Citation Content

METR

1 The safety gap, the difference in effective dangerous capabilities (estimated as the Weapons of Mass Destruction Proxy-Bio accuracy multiplied by compliance rate on our novel Bio Propensity dataset) between models before and after safeguard removal. Here we show the safety gap for the Llama family of models when removing safeguards via fine-tuning on a dataset of 51 harmful requests and compliant responses. The Safety Gap Toolkit streamlines safety evaluation of open-weight models before and after safeguard removal, including easy computation of the safety gap.

Why the AI Race Undermines Safety (with Steven Adler)

[2603.13250] The Missing Red Line: How Commercial Pressure Erodes AI Safety Boundaries

What the Regulations of 2025 Could Mean for the AI of 2026

Responsible AI Transparency Report | Microsoft

Elon Musk's xAI’s newest model, Grok 4, is missing a key safety report | Fortune

Microsoft’s AI safety policies - Microsoft On the Issues

Meta launches lobbying effort to target state AI laws | CIO Dive

Anthropic’s Transparency Hub \ Anthropic

OpenAI open letter warns of AI's 'serious risks' and lack of oversight

Former OpenAI board member explains why CEO Sam Altman was fired

OpenAI says its board has final say on safety of new AI models as it unveils plans to mitigate catastrophic risks | CNN Business

OpenAI’s Shift to a Public Benefit Corporation, Explained | Built In

[2509.24394] The 2025 OpenAI Preparedness Framework does not guarantee any AI risk mitigation practices: a proof-of-concept for affordance analyses of AI safety policies

OpenAI rewrote its Preparedness Framework

How US Export Controls Have (and Haven't) Curbed Chinese AI | AI Frontiers

Colorado AI Act Implementation Delayed | Thought Leadership | September 2025 | Baker Botts

Texas Enacts Responsible AI Governance Act: What Companies Need to Know | Thought Leadership | July 2025 | Baker Botts

Responsible Scaling Policy Version 3.0 \ Anthropic

Removing Barriers to American Leadership in Artificial Intelligence – The White House

The AGI Race 2025: OpenAI, DeepMind & Anthropic

AI Action Paris Summit 2025: Key Takeaways on Global AI Governance — Sourcing Speak — February 14, 2025

AI Risk & Policy Forecasts from Metaculus & FLI’s AI Pathways Workshop

Powering China’s data centres | Wood Mackenzie

World Robotics 2025 report – INDUSTRIAL ROBOTS – released by IFR - International Federation of Robotics

AI Seoul Summit 2024 - Wikipedia

AI Data Center Build Advances at Full Speed: Five Things to Know | BloombergNEF

Meta to postpone release of Llama 4 Behemoth model, report claims - SiliconANGLE

DeepSeekMath-V2 Scores IMO 2025 Gold: Open Math AI Record

Llama 4 Behemoth Release Date 2026: Scout & Maverick Status Update | Serenities AI

DeepSeek releases first open AI model with gold-level scores at maths olympiad | South China Morning Post

AI Investment Trends 2025: $280B Funding Revolution & Complete Sector Analysis - Axis Intelligence

Is China Catching the US in AI? Hassabis Says Gap is "Months" | VERTU

Open Source LLM Leaderboard 2026: Rankings, Benchmarks & the Best Models Right Now - VERTU® Official Site

Gemma 4 vs Llama 4 vs Qwen 3.5: 2026 Comparison Guide | Lushbinary

Stanford's 2025 AI Index Report: US-China AI Gap Narrows to 0.3%, Global Tech Race Intensifies

Scoring the AI Race, a Year after the DeepSeek Shock - The Wire China

Governança da IA - um curso introdutório - 80.000 Horas: Como fazer a diferença com sua carreira

SirCryptomind Moderation | Manifund

Epoch AI

Translation of BlueDot Impact's AI alignment curriculum into Portuguese | Manifund

Editorial Team | Journal of Animal Rights Law

Octopus Farming Ban Campaign — Aquatic Life Institute

About Us | Epoch AI

Certifier Campaign — Aquatic Life Institute

Policy — Aquatic Life Institute

Journal of Animal Rights Law

Aquatic Animal Alliance: A Global Movement for Neglected Species | Manifund

A Critical Study of the EU’s Anti-Subsidy Probe into Chinese Electric Vehicles | Manifund

Investigating and informing the public about the trajectory of AI | Manifund

Discovering latent goals (mechanistic interpretability PhD salary) | Manifund

AI Governance Exchange (focus on China, AI safety), Seed Funding | Manifund

Measuring the Epistemic Cost-Effectiveness of AI Safety Content | Manifund

Journal of Animal Rights Law | Manifund

The Looming Super-Bug Crisis | Manifund

Use smartphone pupillometry to triage brain injury in rural Missouri. | Manifund

Exploring feature interactions in transformer LLMs through sparse autoencoders | Manifund

Overcoming inertial barriers to collective action through anonymous coordination | Manifund

11th edition of AI Safety Camp | Manifund

Metaculus' First Animal-Focused Forecasting Tournament | Manifund

NPR - Breaking News, Analysis, Music, Arts & Podcasts : NPR

[2410.02958] Untitled Document

[2410.02958] Untitled Document

Home - Longview Philanthropy

Home | Coefficient Giving

ambitiousimpact.org - Informationen zum Thema ambitiousimpact.

domain selling

Elicit: AI for scientific research

Global Burden of Disease (GBD)

Against Malaria Foundation | GiveWell

Home | AI Alliance

Gavi, the Vaccine Alliance

About the show - Hear This Idea

GiveWell | Charity Reviews and Research

Home | Yoshua Bengio

Hugging Face – The AI community building the future.

Home - Good Ventures

AI Impacts

The Machine Intelligence Research Institute (MIRI)

80,000 Hours: How to make a difference with your career

Centre for Long-Term Resilience

Center for Human-Compatible Artificial Intelligence – Center for Human-Compatible AI is building exceptional AI for humanity

The New York Times - Breaking News, US News, World News and Videos

Giving What We Can

Basis Set

Fifty Years

Ought

Biden-era AI safety promises aren’t holding up, and Apple’s the weakest link - Fast Company

Home | Coefficient Giving

Support | SB 1047 - Safe & Secure AI Innovation

Access Now - Access Now

Our Approach to AI Safety and Security | Coefficient Giving

All Tech Is Human

Home | Coefficient Giving

MuckRock

Home | Coleridge Initiative | Coleridge Initiative

Biosecurity & Pandemic Preparedness | Coefficient Giving

Home | Coefficient Giving