OpenAI is donating $5 million to a mathematics-focused initiative run by the Simons Foundation, part of a $15 million program that also counts Anthropic among its contributors. The money will fund workshops, conferences and special programs to help mathematicians examine, understand and build on results produced by AI, including the results OpenAI itself recently published.
The donation is the clearest sign yet that the company is trying to repair its relationship with the mathematics community, which has been openly hostile since OpenAI began dropping large volumes of AI-generated research on the field. The controversy began last month when OpenAI published what it called milestone work on the Navier-Stokes equations, problems that have resisted mathematicians for 90 years, prompting accusations that the company may have relied on unpublished human work to get there.
Tensions escalated last week when OpenAI announced that an experimental, unreleased model had produced more than 372 new results representing substantial work on some of the most important open questions in mathematics and computer science. The single-drop publication model drew particular fury. Cambridge professor Henry Wilton wrote that if OpenAI wanted to destroy the mathematical community, the announcement would be a great way to go about it, and called the publication a basic financial injustice because the company's models learned to reason by copying reasoning the community had made available for free.
Mathematicians argued that publishing everything at once, without giving the field time to verify the work, undermined the very processes that allow mathematics to check itself. Ilya Kazachkov of the University of the Basque Country said the magnitude of the announcement indicated that changes to the profession were likely to be profound. Even OpenAI's original, deliberately low-key blog post, titled sharing AI progress in mathematics, acknowledged that the release would be contentious and committed to helping experts work through its consequences.
The Simons Foundation, endowed by mathematician and investor Jim Simons' family foundation, will manage and award the funding, including through an open call for proposals supporting workshops and initiatives that help researchers evaluate recent AI-assisted results, learn new tools and shape collective standards for AI use. OpenAI said it will also provide access to its models and computing resources to Simons-supported researchers.
Foundation president David Spergel framed the moment as an inflection point, saying it is time to look at mathematics as a whole rather than a collection of proofs and theorems, and that the community needs to forge the best path forward as AI reshapes the research landscape. OpenAI, in a post on X, said the goal is to support the mathematics community in turning AI progress into deeper human understanding and new discoveries.
Whether a $5 million check buys goodwill or merely buys time depends on what changes next. The community has said it needs time, verification infrastructure and new norms, and the Simons program funds exactly that. For OpenAI and Anthropic, whose models are trained on the openly shared reasoning of the very people now protesting, the initiative is also a bet that collaboration can succeed where a 372-result bombshell conspicuously did not.
Comments (0)
Log in to join the discussion
Log InNo comments yet