AI news · October 16, 2026
Simons Foundation starts a $15M programme for mathematics and AI
Simons, Anthropic and OpenAI committed $15M for mathematicians to evaluate AI results and build shared methods, including $5M from OpenAI and access to models and compute.
- combined resources
- $15M
- OpenAI contribution
- $5M
The Simons Foundation announced on October 15 a $15 million programme with Anthropic and OpenAI for mathematics and AI. The foundation will manage an open call and support workshops, shared standards, evaluation of AI-assisted results and access to new tools. OpenAI separately said it will contribute $5 million, models and computing resources. The programme builds on the foundation's earlier support for Lean and other mathematical work.
This is more than a grant headline: it puts independent checking and shared methods beside model access. The money does not show that AI-generated mathematics is correct, and the final award process is not described in full. For builders, the useful pattern is to publish the proof or result, its machine-checkable form, the compute used and the human review path. That makes a research tool easier to trust and easier to challenge.
What you can do with it
Follow the Simons open call and OpenAI's public mathematics repository. For a proof or research assistant, store the prompt, model version, compute estimate, citations and Lean formalisation where possible. Test one result with an independent reviewer before calling it verified.
Our take
The programme is a positive distribution signal for tools that help mathematicians check AI work, not proof that model-made mathematics is ready. The best opportunity is in verification, records and shared review. Funding and access will matter only when they produce inspectable results.
Links ProgrammeOpenAI update
Source: Simons Foundation — Made With Models writes the brief; the reporting is theirs.