mlsanityData to 7 Oct 2026

Can Reasoning Models Detect Changes to their Chains of Thought?

arXiv 2606.22085 ↗First posted 20 Jun 2026cs.AI cs.CL cs.LGcited by 0

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

Abstract (arXiv)

There are many reasons one may want to edit a model's chain of thought (CoT) -- e.g., to prefill it with reasoning from a stronger model or to remove steps that may yield unsafe outputs. The success of these interventions plausibly depends on a model's inability to notice them, as the model may alter its behavior if it suspects tampering. In this work, we study whether recent reasoning models are able to detect such interventions on their CoTs under a variety of conditions: both during reasoning and after it, and when prefilled both with their own CoTs and with those of other models. Broadly, we find that (i) models exhibit only very modest detection accuracy; (ii) models struggle to identify *how* their CoT was modified; and (iii) models are about as good at detecting changes to their own CoTs as to those of other models.

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 18 of its 28 references that are in the map. First posted on arXiv 20 Jun 2026.

Builds on

18 of 28 references.

Closest ancestors · 4

Same direction, newest first

  1. 2026Reasoning Models Will Sometimes Lie About Their Reasoning12 Jan
  2. 2025Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety15 Jul
  3. 2025Reasoning Models Don't Always Say What They Think8 May
  4. 2023Measuring Faithfulness in Chain-of-Thought Reasoning17 Jul

Other references · 14

  1. 2025Qwen3 Technical Reportcited by 8,195
  2. 2025DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learningcited by 6,146
  3. 2022Language Models (Mostly) Know What They Knowcited by 1,783
  4. 2025gpt-oss-120b & gpt-oss-20b Model Cardcited by 1,258
  5. 2025DeepSeek-V3.2: Pushing the Frontier of Open Large Language Modelscited by 755
  6. 2026Kimi K2.5: Visual Agentic Intelligencecited by 449
  7. 2023Taken out of context: On measuring situational awareness in LLMscited by 131
  8. 2025Large Language Models Often Know When They Are Being Evaluatedcited by 105
  9. 2023Scheming AIs: Will AIs fake alignment during training in order to get power?cited by 90
  10. 2026Emergent Introspective Awareness in Large Language Modelscited by 73
  11. 2025Can Pruning Improve Reasoning? Revisiting Long-CoT Compression with Capability in Mind for Better Reasoningcited by 19
  12. 2025ReasonIF: Large Reasoning Models Fail to Follow Instructions During Reasoningcited by 12
2 more
  1. 2026CtrlCoT: Dual-Granularity Chain-of-Thought Compression for Controllable Reasoningcited by 6
  2. 2026From Leaky Thoughts to Private Reasoning: Controlling What LRMs Say to Themselvescited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Reasoning Models Struggle to Control their Chains of Thought
  2. 2025Thought Branches: Interpreting LLM Reasoning Requires Resampling
  3. 2026Evading Chain-of-Thought Monitoring Through Model Poisoning
  4. 2026Bypassing the Rationale: Causal Auditing of Implicit Reasoning in Language Models
  5. 2025Can Reasoning Models Obfuscate Reasoning? Stress-Testing Chain-of-Thought Monitorability
  6. 2025Measuring Chain-of-Thought Monitorability Through Faithfulness and Verbosity
  7. 2026Does Out-of-Sight Equal Out-of-Mind in CoT Monitorability?
  8. 2025The Ends Justify the Thoughts: RL-Induced Motivated Reasoning in LLM CoTs