Working to bridge the gap between frontier AI models and the level of cybersecurity they need by connecting professionals to high-leverage opportunities in AI security.

A collaborative research programme between the Leverhulme Centre for the Future of Intelligence and the Centre for the Study of Existential Risk at the University of Cambridge, focused on the global risks, governance, and long-term safety of advanced AI.

A research initiative at the University of Oxford's Martin School that combines technical AI expertise with deep policy analysis to understand and mitigate lasting risks from AI through governance research, decision-maker education, and training future technology governance leaders.

An AI safety research group led by David Krueger at the University of Cambridge's Computational and Biological Learning Lab (2021-2024), focused on technical AI alignment, deep learning safety, and reducing existential risk from advanced AI.

AI-Plans is a platform for discovering, critiquing, and advancing AI alignment strategies, hosting a contributable compendium of alignment plans and running community research events.

Workshop Labs is a public benefit corporation building billions of personalized, privacy-preserving AI models with a mission to keep humans empowered as AI advances.

OpenAI is an AI research and deployment company working to ensure that artificial general intelligence benefits all of humanity. It is the creator of ChatGPT, GPT-4, and a wide range of frontier AI models.

A UK-based research and advocacy think tank that combines complexity modelling, expert elicitation, and democratic deliberation to improve policymaking around existential and catastrophic risks.

A global non-profit building AI safety governance capacity across Asia through policy research, training, and multi-stakeholder dialogue, starting in Southeast Asia.

Coordinates and supports rationality-focused community meetup groups worldwide, serving as a hub for ACX (Astral Codex Ten), LessWrong, and broader rationality community organizers.

LawZero is a nonprofit AI safety research organization founded by Yoshua Bengio to develop safe-by-design AI systems that cannot act autonomously or pursue hidden goals.

UC Berkeley's Center for Long-Term Cybersecurity (CLTC) is a research and collaboration hub advancing future-oriented cybersecurity research, policy, and education, with a growing focus on AI safety governance and risk management for frontier AI systems.

Meridian Cambridge is an independent research and incubation hub in Cambridge, UK focused on AI safety, biosecurity, frontier-risk policy, and institutional design. Formerly Effective Altruism Cambridge CIC, it hosts the Cambridge AI Safety Hub, biosecurity and governance hubs, research labs, and fellowships.

An international advocacy organization devoted to reducing global catastrophic risk from all threats and hazards, working with governments worldwide to enact policies that address existential and catastrophic risks.

AI Safety Argentina (AISAR) is a 6-month research scholarship program based at the University of Buenos Aires that connects Argentine students with mentors to conduct AI safety research.

The Preamble Windfall Foundation is a small Pittsburgh-based 501(c)(3) that supports animal welfare research and philanthropy guidance, notably through the Planetary Animal Welfare Survey (PAWS) project.

Sincxpress Education is a STEM education company founded by Dr. Mike X Cohen that produces online courses and textbooks on applied mathematics, deep learning, and mechanistic interpretability for AI safety. Its courses have reached over 300,000 learners worldwide.

Partnership on AI is a global multi-stakeholder nonprofit that brings together industry, civil society, and academia to address the social implications of AI and promote responsible development and deployment.

FutureSearch is an AI forecasting startup that deploys teams of LLM agents to research, analyze, and forecast across structured data, emphasizing legible reasoning behind predictions.

A Substack blog by PhD mathematician Robert Huben documenting his Open Philanthropy-funded year of AI safety research and writing, covering mechanistic interpretability, AI risk, and related topics.

The Woodrow Wilson International Center for Scholars is a congressionally chartered, nonpartisan think tank in Washington, DC that bridges the world of ideas and the world of policy through research, analysis, and scholarship on global affairs.

High Impact Professionals (HIP) helps experienced mid-career and senior professionals transition into high-impact roles and commit to effective giving across global health, animal welfare, and global catastrophic risk reduction. Through its Impact Accelerator Program, Talent Directory, and HIP Pledge Club, HIP channels professional talent and financial resources toward the most pressing global problems.

AI2050 is a philanthropic initiative of Schmidt Sciences that funds exceptional researchers worldwide working on the hard problems required for AI to be hugely beneficial to society by 2050.

ChinaTalk is a podcast and newsletter covering China, technology, and US policy, founded by Jordan Schneider. It serves as a hybrid think tank and media outlet providing non-partisan analysis on US-China relations and emerging technology.

EA Netherlands (Effectief Altruïsme Nederland) is the national effective altruism community-building organization for the Netherlands, running introductory programs, supporting local groups, and hosting major EA events.

Humans in Control is a nonpartisan grassroots movement working to protect people and future generations from the risks of unchecked AI through advocacy, coalition-building, and state-level policy campaigns.

A nonprofit research institute applying foresight methods to anticipate and mitigate societal-scale risks from advanced artificial intelligence. TFI produces rigorous research for policymakers and decision-makers working to prevent catastrophic AI outcomes.

The AI Safety Communications Centre (AISCC) connects journalists to AI safety experts and resources, helping improve media coverage of AI risks and safety issues.

The University of Virginia is a major public research university in Charlottesville, Virginia, with faculty and programs conducting AI safety and alignment research.

A Substack newsletter by Ajeya Cotra exploring AI capabilities, timelines, and the societal implications of increasingly autonomous AI systems.

Horizon Events is a Canadian non-profit that advances AI safety R&D by organizing high-impact events, including the AI Safety Unconference series and monthly Guaranteed Safe AI Seminars.

The self-study section of AISafety.com curates courses, textbooks, and reading lists for independent learning in AI safety, covering both technical alignment and AI governance.

The personal newsletter of Miles Brundage, former Head of Policy Research at OpenAI, covering independent AI policy research and governance.

OAISI is a student- and researcher-led community at the University of Oxford committed to reducing catastrophic risks from advanced AI. It runs technical and governance programmes to support existing researchers and introduce new Oxford talent to AI safety work.

A bi-weekly newsletter by Concordia AI covering technical AI safety research, governance, and policy developments in China, aimed at bridging the knowledge gap between China's AI safety ecosystem and the global community.

Stanford University's interdisciplinary research center tackling critical security challenges, including AI governance, nuclear risk, biosecurity, and emerging technology policy.

Yale University is a private Ivy League research university in New Haven, Connecticut, home to several AI safety and governance research programs, including the Schmidt Program on AI and National Power, the Center for Algorithms, Data, and Market Design (CADMY), and the Digital Ethics Center.

A remote, non-profit research group focused on mechanistic interpretability of deep learning models, developing causal abstraction frameworks, open-source course materials, and mentorship programs for the AI safety community.

An Ivy League research university in Philadelphia with multiple programs relevant to AI safety, including formal verification of autonomous systems, AI governance research, and AGI international security analysis.

The University of Maryland, College Park is a flagship public research university conducting extensive AI safety, trustworthy AI, and responsible AI research through multiple interdisciplinary institutes and centers.

A program within EURAIO (European Responsible Artificial Intelligence Office) that convenes expert summits and develops frameworks to address AI-driven psychological manipulation and protect civil liberties from autonomy-eroding AI systems.

A hardware security startup developing tamper-proof enclosures for AI chips to prevent physical attacks on AI hardware and enable international AI governance through verifiable compliance mechanisms.

An R&D lab building tools to map how ideas spread online, helping communities understand their information landscape and defend against coordinated manipulation.

A research project developing non-maximizing, aspiration-based designs for AI agents that avoid objective function maximization, arguing that such optimization is inherently unsafe in sufficiently capable AGI systems.

Trajectory Labs is a nonprofit coworking and events space in downtown Toronto dedicated to AI safety research and community building. It provides free workspace, weekly events, and a peer network to grow Toronto's AI safety ecosystem.

A leading Israeli research university home to the Governance of AI Lab (GOAL), which conducts cross-disciplinary research on AI governance, legal alignment, and the safe development of advanced AI systems.

One of the world's oldest and most prestigious research universities, Oxford has been a central hub for AI safety and existential risk research through institutions like the Future of Humanity Institute (FHI) and the Oxford Martin AI Governance Initiative (AIGI).

5050 is a free 12-14 week company-builder program run by Fifty Years that helps scientists, researchers, and engineers become deep-tech startup founders, with a dedicated AI safety track.

Straumli is an AI safety company that offers managed auditing and self-serve evaluations to help AI developers identify misuse risks and ship safer models faster.