student models. The value of capability transfer has motivated many deployed systems with reasoning models to hide raw internal traces and expose at most summaries and answers to users. As a result, we ask whether such interface-level trace hiding prevents users from obtaining useful reasoning supervision through prompting. We study this question with Reasoning Exposure Prompting (REP), a lightweight in-context elicitation method that uses shadow-model-generated demonstrations wrapped in auxiliary code-like formats to raise user-visible reasoning traces from a victim model. Across the common reasoning dataset, different victim models, and different student model distillation, REP substantially increases similarity between exposed and REP-conditioned internal traces while preserving useful reasoning signals.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Placed in Prompting and chain-of-thought reasoning, one of 17 directions in LLM reasoning. Builds on 14 of its 58 references that are in the map. First posted on arXiv 30 May 2026.
Builds on
14 of 58 references.
Closest ancestors · 3
Same direction, newest first
- 2025Reasoning Models Don't Always Say What They Think8 May
- 2025Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation14 Mar
- 2023Measuring Faithfulness in Chain-of-Thought Reasoning17 Jul
Other references · 11
- 2021Training Verifiers to Solve Math Word Problemscited by 10,109
- 2025Qwen3 Technical Reportcited by 8,820
- 2024Qwen2.5 Technical Reportcited by 4,942
- 2025Large Language Diffusion Modelscited by 912
- 2025Dream 7B: Diffusion Large Language Modelscited by 544
- 2023Detecting Language Model Attacks with Perplexitycited by 431
- 2023Orca: Progressive Learning from Complex Explanation Traces of GPT-4cited by 379
- 2025SALT: Steering Activations towards Leakage-free Thinking in Chain of Thoughtcited by 11
- 2026How to Steal Reasoning Without Reasoning Tracescited by 8
- 2026Chain-of-Sanitized-Thoughts: Plugging PII Leakage in CoT of Large Reasoning Modelscited by 5
- 2026Safer Reasoning Traces: Measuring and Mitigating Chain-of-Thought Leakage in LLMscited by 1
Built on by
Citing papers per year: 2026: 2.
2 papers.
Similar, not cited
Close in wording, no citation either way
- 2025Information-Preserving Reformulation of Reasoning Traces for Antidistillation
- 2026Explaining Too Much? Understanding How Large Language Model Reasoning Traces Influence Performance and Metacognition
- 2025Trace-of-Thought Prompting: Investigating Prompt-Based Knowledge Distillation Through Question Decomposition
- 2025ReasonIF: Large Reasoning Models Fail to Follow Instructions During Reasoning
- 2024Code Prompting Elicits Conditional Reasoning Abilities in Text+Code LLMs
- 2025The Challenge of Teaching Reasoning to LLMs Without RL or Distillation
- 2026Backtracking When It Strays: Mitigating Dual Exposure Biases in LLM Reasoning Distillation
- 2025In Their Own Words: Reasoning Traces Tailored for Small Models Make Them Better Reasoners