The AI Safety Fund supports independent research addressing frontier-model safety and security. On December 11, 2025, Frontier Model Forum announced 11 grantees receiving more than $5 million. FMF took over management after Meridian Institute announced its June 2025 wind-down. The reviewed application page is closed; the later announcement says remaining funds will support narrowly scoped urgent research.
Programs
Who funds them, what they fund and how to apply.
AI x Animals Pooled Fund is a targeted, experimental pooled fund supporting projects that help animal advocates respond to AI. The 2026 request for proposals had more than $300,000 available, expected to support 5-20 projects, and prioritized advocate tools, field-building, and actionable research.
The Bureau of South and Central Asian Affairs posted the Artificial Intelligence for American Competitiveness and Economic Security (AI-ACES) in South Asia opportunity (#DFOP0018586) on grants.gov, with a deadline of August 2, 2026.
AISI's Alignment Project funded research to prevent advanced AI systems from behaving dangerously, intentionally or accidentally, with awards up to £1 million per project. Its first round closed on September 10, 2025 and awarded over £27 million to over 60 projects. AISI lists it under past programmes and says it is unlikely to run a similar program in 2026.
AISI's Challenge Fund supported UK and international academic institutions and non-profit organizations researching safeguards, control, alignment, and societal resilience to advanced AI risk, with awards up to £200,000 per project. Closed with no current open applications.
AlgoGrants was the Algorand Foundation's ecosystem funding programme during 2020-2022. Official historical reporting identifies it as the main support vehicle before Algorand Ventures began in late 2022. Current Ventures and xGov funding are separate programmes; no current intake is verified for this legacy record.
Animal Microgrants was a 2026 funding-circle call for individuals and very small organizations working on farm-animal advocacy. Typical grants were £500-£2,500, with exceptional requests up to £5,000; applicants were expected to hear back by the end of March.
An Arbitrum Foundation program that subsidized third-party smart contract audits with $10M in ARB over 12 months. Applications opened 1 August 2025 and closed 31 July 2026. Its successor, the Arbitrum Security Program, opened applications on 28 September 2026.
Arbitrum DDA used community-selected domain allocators and Questbook to award milestone-based grants. The program manager closed Season 3 submissions on March 17, 2026 and moved the program into lower-capacity milestone review, disbursement and impact reporting. The Arbitrum Foundation now lists the Arbitrum DAO Grant Program as complete and inactive.
Arbitrum DAO Grant Program offered milestone-based funding to builders, dApps, technical infrastructure projects, adoption initiatives, and community growth efforts on Arbitrum One and Arbitrum Nova. Season 3's active submission phase closed on March 17, 2026. The program then entered a lower-capacity phase for remaining application reviews, milestone checks, and disbursements through September 2026.
A one-time, DAO-approved incentive round that allocated its 50M ARB budget to 30 protocols on Arbitrum to bootstrap usage and liquidity. Recipients committed to incentive-distribution plans (e.g. liquidity mining, fee rebates) rather than open-ended grants; funded via an Arbitrum DAO governance proposal.
A follow-on, bridging round of the Short-Term Incentive Program that extended incentive funding to 39 protocols on Arbitrum after STIP Round 1 concluded, approved via Arbitrum DAO governance to avoid a funding gap while a longer-term incentive framework was designed.
A concluded Arbitrum DAO program, run on Questbook, distributing 5M ARB worth of grants to projects building with Stylus, marked "Program Complete" on the Arbitrum Foundation's grants page.
The Artificial Superintelligence Alliance offered $50,000 for development of an AI-native IDE for ASI:Chain. Proposals were submitted through DeepFunding by February 13, 2026.
Who Let The Docs Out ran the Automation & Humanity Documentary Fund for independent nonfiction films examining AI safety, alignment, autonomous systems, and their social effects. Its Spring 2026 research grants offered $8,000 for feature projects and $4,000 for short films, alongside monthly virtual sessions.
Amazon Research Awards' Spring 2026 call supported academic post-training research using AWS Trainium. The May 13 submission deadline has passed, with decisions scheduled for August 2026; selected teams may receive up to $50,000 or $250,000 and beyond in AWS promotional credits.
AWS Imagine Grant is an annual program for registered nonprofits using cloud technology to advance their missions. The 2026 U.S. cycle offers three award categories, including up to $200,000 in unrestricted support and $100,000 in AWS credits for the Pathfinder Award. Public applications closed June 5 and the selection process continues through December 2026.
Boundless San Francisco is an open call run by Lever for Change with Crankstart, a San Francisco-based family philanthropy, as donor partner, for projects that primarily benefit San Francisco. The total is up to $150 million, with up to three awardees each receiving up to $50 million, and up to seven finalists each receiving $250,000 to $500,000 from Crankstart plus technical assistance. Organizations anywhere in the world could apply in any subject area, and applicants could request grant funding or flexible, below-market capital such as a low-interest loan. The call launched August 5, 2026, registration closed September 11, 2026, and applications closed October 15, 2026, extended from October 9. Review has three steps: administrative review, participatory review where applicants score each other's proposals, and an external evaluation panel. Finalists and awardees are selected in 2027.
Status describes whether a program is running. Application methods are recorded separately. Read the funder’s terms before applying.