FundingBench
← all funders

Coefficient Giving — Navigating Transformative AI Fund

typical check size
$25k (research expenses) to multi-million multi-year org support; median on the order of $200k–$400k (approximate)
yearly spend
~$46–50M/year on AI safety & governance through this fund (2025, per Inside Philanthropy; approximate — Open Phil's historical AI spend has been higher some years)
grants per year
~100–200/year (440+ total through the fund as of 2026; approximate)
as of
2026-08 (approximate)

Thesis

Coefficient Giving (formerly Open Philanthropy, renamed 2025) runs the Navigating Transformative AI Fund to reduce the risk of AI-caused global catastrophes and help society prepare for rapid AI advances. It is the successor to Open Philanthropy's technical AI safety and AI governance & policy programs, and spans research to make AI systems more trustworthy and controllable, AI governance and policy work, and capacity-building for the AI safety field.

The fund practices hits-based giving: it accepts that many grants will fail if the expected value of the wins is high enough. It favors legible cost-effectiveness reasoning, strong teams over polished proposals, and is comfortable being the majority funder of an organization. It funds academics (research expenses, startup packages), nonprofits (seed and multi-year operating support), and individual researchers, and has made 440+ grants through the fund. It generally avoids work that mostly accelerates AI capabilities, and scrutinizes theory-of-change carefully for policy/advocacy work.

Backed primarily by Dustin Moskovitz and Cari Tuna, it is by far the largest funder in AI safety, so its effective bar is 'is this among the best marginal uses of ~$50M/year' rather than scarcity of capital. Speed and founder quality matter; it increasingly writes rolling grants rather than batching into rounds.

Grantmakers

  • Peter Favaloro — Leads technical AI safety grantmaking; economics PhD, formerly at Open Philanthropy in the same role. (approximate — team details compiled quickly)
  • Luke Muehlhauser — Longtime lead of AI governance and policy grantmaking, formerly executive director of MIRI. (approximate)

Process

Rolling recommendations rather than fixed rounds; hits-based giving philosophy; comfortable as majority funder; separate RFPs for specific research directions. NOTE: this rubric is a best-guess approximation compiled 2026-08 without exhaustive verification.

Sources