UK AI Safety Institute Awards £27M to 60 Alignment Research Teams
March 7, 2026 · 2 min read
Arthur Griffin
The UK AI Safety Institute has awarded its first 60 grants under the Alignment Project, distributing £27 million to research teams across eight countries working on making AI systems safer and more predictable. A second funding round opens this summer.
From 800 Applications to 60 Funded Projects
The competition was fierce. Over 800 applications from 466 institutions across 42 countries competed for the inaugural round. The selected projects span mathematics, learning theory, economics, and cognitive science — reflecting the institute's view that alignment is not purely a computer science problem.
Three projects illustrate the range. LawZero's "Scientist AI" initiative is building safe-by-design systems that improve how AI assesses information reliability. Yale and MIT economists are developing governance frameworks that balance capability gains against safety constraints. Stanford researchers are pursuing "Precision Optimisation," using energy-conserving methods to make AI training behavior more predictable.
A Growing Coalition of Tech Giants and Governments
What started as a £15 million program backed by the Canadian AI Safety Institute, CIFAR, Schmidt Sciences, AWS, and Anthropic has nearly doubled in size. OpenAI contributed £5.6 million, with Microsoft, Australia's AI Safety Institute, and several philanthropic organizations joining the coalition.
The expansion reflects a rare consensus: major AI developers, government safety bodies, and academic funders all agree that alignment research needs dedicated, well-funded attention — even as they compete fiercely on capabilities.
Second Round Opens This Summer
Deputy Prime Minister David Lammy announced the grants alongside AI Minister Kanishka Narayan at the AI Impact Summit in India. The Alignment Project website will post details for the second funding round when applications open.
Researchers planning proposals should note the first round's emphasis on interdisciplinary approaches. Pure technical proposals competed against work drawing on economics, philosophy, and institutional design — and the funders clearly rewarded breadth. Grant seekers tracking international AI funding can monitor these opportunities through Granted, which indexes AI-related funding across federal and international sources.