Simons Foundation, Anthropic and OpenAI Put $15M Toward Helping Mathematicians Adapt to AI

Simons Foundation commits $15 million to help mathematicians adapt to AI

The Simons Foundation has launched a set of initiatives designed to help the mathematics community navigate the rapid rise of artificial intelligence in research. Under the arrangement, the foundation is partnering with Anthropic and OpenAI, with the three organizations committing a combined $15 million in funding and resources. The Simons Foundation will manage the program and award the money.

What the funding will support

According to the Simons Foundation, the money will be distributed in part through an open call for proposals. Grants will back workshops and initiatives aimed at four broad goals: helping researchers evaluate recent AI-assisted results, training them to learn and use new tools, giving them venues to share perspectives, and shaping collective standards for how AI is used in the field.

Foundation President David Spergel framed the effort as a way for mathematicians to come together and choose their own path as AI reshapes the research landscape. He argued that the moment calls for viewing mathematics as a whole — as a shared understanding built by human curiosity — rather than simply a stockpile of theorems.

Tools and compute from the AI labs

Beyond the grant funding, Anthropic and OpenAI will provide AI tools and computing resources to researchers supported by the foundation. That support is intended to reach beyond mathematics into the basic sciences more broadly.

The foundation casts itself as a natural steward for the effort. It was an early backer of machine learning for science, and it funded development of the Lean proof assistant — the tool that has reshaped how many mathematicians work and made AI-assisted proofs feasible. It has also previously convened workshops, lectures and collaborations where mathematicians examined AI's growing role in their discipline.

Community control as the stated priority

Spergel emphasized that the program is meant to be community-driven, saying decisions about how AI tools are evaluated, adopted and used should sit with mathematicians as a group rather than with any single company or institution. That framing is notable given that two of the industry's largest AI developers are co-funding the effort and supplying the very tools involved.

Why it matters

Mathematics has become one of the clearest test cases for how AI is changing research itself. Proof assistants and AI systems can now contribute meaningfully to mathematical work, which raises hard practical questions: how AI-assisted results get checked, what skills researchers need, and what norms should govern verification and credit.

The $15 million program is a bet that those norms are best set by the research community rather than left to the companies building the tools. If the structure works — a scientific funder administering community-driven grants while AI labs contribute software and compute — it could become a template for other disciplines confronting the same shift. And because the tooling support extends into the basic sciences, the foundation appears to view mathematics not as an isolated case but as the first of many fields to face these questions.