Navigators Incubator

Navigators Incubator — Cohort 1: Securing Transformative AI

Runs a selective AI security fellowship for small teams tackling critical risk problems.

Eligibility · Global — open worldwide

The Navigators Incubator, run by Coefficient Giving under the Navigating Transformative AI Fund, is a selective fellowship program for individuals and small teams working on AI security. Cohort 1 focuses specifically on Securing Transformative AI, drawing from Coefficient Giving's AI Infrastructure Security Shortlist of priority problems. The six target problem areas are: securing AI models from exfiltration of model weights; securing AI integrity against data poisoning and tampering; disconnected model safety (ensuring models remain safe when operated offline or in isolated environments); secure compute verification; protocols for using compromised AI systems (AI control); and detecting and responding to rogue AI deployments. The incubator's explicit long-term goal is for participants to develop ambitious grant proposals of $10 million or more and launch scaling projects in these areas.

The fellowship provides a salary matched to a generous benchmark or transition cap, full coverage of coding tools, compute, and research expenses, relocation support for San Francisco, London, and Washington D.C., coworking access (including Constellation in the Bay Area), and a dedicated research and conference travel budget. Program duration ranges from six months to two years. Individual applicants and teams of up to three people may apply; part-time arrangements are possible for exceptional candidates. Institutional applicants — organizations, universities, and research organizations as entities — are not eligible; this program is individual and team applicants only. No visa sponsorship is available, though remote work is supported.

Applications are submitted via an Ashby-hosted form linked from the Navigators Incubator page at coefficientgiving.org. Coefficient Giving launched this incubator in 2026 as a new mechanism to rapidly develop specialized talent in AI security areas it considers critically underfunded. Strong applicants demonstrate deep technical expertise relevant to one or more of the six problem areas and a credible plan for producing either research outputs or practical tools within the fellowship period. The program reflects Coefficient Giving's role as an early and ongoing backer of AI safety infrastructure since 2015.