Chain-of-Thought Oversight Requires Separate Evidence for Faithfulness and Monitorability
cs.AI, cs.CR
Submitted: 2026-11-01
Updated: 2026-11-01
Code: https://github.com/Sichao-Li/cot-oversight-evidence
Terminology
Sources
- Understanding intermediate layers using linear classifier probes
- Analyzing and Improving Chain-of-Thought Monitorability Through Information Theory
- Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation
- Purple Llama CyberSecEval: A Secure Coding Benchmark for Language Models
- CyberSecEval 2: A Wide-Ranging Cybersecurity Evaluation Suite for Large Language Models
- Reasoning Models Don't Always Say What They Think
- When Chain of Thought is Necessary, Language Models Struggle to Evade Monitors
- How to use and interpret activation patching
- Faithfulness as Information Flow: Evaluating and Training Faithful Chain-of-Thought Reasoning
- Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety
- Measuring Faithfulness in Chain-of-Thought Reasoning
- STREAM (ChemBio): A Standard for Transparently Reporting Evaluations in AI Model Reports
- Measuring Chain-of-Thought Monitorability Through Faithfulness and Verbosity
- Qwen2.5 Technical Report
- Audit Cards: Contextualizing AI Evaluations
- Investigating CoT Monitorability in Large Reasoning Models
Related papers
- Multi-Objective Structured Pruning of LLMs for Latency and Model Size Optimization
- MAVEN-T: Reinforced Heterogeneous Distillation for Real-Time Multi-Agent Trajectory Prediction
- Model Discovery Agent: LLM-assisted Bayesian experiment design for data-efficient discovery of mechanistic world models
- Embeddings for Preferences, Not Semantics
- Can We Trust Item Response Theory for AI Evaluation?
- Diagnosing Task Insensitivity in Language Agents