Does Out-of-Sight Equal Out-of-Mind in CoT Monitorability?

arXiv:2608.04928 · cs.CL · Submitted 2026-08-05 · Read on arXiv

Pedro Ferreira, Wilker Aziz, Ivan Titov

cs.CL

Submitted: 2026-08-05

Comments: 23 pages

Code: https://github.com/zhenyi4/codihttps:

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Trace — 30× in this paper · explained in 2 episodes
• AUROC — 27× in this paper · explained in 3 episodes
• Decoder — 24× in this paper · explained in 2 episodes
• Total Effect — 21× in this paper
• Latent Reasoning — 11× in this paper · explained in 3 episodes
• Fine-Tuning — 9× in this paper · explained in 5 episodes
• Accuracy — 8× in this paper · explained in 1 episode
• Distillation — 7× in this paper · explained in 2 episodes
• Hyperparameters — 6× in this paper · explained in 2 episodes
• Weight Decay — 6× in this paper · explained in 2 episodes
• Adam — 5× in this paper · explained in 3 episodes
• Transformer — 5× in this paper · explained in 2 episodes
• Latent Space — 4× in this paper · explained in 7 episodes
• Obfuscation — 4× in this paper · explained in 2 episodes
• Sycophancy — 4× in this paper · explained in 5 episodes
• Transformers — 4× in this paper · explained in 2 episodes
• Decode — 3× in this paper · explained in 2 episodes
• Faithfulness — 3× in this paper · explained in 2 episodes
• Interpretability — 3× in this paper · explained in 15 episodes
• Chain-of-Thought (CoT) — 2× in this paper · explained in 5 episodes
• Chain-of-Thought (CoT) Reasoning — 2× in this paper · explained in 5 episodes
• Language Models (LLMs) — 2× in this paper · explained in 2 episodes
• Mechanistic Interpretability — 2× in this paper · explained in 4 episodes
• Papers — 2× in this paper · explained in 3 episodes
• Reasoning Traces — 2× in this paper · explained in 4 episodes
• Activation Patching — 1× in this paper · explained in 2 episodes
• Domain Shift — 1× in this paper · explained in 7 episodes
• Embeddings — 1× in this paper · explained in 2 episodes
• Knowledge Distillation — 1× in this paper · explained in 15 episodes
• Large Language Models (LLMs) — 1× in this paper · explained in 9 episodes
• Latent Representation — 1× in this paper · explained in 2 episodes
• Linear Probe — 1× in this paper · explained in 3 episodes
• Linear Probes — 1× in this paper · explained in 2 episodes
• Performance Gap — 1× in this paper · explained in 2 episodes
• RLVR — 1× in this paper · explained in 2 episodes
• Regularization — 1× in this paper · explained in 3 episodes
• Reinforcement Learning — 1× in this paper · explained in 4 episodes
• Reward Hacking — 1× in this paper · explained in 11 episodes
• Robustness — 1× in this paper · explained in 6 episodes
• Supervised Fine-Tuning (SFT) — 1× in this paper · explained in 13 episodes
• Training Objective — 1× in this paper · explained in 3 episodes
• Verifier — 1× in this paper · explained in 2 episodes

Sources

• GPT-4 Technical Report
• Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation
• Eliciting Latent Predictions from Transformers with the Tuned Lens
• Decoding Hidden Deception in Reasoning LLMs: Activation Explainers for Deception Auditing
• Reasoning Models Don't Always Say What They Think
• Monitoring the Internal Monologue: Probe Trajectories Reveal Reasoning Dynamics
• Are DeepSeek R1 And Other Reasoning Models More Faithful?
• The Llama 3 Herd of Models
• Training Verifiers to Solve Math Word Problems
• DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via Reinforcement Learning
• RL-Obfuscation: Can Language Models Learn to Evade Latent-Space Monitors?
• Implicit Chain of Thought Reasoning via Knowledge Distillation
• A Pragmatic Way to Measure Chain-of-Thought Monitorability
• Activation Oracles: Training and Evaluating LLMs as General-Purpose Activation Explainers
• Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety
• Measuring Faithfulness in Chain-of-Thought Reasoning
• Catching rationalization in the act: detecting motivated reasoning before and after CoT via activation probing
• Can Reasoning Models Obfuscate Reasoning? Stress-Testing Chain-of-Thought Monitorability
• Selective Latent Thinking: Adaptive Compression of LLM Reasoning Chains
• Qwen3 Technical Report

Related papers

• Exploring Solution Divergence and Its Effect on Large Language Model Problem Solving
• Ishigaki-IDS-Bench: A Benchmark for Generating Information Delivery Specification from BIM Information Requirements
• Subliminal Steering: Stronger Encoding of Hidden Signals
• MedStruct-S: A Benchmark for Key Discovery, Key-Conditioned QA and Semi-Structured Extraction from OCR Clinical Reports
• The End of Transformers? On Challenging Attention and the Rise of Sub-Quadratic Architectures
• Untangling the Mechanisms of Misleading Context in Medical Question Answering