grantmaking.ai is an open platform to discover, evaluate, and fund high-impact AI
safety work. We bring every organization, person, project, fund, and grant we can
find into one place, layer trusted signal on top, and make it easy for the people
with money to give and the people doing the work to find each other.
Our mission is to build the grantmaking infrastructure that AI safety is missing.
The ecosystem is about to receive billions of dollars from a new wave of funders,
and the system they are inheriting was never built for them. Recommendations live
in private emails, Substack posts, and Slack threads. Funding status goes stale
the day it is posted. Small, high-upside projects fall through the cracks because
no one has the time to look. We think that information should be open, current,
and in one place. And we think funding decisions are better when they happen in
the open.
What it is
Three things working together.
A public directory. A searchable database of AI safety
organizations, people, projects, funds, and grants, drawn from public sources and
kept current. Filter by cause area, funding status, team size, or endorsements,
and trace who has funded whom.
A live map of who needs funding. The
page surfaces the projects that are actively fundraising right now, including what
they are working on and how much they need, so a funder with limited time can go
straight to live opportunities.
A signal layer. Comments, upvotes, and named endorsements from the community,
plus private reviews from the expert reviewers we work with. A public
activity feed shows the latest discussion and applications. This is
the part a plain database is missing: not just the data, but informed takes on
what is worth funding and why.
On top of that, grantees can apply for funding once and be seen by every
funder on the platform, and everything public is available through a
public API, including a keyless tier and an /llms.txt for AI
agents.
Who it is for
Grantees are researchers, practitioners, and small teams doing concrete work
to reduce x-risk from AI. List your project, keep your funding status
live, and apply once instead of rewriting the same application for every funder.
We are especially built for the work traditional funders overlook: small asks,
early-stage bets, and short runways.
Funders are individual donors, angels, and small funds looking for high-impact
AI safety work to support. This is the tier the big funds are least designed to
serve, and the one where a single grant can change what happens next. Find live
funding gaps, reuse other people's diligence, and lean on trusted signal instead
of rebuilding the map yourself.
The community is everyone in AI safety who wants to share takes, evaluate
projects, and help good work get funded. Use the database and API freely, and
weigh in with comments and endorsements. Some members go on to become paid
reviewers or regrantors.
For funders: fund the work others miss
We do not try to replace the major AI safety funders. For large, legible grants,
they are often the right first stop. We fill the layer underneath: small, public,
high-upside opportunities where current funding status is hard to find, expert
judgment is scattered, and a grant of $20k to $50k can change what happens next.
If you have money you want to move into AI safety, whether you are giving for the
first time after a liquidity event or you have been doing this for years, tell us
your funding thesis and we will help you find the right route. Sometimes that is
grantmaking.ai, and sometimes it is not.
Most of what you see started as public information that we gathered, cleaned, and
linked together in one place. Live grant data flows in continuously from:
Manifund
the Long-Term Future Fund (via EA Funds)
the Survival and Flourishing Fund (SFF)
Coefficient Giving (formerly Open Philanthropy)
We also build out organization, person, and project profiles from public websites
and directories like AISafety.com. Because not everything labeled "AI" is about
reducing x-risk, we use language models to filter for relevance and to
suggest cause-area tags, then keep a human in the loop.
Every profile carries its provenance: where a piece of data came from, when it was
last updated, and whether it was scraped, AI-assisted, or provided by the people
involved. When someone who manages a profile edits it, their version takes
precedence, but we keep the original underneath. If something is wrong or out of
date, the people closest to it can claim their profile and correct it.
Claimed and verified profiles
These are two different ideas, and both show up on profiles.
Claimed means a profile is managed by someone at that organization or project.
Claiming an entity lets you edit it and control whether your funding details are
public or private. It happens automatically when you apply for funding, or on
request.
Verified means a specific section of a profile's data has been checked. A
verified badge shows who verified it, when, and how. It expires automatically the
moment the underlying data changes, so a badge always reflects the current data
rather than a stale snapshot. Separately, comments from reviewers we have vetted
carry a reviewer badge, so you can tell an expert take from an organic one.
Public and private data
The platform is public by default. The directory, funding asks, public comments,
endorsements, and approved grants are open to anyone, both on the web and through
the API.
Some things are sensitive, so we let them stay private:
Funding details like runway, budget, and specific amounts. A project or
organization can mark these private, though the fact that they are actively
fundraising always stays public.
Private comments, including candid reviewer and funder notes and
confidential concerns, kept off the public page.
References and reviewer scores, which are always private to reviewers.
Private data is visible only to verified funders, reviewers, and admins. Everyone
else can see that private data exists, but not its contents, so totals and
discussion structure stay honest without exposing the details. The same rules
apply through the API, where private fields come back empty and private comments
come back redacted.
Want access to the private layer? It is reserved for approved funders and
reviewers. Email
hi@grantmaking.ai
or to get verified.
The launch round
We are kicking off the platform with a $1M grant round for work that reduces
x-risk from AI.
Grants of $5k–$50k
For individuals, small teams, and organizations
Focused strictly on reducing x-risk from AI. Technical safety, governance,
field-building, and evaluations are all in scope, as long as the path to
reducing x-risk is clear.
A short application with a human-written theory of impact
Reviewed by a panel of AI safety practitioners and grantmakers, with decisions
made on a rolling basis, so strong applications can be funded before the round
closes
Fiscal sponsorship through Manifund, so you can receive a grant without
setting up your own legal entity
The priority deadline is July 13, 2026, and every application
submitted before it is guaranteed a review. The round is a way to launch the
platform and seed it with live opportunities, not a one-time event. We intend to
keep funding AI safety work well beyond it.
Marcus Abramovitch:
Managing Partner at AltX, former professional poker player, and
prediction-market trader, who has personally donated around $1.5M to
high-impact causes.
We are expanding the directory, deepening the evaluation tools, and growing the
public API so the whole ecosystem can build on the data. The long-term goal is
durable infrastructure for moving millions of dollars into high-impact AI safety
work, quickly and transparently and in the open. This page will keep pace as the
platform grows.