The Simons Foundation said on October 15 that it, Anthropic and OpenAI will commit a combined $15 million in funding and resources to help mathematicians assess AI-assisted research and decide how to use the tools. Simons will manage and award the funding, including through an open call for proposals. The announcement is a commitment to future support, not a report that the money has been awarded or that applications can already be submitted.

The foundation plans to support workshops and other initiatives where researchers can evaluate recent AI-assisted results, learn to use AI tools, share perspectives and shape standards for the field. Anthropic and OpenAI also plan to provide tools and computing resources to researchers supported by Simons, including researchers in basic science beyond mathematics.

The big change

  • What changed: A research funder and two AI labs have committed resources for mathematicians to examine AI-assisted work and develop practices for using it. Simons says it will manage the awards, giving the mathematical community a route to organize that work through proposals and workshops.
  • Why it matters: Evaluating a mathematical result takes expert time and shared standards. The proposed support gives researchers and institutions a way to seek resources for that work as AI-assisted results enter the field. Its effect depends on which projects Simons funds and what they establish.
  • What to watch: The foundation has announced an open call but has not published its timing or award details in this announcement. Those terms, followed by the funded work, will show how the commitment is put into practice.

What the commitment covers

The Simons announcement describes $15 million in combined funding and resources from the three partners. That wording matters for institutions planning research: it does not say that $15 million in cash is immediately available to applicants. The release gives no allocation among workshops, other initiatives and the promised access to tools and compute, and it names no awardees.

Simons says it will manage and award the funding, including through an open call for proposals. The stated aims cover two kinds of work. One is assessing recent AI-assisted mathematical results. The other is helping researchers learn the tools, compare views and set collective standards for their use. The announcement does not prescribe a review method or define a threshold for accepting a result.

Simons Foundation president David Spergel said the mathematics community should drive discussions of how the tools are evaluated and used. That stated governance aim gives the foundation a role beyond transferring resources. Whether the eventual program gives researchers a useful, independent forum will be clearer when Simons publishes the call and its selection terms.

What remains open

The October 15 release does not give an application deadline, eligibility criteria, grant sizes, a timetable for workshops or a list of projects already funded. Researchers considering a proposal therefore have the program's aims but not yet the terms needed to plan an application. The announcement also does not report independently reviewed mathematical findings produced by this program; its proposed review and standards work still lies ahead.

OpenAI separately said it would contribute to the Simons initiative in an October 15 update to its mathematics post. That statement describes OpenAI's participation. It does not establish that the Simons program has reviewed, validated or arisen from any particular OpenAI release. Simons' own announcement frames the effort across recent AI-assisted results and the wider mathematics community.

Sources & further reading

  • Simons Foundation announcement, October 15, 2026: The primary account of the three-party commitment, Simons' management of awards, the planned open call, the aims for workshops and the promised tools and compute. It is an announcement of plans, not an award register.
  • OpenAI mathematics post, updated October 15, 2026: Confirms OpenAI's own stated contribution and planned model and compute access for Simons-supported researchers. Its discussion of OpenAI results is the company's account and does not show that this program has reviewed those results.