AI safety research links

I.Frontier labs

Anthropic

Research

• Research
• News
• Alignment Science
• Interpretability (Transformer Circuits)
• Frontier Red Team

Model docs

• System cards
• Claude’s constitution

Safety framework

• Responsible Scaling Policy, risk reports, safety roadmap

Landmark papers

• Agentic misalignment2025
• On the biology of a large language model2025
• Auditing language models for hidden objectives2025
• Alignment faking2024, with Redwood
• Scaling monosemanticity2024
• Sleeper agents2024

OpenAI

Research

• Research
• Safety
• Alignment Research Blog

Model docs

• Deployment Safety Hub (system cards, safety evaluations)
• Model Spec

Safety framework

• Preparedness Framework v2 (PDF)

Landmark papers

• Monitoring reasoning models for misbehavior2025
• Weak-to-strong generalization2023

Google DeepMind

Research

• Blog
• Google Research blog
• Safety Research (Medium)

Model docs

• Model cards

Safety framework

• Frontier Safety Framework and per-model FSF reports

Landmark papers

• An approach to technical AGI safety and security2025
• Model evaluation for extreme risks2023

For other developers, METR maintains links to their safety frameworks, including prior versions.

II.Independent research organizations

Evaluations and red-teaming

• METR Research Updates Key paperMeasuring AI ability to complete long tasks2025
• UK AI Security Institute Research Blog Inspect (eval framework)
• US CAISI (NIST) Overview Research blog
• Apollo Research Research Blog Key paperFrontier models are capable of in-context scheming2024
• Epoch AI Blog Gradient Updates Benchmarking hub
• Irregular
• Palisade Research
• SecureBio

Alignment, control and interpretability

• Redwood Research Research Blog Key paperAI control2023
• Truthful AI Key paperEmergent misalignment2025
• Goodfire
• Transluce
• FAR.AI
• ARC
• Timaeus
• Center for AI Safety Key papersWMDP benchmark2024  Representation engineering2023
• CHAI (Berkeley)
• MIRI

Governance and strategy

• GovAI Research
• CSET (Georgetown) Publications
• RAND Technology and Security Policy center
• IAPS Research
• AI Futures Project Key paperAI 20272025
• CLTR
• SaferAI

III.Reference

Capabilities

• Epoch Capabilities Index
• Epoch benchmarking hub
• Epoch trends

Safety frameworks

• METR frontier AI safety policies
• Common elements of frontier AI safety policies

Lab scorecards

• AI Safety Index (Future of Life Institute)

Evidence synthesis

• International AI Safety Report

Model docs

• Anthropic system cards
• OpenAI Deployment Safety Hub
• Google DeepMind model cards

Eval tooling

• Inspect

Incidents

• AI Incident Database

Forums

• AI Alignment Forum
• LessWrong

Training

• MATS
• ARENA
• BlueDot Impact

IV.Foundational and cross-lab papers

• Concrete problems in AI safety Amodei et al., 2016
• Risks from learned optimization in advanced machine learning systems Hubinger et al., 2019
• Toy models of superposition Elhage et al., 2022
• Universal and transferable adversarial attacks on aligned language models Zou et al., 2023
• Gradual disempowerment Kulveit et al., 2025
• Chain of thought monitorability: a new and fragile opportunity for AI safety Korbak et al., 2025

V.The OpenAI–Hugging Face incident

Hugging Face

• Security incident disclosure16 July 2026
• Be ready before the attack: a practical guide to self-hosting an open model for cyber defense20 July 2026
• Anatomy of a frontier lab agent intrusion: a technical timeline of the July 2026 incident27 July 2026

OpenAI talk

• The ‘Breaking’ news: the OpenAI–Hugging Face incidentBlack Hat USA, August 2026 (video)

Reports

• OpenAI and Hugging Face partner to address security incident during model evaluationOpenAI, 21 July 2026, updated 26 August
• Hugging Face incident investigation report (PDF) and summaryMETR and Redwood Research, 26 August 2026
• Incident technical report (PDF)OpenAI, 26 August 2026
• The Hugging Face incident and the road aheadOpenAI, September 2026

Public evidence

• Revealing the details of how OpenAI agents hacked Hugging FaceForman et al., 25 September 2026

VI.Researchers’ blogs

• Ajeya Cotra Planned Obsolescence
• Boaz Barak Windows on Theory
• Chris Olah colah.github.io
• Dean Ball Hyperdimensional
• Gwern Branwen gwern.net
• Helen Toner Rising Tide
• Holden Karnofsky Cold Takes
• Jan Leike Musings on the Alignment Problem
• Joe Carlsmith joecarlsmith.com
• Lennart Heim blog.heim.xyz
• Lilian Weng Lil’Log
• Miles Brundage Miles’s Substack
• Neel Nanda neelnanda.io
• Paul Christiano AI Alignment
• Peter Wildeford The Power Law
• Simon Willison Simon Willison’s Weblog
• Steven Byrnes LessWrong posts
• Victoria Krakovna vkrakovna.wordpress.com
• Yoshua Bengio yoshuabengio.org
• Zvi Mowshowitz Don’t Worry About the Vase

VII.Podcasts and newsletters

Podcasts

• 80,000 Hours Podcast
• 80k After Hours
• AXRP
• Dwarkesh Podcast
• FLI Podcast

The Ezra Klein Show (AI safety episodes)

• Intelligence isn’t power Arvind Narayanan, October 2026
• ‘This is nuts.’ An OpenAI insider explains why he quit David Robinson, October 2026
• Bill Gates’s blunt warning on A.I. Bill Gates, September 2026
• The ‘But China!’ dilemma driving the A.I. race Matt Sheehan, September 2026
• The A.I.s are already out of control Helen Toner, August 2026
• How afraid of the A.I. apocalypse should we be? Eliezer Yudkowsky, October 2025
• The government knows AGI is coming Ben Buchanan, March 2025
• What if Dario Amodei is right about A.I.? Dario Amodei, April 2024
• Freaked out? We really can prepare for A.I. Kelsey Piper, March 2023
• Is A.I. the problem? Or are we? Brian Christian, June 2021

Newsletters

• Import AI
• Transformer
• AI Safety Newsletter (CAIS)