TL;DR: Schmidt Sciences is a philanthropic research-funding organization founded in 2024 by Eric and Wendy Schmidt — it is not the same thing as the Schmidt Science Fellowship, which is one specific postdoctoral career-development program that Schmidt Sciences funds. Schmidt Sciences organizes its giving around five research areas, one of which — AI & Advanced Computing — funds a distinct portfolio of AI safety and "trustworthy AI" grant programs, including the AI2050 Fellows program and periodic competitive research-grant RFPs such as the Science of Trustworthy AI initiative. This guide explains how the organization is structured, what its AI-safety funding programs actually are, and what a research administrator needs to know before advising a faculty applicant.
What Schmidt Sciences is, and how it differs from Schmidt Futures and the Schmidt Science Fellowship
Schmidt Sciences was founded in 2024 by Eric and Wendy Schmidt as a nonprofit philanthropic organization dedicated to advancing scientific discovery. It describes its mission as working "toward a healthy, resilient, secure world for all" and organizes its funding across five research areas: AI & Advanced Computing, Astrophysics & Space, Biosciences, Climate, and Science Systems. It is one of several philanthropic organizations and initiatives created and funded by Eric and Wendy Schmidt, alongside entities such as Schmidt Family Foundation and Schmidt Ocean Institute.
The naming is genuinely easy to confuse, and research administrators fielding a faculty inquiry should be precise about which entity is actually involved:
- Schmidt Futures was the Schmidts’ earlier philanthropic venture, active from roughly 2017 onward, that originally launched programs including the Schmidt Science Fellowship (in partnership with the Rhodes Trust, 2018) and the AI2050 initiative (2022).
- Schmidt Sciences is the 2024 successor organization — effectively a spin-off/restructuring of the Schmidts’ scientific philanthropy into a dedicated science-funding entity — that now runs both of those legacy programs and has since launched additional AI-safety-specific grant competitions of its own.
- The Schmidt Science Fellowship is a single program within that portfolio: an individual, internationally-open postdoctoral fellowship built around a mandatory cross-disciplinary field pivot (see the dedicated guide linked above for eligibility, stipend, and nomination details). It is a career-development award for an individual researcher, not a research-project grant to an institution or a team.
The programs covered in the rest of this guide — AI2050 and the Science of Trustworthy AI RFP line — are structurally different from the Fellowship: they fund research projects and named investigators (sometimes multi-institution teams) working on AI safety and AI governance questions, rather than an individual’s discipline-switching postdoctoral placement. A research office should not assume that eligibility rules, nomination gates, or award mechanics for the Fellowship carry over to these programs, or vice versa — they are administered as separate competitions with separate terms.
The AI & Advanced Computing research area
Schmidt Sciences groups its AI-related giving under an AI & Advanced Computing research area, which the organization describes as supporting AI research intended to accelerate scientific breakthroughs while also addressing the risks and governance questions AI raises. Within that area, the organization’s AI-safety-specific work sits under what it calls its AI Safety Science program — funding aimed specifically at the technical foundations of AI trustworthiness, alignment, and oversight, distinct from the organization’s other AI grantmaking aimed at using AI as a scientific tool (for example, AI applied to biosciences or climate research).
AI2050: the flagship AI-safety fellows program
AI2050 is Schmidt Sciences’ longest-running AI-safety initiative, conceived and co-chaired by Eric Schmidt and James Manyika. It originated in 2022 under Schmidt Futures, publishing an initial working list of "Hard Problems in AI" — technical capability questions, responsible-deployment questions, and questions about harnessing AI for societal benefit — to guide the program’s investments, and has continued under the Schmidt Sciences banner since the 2024 reorganization.
Structurally, AI2050 runs as a fellows program rather than a general open RFP:
- It funds two fellow tracks: Senior Fellows (established researchers) and Early Career Fellows (postdoctoral and pre-tenure researchers), each working on one or more of the program’s identified hard problems.
- Fellows are selected in periodic cohorts. Reporting on the most recent publicly announced cohort (November 2025) put the group at 28 new fellows sharing more than $18 million in funding, bringing the program’s cumulative total to 99 fellows across roughly 42 institutions and eight countries.
- Unlike the Schmidt Science Fellowship, AI2050 fellows are working researchers proposing a specific AI-safety or AI-governance research agenda within their existing field, not early-career researchers required to pivot into a new discipline.
Because cohort size, funding totals, and the current list of hard problems change from cycle to cycle, an institution advising a prospective nominee or applicant should confirm current program status directly on Schmidt Sciences’ own AI2050 site rather than rely on figures from a prior cohort.
The Science of Trustworthy AI RFP and similar competitive grant lines
Separately from the standing AI2050 fellows program, Schmidt Sciences periodically opens competitive, proposal-based research-grant RFPs targeting specific AI-safety research questions. The 2026 cycle of its Science of Trustworthy AI RFP is a useful illustration of how these are structured (verify current terms against the live opportunity page before advising an applicant, since RFP cycles open and close on their own schedule and this specific cycle’s May 2026 deadline has already passed as of this writing):
- Two funding tiers: a Tier 1 track offering awards up to $1 million over one to three years for focused investigations, and a Tier 2 track offering $1 million to $5 million or more over a comparable period for larger, multi-year, interdisciplinary collaborations.
- Three research aims: (1) characterizing why and how frontier AI systems develop misaligned objectives that fail under distribution shift; (2) developing evaluations with genuine predictive validity and interventions that shape how AI systems learn; and (3) extending meaningful human oversight to systems with superhuman capabilities, including risks specific to multi-agent AI interactions.
- Eligibility: open to universities, research institutions, and nonprofit organizations worldwide, with multi-institution and international teams explicitly permitted, and multiple proposals per applicant allowed.
- Indirect cost cap: indirect (facilities-and-administrative) costs are capped at 10% of total direct costs — substantially lower than most US federally negotiated F&A rates, and a figure a sponsored-programs office should flag early in proposal budgeting rather than discover at award stage.
Schmidt Sciences has also signaled newer, related competitions in the same research area (for example, an RFP framed around scaling AI safety research for multi-agent AI systems). Because the specific scope, deadlines, and award terms of any given RFP change between cycles and this guide cannot track them in real time, treat any single cycle’s details as illustrative of the program’s general shape rather than as a standing, always-open opportunity, and confirm current live RFPs directly at schmidtsciences.org.
What this means for a research administration office
Schmidt Sciences funding sits outside federal award frameworks entirely, which has concrete administrative consequences distinct from handling an NIH or NSF award:
- Not Uniform Guidance (2 CFR 200) funding. Because Schmidt Sciences is a private philanthropic funder, its awards are not subject to the federal cost-principle and administrative-requirements framework that governs federal grants — the funder sets its own terms, including the aggressively capped indirect-cost rate noted above, and an institution’s federally negotiated F&A rate does not automatically apply.
- Non-federal award handling for effort, IP, and outside-activity policy. As with any privately funded award, the receiving institution needs to apply its own institutional policy for effort commitment, outside-activity disclosure, and intellectual-property terms, since a private funder’s award agreement may set different default terms than a federal award’s.
- Multi-institution and international team eligibility on the RFP-style programs means a sponsored-programs office may need to coordinate a subaward or collaboration agreement with partner institutions in other countries — a different administrative workflow than a purely domestic federal award.
- Distinct nomination/application paths. AI2050 operates as an invitation/nomination-adjacent fellows program with its own selection process, while the RFP-style competitions accept direct proposals from eligible institutions. Confirm which pathway applies before building an internal routing timeline.
Frequently asked questions
Is Schmidt Sciences the same organization as Schmidt Futures?
No, not exactly. Schmidt Futures was the Schmidts’ original philanthropic venture, active from roughly 2017. Schmidt Sciences, founded in 2024, is the successor organization that now carries forward Schmidt Futures’ science-funding programs, including AI2050 and the Schmidt Science Fellowship, under a dedicated science-philanthropy structure.
Is Schmidt Sciences the same thing as the Schmidt Science Fellowship?
No. The Schmidt Science Fellowship is one specific postdoctoral career-development program that Schmidt Sciences funds (in partnership with the Rhodes Trust). Schmidt Sciences is the broader parent philanthropic organization, which also funds separate AI-safety research-grant programs (AI2050, the Science of Trustworthy AI RFP line) that have entirely different eligibility rules, funding structures, and application processes from the Fellowship. See the dedicated Schmidt Science Fellowship guide for that program specifically.
What is AI2050?
AI2050 is Schmidt Sciences’ flagship AI-safety fellows program, originally launched in 2022 under Schmidt Futures. It funds Senior Fellows and Early Career Fellows working on a defined set of “Hard Problems in AI” spanning technical AI capability, responsible deployment, and AI’s societal impact.
How much funding does the Science of Trustworthy AI RFP offer?
The 2026 cycle offered a Tier 1 track of up to $1 million and a Tier 2 track of $1 million to $5 million or more, both over one to three years, with indirect costs capped at 10% of direct costs. Terms are set fresh for each RFP cycle, so confirm current figures directly with Schmidt Sciences before budgeting a proposal.
Who is eligible for Schmidt Sciences’ AI-safety grant programs?
Eligibility varies by program. The RFP-style competitions have generally been open to universities, research institutions, and nonprofit organizations worldwide, including multi-institution and international teams. AI2050 selects individual fellows (senior and early-career researchers) rather than accepting open institutional proposals. Always check the specific program’s current eligibility criteria before advising an applicant.
Related CASRAI resources
For the individual postdoctoral fellowship within this same funder’s portfolio, see the Schmidt Science Fellowship guide. For the underlying concept these programs fund research toward, see the dictionary entry on Trustworthy AI. For comparable philanthropic funder profiles already covered in this cluster, see the Arnold and Mabel Beckman Foundation Grants guide. For the broader landscape of grants-management topics, start at the Grants Management pillar page.
