Foresight Institute AI for Science & Safety Nodes Grants 2026: Up to $100,000 for AI Safety Research
Overview
The Foresight Institute is accepting proposals for its AI for Science and Safety Nodes Request for Proposals 2026, which funds open-source research on AI safety, coordination, governance and human empowerment. Projects must produce open-source outputs and show potential to reduce existential risks from increasingly capable AI systems within the next one to three years. The 2026 call focuses on coordination and accountability.
What you get
Grants range from US$30,000 to US$100,000. Smaller grants typically go to Human Empowerment and AI Insurance and Open Governance projects, while larger awards are generally for broader coordination and decentralised alignment work. Both short-term and multi-year projects can be funded. Larger projects may receive milestone-based tranches, but most grants are paid as a single lump sum. Overhead of up to 10 percent of direct research costs is permitted when it supports the project.
Who can apply
Applications are open globally to individual researchers, research teams, nonprofits, for-profit organisations, independent technical experts and interdisciplinary collaborations. For-profit applicants should explain why grant funding is necessary. Foresight prioritises applicants who want to take part in the research community through in-person collaboration, research sprints or regular engagement.
Requirements
There are three priority themes. Supercollaboration and Decentralised Alignment covers coordination protocols, distributed oversight, verification and open coordination platforms for multi-agent AI systems. Human Empowerment covers when humans should keep decision authority versus delegate to AI. AI Insurance and Open Governance covers risk assessment for insurers, incident reporting, catastrophe bonds, liability mechanisms and developer mutuals.
Proposals are assessed on fit with a focus area, potential to reduce existential risk, feasibility within short AI timelines, execution capability, high-risk high-reward potential, efficient use of resources and compute, and commitment to fully open-source code, data and outputs. Successful applicants complete due diligence including an itemised budget, project plan and organisational documents, and submit brief progress reports.
How to apply
Complete the official online application form, naming the focus area, explaining the expected impact on AI safety, setting out milestones and deliverables, and showing why the team can execute.
Key dates
The deadline is 31 October 2026 at 23:59 PDT. Decisions are typically made about three months afterwards.
Explore more: Technology & AI opportunities