In the past three years, Manifund has run a regranting program, in which external donors who are less enmeshed in AI safety can delegate a budget to people with expertise, good track records, and wide networks.
This year, we’re also offering the option to join as a self-funded regrantor. If you have a significant amount of money to donate and you want to make the decisions yourself, you can join us and regrant through Manifund.
So, why regrant through Manifund instead of just donating yourself?
Logistics: we handle the “last-mile problem”
We think many of the best regrants aren’t to 501c3s—travel expenses, career transition funds, funding for independent researchers, early-stage organizations, and more. If you want to make any of these grants from a DAF, the grantee will have to go through the process of getting fiscal sponsorship on their own.
When you regrant through Manifund, we handle all the fiscal sponsorship logistics. We also get grant agreements signed, actually make the wire or ACH transfers to pay out the money, and generally handle all the annoying details.
After the first ACX Grants round in 2021, Scott Alexander noted:
Then you have to actually give people money.
You know how, whenever there’s a debate about cryptocurrency, some crypto fanboy gushes about how it makes sending money so much easier? And if you’re like me, you think “yes, but right now you can just enter a number into Paypal, that already seems pretty easy to me”?
I take it all back. The crypto future can’t come soon enough. Sending money is terrible.
Instead of crypto, Scott opted to have Manifund handle payouts for the future rounds of ACX Grants.
Transparency: we signal boost your grantees
Gavin Leech’s Shallow review of AI safety 2024 started as an $8000 Manifund regrant from Neel Nanda. Austin mentioned it in a post on EA Forum, and that led Coefficient Giving to fund the rest.
Our regrantors Evan, Marcus and Rachel’s donation rationales to Timaeus were helpful public evidence when they were just starting off and applying to larger funders like SFF.
We think knowing what people are funding is really valuable—being more aware of what’s being funded and why helps other funders fill in gaps and lets founders and researchers know what projects to prioritize. Having information about grantees, amounts, and the donors’ reasoning live on Manifund provides an easily accessible way for people to get this information, which we view as a public good.
For-profits: we facilitate investments, loans, and purchases
Most DAFs only support donations to other established 501c3 charities. However, we believe that many impactful opportunities exist within forprofit entities.
We think investments and loans are a better way for donors to send money to forprofits (compared to nondilutive grants), especially at six figures and up. These encourage fiscal discipline, better credit allocation, and capital efficiency. Our general policy is to return proceeds from an exit or repayment back to the regrantor’s Manifund account.
For example, Jueyan Zhang of AISTOF has structured a $400k investment into Equistamp, a $350k investment into Seldon, and a $400k loan to Ink & Switch. Peter from Ink & Switch said:
This loan will give me and our team so much peace of mind, allowing us to focus on our work instead of constantly juggling cash-flow.
We’re also excited for donors to offer purchase agreements or advance market commitments for specific kinds of scoped AI safety work. The idea is to provide revenue or a prize for a concrete outcome, rather than unconditional investment. Here’s one example of a $10k purchase agreement for 3 longform essays from The New Critic.
Dealflow: we find you opportunities to consider
Neuronpedia started as a $2500 Manifund proposal, got the attention of regrantors, and went on to become interpretability infrastructure used by DeepMind and Anthropic. It’s great when regrantors bring their own dealflow, but the regrantor program itself also brings proposals to Manifund you might not otherwise have seen.
Historically, Manifund has been an open platform where anyone can apply. In the future, we plan on being more proactive about sourcing new opportunities, eg with a scouting/referral program, or by building a database of every single person in AI safety and using LLMs to identify who should apply.
You can see this in the comments of Timaeus’s first proposal; Evan Hubinger funded it, and based on his funding and endorsement Rachel Weinberg and Ryan Kidd added funds from their own pot as well. Regranting brings with it a ready-made community of people with similar interests to talk about grants, exchange ideas, and encourage you to actually do the thing. In addition to the public comments on Manifund, we also have a Discord and (optional) dinners and meetups.
Flexibility: we test out weird funding mechanisms
Manifund has experimented with: impact markets; retroactive funding; quadratic funding + airdropped microgrants; dominance assurance contracts; essay contests; prediction markets; AI-powered grantmaking. (If Vitalik Buterin likes it, we’ve probably tried it.)
“How to optimally allocate dollars towards public goods” remains an unsolved problem, and we plan to continue trying things out, publishing our findings, and integrating our learnings into the way we support grants. If you want to experiment with any of this in your grantmaking, Manifund is a great place to do it.
Why you might not want to regrant
- Regrants are public, so if you don’t want your donations public it probably isn’t a good fit for you.
- Regrants flow through our 501c3, so we can only make grants consistent with a 501c3 charitable purpose; notably, political donations and some kinds of lobbying are prohibited.
- Ultimately, regrantors give recommendations, and the decision on whether to make the grant rests with Manifund. We extend great deference to our regrantors’ judgment, but are legally required to ensure grants comply with our charitable mission.
Next steps
If you have at least $50k to donate in 2026 and you’re interested in self-funded regranting, email [email protected]! Let us know what kinds of things you’d want to fund and the rough amount you’re looking to donate. We’re especially excited about hearing from those who have dealflow or networks that are currently undercovered.
