LawAI builds legal capacity in a field that has very little of it, from hubs in Washington DC and Cambridge, UK. We fund its EU team. Its summer fellows have gone on to the EU AI Office, the UK AISI and the US Department of Commerce. With Cambridge's Programme on AI Science & Policy it publishes the Cambridge Commentary, the first scholarly commentary devoted to the AI Act's GPAI model rules.
Astralis finds and funds the most promising yet neglected work to make advanced AI secure and beneficial for all.
What we do
We think advanced AI will be the most consequential development of this century, for better and for worse. The research, institutions and agreements that will shape it are being built now, and what is built now will last. Very few funders pay for that work, and outside the US, almost none.
Astralis exists to close part of that gap, making grants where funding is least likely to arrive otherwise. We pool money from donors in several countries, none of it from major AI companies, and focus where our international and independent position is especially useful — currently AI governance and strategy in Europe and Asia. Our main vehicle is our pooled fund, Shared Horizons. Where it helps, we also support grantees with strategy and introductions.
Focus areas
We look for important work that others are unlikely to fund, and where being an independent European, multi-donor foundation gives us leverage. Our current priorities are:
AI safety capacity and international cooperation in Asia.
Asia will be central to how advanced AI is developed and governed, but safety-focused funding and institutions remain concentrated in the West. By our own mid-2026 estimate, philanthropic funding for frontier AI safety in China runs at roughly $20 million a year, against more than $1 billion in the rest of the world. We support organisations building AI safety capacity in the region, including in China, and the channels that keep researchers and institutions worldwide in contact.
AI progress, safety, and governance capacity in Europe.
European institutions are making consequential decisions about advanced AI, and their rules travel: the EU's AI Act and Code of Practice shape how companies behave well beyond Europe. As a European funder, we are well placed to build the independent expertise — legal, technical and policy — needed to make those rules work, and to make the case that safety and progress are compatible.
These are our current priorities, not the limits of what we fund.
We are open to backing neglected work wherever the case is strong and expect to open new areas as we grow. We are currently scoping AI company oversight, and with Macroscopic Ventures we've launched a $10 million request for proposals on the role of middle powers in the AI transition; the deadline for applications is Friday 23rd October.
Selected grants
Institute for Law & AI
SASH
SASH, the Singapore AI Safety Hub, conducts practical research and trains AI safety and security talent across the Asia-Pacific. Its research spans agent governance, cybersecurity and AI verification. Its Singapore AI Safety Fellowship brings East and West researchers together on joint projects, and its Frontier AI Security Training upskills experienced cybersecurity professionals in AI security. Astralis is one of several funders.
Simon Institute for Longterm Governance
The Simon Institute works from Geneva on the international coordination needed to reduce extreme risks from frontier AI, with a focus on sustaining contact between Western and Chinese institutions. Our grant funded the senior team that carries out that work. The institute has briefed diplomats in Geneva and New York dozens of times, and published recommendations for the UN's scientific panel on AI.
How we work
International perspectives
Our team spans several countries, and we grant through legal entities in the UK and Sweden. That lets us complement AI safety work, which today is concentrated in the US, and brings us grants others find harder to assess and fund: a grantee working on something as sensitive as AI governance may prefer funding from donors in its own region.
Ambition over certainty
We back grants for the size of the counterfactual difference they could make, not for how safe a bet they are. That means accepting that many grants will not work out, because most of our impact will come from the few that do. It does not mean accepting downside risk, which we assess separately.
Expert-driven decisions
Our grantmakers often come from the fields they now fund, and they source their own grants rather than waiting for proposals to arrive. They own their portfolios: the person who investigates a grant is the one who argues for it and works with the grantee afterwards. For each grant we set out the case and the main uncertainties, and test our reading against outside experts. Beyond assessing whether to make a grant, we look at how we can reduce downsides and raise upsides.
Independent, multi-donor funding
We pool money from well over a dozen donors across several countries, and no single donor dominates our fund. We take no money from major AI companies or their employees, from AI industry associations or lobbying groups, or from governments, government-funded organisations or political parties active in AI policy — our gift acceptance policy sets this out in full.
"It's critical that the world is prepared for the new global risks and opportunities that come with more powerful AI, yet effective international governance remains a significant gap. I chose to anchor this initiative because Astralis offers a strategic, efficient way to broaden participation. It's a rare opportunity to sit at the front lines of shaping how AI impacts the world."
— Jan Beckers, philanthropist and tech investor, GTR Foundation and BIT Capital
People
Astralis is a small team of grantmakers and operators, working remotely across several countries, including the UK, the US and Sweden. People here have founded or run organisations before coming to grantmaking, and have arrived from BCG, GiveWell and Concordia AI among others. We expect to be hiring in operations and in our donor-facing work over the coming months.
Contact
Astralis does not accept unsolicited funding requests. We are glad to hear from anyone interested in funding this work, partnering with us or putting an idea to us — write to ops@astralisfoundation.org. If you are interested in working at Astralis, see Careers.