Cover Image

Chain of Thought Monitorability: A New and Fragile Opportunity for AI Safety

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Korbak, Tomek, Balesni, Mikita, Barnes, Elizabeth, Bengio, Yoshua, Benton, Joe, Bloom, Joseph, Chen, Mark, Cooney, Alan, Dafoe, Allan, Dragan, Anca, Emmons, Scott, Evans, Owain, Farhi, David, Greenblatt, Ryan, Hendrycks, Dan, Hobbhahn, Marius, Hubinger, Evan, Irving, Geoffrey, Jenner, Erik, Kokotajlo, Daniel, Krakovna, Victoria, Legg, Shane, Lindner, David, Luan, David, Mądry, Aleksander, Michael, Julian, Nanda, Neel, Orr, Dave, Pachocki, Jakub, Perez, Ethan, Phuong, Mary, Roger, Fabien, Saxe, Joshua, Shlegeris, Buck, Soto, Martín, Steinberger, Eric, Wang, Jasmine, Zaremba, Wojciech, Baker, Bowen, Shah, Rohin, Mikulik, Vlad
Format:	Preprint
Published:	2025
Subjects:	Artificial Intelligence Machine Learning
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Lessons from Studying Two-Hop Latent Reasoning
by: Balesni, Mikita, et al.
Published: (2024)

How to evaluate control measures for LLM agents? A trajectory from today to superintelligence
by: Korbak, Tomek, et al.
Published: (2025)

Towards evaluations-based safety cases for AI scheming
by: Balesni, Mikita, et al.
Published: (2024)

Large Language Models can Strategically Deceive their Users when Put Under Pressure
by: Scheurer, Jérémy, et al.
Published: (2023)

The Reversal Curse: LLMs trained on "A is B" fail to learn "B is A"
by: Berglund, Lukas, et al.
Published: (2023)

Monitoring Reasoning Models for Misbehavior and the Risks of Promoting Obfuscation
by: Baker, Bowen, et al.
Published: (2025)

A sketch of an AI control safety case
by: Korbak, Tomek, et al.
Published: (2025)

Me, Myself, and AI: The Situational Awareness Dataset (SAD) for LLMs
by: Laine, Rudolf, et al.
Published: (2024)

Frontier Models are Capable of In-context Scheming
by: Meinke, Alexander, et al.
Published: (2024)

Practical challenges of control monitoring in frontier AI deployments
by: Lindner, David, et al.
Published: (2025)

Evaluating Frontier Models for Stealth and Situational Awareness
by: Phuong, Mary, et al.
Published: (2025)

Realistic honeypot evaluations for scheming propensity
by: Krakovna, Victoria, et al.
Published: (2026)

A Pragmatic Way to Measure Chain-of-Thought Monitorability
by: Emmons, Scott, et al.
Published: (2025)

AI Behind Closed Doors: a Primer on The Governance of Internal Deployment
by: Stix, Charlotte, et al.
Published: (2025)

AI Control: Improving Safety Despite Intentional Subversion
by: Greenblatt, Ryan, et al.
Published: (2023)

Training Agents to Self-Report Misbehavior
by: Lee, Bruce W., et al.
Published: (2026)

Safety Cases: A Scalable Approach to Frontier AI Safety
by: Hilton, Benjamin, et al.
Published: (2025)

Looking Inward: Language Models Can Learn About Themselves by Introspection
by: Binder, Felix J, et al.
Published: (2024)

Async Control: Stress-testing Asynchronous Control Measures for LLM Agents
by: Stickland, Asa Cooper, et al.
Published: (2025)

Gram: Assessing sabotage propensities via automated alignment auditing
by: Lindner, David, et al.
Published: (2026)

An Approach to Technical AGI Safety and Security
by: Shah, Rohin, et al.
Published: (2025)

Honesty to Subterfuge: In-Context Reinforcement Learning Can Make Honest Models Reward Hack
by: McKee-Reid, Leo, et al.
Published: (2024)

Das Berlin Max Webers
by: Aldenhoff-Hübinger, Rita, et al.
Published: (2026)

Adsorption design for wastewater treatment / David O. Cooney
by: Cooney, David O

X-ray reflectivity study of a W/Si multilayer grating
by: P. Mikulík
Published: (2001)

Uncovering Latent Human Wellbeing in Language Model Embeddings
by: Freire, Pedro, et al.
Published: (2024)

The Persistent Ambiguity of Adverse Drug Reactions
by: Edward M. Sellers, et al.
Published: (2025)

Stress Testing Deliberative Alignment for Anti-Scheming Training
by: Schoen, Bronson, et al.
Published: (2025)

Aligning language models with human preferences
by: Korbak, Tomasz
Published: (2024)

When Chain of Thought is Necessary, Language Models Struggle to Evade Monitors
by: Emmons, Scott, et al.
Published: (2025)

A Benchmarking Suite for Flexible Job Shop Scheduling Problems with Worker Flexibility under Uncertainty
by: Hutter, David, et al.
Published: (2025)

Frequency Scaling Laws for Flat Plate Wing Active Separation Control
by: Vey, Stefan, et al.
Published: (2025)

Safety case template for frontier AI: A cyber inability argument
by: Goemans, Arthur, et al.
Published: (2024)

Red novae, stellar mergers in binary and triple systems, and bipolar nebulae
by: Kaminski, Tomek
Published: (2024)

Emergent Compositional Communication for Latent World Properties
by: Kaszyński, Tomek
Published: (2026)

In Defiance: 20 Abolitionists You Were Never Taught in Schools. By TomWeiner and AmilcarShabazz. Northampton, MA: Interlink/Olive Branch Press, 2025. 248 pp. $25.00 (paperback). ISBN: 978‐1‐62‐371661‐5
by: Beverly Tomek
Published: (2025)

Introduction to AI Safety, Ethics, and Society
by: Hendrycks, Dan
Published: (2024)

Atoms and Molecules as Quantum Attosecond Processors
by: Farhi, Asaf
Published: (2025)

On refinements of two-term Machin-like formulas
by: Farhi, Bakir
Published: (2026)

Expansion of a bivariate symmetric mean in the neighborhood of the first bisector
by: Farhi, Bakir
Published: (2025)