Request for Proposals: Coordination and Accountability

Deadline: 31-Oct-2026

The Foresight Institute is inviting proposals for projects that advance coordination, accountability, and human empowerment in the development and governance of increasingly capable artificial intelligence systems. The AI for Science & Safety Nodes RFP: Coordination and Accountability supports research and development focused on multi-agent cooperation, decentralized alignment, human oversight, AI insurance, incident reporting, and open governance. Grants typically range from $30,000 to $100,000, with higher funding generally available for projects focused on supercollaboration and decentralized alignment and smaller awards for Human Empowerment and AI Insurance & Open Governance. Individuals, teams, nonprofits, and for-profit organisations can apply, while funded work must be open-sourced. The programme supports both short-term and longer-term projects and prioritizes work with potential to reduce existential risks from AI. The AI for Science & Safety Nodes RFP: Coordination and Accountability is part of the Foresight Institute’s broader effort to address the opportunities and risks associated with increasingly capable AI systems. The call focuses on research and development that can help: The RFP focuses on three major areas. This area supports research into systems that enable multiple AI agents and humans to cooperate safely and effectively.

Potential areas include:

The goal is to develop approaches that can support robust coordination without relying entirely on centralized control. This area focuses on determining when humans should retain control and when delegating tasks to AI systems may be beneficial.

Research may explore:

The focus is on strengthening human agency while enabling useful AI delegation where appropriate. This area supports mechanisms that can improve independent assessment, accountability, and incentives for reducing AI-related risks.

Potential work may include:

Typical grants range from $30,000 to $100,000. Funding levels may vary depending on the research area and project scope. Projects focused on supercollaboration and decentralized alignment may receive higher grant amounts within the typical funding range.

Projects focused on:

may receive smaller grant amounts depending on their scope and requirements. The programme supports both short-term and longer-term research projects. Grants are generally paid as a lump sum. For larger projects, funding may instead be provided in tranches linked to milestone completion. This structure allows larger research efforts to receive continued funding as agreed project milestones are achieved. Overhead costs of up to 10% of direct research costs are eligible when they directly support the funded work. Applicants should ensure that any requested overhead is clearly connected to the proposed research activities. All funded work must be open-sourced. Applicants should therefore consider from the beginning how their research outputs, tools, protocols, or other project results can be made available openly in accordance with the programme’s requirements. The RFP is open to a broad range of applicants.

Eligible applicants may include:

Applicants should have the capability to execute the proposed project and demonstrate a strong connection between their expertise and the RFP’s research areas. Proposals are evaluated against several factors.

Key considerations include:

Active participation in the Nodes research ecosystem is strongly prioritized. Applicants should clearly identify the problem they intend to address and explain how their proposed work fits within one or more of the three RFP areas.

A strong proposal should explain:

Applicants should consider the following programme requirements: Selected grantees must pass the programme’s due diligence checks before receiving funding. Funded projects are also subject to basic reporting requirements.

These include:

Applicants should therefore have appropriate systems in place to track project progress and expenditure. As AI systems become increasingly capable, coordination between humans and AI agents, accountability mechanisms, and meaningful human oversight can become important areas of research. The RFP supports technical and governance-oriented approaches that explore how humans and AI systems can coordinate, how responsibility and monitoring can be strengthened, and how independent mechanisms can contribute to reducing AI-related risks. By requiring funded work to be open-sourced, the programme also aims to contribute research and tools to the broader ecosystem rather than limiting results to a single organisation. It is a Foresight Institute funding call supporting research and development focused on AI coordination, accountability, human empowerment, decentralized alignment, AI insurance, and open governance. Projects addressing supercollaboration and decentralized alignment may receive higher funding within the typical range. The three areas are Supercollaboration and Decentralized Alignment, Human Empowerment, and AI Insurance & Open Governance. Yes. Individuals, teams, nonprofit organisations, and for-profit entities may apply. Yes. All funded work must be open-sourced. Yes. The programme can support both short-term and longer-term projects. Overhead costs of up to 10% of direct research costs may be eligible if they directly support the funded work. Grants are usually paid as a lump sum. Larger projects may receive funding in tranches tied to milestone completion. The Human Empowerment area focuses on mechanisms and frameworks that help determine when human control is beneficial and when delegating tasks to AI systems may be preferable. This area supports work involving independent technical assessment, incident reporting, and incentive mechanisms designed to reduce AI-related risks outside traditional laboratory or government oversight. Proposals are evaluated for alignment with the RFP, potential impact on reducing existential AI risks, feasibility wit