AI Safety Funder Bulletin

Last updated October 15, 2026

This is a digest of funders in the AI safety space. The goal is to give an overview of who is funding in the space that includes the most relevant information for people seeking funds, looking to donate, or looking to work in grantmaking.

These tables are a rough summary of the numbers in the digest and the numbers are often best guesses based on public info; more detail is in the writeup and the footnotes. For funders that do work in multiple cause areas, it only considers grantmaking and staff focused on AIS. Some of the funding may repeat between rows—e.g. Coefficient donating to BlueDot or AISTOF making grants through Manifund—but this shouldn't substantially change the big-picture numbers.

This is intended to be a living document that updates regularly. If you have any corrections, please reach out!

At a glance

Funder name	$ donated in 2025	# grants made in 2025	Expected $ donated in 2026	FTE	General applications	Open RFPs	Accepting donations
Coefficient Giving	$409M1	2322	$1B3	75	—	✓	✓
OpenAI Foundation	$0	0	$250M4	3	—	—	—
Longview Philanthropy	$60M5	—	$200M6	17	—	—	✓
Macroscopic	$30M7	—	$100M8	11	—	✓	—
Astralis Foundation	—	—	$100M9	12	—	✓	✓
SFF	$35M10	8811	$64M12	10	✓	✓	—
Lightcone Commons	$0	0	$50M13	2	✓	—	✓
AISTOF	~$15M14	—	$40M15	1	—	—	—
Schmidt Sciences	$10M16	2717	$20M18	2	—	—	—
BlueDot Impact	$019	019	$10.1M20	1	✓	—	—
GTR AI Safety Fund	—	—	~$10M21	2	✓	—	✓
Manifund	$6M22	14423	$8.8M24	2	✓	—	✓
Transformative AI Fund (formerly LTFF)	$1.2M25	2025	$4M26	1	✓	—	✓

Funder profiles

These are roughly sorted by money moved per year.

▸Coefficient Giving

Website
Background:
- Previously Open Philanthropy, it grew out of a partnership between GiveWell (founded in 2007 by Holden Karnofsky and Elie Hassenfeld) and Good Ventures (foundation started in 2011 by Cari Tuna and Dustin Moskovitz). They're mainly funded by Tuna and Moskovitz, but looking to work with more donors.
Thesis:
- The biggest funder in the space by far; AI safety is one of a dozen cause areas they make grants in. Grants are usually sourced through their own research rather than applications.
By the numbers:
- donations: allocating $1b to catastrophic risks in 2026
- grant sizes: $12k to $160m
- number of grants: the Navigating Transformative AI Fund lists 189 in 2025 ($324.3m) and 75 so far in 2026 ($240.6m); the GCR Opportunities Fund lists 43 ($85.0m) and 13 ($46.3m)
- note that the public grants database understates their AI giving: their $160m grant to Resolution, announced in July 2026, does not appear in it, and nor do the Redwood Research grant of more than $70m over two years or the $10m for an institute being established by Fields Medallist Jacob Tsimerman, both named in their September 2026 scaling post. In September 2026 they said they committed $351m across technical AI safety, security, and fieldbuilding in 2025 — more than the $324.3m the database shows for the whole Navigating Transformative AI Fund.
- staff: 217 total. 43 of the 119 staff working on grants are on AI-focused teams, and I estimated 75 total by amortizing the staff working on operations, communications, and partnerships.
Get involved:
- Apply for funding:
  - They have open RFPs for:
    - capacity-building (e.g. training and mentorship programs, events, groups, coworking spaces, media, online infrastructure, career advising) — decisions within 3 months
    - programs and events on global catastrophic risk, effective altruism, and related topics — decisions within 3 months
    Their career development and transition funding RFP has closed: that program moved to BlueDot Impact on September 4, 2026.
  - Project Tailwind (launched September 9, 2026) is a call for founders to start new AI safety organizations targeting catastrophic risks, with a list of initiatives they want funded. Funding runs from pre-seed ($200k-$2m) through seed ($2m-$20m) to scale ($20m-$200m+); express interest here. They automatically filter out submissions unrelated to catastrophic risks from AI, or that appear AI-generated. There is no deadline, though they say they may stop accepting submissions if they receive a large volume.
  - Their GCR capacity building team also takes a general application for work relevant to their goals that doesn't fit one of the programs above.
- Donate: they are recruiting funders giving > $250k/year; reach out to partnerwithus@coefficientgiving.org
- Apply for a job: as of October 15, 2026 their careers page says “There are no open roles at this time.” You can still submit a general expression of interest, and the page now also points would-be founders at Project Tailwind

▸OpenAI Foundation

Website
Background:
- A nonprofit that owns a large stake in OpenAI, spun off in 2025. Their AI resilience team is run by OpenAI cofounder Wojciech Zaremba.
Thesis:
- They now describe three priority programs: life sciences and curing diseases, AI resilience, and civil society and philanthropy. They treat the economic impacts of AI as “part of the broader AI resilience agenda” but say that, “given the scale of the economic transition,” they are developing that work as a separate program. Within AI resilience, they're focused on bio-resilience, cyber-resilience, AI model safety, and AI's impact on young people.
By the numbers:
- donations: $130m+ in AI resilience grants that they said in June 2026 they were working to finalize
- staff: 3 publicly named in AI resilience — Wojciech Zaremba, Divya Siddarth, and Dan Wattendorf, whom Zaremba announced in 2026 as Head of Bio-Resilience; they said in August 2026 that the team is “still small”
Recent updates:
- The Foundation still hasn't published a list of the AI resilience grants it described as being finalized in June. The ones that have surfaced were announced by the grantees. The Child Mind Institute said on July 22, 2026 that it had launched a research initiative on youth mental health and AI chatbots “with support from the OpenAI Foundation,” without naming an amount; SecureBio Detection said in August 2026 that it had received $17.2m to cut its pathogen detection time from fourteen days to three. Inside Philanthropy reported in September 2026 that the Foundation has no single publication schedule for its grants.
- On September 9, 2026 they appointed Paul Christiano — founder of the Alignment Research Center, who led alignment research at OpenAI from 2017 to 2021 and is now a Senior Tech Advisor at NIST's CAISI — to the Foundation Board, as a non-voting observer on the OpenAI Group PBC Board, and to the Foundation Board's Safety and Security Committee chaired by Zico Kolter. He recuses himself from all OpenAI-related matters and model evaluations.
- In August 2026 they said they are hiring for more than 20 roles and aim to invest at least $1b over the next year.
Get involved:
- Apply for a job: they're hiring for 14 roles, all in San Francisco, listed here. The most AI-safety-relevant are the two AI Resilience roles, a Program Officer for AI Model Safety and a Program Officer for AI Resources; the Program Director for Formal Methods and Chief of Staff for AI Resilience postings listed in September 2026 have since been delisted.
- Apply for funding: the Foundation itself has no application route, and has not published one.

▸Longview Philanthropy

Website
Background:
- Founded in 2018 by barrister Natalie Cargill, who is now listed simply as Founder; Simran Dhaliwal is CEO.
Thesis:
- Primarily a donor advisor: they design giving strategies for major philanthropists and move most money via grant recommendations, alongside their own funds (Frontier AI Fund, Digital Minds Fund, Nuclear Weapons Policy Fund, and the Emerging Challenges Fund, which they are now closing).
By the numbers:
- donations: they directed over $75m in 2025 and say they expect 2026 to be “substantially bigger”; their (now removed) hiring materials earlier said they aimed to move $200m in 2026. The only per-fund figure they publish is for the Frontier AI Fund, which “supported 18 organizations” in its first nine months, December 2024 to September 2025.
- staff: 34 listed as of October 15, 2026, 10 of them in the AI program and 8 of those with AI in the title; around half the org working on AIS gets 17.
Get involved:
- Apply for funding: they post periodic RFPs here, though none are open right now; otherwise opportunities are sourced via proactive research
- Donate: contact form; they're interested in donors giving at least $1m/year for advisory services, and their private funds are open to donors giving over $100k
- Apply for a job: their careers page lists no open roles and takes general applications only.

▸Macroscopic

Website
Background:
- Swiss nonprofit founded by Ruairí Donnelly, Jonas Vollmer, David Althaus, and Daniel Kestenholz in 2019. Formerly Center for Emerging Risk Research and Polaris Ventures.
Thesis:
- Within AI safety, they're focused on preventing AI misuse, AI welfare, and cooperation between advanced AI systems. They also donate to reason & democracy and animal welfare and do for-profit investing in their areas of interest.
By the numbers:
- donating: up to $100m this year
- grant sizes: $100k to $15m
- staff: 16 listed on their about page as of October 15, 2026, up from 14. AI governance is the only named AI program and has one grants director; one grants associate works on animal welfare and three roles are pure operations.
Recent updates:
- On September 23, 2026 they and Astralis Foundation launched a joint Middle Powers and Transformative AI RFP, run by Astralis in collaboration with Macroscopic. It will allocate around $10m, mostly in grants of $100k to $2m over 6 to 24 months, for work that helps middle powers gain and use leverage over frontier AI development so that it is safer and more broadly beneficial. Deadline October 23, 2026, reviewed on a rolling basis, with a substantive response aimed at within 4 weeks and a final decision within 8 weeks of applying.
Get involved:
- Apply for funding: the Middle Powers RFPis open to individuals, informal teams, universities, fiscally-sponsored organizations and nonprofits, including nonprofit projects inside for-profits (other for-profits case by case), until October 23, 2026 — apply here. Otherwise you can email info@macroscopic.org, though most grants are sourced through proactive research and they don't respond to most proposals
- Donate: they're not seeking donations, but they're happy to advise those donating > $100k
- Apply for a job: no current open roles, but you can express interest here

▸Astralis Foundation

Website
Background:
- A UK not-for-profit with a Swedish sister foundation, Stiftelsen Astralis; the team works remotely across the UK, US, and Sweden. They pool money from “well over a dozen donors across several countries,” with no single donor dominating, and take no money from major AI companies or their employees, governments, or political parties active in AI policy. Their main vehicle is a pooled fund called Shared Horizons.
Thesis:
- They fund neglected work to make advanced AI secure and beneficial, focused on AI safety capacity and international cooperation in Asia, including China, and AI progress, safety and governance capacity in Europe. They are also scoping AI company oversight. Named grantees include the Institute for Law & AI's EU team, the Singapore AI Safety Hub, and the Simon Institute for Longterm Governance.
By the numbers:
- donations: an August 2026 job posting said they had mobilized more than $60m in 2026 and planned to reach around $100m by year end, after raising and deploying more than $30m to more than 15 grantees in their first 18 months; their own site publishes no totals
- staff: 12 listed on their team page, plus 6 board members across the UK and Swedish entities
Get involved:
- Apply for funding: they don't take unsolicited requests, but the Middle Powers and Transformative AI RFP, which they run with Macroscopic (details above), is open until October 23, 2026. You can also email ops@astralisfoundation.org about funding, partnering, or ideas.
- Donate: they take donations from qualifying donors under their gift acceptance policy; get in touch at the same address.
- Apply for a job: no open roles listed, but you can submit an expression of interest via their careers page.

▸Survival and Flourishing Fund (SFF)

Website
Background:
- Founded in 2019 and funded by Jaan Tallinn, who was joined by Dustin Moskovitz as a second Funder in the 2026 Main Round — the first SFF round run for more than one Funder.
Thesis:
- SFF is a virtual fund focused on organizing grant processes to support the long-term survival and flourishing of sentient life. Most grants go to reducing AI x-risk. Jaan Tallinn's priorities include efforts to restrict AI—datacenter certifications, speed limits, liability laws, labeling requirements, veto committees, and off-switches—as well as constructive efforts to set examples for the positive use of AI—AI assistance for human intelligence, AI healthcare tech, positive moralities, safety specs for AI, and hardware-level AI controls.
By the numbers:
- their homepage chart puts annual totals at $33m (2023), $41m (2024) and $35m (2025), rising to $64m in 2026, and says SFF “has organized ~$216MM in philanthropic gifts and grants” to date; in 2025 they recommended $34.33m across 88 grants, and the 2026 Main Round recommended $63.13m across 115 grants. Both the lifetime total and the 2026 figures have been revised down since October 1, 2026
- staff: 10 people listed at Survival and Flourishing Corp, plus a 2-person board (Andrew Critch and Eric Rogstad). Recommendations are done by part-time recommenders — twelve of them across three tracks in 2026, two anonymous.
Recent updates:
- The 2026 Main Round, split into Main, Freedom, and Fairness tracks, closed on April 22, and its recommendations were published on September 30, 2026 and have since been revised down. As of October 15, 2026 the page shows $63.13m recommended across 115 grants to 114 organizations, with $63.64m expected to be distributed in association with the round — roughly double their own $14-28m estimate, but about $1.4m below the figures the page carried two weeks earlier, and it warns that “Details of some recommendations may change later.” Jaan Tallinn funds $34.10m of it and Dustin Moskovitz $29.03m. The track totals inverted their expectations: the Freedom Track came in at $31.22m against a $2-4m estimate, making it the largest track, with the Main Track at $13.49m and the Fairness Track at $8.02m; a further $10.40m is attributed to a Synthetic Mean Recommender. In a September 9, 2026 update, SFC CEO Ethan Ashkie said the round's process “created confusion and uncertainty, backlash from some Evaluators, delays in grant decisions, and ultimately meant that some applications which SFF Evaluators recommended funding for were not facilitated by SFC.” SFC says it does not want to work with people who spread messages advocating for violence, or to recommend funding for work that would foreseeably inflame and coarsen public AI discourse.
- On October 3, 2026 SFF publicly withdrew its grant recommendation to John Wentworth, saying that after the September 30 announcement “public statements by Mr. Wentworth came to our attention which we found objectionable, disturbing, and contrary to the values of SFF,” and that the Funders would not fulfil the remainder of the recommendation; an earlier Speculation Grant had already been disbursed. He no longer appears in the 2026 recommendations.
- They have announced a $5-20m ISPA Inspection Prize — now the top item on their announcements list — for initiating “Independently Supervised Peer-Assisted” inspections between frontier AI labs, aimed at pre-deployment security risks like lab leaks and sandbox exfiltration. Jaan Tallinn will contribute at least $5m, plus another $5m at a 1:2 matching rate with other funders. See Get involved below.
- They added three themed rounds for 2026 with $2-4m each — climate change (closed June 10), animal welfare (closed June 24), and human self-enhancement and empowerment (closed July 8). Recommendations for these are expected in November 2026 and have not yet been published.
- Their programs list now includes an Advisory Organizations Panel, “a set of trusted outside organizations that SFC consults for non-binding input on proposed grants where the balance of potential positive and negative impacts is in live debate” — a response to the 2026 process problems described above. It was formed in 2026 by invitation and has nine participants, of whom four are named (the Center for AI Safety, CHAI, FAR AI, and the Foundation for American Innovation); they were asked to review 30-35 proposed grants and give a positive, negative, or abstaining opinion. SFC says the panel has no decision-making power and no formal role in the S-Process, and that earlier communications wrongly depicted it as a body that could approve or veto grants.
Get involved:
- Apply for funding: instructions for applying to a Speculation Grant, and the ISPA Inspection Prize is open with no deadline (there is no formal prize application, but an eligibility inquiry form for optional advance vetting, and SFF will consider grants to support prospective inspections)
  - Speculation Grants are faster grants made outside the S-Process timeline. Around 40 people serve as Speculators, each holding a budget of roughly $400-500k to approve them from.
  - Submitting an application also puts you in consideration for the next S-Process Grant round. In fact, being awarded a Speculation Grant is how you guarantee eligibility for a round, and SFF says over 95% of applications evaluated in past rounds received one.
- Donate: No page states a policy on cofunders either way, but they did add a second Funder in 2026; consider reaching out to sff-contact@googlegroups.com.
- Apply for a job: Survival and Flourishing Corp is hiring five roles — four engineering roles (full-stack, security, design, and QA/testing) at $250k-$350k, plus a Product Manager at $100k-$150k. All are remote, with 2-4 in-person team weeks a year in the Bay Area.

▸Lightcone Commons

Website
Background:
- Announced by Oliver Habryka in July 2026. Their tentative first-round funder list is Jaan Tallinn ($10m, conditional on $10m coming from other funders), Dustin Moskovitz (~$5m for the first round, $10m over the first year if it goes well), two anonymous donors ($3m+ and $2m+), the LTFF and the AI Risk Mitigation Fund (~$2m each), Timothy Telleen-Lawton (~$200k), and Andreas Stuhlmüller (~$100k).
Thesis:
- Lightcone Commons is aiming for near-feature-parity with the S-process used by SFF, with several parts redesigned to make the application and evaluation processes less time-intensive, and a round every three months.
By the numbers:
- first round: originally expected to be around $15-25m, but as of October 12, 2026 Lightcone Commons looks set to advise on around $50m
- fees: 5% on top of grants — 3% to Lightcone Commons and 2% to evaluators, lowered for funders moving large amounts
- staff: recommendations done by part-time evaluators, currently Zvi Mowshowitz, Yafah Edelman, Eliezer Yudkowsky, Nate Soares, Caleb Parikh, Elizabeth Van Nostrand, and Oliver Habryka, with Katja Grace confirmed for future rounds
Recent updates:
- Applications for the first round closed on August 23, 2026, and recommendations are expected around October 23, 2026; as of October 15, 2026 none had been published, and the site still describes the round in the future tense. Applications are still open on a rolling basis: anything submitted now rolls into the second round, which Nathan Young says will run before Christmas, earlier than the site's stated response date of around January 23, 2027. No deadline for that round has been published yet. Oliver Habryka said in September 2026 that he hopes to move to monthly rounds, probably in Q1 2027, cutting average decision time to 2-3 weeks.
- The Corrigibility Research Fund, a Lightcone Infrastructure program managed by Max Harms that will award at least $200k in grants and prizes for corrigibility research in 2026, is moving onto Lightcone Commons for its second round, with applications due October 31, 2026. Its first round was run over email and awarded $132k to eight of more than 100 applicants. Separately it has paid out retroactive prizes for existing work: $27k in July 2026 and a further $48k announced on September 30, 2026, for $75k so far, with more than $60k more hoped for in December 2026. So the fund has already passed its “at least $200k” floor for 2026. Note that October 31 is the CRF's own deadline, not a Lightcone Commons round deadline.
- In August 2026 the Long-Term Future Fund announced it was closing, and that it would spend down around $2.8m on existing applicants and longtermist projects, primarily through this first round. Lightcone's own funder list puts the LTFF contribution at around $2m; the LTFF's closure post says Lightcone agreed to waive its 3% platform and 2% evaluator fees on it.
Get involved:
- Apply for funding: here
  - They also import applications from grantmaking.ai, Manifund, and LTFF.
- Donate: They have no strict requirement, but ask that you assign some non-trivial probability to distributing at least $50k in the coming round; book an onboarding call here.

▸AI Safety Tactical Opportunities Fund (AISTOF)

Background:
- Founded by JueYan Zhang (former BlackRock PM) in 2023
Thesis:
- A multi-donor fund focused on moving fast to fill time-sensitive gaps.
By the numbers:
- funds raised: > $30m, per their own description
- grants made: > 150 since September 2023
- their Manifund account, which is only one of their channels, shows $1.75m across 17 grants in 2025 and $2.77m across 28 grants in 2026 through October 15; their most recent grant there was on October 4, 2026
- Manifund's Trace database estimates their overall giving at $15.0m in 2025 and $25.0m in 2026, most of it as a lump estimate rather than itemised grants ($47m all-time across 49 records, of which it itemises 9%)
- they are one of the named backers of the UK AI Security Institute's Alignment Project, a pooled fund of over £27m, alongside Schmidt Sciences, Halcyon Futures, Anthropic, OpenAI and others; no per-funder amount is published
- staff: ~1 FTE
Get involved:
- Apply for funding: there's no open application, but consider posting a proposal on Manifund. A site has appeared at aistof.org but as of October 15, 2026 it serves only a “Coming Soon” placeholder
- Donate: there's no formal way to do so, but you could reach out to JueYan on LinkedIn

▸Schmidt Sciences

Website
Background:
- Science-focused foundation funded by Eric and Wendy Schmidt.
Thesis:
- They fund several different areas of science, focused on academic research. Their broader AI portfolio is more focused on beneficial AI in a broad sense and on accelerating AI capabilities, but they also fund research on AI safety.
By the numbers:
- donations: $10m to AI safety in 2025
- staff: 11 listed under “AI & Advanced Computing” on their team page, mostly not focused on AI safety, estimate 2 in AIS. Michael Belinsky, who led the AI Safety Science program and directed the AI institute, left for the OpenAI Foundation in July 2026; Mark Greaves oversees the AI group while they search for a new leader.
- their AI safety work now runs under the “Science of Trustworthy AI” name; the old AI Safety Science page redirects there. That program now also lists an advisory board: Percy Liang, Yonadav Shavit, Ajeya Cotra, JueYan Zhang (who runs AISTOF, above), and Mark Greaves.
Recent updates:
- All three of their 2026 AI safety RFPs have now closed: the science of trustworthy AI (May 17), interpretability (May 26), and multi-agent safety (August 9). Decisions on the first two were due in summer 2026 and the third in autumn 2026; as of October 15, 2026 no 2026 awardees have been announced for any of them, so two of the three decision windows have now slipped. The year filter on their awardees directory still stops at 2025.
- In September 2026 they launched GlossoGen, an open-source platform for studying the languages AI agents evolve to communicate with each other, framed around keeping agent communication monitorable. No new funding was attached.
Get involved:
- Apply for funding:
  - Their joint RFP on multi-agent safety with Google DeepMind, ARIA, the Cooperative AI Foundation, and Google.org closed on August 9, 2026 per Schmidt's own portal, though Google DeepMind and the Cooperative AI Foundation both gave the deadline as August 8. The DeepMind announcement is the source for the headline figure of up to $10m across all the funders; awards are tiered at up to $300k (tier 1) and $300k-$1m (tier 2) over 1-2 years.
  - They have no AI safety RFP open right now, and otherwise don't accept unsolicited proposals. Their interpretability page no longer points to a future funding round; it now describes a red-team versus blue-team interpretability competition they are running with NDIF and David Bau at Northeastern University, with no application route.
- Apply for a job: They are hiring for scientists, program staff, and fellows in AI here.

▸BlueDot Impact

Website
Background:
- Founded in 2022 as an AI safety training organization. They started making grants at scale in 2026.
Thesis:
- They give fast grants to people and projects working in AI safety and biosecurity, aimed at new projects and individuals.
By the numbers:
- donated: $7.97m across 1,163 grants when checked on October 15, 2026, almost all of it in 2026; their newer Seed Grants program publishes no running total, so is not included
- staff: 17 listed, two of them “Talent Investors” but none with a grants title; their own figures disagree — their CEO said 15 in a September 12, 2026 post and their join-us page says “a team of 16, growing to 20” — and the CEO said they could grow 3-5x over the next year
- funders: Coefficient Giving made them a $25.6m three-year general support grant in 2025; their CEO said in September 2026 that they have raised $71m in total
Recent updates:
- In September 2026 Coefficient Giving handed over its career development and transition funding program to BlueDot, with a grant to support it; Coefficient now directs those applicants to BlueDot's two grant programs.
- In a September 12, 2026 post, their CEO said they have raised $71m, are 15 people and could grow 3-5x in a year, make grants ranging from $100 to $200,000, and are building a downtown San Francisco campus for more than 200 people.
Get involved:
- Apply for funding: they have two grant programs:
  - Career Transition Grants of up to $200k, generally starting at $20k, for people switching to work full-time on AI safety or biosecurity — application: 45 minutes, decision time: 20 days
  - Rapid Grants: up to $20k for AI safety or biosecurity work, now explicitly including events, community and travel as well as projects — application: about 15 minutes, decision time: 5 days on average
  - Seed Grants of $20k-$200k over 6-12 months, to get a new AI safety or biosecurity project or organization off the ground — application: about 45 minutes, decision time: about 4 weeks. They also offer introductions to follow-on funders including Coefficient Giving and Longview Philanthropy, and can fund for-profits in some cases. No grant count or total is published for this program. Their Missions page lists concrete projects they would like someone to take on with Seed Grant funding; the ten listed as of October 2026 are all biosecurity, with AI safety ones said to be coming, which is why it is not counted as an open RFP in the table above.
  - They also run Incubator Week, a 5-day expenses-paid program in San Francisco offering up to $200k in funding if they back your pitch, paid within two weeks of good progress. The next cohort runs November 16-20, 2026, with applications due November 11. Past cohorts total 50 participants and $10m+ in external funding raised.
  - In October 2026 they and Entrepreneurs First announced Fulcrum, an AI safety entrepreneurship program starting with a 7-day remote challenge, where people who finish “will be considered for grants from BlueDot” as well as Entrepreneurs First fellowships and investments. No amounts or deadlines have been published.

▸GTR AI Safety Fund

Website
Background:
- The grant program of the Berlin-based Global Technology Risk Foundation gGmbH, which German tech investor Jan Beckers (founder of BIT Capital) cofounded with his sister Lisa Beckers, its Managing Director, in 2023. On September 14, 2026 Beckers committed €50m over three to seven years and launched the fund. He is also quoted on Astralis's homepage.
Thesis:
- They fund people rather than set projects: individuals and early-stage organizations at any career stage, with no geographic focus, working on loss of control, cybersecurity, CBRN, power concentration, international agreements and verification, and AI's gradual effects on human development and social behavior. They offer strategic, operational and network support alongside money. Grantees since 2023 include Apollo Research, the KIRA Center, The AI Whistleblower Initiative, and Safe AI Germany.
By the numbers:
- donations: €50m over three to seven years
- grant sizes: not published; from initial research or compute to multi-year organization building
- staff: 4 listed — Jan Beckers, Lisa Beckers, an advisor, and a chief of staff
Get involved:
- Apply for funding: rolling applications, a 1–3 page proposal plus CV, with initial feedback targeted within 7 days. Nominations go to grants@gtr-foundation.org.
- Donate: they welcome “additional funding from like minded individuals.”

▸Manifund

Website
Background:
- Founded in 2023 by Austin Chen (formerly Manifold Markets).
Thesis:
- Manifund is an open platform where everything is public, grants can be turned around in days, and regrantors make independent calls.
By the numbers:
- in 2026 so far (through October 14):
  - $6.91m donated
  - ~162 projects funded
- grant sizes between $0-$500k
- staff: ~2 FTE
Get involved:
- Apply for funding: make a public project proposal at manifund.org. They also run the Falcon Fund, launched on April 24, 2026 and led by regrantor Marcus Abramovitch, which renewed its call for proposals on September 29, 2026 — $500k, grants primarily $25k-$150k, rolling with no deadline, decisions on a one-week timescale, and $140k given away across 5 grants as of that post — but it funds the intersection of animal welfare and transformative AI rather than AI safety as such, so it is not counted as an open RFP in the table above.
- Donate: you can donate to projects yourself, or donate to regrantors here

▸Transformative AI Fund (TAIF), formerly the Long-Term Future Fund (LTFF)

Website
Background:
- The LTFF started in 2017 as a project of Centre for Effective Altruism. In August 2026 EA Funds closed the LTFF and launched the Transformative AI Fund in its place, with Lowe Lundin as full-time Head of Fund. The LTFF's outgoing managers said the closure was partly about institutional friction around grants to individuals, for-profits, and policy-adjacent work; several of them plan to keep granting part-time through the AI Risk Mitigation Fund and Lightcone Commons.
Thesis:
- Early-stage grants to individuals, new organizations, and existing organizations with new projects. Primary focus is technical AI safety and AI governance, including post-AGI governance, plus field-building and forecasting.
By the numbers:
- the LTFF donated around $5-6m/year through 2024, but only $1.2m in 2025 and $0.3m in 2026 as recorded in the public grants database; EA Funds' own payout chart shows more for 2026, $1.52m across 18 grants
- number of grants: 100-200/year through 2024, but 20 in 2025 and 6 in 2026 in the public database, which runs through 2026 Q2
- grant size: typically $10k-$150k, rarely above $300k — now TAIF's stated policy, not just LTFF's historical pattern
- staff: 1 full-time (Lowe Lundin); advisors Caleb Parikh and Catherine Low are part-time
- stated response time: 6-8 weeks
Recent updates:
- No narrative payout report has been posted since the one covering May 2023 to March 2024, though the grants databaseis updated through 2026 Q2. TAIF says it will publish its first quarterly report before the end of 2026. As of October 1, 2026 no TAIF grants appear in that database and the fund page lists no payout reports, so over two months after launch none of its grantmaking is public yet. In fairness, the database holds no 2026 Q3 round for any EA Fund, and TAIF's first grants would fall in that quarter.
- The LTFF is spending down around $3.7m: roughly $2.8m to existing applicants, mostly through the first Lightcone Commons round, and around $0.9m as a seed grant to TAIF.
Get involved:
- Apply for funding: use this form; applications are rolling. The LTFF is no longer taking new applications, and live LTFF applications were imported into the first Lightcone Commons round.
- Donate: donate to TAIF here; TAIF is actively fundraising beyond its ~$1m seed grant
- Apply for a job: the Associate Program Officer search has closed. CEA's job board has no TAIF-specific role open; its grantmaking openings as of October 15, 2026 are Head of the EA Infrastructure Fund and a Grants Associate posted October 15, 2026, a grants-operations role spanning CEA's grantmaking rather than TAIF specifically. The Head of Grantmaking Operations posting has since been delisted.

▸Not included

NVIDIA: On October 13, 2026 they committed compute “valued at $250 million” over five years to UK agentic AI safety and security research, delivered through UK-based NVIDIA Cloud Partners, with a portion going to the Oxford-led BOLD, the UCL-led SOFAIR, and Imperial College London and the rest to the wider UK community.
Mercor AI Safety Fund: Announced September 15, 2026. “Mercor is committing $5 million to fund safety research. Award size and mix depend on the project,” covering researcher hours, API costs, conference stipends, and the time of experts from Mercor's platform. Named interest areas are misalignment, sandbox escape, evaluation awareness, interpretability, oversight and control, and red-teaming methodology. Applications are rolling with no deadline, but eligibility is narrow: applicants must be based in the US or UK, and awards go to the researcher's institution — universities, nonprofit research organizations, public benefit corporations, and small startups “that do not sell AI training data.” Mercor is a for-profit company, and the award terms carry commercial strings: grantees keep ownership but grant Mercor a non-exclusive licence to use the work commercially; for benchmark and evaluation projects Mercor gets early access plus “the exclusive right to build and host a private held-out test set”; and grantees agree not to work with specified Mercor competitors on substantially similar benchmarks during the award and for 12 months afterward. Grantees are expected to publish under an open licence permitting commercial use.
Halcyon Futures: An incubator-grantmaker for new organizations in AI safety, cybersecurity and biosecurity. Their site has been redesigned since our last check and the ceilings have gone up: career transition grants up to $500k (unchanged), nonprofit seed grants up to $3m (was $1m), and VC investments “up to $5M, and sometimes much more” (was $1m). The “apply on a rolling basis” language and the request for founders are no longer on the site; the only route given is email. They now publish three cumulative stats — 30 new orgs and companies launched, $500m in follow-on funding raised, and 200 career transitions supported. They also appear as a cofunder of the UK AI Security Institute's Alignment Project.
Navigation Fund: Jed McCaleb's foundation. At one point they announced they were giving $20m/year to AI safety, but this seems to have not materialized and it's now gone from their website.
AI Risk Mitigation Fund: They announced a spinoff from LTFF in 2023, but haven't made any updates or grant announcements on their website. They're one of the funders of Lightcone Commons, contributing around $2m to its first round. Their apply page says they have no set date for opening applications and currently fund only opportunities they source themselves.
FLI: Their 2025 grants page, published in July 2026, lists eight grants totalling $7.08m, led by $2.3m to the Center for AI Risk Management & Alignment and $1.95m to IASEAI, with $1m to HITRECORD and around $0.5m each to several others. Their PhD fellowship is now paused — they aren't accepting applications in fall 2026 while they reassess the program. Their one open program is the Digital Media Accelerator, which funds AI safety content creators on a rolling basis and doesn't publish grant sizes. A September 2026 update to that page says they are prioritizing content on the AI industry's “race-to-replace,” escalating harms from advanced systems, and loss-of-control risk, and warns that “review timelines for applications are longer than normal at the moment.” Their grantmaking page otherwise lists no open programs and says they do not accept unsolicited requests.
Frontier Model Forum AI Safety Fund: It was established in October 2023 as a “$10 million+” initiative. Most of this was distributed in 2024 and 2025 — their most recent grantees were announced in December 2025, and there have been no 2026 grantees or news — and they now appear to be winding down, saying their remaining funds “will be used to support narrowly-scoped research projects that target urgent bottlenecks.” The fund was initially administered by the Meridian Institute, and the Forum has managed it directly since Meridian announced in June 2025 that it was closing.
Foresight AI for Science & Safety Nodes RFP: Their standalone AI for Safety and Science program page now redirects here, so the Nodes RFP is the whole of their current AI offering. The RFP (deadline October 31, 2026, 23:59 PDT) funds work on local compute, coordination and accountability, and AI-first science, “typically $30,000 – $100,000 for this RFP but larger amounts are possible for the right project”, with overhead capped at 10% and review about three months after the deadline. All funded work must be open-sourced. They strongly prefer applicants who will be in-person members of their SF or Berlin hubs; they have also added a community xNode in Cape Town. Applicants can opt in to having their application shared with Lightcone Commons for additional funding consideration.
Astera: Their webpage mentions that they're primarily backing Simplex.
grantmaking.ai: New initiative housed under Manifund, with an initial $1m funded by Anton Makiievskyi, see launch post. That round closed in July 2026; per their results page, all $1m was distributed to 33 projects out of 581 applications, averaging $30,303, with 91 reviewer-endorsed and 143 screened out. They published a retrospective on August 14, 2026 saying they have “many ideas for the future” but that plans will take time to finalize. No second round has been announced.
AI Forge: A joint DARPA/NSF program launched on June 1, 2026, working with CAISI at NIST, to fund research on AI interpretability, AI control, and adversarial robustness. Per the program's RFI (SAM.gov notice DARPA-SN-26-80 — the figures are not on the DARPA program page), its “Project Ventures” are “university-led efforts ranging from $750K to $3M or higher” spanning up to one year, targeted at US universities including military service academies, with key personnel expected to be US citizens or permanent residents. That RFI closed on June 22, 2026 and drew nearly 130 responses; no follow-on solicitation or awardees have been announced. The forum was slated to launch in summer 2026 and its administering nonprofit still has not been publicly named as of October 2026.
ILIAD: They fund new research organizations that meet a high bar of scientific rigor. As of October 15, 2026 their funding page lists only a rolling open call for research projects — the earlier request for proposals on epistemics-raising projects and theory-driven research on ReLU networks is no longer posted — and they still don't publish grant sizes or deadlines. Support may also be non-financial (research advice, introductions, operational help).

▸Notes

Using the public grants lists on the fund pages under https://coefficientgiving.org/funds for the Navigating Transformative AI Fund ($324.3M in 2025) and the GCR Opportunities Fund ($85.0M), totalling $409.3M as of October 15, 2026. In the two weeks to October 1, 2026 the GCR Opportunities figure fell from $97.0M, so grants appear to get re-dated or reclassified as well as added. I don't know what the total of nonpublic AIS grants was; their 2025 letter from the CEO says they directed over $1B across all cause areas in 2025.
Filtering the grants list on each fund page under https://coefficientgiving.org/funds by year gives 189 grants in 2025 for the Navigating Transformative AI Fund and 43 for the GCR Opportunities Fund. These counts still grow as back-dated grants are published: over the two weeks to October 15, 2026 the GCR Opportunities 2026 count went from 12 grants ($45.1M) to 13 ($46.3M), while the Navigating Transformative AI 2026 count held at 75.
Per Luke Muehlhauser's post.
Estimate: as of June 2026 they said they were working to finalize more than $130M in grants through their AI resilience program, “to be shared publicly soon and with more to come,” and planning to invest more than $1B across several programs over the next year. The Foundation has still not published an AI resilience grantee list. Inside Philanthropy reported on September 29, 2026 that the Foundation had announced seven commitments totalling $815M across all programs and had “disbursed nearly $200 million” in grants, with over $750M in grant agreements approved; it notes the $1B one-year goal counts approved grant agreements rather than cash disbursed.
Per this post from a Longview team member, they directed over $60M in 2025, more than 2x their 2024 figure. Their August 2026 COO job posting said “We directed over $75 million to high-impact organizations in 2025, and expect 2026 will be substantially bigger,” but that figure covers all of Longview's giving rather than AI safety specifically. That posting has since been delisted and its URL now redirects to their careers page.
Longview's hiring materials said “In 2026, we aim to move $200 million — making us the second largest funder in that field.” That page has since been taken down; an archived copy survives. Re-checked across their sitemaps on October 15, 2026: no live Longview page states a 2026 target. Their now-delisted COO posting said only that they expect 2026 to be “substantially bigger” than 2025.
https://forum.effectivealtruism.org/topics/macroscopic-ventures
Macroscopic's grants page says “This year, we plan to give up to $100m in total to organizations and individuals in our focus areas.” Their job posting says they are deploying “up to $100+ million this year.”
An Astralis job posting dated August 2026, archived on a third-party job board, says: “In 2026 alone, we have already mobilized >$60M, and we plan to mobilize a total of ~$100M by the end of the year.” The same posting says they raised and deployed more than $30M to more than 15 grantees in their first 18 months, and that their goal for 2027 is to make $100M in grants. Their own site publishes no totals.
https://survivalandflourishing.fund/2025/recommendations
https://survivalandflourishing.fund/2025/recommendations
SFF's 2026 Main Round recommendations say “The total funding recommended in this round is $63.13MM, exceeding our $14MM–$28MM estimate... The total funding expected to be distributed in association with this round is $63.64MM.” Those figures were revised down from $64.58MM and $65.08MM after the round was announced on September 30, 2026; the page itself warns that “Details of some recommendations may change later.” That is the Main Round alone; the three themed rounds, estimated at $6–12MM in total, are still pending. Their application page had estimated “$20MM - $40MM in funding will collectively be distributed across all rounds and tracks.”
Nathan Young, posting on October 12, 2026: “Lightcone Commons looks set to advise on $50m this round. We are having a second round before Christmas.” This is money funders are expected to distribute on Lightcone Commons' recommendations, not money Lightcone holds.
Estimate from Manifund's Trace database, which as of October 15, 2026 puts AISTOF's 2025 giving at $15.0M across 14 recorded grants ($13.5M of it an explicit estimate rather than itemised grants) and its 2026 giving at $25.0M across 30 ($22.2M of it an estimate). Trace's year totals are fixed lump sums, so publishing another itemised grant moves money out of the estimate line and raises the record count without changing the year total. AISTOF publishes no figures of its own.
Private communications
https://www.schmidtsciences.org/new-10-million-ai-safety-science-program-launched-for-foundational-research/
The February 2025 announcement says they selected 27 projects, and a July 2025 release puts it at “more than $10 million in grants to 27 research projects.” Their Science of Trustworthy AI page also lists a second cohort of 21 inference-time-compute projects, for 48 in total, but publishes no award date or dollar total for those, so they are not counted here.
Estimate: totals for their 2026 AI safety RFPs were not published. The live page for the 2026 Science of Trustworthy AI RFP has since been taken down; an archived copy survives.
Treated as zero: their published list of rapid grants runs from April 2024 and shows only 17 grants totalling $9,516 in 2025, plus 10 grants totalling $2,496 in 2024, negligible at the scale of this table. They began granting at scale in 2026, and Career Transition Grants only launched in May 2026.
Their grant pages reported 1,077 rapid grants ($3,030,964) and 86 career transition grants ($4,939,775) when checked on October 15, 2026 — the pages publish running totals with no as-of date — of which about $7.97M is 2026 giving, extrapolated here to a full year. Rapid grants grew by 181 grants and $560k in the two weeks from October 1; the career transition figure did not move. Their newer Seed Grants program publishes no running total, so it is not in this figure. Note also that the browsable grantee list is incomplete: it held 772 entries totalling $1.74M against the 1,077-grant headline.
Estimate: Jan Beckers committed €50M over three to seven years in September 2026, or roughly €7–17M a year. They publish no annual budget, grant totals, or grant counts.
Around $5.9M donated in calendar 2025, per the data behind manifund.org/about (the page itself displays only all-time totals).
Distinct projects that received funding in 2025, per the data behind manifund.org/about.
Around $6.91M donated between January 1 and October 14, 2026, per the data behind manifund.org/about, extrapolated to a full year.
LTFF grants recorded for 2025 in the EA Funds grants database. That database is incomplete: it lists 693 LTFF grants totalling $30.2M all-time, while the fund's own successor announcement says LTFF made 820+ grants totalling just under $35M since 2017. The payout chart on EA Funds' own fund pages shows the higher figure — 823 grants totalling $35.5M as of October 15, 2026 — so roughly 130 grants and $5.3M are missing from the public database. That chart no longer renders on an LTFF or Transformative AI Fund page, but the Long-Term Future series is still carried on EA Funds' other fund pages. For 2025 specifically the two sources agree exactly, at 20 grants and $1,156,340.
The LTFF closed in August 2026 with around $3.7M left: roughly $2.8M for existing applicants, mostly through the first Lightcone Commons round, and around $0.9M seeding the Transformative AI Fund, which is fundraising for more.

Other resources

See another database of AIS funding at https://aisafety.com/funding and subscribe to their newsletter at https://aisafetyfunding.substack.com/. Manifund also launched Trace in August 2026, a database tracking $3.5b of AI safety funding across 4,900+ grants and 800+ funders as of October 15, 2026, with dated grants going back to 2007. It is available as a website, an API, an MCP endpoint, and a bulk download.