Greetings,

We are excited to share this joint funding opportunity from Schmidt Sciences, Google DeepMind, Google.org, the Cooperative AI Foundation, and the Advanced Research and Invention Agency (ARIA). The Scaling AI Safety for a Multi-Agent World request for proposals (RFP) aims to catalyze foundational technical research on the question of how to ensure safety in a world with millions of interacting AI agents, built and deployed by many different actors. More information about the award can be found in the synopsis and links below. For a full list of currently open opportunities that have been announced through our office, please visit our new Current Opportunities web page. As a reminder, our Research Development team is available to provide proposal development support to FAS ladder faculty.

Scaling AI Safety for a Multi-Agent World

OSP Deadline: July 31, 2026 at 9:00 AM ET
Departments or areas may require additional time for proposal review and submission. Please discuss a timeline with your departmental or FORA grants administrator.
Sponsor Deadline: August 8, 2026 by 11:59 PM Anywhere on Earth
Award Amount:
- Tier 1: Up to $300,000 over 1-2 years
- Tier 2: $300,000-$1,000,000 over 1-2 years
  For projects funded by Schmidt Sciences, indirect costs must not exceed 10%. This rate falls short of the 15% overhead required by FAS/SEAS policy. Please discuss this with your grants administrator before preparing an application.
- Google.org is one of the funding partners supporting this call. Principal Investigators (PIs) who are interested in applying to this RFP must discuss the potential impact of the Intellectual Property (IP) award terms with OTD before submitting an application. For more details, please review the Additional Information and Resources section below.

Program Overview

Schmidt Sciences, together with Google DeepMind, Google.org, the Cooperative AI Foundation, and the Advanced Research and Invention Agency (ARIA), has announced the Scaling AI Safety for a Multi-Agent World RFP. As AI agents become more capable, many agents, built and deployed by different actors, interact across shared digital environments. These interactions create new opportunities for cooperation and shared benefit, but also new system-level risks that may not be visible when evaluating individual models in isolation. This funding call aims to accelerate the foundational scientific research needed to to understand safety and risk through a system-level, multi-agent lens – developing methods to analyze emergent collective dynamics, building infrastructure for trustworthy interaction between agents, and creating scalable approaches for monitoring and control of increasingly complex networks of AI systems.

Applicants are invited to apply to either or both funding tiers: Tier 1 (up to $300,000) or Tier 2 ($300,000-$1,000,000). Project durations can range from one to two years. Tier 1 aims to support exploratory research projects, pilot studies, or focused technical investigations, whereas Tier 2 targets more ambitious or collaborative projects.

Priority Areas

Sandboxes and Testbeds for studying realistic multi-agent interactions
The Science of Agent Networks, including emergent collective capabilities, cascading failures, and dangerous population-level properties
Trustworthy Agent Infrastructure, including identity, verifiability, reputation, communication, and commitment
Multi-Agent Oversight and Control, including detection, attribution, security, and intervention methods for deployed agent populations
Applicants should refer to the RFP for a list of Out of Scope topics

Eligibility Criteria

The following recipients are eligible to apply:
- Universities or Research Institutions (Including International Institutions), including applications led by individual researchers or Project Teams
- Other Non-Profit Research Organizations or Agencies
- Multi-Institution Collaborations Between Universities and/or Other Non-Profits
- Universities and/or Other Non-Profits in Partnerships with Industry or For-Profit Organizations; however, for-profit entities may not lead projects.
Research teams based at institutions anywhere in the world are eligible to apply. Schmidt Sciences must evaluate each international direct funding applicant’s eligibility for charitable funding under US law.
Collaborations across institutions and geographic regions are welcome. Project teams may include researchers from multiple institutions and multiple countries, and there are no restrictions on the geographic distribution of collaborators.
- Normally, one institution should be designated as the lead host institution for the award, which will typically receive the funds and manage any subawards to partner institutions.

Selection Criteria

Research Agenda Fit.
Scientific Quality and Rigor.
Potential Impact.
Philanthropic fit.
Feasibility and Scope.
Team Expertise.
Cost Appropriateness.
Additional funder-specific considerations. Applications will be considered by the parties jointly supporting this funding call. If a proposal is selected for funding, the specific party or parties funding the project may provide additional limitations or guidance applicable to the award, which would be documented, if agreed upon, in the final award documentation.

Additional Information and Resources

Google award terms have presented challenges for Harvard University and many of our peer universities across the country. While we are unsure whether this specific program may be subject to these problematic Google terms, we wanted to inform you in advance about the potential for such terms, as they do not align with Harvard’s Openness in Research and non-discrimination Policies and jeopardize the Fundamental Research Exclusion, which is essential to the open and collaborative nature of our research environment. Should these clauses be included and Google refuses to negotiate, we may not be able to accept the award. Principal Investigators (PIs) who are interested in applying to this RFP must discuss the potential impact of the Intellectual Property (IP) award terms with OTD before submitting an application. If a proposal is submitted under these circumstances, Harvard may be unable to accept an award if the sponsor declines to negotiate problematic terms.

Applications are submitted in the Schmidt Sciences application portal. Notification of decisions will occur in the fall of 2026.

For more information, see FAQ or contact multiagentsafety@schmidtsciences.org.

Questions from FAS or SEAS faculty about this opportunity may be directed to research_development@fas.harvard.edu.

MAILING LISTS

Sign up to join the mailing list for FAS Funding Focus, a resource series tailored to help FAS faculty and staff navigate the ever-evolving landscape of grants and funding. Previous editions of FAS Funding Focus (FFF) are accessible in the FFF archive. To receive additional announcements from FORA, including the monthly On Your RADAR (Research Administration Developments & Announcements Rundown) newsletter and invitations to the FORA Forum, a monthly meeting for Harvard research administrators, please sign up for the FORA listserv here.

RESOURCES

Research Development maintains a Sample Proposal Library of successful proposals from recent award recipients that are available to FAS and SEAS faculty by request; please email Research Development to request to view a proposal. For more information on Research Development's resources—including lists of internal funding programs, early career funding programs, and sabbatical fellowships—please visit the Funding Opportunities web page.

SEARCH FOR OPPORTUNITIES WITH PIVOT

All Harvard affiliates can search for opportunities and set up alerts using Pivot, a searchable database of federal and private funding opportunities. Information on Pivot, including a one-page user guide, and other funding opportunity databases available to the Harvard community can be found on the Funding Databases web page.

PIVOT CURATED SEARCHES

Curated funding opportunity lists are funding search topics of broad interest. The following curated lists can be further refined in Pivot based on your funding needs. To view the results you must be on a Harvard network or logged in to your Pivot account. Information on setting up a Pivot account can be found in our one-page user guide.

Data Science, Computer Science, and Artificial Intelligence

Engineering and Technology

Quantum Science and Engineering

Additional curated lists are available on the Funding Databases web page. We continue to add to and refine these lists based on feedback. Please feel free to reach out to us at research_development@fas.harvard.edu with any suggestions you may have.