Survival and Flourishing Fund (SFF)

SFF exists to "support the long-term survival and flourishing of sentient life," and in practice the large majority of its money goes to reducing existential risk from advanced AI. It is not a foundation with a program staff and a house view; it is a grant-recommendation process run for its funders — overwhelmingly Jaan Tallinn (Skype co-founder), with additional funders like Jed McCaleb and the Future of Life Institute. Since 2019 it has coordinated roughly $152M in gifts, and in recent years distributes on the order of $30–40M per year across ~90–100 organizations.

The worldview anchoring the money is Tallinn's: his stated goal is to "maintain and improve humanity's ability to exist and enjoy existing," with the current decade's focus on AI extinction risk. He favors restrictive measures on frontier AI — datacenter certification and international oversight, legally mandated speed limits on AI computation, liability laws (including "Know Your Developer" for open-source), mandatory labeling of AI interactions, veto committees for large deployments, and global off-switch capacity — alongside constructive measures: collective-intelligence tools for human cooperation, guaranteed-safe AI with quantifiable safety specs, hardware-level controls, and AI for healthcare. SFF rounds since 2024 have also run explicit Freedom (avoiding concentration of power, protecting speech and individual liberty) and Fairness (AI empowering the disempowered, resisting monopoly, inclusive governance) tracks, and 2026 added theme rounds in climate change, animal welfare, and human enhancement — but the Main Track (AI x-risk broadly) gets the large majority of dollars.

Culturally, SFF's recommenders are drawn from the rationalist/x-risk community (past and current recommenders include Zvi Mowshowitz, Nate Soares, Malo Bourgon, Scott Alexander, Oliver Habryka, Eliezer Yudkowsky as a Speculator), so proposals are read by people who take AGI ruin scenarios seriously, are skeptical of pure capabilities-adjacent work, and reward legible theories of change about x-risk reduction — technical alignment research, AI governance and policy advocacy (including "slow down / regulate" advocacy that other funders shy away from), field-building like MATS, and biosecurity (e.g. SecureDNA). The S-Process is explicitly designed "to generally favor funding things that at least one Recommender is excited to fund, rather than things that every Recommender is excited to fund" — so a proposal that strongly excites one credible evaluator can win even if others are lukewarm, which makes SFF friendlier to weird, speculative, or contrarian bets than consensus-driven funders. What it avoids: individuals without fiscal sponsors (orgs only), work outside the survival-and-flourishing frame, and — per Tallinn — things outside his stated priorities (with carve-out exceptions for climate and life extension). Grants range from ~$5k to ~$1.6M per round, so it writes many mid-six-figure checks rather than a few mega-grants.

Decisions run through the S-Process ("Simulation"): each round, 4–12 part-time Recommenders (subject-matter experts, many from the rationalist/x-risk community) read applications and build marginal-value functions over funding levels for each org in a simulation app, refined over multiple meetings; an algorithm aggregates the curves (including a "Synthetic Mean Recommender") and allocates each funder's budget. Funders — chiefly Jaan Tallinn — always have final say. The process deliberately favors things at least one Recommender is excited about rather than requiring consensus, so single-champion dynamics matter. Since 2024, rounds run parallel tracks (Main, Freedom, Fairness; 2026 adds Climate, Animal Welfare, and human-enhancement theme rounds). Applications are rolling, but evaluation happens in annual rounds (2026 Main Round: apply by April, recommendations ~September, money by year-end). Every applicant must first receive a Speculation Grant to be guaranteed S-Process eligibility (95%+ historically approved): individual Speculators (~dozens of trusted community members, each with ~$400k default budgets from a pool that has grown to ~$20M) can approve fast grants — sometimes within a week — which are retroactively netted against later S-Process recommendations, and Speculators' future budgets rise or fall based on how their picks score in the next round. Eligibility: registered charities almost anywhere, for-profits in US/UK/Canada/Australia under stricter conditions; no unsponsored individuals. There is no stated rubric beyond alignment with the survival-and-flourishing mission and Tallinn's published priorities; matching pledges (e.g. 2:1) are often attached to incentivize other donors.