1,000+ Opportunities
Find the right grant
Search federal, foundation, and corporate grants with AI — or browse by agency, topic, and state.
The AI Safety Fund is the grantmaking arm of the Frontier Model Forum, and its notable feature is who funds it: Anthropic, Google, Microsoft and OpenAI as founding frontier-lab members, alongside the Patrick J. McGovern Foundation, the David and Lucile Packard Foundation, Schmidt Sciences and Jaan Tallinn.
That composition is the fund's defining characteristic in both directions - it gives grantees proximity to frontier model developers and their threat models, and it means researchers whose work requires independence from those developers should think carefully about the positioning.
The fund is capitalised at over 10,000,000 dollars and has run at least two completed grant rounds; a recent cohort of 11 grantees shared more than 5,000,000 dollars, selected from over 100 competitive proposals, implying an acceptance rate around 10 percent. Scope is deliberately narrow and applied.
The fund states it supports narrowly-scoped research projects that target urgent bottlenecks, organised into workstreams covering AI-Bio, AI-Cyber, AI-Nuclear and AI Security, with prior RFPs also issued in AI Agent Evaluation and Synthetic Content. This is not general alignment theory funding - proposals that do not map onto a specific, near-term evaluation or mitigation bottleneck in one of those domains are unlikely to fit.
The practical difficulty for applicants is that the fund does not publish per-grant amounts, eligibility criteria, or a standing open call with dates; RFPs are issued periodically by topic. Anyone interested should contact info@frontiermodelforum. org or monitor aisfund.
org and the Frontier Model Forum updates page rather than waiting for a general solicitation, since the topic-specific RFP windows are the only entry point and they are not always widely advertised.
Get a weekly digest of new grants like this
A free weekly digest of new foundation and federal funding opportunities as they're added to Granted. Unsubscribe anytime.
Or search similar grants →According to the current listing, eligibility includes: Detailed eligibility criteria are not published on the fund's public page. The fund supports narrowly-scoped research projects targeting urgent bottlenecks in AI safety and security, with workstreams covering AI-Bio, AI-Cyber, AI-Nuclear and AI Security; prior requests for proposals were issued in Biosecurity and Cybersecurity, and in AI Agent Evaluation and Synthetic Content. Grants are issued through periodic topic-specific RFPs rather than a standing open call, so there is no permanent deadline; applicants should monitor aisfund.org and the Frontier Model Forum updates page for new RFP windows. The fund is capitalised at more than 10,000,000 US dollars; a recent round awarded more than 5,000,000 dollars across 11 grantees selected from over 100 proposals. Per-grant minimum and maximum amounts are not disclosed. The fund was initially managed and overseen by the Meridian Institute. Funders include Anthropic, Google, Microsoft, OpenAI, the Patrick J. McGovern Foundation, the David and Lucile Packard Foundation, Schmidt Sciences and Jaan Tallinn - researchers whose work depends on demonstrable independence from frontier model developers should weigh that before applying. Enquiries go to info@frontiermodelforum.org. Because award amounts, eligibility and application procedures are not published, contact the fund directly to confirm current opportunities before planning a submission. Confirm the full requirements in the official notice before applying.
The current listing shows the fund is capitalised at more than 10,000,000 US dollars. Per-grant amounts are not published. The one hard data point available is that a recent cohort of 11 grantees shared more than 5,000,000 dollars, implying an average of roughly 450,000 dollars per grant. amount_min of 100,000 and amount_max of 900,000 US dollars are INFERRED from that cohort average and the typical spread of narrowly-scoped research grants around such an average; they are not AI Safety Fund figures and should not be used for budgeting. The fund describes what it supports as narrowly-scoped research projects targeting urgent bottlenecks, which suggests awards in the hundreds of thousands rather than multi-million range, but the authoritative per-grant amounts, eligibility rules and application process are only available from the fund directly at aisfund.org or via info@frontiermodelforum.org. Verify award ceilings, matching requirements, and allowable costs in the official notice.
Frontier Model Forum AI Safety Fund Research Grants for AI-Bio, AI-Cyber, AI-Nuclear, AI Security, Agent Evaluation and Synthetic Content is funded by AI Safety Fund (AISF), convened by the Frontier Model Forum with funding from Anthropic, Google, Microsoft, OpenAI, the Patrick J. McGovern Foundation, the David and Lucile Packard Foundation, Schmidt Sciences and Jaan Tallinn. Verify program details on the funder's official page before applying.
Start from the official opportunity page linked in this listing — it carries the sponsor's submission instructions.
The Sony Research Award Program is Sony's principal channel for funding external academic research, and the Focused Research Award is its collaborative track: up to 150,000 dollars for focused joint research between a university or research institute and Sony. The 2026 research areas are unusually broad for a corporate program and cover most of the modern AI stack alongside Sony's hardware interests - AI and large language models, computer vision, machine learning, robotics, human-computer interaction, affective computing, speech and language technologies, audio technologies, RF sensing, wireless communications, cybersecurity, sports technology, digital humans, generative AI, content creation and neural rendering, plus device-level areas including MicroLED and optical metasurfaces. For AI and robotics groups this makes it one of the wider corporate calls available, though the breadth is deceptive: Sony funds work that connects to its own research agenda, and proposals are strongest when there is an identifiable Sony research counterpart for the collaboration. Eligibility is tightly drawn around the principal investigator rather than the institution. The PI must be a full-time faculty member or researcher at a recognized institution - Assistant Professor, Associate Professor, Professor or equivalent researcher - and must be able to supervise PhD students. Co-PIs are permitted but must be from the same institution and meet the same requirements, which rules out the cross-institutional consortia common in public funding. The 2026 deadline is 15 September 2026 at 11:59pm Pacific, with a separate India-specific time given as 16 September 2026. Submission guidelines are on the Sony Research Award Program site.
The Meta Research PhD Fellowship supports doctoral students conducting research in areas central to Meta's technical agenda, with several of its annual award tracks dedicated to artificial intelligence and machine learning. Current AI-relevant tracks include AI System Hardware/Software Co-Design (high-performance AI algorithms spanning model compression, numerical optimization, benchmarking, and distributed inference and training), Applied Statistics (bias and variance estimation and correction in models and datasets, uncertainty quantification), AR/VR Human Understanding (efficient ML techniques that run on AR/VR devices), and Programming Languages (program synthesis, probabilistic and differentiable programming). Recipients receive two years of paid tuition and fees, a $42,000 annual stipend covering living expenses and conference travel, and a paid visit to Meta headquarters for the annual Fellowship Summit; the award carries no intellectual property claim on the student's research. The 2027 cycle opened August 3, 2026 with applications closing September 20, 2026, reference letters due in October 2026, and winners notified in January 2027.
OpenAI's AI and Teen Development Research Grant Program makes up to 5,000,000 US dollars available in grants of up to 1,000,000 dollars for independent research on how AI affects adolescent development and wellbeing. Applications opened 8 September 2026 at 8:00 AM PDT and close 6 October 2026 at 11:59 PM PST, with notifications by 13 November 2026, and submission is through the SurveyMonkey Apply platform. The published scope has four strands: teen usage patterns and developmental outcomes; AI's potential effects on young people's lives; the factors that cause those effects to differ across populations and contexts; and design interventions and safety mechanisms. The inclusion of the fourth strand is what separates this from a pure effects-measurement programme, since it invites work that proposes and tests product-level safeguards rather than only documenting harm or benefit. The third strand is equally notable: OpenAI is asking specifically about heterogeneity of effect, which is an implicit acknowledgement that aggregate findings about teen AI use have limited policy value. Review is rolling and conducted by internal researchers and outside experts. Non-profit applicants are preferred and for-profit organisations are explicitly not prioritised, and indirect costs are capped at 10 percent of the grant, both of which shape who can realistically compete.
Schmidt Sciences opened a 'Scaling AI Safety for a Multi-Agent World' program with awards up to $1 million and an August 8, 2026 deadline. Against a backdrop of federal research slowdowns, here is why private science philanthropy matters more than ever, how these funders differ from federal agencies, and how researchers should approach them.
Read articleSchmidt Sciences' 2026 Science of Trustworthy AI RFP closes May 17 with two funding tiers — up to $1M (Tier 1) and $1–5M+ (Tier 2) over 1–3 years, with a 10% indirect cost cap. The three research aims target misalignment under distribution shift, predictive-validity evaluations, and oversight of superhuman systems. Here is why the structure favors team-based proposals.
Read articleDPA26BZ06-DV026 offers $300,000 at Phase I or $1,800,000 as a single Direct-to-Phase-II tranche with no options. The deliverable is a simulated auction market that measures whether AI agents deceive, collude, or manipulate the humans they serve — measured entirely from the outside. Here is the 90% efficiency gate, the team composition most bidders will get wrong, and why this topic sits in DARPA's biology office.
Read article