
Manifund is launching an AI safety fund, run by Austin Chen and Caroline Ellison. In one sentence, our goal is to make fast, transparent, early-stage grants for AI safety.
There are lots of people who want to give money to AI safety and lots of projects that need money, but matching these up is far from a solved problem. A lot of current funding opportunities are optimized for large donors and grantees, are private about information, and move slowly. The Sunlight Fund wants to fill a gap in the funding ecosystem.
We believe given rapid AI progress, delays of months waiting for funders to decide on grants are quite costly. We’ll make grants within 1 week or less. From working on Manifund, we have experience in getting money out the door to grantees quickly.
All applications, grants, and reasoning will be public in real-time, so you can see exactly what grants are being made, when, and why. This allows donors to see what we’re doing rather than needing to blindly trust us. We also think this is a public good—it’s informative to others to see what projects are getting funded.
We think the best way to donate effectively is to make fast grants to early-stage opportunities as they come up. Large, established orgs with plenty of runway don’t need the money as much as projects that are just starting up do, and existing funders already serve them well. See When you donate can matter more than where for more thoughts on this.
Currently we expect check sizes to be in the range of $1k-$50k—this may grow if we get more funding.
Our grants will be made within the basic framework that transformative AI is likely coming soon, and making it go well is very important. Most grants will probably target preventing existential risk from AI, but could also include things like the welfare of digital minds or making the post-transformative AI future go well.
I’m excited about grants funding:
All applications are made as public project proposals on Manifund.
We aim to get back to all applicants in 1 week or less, up to the end of the year. At that point, we’ll reevaluate.
Applications open soon.
We’ve heard from many people lately who fit this description:
“I’m planning to donate $10-100k this year. In the past, I’ve donated to GiveWell: I trust them and I think they do good work identifying the top charities. I’m increasingly convinced that mitigating risk from advanced AI is a better donation target, but there’s no GiveWell equivalent and I don’t know what to donate to. I don’t want to spend a ton of effort evaluating opportunities. I get the sense that there’s lots of money floating around, and more coming, and maybe AI safety orgs just don’t need my money?”
We’re hoping the Sunlight Fund can be an answer to that.
You can also see Austin’s grantmaking history as a regrantor at his profile: https://manifund.org/Austin—whether you would have wanted to support his grants in the past might be a good indicator of whether you want to donate to this fund.
You can donate at the top of the page: Manifund accepts donations via credit card, ACH, wire transfer, crypto, and DAFs.
Manifund will take a 5% fee on donations to the fund to cover grantmaking, operations, and fiscal sponsorship.
Sunlight Fund hasn’t made any grants yet. Each one will appear here, with how much it was for, as soon as it’s made.
Sunlight Fund isn’t taking applications yet.