TL;DR

The UK AI Safety Institute has announced its first 60 grant recipients under the Alignment Project, a global programme to advance AI alignment research. Total funding has reached £27 million — up from £15 million at launch — with new partners including OpenAI (contributing £5.6 million), Microsoft, and Australia’s AI Safety Institute joining the coalition.

What Happened

The Alignment Project launched in July 2025 with backing from an international coalition including the Canadian AI Safety Institute, CIFAR, Schmidt Sciences, AWS, Anthropic, and UK Research and Innovation. The first funding round received over 800 applications from 466 institutions across 42 countries.

After assessing submissions for relevance, feasibility, innovation, and team capability, 101 proposals were shortlisted. The final 60 projects span mathematics, learning theory, economics, cognitive science, and other fields.

Among the funded projects: LawZero, a non-profit founded by Yoshua Bengio, is developing “Scientist AI” — a system designed to track data provenance and use a prover-verifier setup to keep reasoning transparent. Yale and MIT economists are applying mechanism and information design to create frameworks for governing AI behaviour. Stanford researchers are working on making AI training more predictable through “Energy Conserving Descent,” which could make it easier to steer models toward safer outcomes.

Why It Matters

The programme addresses a practical concern: as AI systems become more capable and autonomous, current alignment methods may not hold up. The breadth of disciplines represented — from economics to physics to cognitive science — reflects a view that alignment progress requires tools from well beyond computer science.

The growing coalition of funders is also notable. OpenAI and Microsoft both joined as new partners in this round, alongside Australia’s AI Safety Institute, signalling international momentum behind alignment research.

Looking Forward

Applications for the next funding round open this summer. The programme aims to remove barriers that have historically limited alignment research by funding people, compute time, and cross-disciplinary collaboration at scale.