Humanity AI's First $18M: What the $500M Coalition's Inaugural Grants Reveal About the AI-Public-Interest Funding Map

On May 12, 2026, Humanity AI — the ten-foundation pooled fund launched in October 2025 with a $500 million five-year commitment — announced its first $18.1 million in grants. The composition of the inaugural cohort is more revealing than the headline number. Twelve organizations received $500,000 each, in the same week, with no apparent competitive solicitation. A separate $3 million grant went to Data & Society to lead an AI Civics initiative with the Digital Public Library of America. And $10 million was set aside for a forthcoming open call expected to launch in summer 2026. For nonprofits trying to read which themes, organizations, and methods the coalition will fund over the next four years, this first cohort is the most concrete signal yet.

It is also the moment to understand how a coalition this large — Doris Duke, Ford, Lumina, Kapor, MacArthur, Mellon, Mozilla, Omidyar, Packard, and Siegel Family Endowment — has decided to operate. Coalitions of this scale rarely succeed at pooled grantmaking; competing internal priorities and divergent program staff usually fragment the work back into parallel single-foundation streams within two years. Humanity AI's design choices, visible for the first time in this cohort, suggest the founders have anticipated that failure mode and built around it.

What $500K Bought, Twelve Times

The nine identified $500,000 grants went to: AI Now Institute, the Center for Democracy and Technology, the Center on Resilience & Digital Justice, the Council on Foreign Relations LEAD AI program, the Distributed AI Research (DAIR) Institute, Kinfolk Tech, Partnership on AI, the Pulitzer Center, Student Defense's SHAPE AI program, and TechEquity. The work spans research and policy on AI accountability (AI Now), civil-rights-focused AI governance (CDT), community-driven AI research (DAIR), labor accountability (TechEquity), institutional AI adoption in higher education (Student Defense), AI-aware journalism (Pulitzer Center), and foreign-policy analysis (CFR LEAD AI).

The $500,000 grant size and the simultaneous announcement timing function as a deliberate signal. The coalition is not making bets on individual flagship organizations at this stage; it is funding a peer network. Twelve organizations of roughly equal weight, working across roughly equal time horizons, each given enough to add two to three people for two years — that is the architecture of a field-building strategy, not a moonshot. Compare this to the typical AI-policy funding announcement of the last three years, where a single foundation will frequently announce a single $5 million to $25 million grant to a single anchor institution. Humanity AI's approach is closer to ecosystem-level grantmaking: small enough grants to spread risk, large enough grants to attract serious staff, given to a deliberately diverse set of organizational forms.

The $3 million Data & Society grant for AI Civics — co-led with the Digital Public Library of America — is the only outlier in the cohort, and the outlier reveals where the coalition is most willing to concentrate. AI Civics is described as community engagement on AI creation, deployment, and use through library partnerships. That description — civic infrastructure, plural-institution participation, the public library as venue — is unmistakably MacArthur Foundation language layered with Mellon Foundation methodology. The coalition has placed its largest first-round bet on a project that pairs technological literacy with the cultural-infrastructure model that Mellon has spent two decades building.

Five Priority Areas, Visible in the Cohort

The Humanity AI initiative was announced in October 2025 with five priority areas: democracy, education, humanities and culture, labor and economy, and security. The inaugural cohort distributes across those areas in revealing proportions. Democracy and civil-society infrastructure dominate — AI Now, CDT, Partnership on AI, CFR LEAD AI, and the Data & Society civic-engagement initiative are all primarily democracy-and-governance grants. Labor and economy gets TechEquity. Education gets Student Defense. Humanities and culture gets Kinfolk Tech, which reframes memory through art and technology. Journalism, which is folded loosely under democracy, gets the Pulitzer Center.

Security — defined in the original announcement as maintaining safety standards in deployments like autonomous vehicles and lending decisions — is absent from the inaugural cohort. Either the coalition has not yet found the right security-focused grantees, or it intends to fund that pillar through different mechanisms, perhaps including the upcoming $10 million open call. Either reading matters for organizations considering an application. Operational AI safety in deployment contexts — algorithmic accountability for credit, hiring, criminal justice — is the area least represented in the first round and likeliest to receive deliberate attention in the second.

The composition also tells nonprofits what not to expect. The coalition's program areas, at least at this stage, do not include direct AI research funding, AI model development, AI safety alignment in the technical sense, or AI compute infrastructure. Humanity AI is funding the social, political, and institutional response to AI, not AI itself. Organizations that have positioned themselves as builders of AI tools — even mission-driven AI tools — are not on the inaugural list, and the program's framing suggests they should not expect to be on subsequent ones either.

The $10M Open Call: What to Watch For

Humanity AI has committed $10 million to an open call launching in summer 2026. The fund has shared little about the criteria, but several inferences are reasonable based on the inaugural cohort and the coalition's design.

First, expect grant sizes in the $250,000 to $1 million range rather than $5 million flagship awards. The inaugural cohort's standardization at $500,000 suggests the coalition is comfortable in that band and probably uncomfortable above $3 million for any single grantee outside of strategically chosen anchors.

Second, expect priority for organizations serving communities most directly affected by AI deployment — workers in algorithmic-managed industries, residents in high-surveillance contexts, students at institutions adopting AI in instruction and evaluation, journalists covering AI accountability. The coalition's repeated framing of "people-centered AI" and "communities as architects, not just users" is not boilerplate; it is a screening criterion.

Third, expect competitive emphasis on organizations that can demonstrate work already in progress. The inaugural grantees are nearly all organizations with three or more years of existing AI-focused work; the coalition has shown no appetite in this round for funding ideas that are still at the concept stage. Open-call applicants whose strongest case is "we propose to build" will be at a disadvantage against applicants who can describe what they have already built.

Fourth, watch for the fiscal sponsor. The pooled fund is administered by Rockefeller Philanthropy Advisors, and RPA will hire the executive director who runs the day-to-day program. The selection of that executive director — and the open call's review committee — will shape the second cohort's identity more than any abstract priority statement.

How This Reshapes the AI-Public-Interest Funding Landscape

The U.S. philanthropic landscape for AI-public-interest work has been, until recently, dominated by Open Philanthropy on the safety-and-alignment side and a fragmented set of individual-foundation grants on the policy-and-civil-society side. Humanity AI's emergence as a $500 million pooled fund on the civil-society side is structurally significant. It does not compete with Open Philanthropy's technical safety portfolio — the two funders do almost no work in the same space — but it does compete with, and probably consolidates, the policy-and-civil-society work that individual member foundations were doing independently.

Ford, Mellon, MacArthur, and Mozilla were each running AI-related programs in 2024 and early 2025 with overlapping grantees and partly redundant strategies. The decision to pool those resources is a bet that coordinated grantmaking will compound the effects of each foundation's individual work. For grantees, the practical consequence is that the multi-foundation cultivation cycle — separately courting program officers at Ford, Mellon, and MacArthur over the same project — is becoming one application to one pooled fund. That is a meaningful reduction in fundraising overhead for grantees and a meaningful increase in funder leverage over which projects get backed.

It also means rejection from Humanity AI will carry weight that a rejection from a single foundation would not. Organizations applying to the summer 2026 open call should treat the application as they would a multi-foundation collaborative — because that is exactly what it is.

For broader context on how AI public-interest funding interacts with the federal research-funding environment in 2026, see Granted's analysis of the AI funding chasm and AI alignment and safety research funding. The story of Humanity AI is not that philanthropy is replacing federal AI funding. It is that ten foundations have decided to fund the parts of the AI story that the federal government has chosen not to fund — and the inaugural $18 million is the map of where they intend to go.