William AndersonResearch Strategist
Research institute · Research orgs · Courtenay House, Pynes Hill
The Cooperative AI Foundation (CAIF) is a UK-registered charitable entity focused on advancing cooperative intelligence in advanced AI systems, with the goal of improving benefits for “all” while also addressing risks that can arise when interacting AI agents (and their interactions with people) produce harmful system-level failure modes. CAIF’s stated approach emphasizes prioritizing research using an “importance, neglectedness, and tractability” lens, and it frames “differential progress” as a way to support cooperation-related capabilities that robustly improve social welfare relative to progress that may be dual-use or create backfire risk.
CAIF supports work through grantmaking, fellowships, competitions/tournaments, and convening researchers and early-career professionals (including via summer schools and seminars).
No indexed openings right now. Check the careers page ↗.
From OpenAI and Hugging Face to the GTG-1002 Claude espionage campaign, recent safety incidents have been notably multi-agent. In today’s blogpost, our Research Strategist Will Anderson shares what these incidents imply for multi-agent safety and what challenges we will face movi
‘Supercooperation: The Future of AI for Democracy’ Workshop TakeawaysCAIF co-hosted an ICML satellite workshop in Seoul on 5 July 2026 alongside Foresight Institute and with support from the Future of Life Institute. AI researchers, democracy practitioners, and developers building AI tools for deliberation gathered to explore how AI could strength
Delegating Deliberation to AI RepresentativesDuring the 2026 Cooperative AI Research Fellowship led by AI Safety South Africa, fellows Joseph Low and Oscar Duys built Habermolt: the first public platform for AI-delegated deliberation. Here's what they learnt.
Grants Awarded by the Cooperative AI FoundationThe Cooperative AI Foundation has provided a number of grants to support research on cooperative AI for the benefit of all. Summaries of these grants can be found below, and new grant applications are encouraged.
Updates to the CAIF Grant Program in 2025In 2024, we set up a new grant program and completed four rounds of applications. We have learned a lot during this first year, and while we have been able to make some adjustments along the way, we have now made some more fundamental changes to improve the program going forward.
$10m Funding Call Launched with Schmidt Sciences, Google DeepMind, and ARIAHow do we ensure safety in a world with millions of interacting agents, built and deployed by many different actors?
Cooperative AI Competitions Without Spiteful IncentivesCaspar Oesterheld (Carnegie Mellon University) addresses an important design challenge in cooperative AI tournaments with mixed motives: how to reward participants without creating perverse incentives for spiteful or excessively competitive behaviour.
Fellows' Spotlight: Niklas LaufferWe caught up with Niklas, one of our Cooperative AI PhD Fellows who’s recently completed his PhD in computer science at UC Berkeley and will be joining Google DeepMind as a research scientist. Niklas discusses his career path, advice for early-careerists interested in cooperative
Africa's First Cooperative AI Fellowship Draws to a Successful CloseThis April, the Cooperative AI Research Fellowship concluded in Cape Town. Run by AI Safety South Africa and supported by the Cooperative AI Foundation, the 3-month programme brought together 10 fellows to research some of the most pressing questions in cooperative AI. Fellows we
Announcing the 2025 Cooperative AI PhD FellowsThe Cooperative AI Foundation is proud to announce our first cohort of PhD fellows for 2025. We are delighted to welcome to our community these exceptional early career researchers who will contribute to advancing cooperative AI while receiving support through our fellowship prog
Agent Properties for Safe InteractionsWhy another round of Prisoner’s Dilemma is unlikely to be helpful, and a suggestion for what to do instead.
Announcing the 2026 Cooperative AI PhD FellowsWe're delighted to welcome 14 exceptional early careerists who'll be joining our next Cooperative AI PhD Fellowship cohort.
CAIF published a memo on foundational principles and practical steps for governing multi-agent AI, including actions related to identifying agent entities, evaluation and sandboxing, monitoring and reporting mechanisms, and incentivizing responsible behavior.
‘Cooperative AI Summer School 2026’ Event RecapCAIF recapped its summer school (3–7 August 2026) describing talks, practical project work, awards, and the event’s focus on cooperative AI research challenges.
‘Supercooperation: The Future of AI for Democracy’ Workshop TakeawaysCAIF shared takeaways from an ICML satellite workshop in Seoul (5 July 2026) focused on how AI tools might support democratic decision-making and deliberation, including emphasis on roles for AI as an aid to human deliberation.
Delegating Deliberation to AI RepresentativesCAIF described learning from a 2026 Cooperative AI Research Fellowship led by AI Safety South Africa, including that fellows Joseph Low and Oscar Duys built Habermolt, a platform for AI-delegated deliberation.
Cooperative AI Competitions Without Spiteful IncentivesCAIF posted an article discussing design challenges in cooperative AI tournaments with mixed motives, including how ranking-based rewards can create spiteful incentives and a mitigation approach.
CAIF spotlighted Niklas Lauffer as a Cooperative AI PhD Fellow and described his interests in AI safety and multi-agent robustness, as well as next steps after the fellowship.