Long-Term Future Fund

The Long-Term Future Fund (LTFF) is a grant programme operated within EA Funds, an infrastructure project of the Effective Ventures Foundation — the parent body of the Centre for Effective Altruism (CEA). Launched in February 2017 as one of the four original EA Funds, it was founded with Nick Beckstead of the Open Philanthropy Project as its sole manager. The fund's stated mission is to make grants reducing global catastrophic risks, with particular emphasis on catastrophic risk from advanced AI, and to support the broader project of improving humanity's long-term prospects. As of 2025, LTFF is managed by Caleb Parikh (Project Lead, EA Funds), Oliver Habryka (CEO of Lightcone Infrastructure), and Linchuan Zhang (Senior Researcher at Rethink Priorities).

Scale and mechanics

LTFF targets a specific stratum of the EA-aligned AI safety funding ecosystem: individuals and small organisations whose work is too large to self-fund but too small or speculative to attract major institutional funders. Grants typically range from $1,000 (short research stipends or compute) to $200,000 (organisational seed funding), with a median of approximately $25,000. Annual grantmaking reached approximately $6.67 million across 263 grants in 2023, with a projected $8.4 million for 2024. Cumulative grants since 2017 exceed $20 million total, with roughly half going to AI safety work directly.

Applications were processed on a fully rolling basis through 2023–2024. In 2025, LTFF shifted to quarterly rounds to "recalibrate funding thresholds given a rapidly changing giving environment" and to provide applicants with clearer feedback and more predictable timelines. The acceptance rate over May 2023–March 2024 was 19.3%. LTFF publishes detailed public writeups for its grant recommendations — a transparency practice that distinguishes it from most philanthropic vehicles in the AI safety space.

The fund uses a "one excited manager" rule: a grant is approved when at least one fund manager is strongly in favour, even if others are neutral. This champion-based logic is designed to surface high-conviction speculative bets that would be averaged away by consensus gatekeeping — the same philosophy embedded in the S-Process used by the Survival and Flourishing Fund.

Focus and grantee mix

Approximately 67% of LTFF grants address AI safety directly, with remaining funding going to biosecurity, forecasting, and other existential-risk areas. Within the AI safety allocation, the fund covers a broad portfolio: technical alignment research (interpretability, robustness, agent foundations); AI governance and policy analysis; career-transition upskilling for researchers entering the field; and field-building for the AI safety community.

The grantee mix is tilted heavily toward individuals — independent researchers, graduate students, early-career practitioners, and content creators building community infrastructure — rather than established organisations. Representative grantees include Gabriel Mukobi, who received a $40,680 stipend for a Stanford computer science master's with an AI safety focus and was subsequently accepted to multiple PhD programmes; and Joshua Clymer, who received $1,500 for GPU compute to run alignment experiments. Podcasts and written content serving the field-building function — AXRP, Robert Miles — have received repeat grants. Organisational grantees tend to be early-stage: Manifold Markets received a $200,000 LTFF grant in 2022 before becoming a widely-used EA forecasting platform.

Position in the corpus funder slice

Within the corpus's EA-aligned AI safety funder cluster, LTFF fills the small-grants entry-point slot — the tier below both Survival and Flourishing Fund and Coefficient Giving. SFF's median grant is approximately $100,000 and its S-Process runs on a semiannual cycle; Coefficient Giving's median exceeds $250,000 and its portfolio logic is institutional in character. LTFF's $25,000 median and quarterly cycle serve a structurally different applicant pool: researchers at the start of their careers, projects at proof-of-concept stage, and work that has not yet accumulated the track record institutional funders require. LTFF's pipeline role is explicit in the field: many grantees subsequently receive larger grants from SFF or Coefficient Giving, or join frontier AI laboratories.

The fund's institutional home within Effective Ventures / CEA also gives it a different character from SFF, which operates as a virtual fund via a donor-advised fund at the Silicon Valley Community Foundation. LTFF is embedded in the organisational infrastructure of the EA movement itself, and its funding sources reflect that: Coefficient Giving is reported to provide 40–50% of LTFF's capital, with direct individual donations making up most of the remainder.