Twenty-five Fields Medal winners warn AI labs threaten mathematical research

Twenty-five winners of the Fields Medal, the most prestigious prize in mathematics, have signed an open letter arguing that AI labs are putting mathematicians' intellectual work at risk as they compete to solve famous unsolved problems. According to TechCrunch, the signatories accept that AI cracking major mathematical challenges could benefit humanity — but only if the resulting solutions can be understood, verified and communicated by the mathematical community and, eventually, the wider world.

How the dispute with OpenAI escalated

The letter arrives amid an increasingly public conflict between mathematicians and OpenAI. TechCrunch reports that Tristan Buckmaster, a professor at NYU, this week accused OpenAI of pressuring him not to credit a collaborator — who works for Anthropic — for their role in solving an important math problem. Buckmaster also questioned whether the company had drawn on that earlier work, carried out through its Codex tool, to produce its own headline proof over a marathon weekend of model inference.

According to TechCrunch, OpenAI then withdrew its sponsorship of a mathematics event at CalTech on Thursday, after researchers at the university criticized the company. The report also notes that OpenAI's proof remains unverified.

The letter's argument

The signatories' complaint is less about whether AI can produce proofs and more about how those proofs are released. As they write, solutions are often "announced in a rush, leaving no time for a proper writeup, the isolation of new methods and ideas, and citing relevant previous work of others."

That rush, they argue, creates severe attribution and plagiarism questions of the kind familiar across creative professions. More fundamentally, they contend that AI-conceived ideas depend on human mathematicians to develop them and fold them into the mathematical canon. Without those people, AI-generated ideas would "never become fully alive," and the human transmission chain between mathematicians would be broken.

A dynamic that rewards secrecy

TechCrunch reports growing unease among mathematicians who suspect their own use of Codex may have been folded into OpenAI's newer models. The broader worry is structural: a frontier lab that spots a promising path to a discovery can spend tens of millions of dollars on LLM inference to beat the original researchers to a proof. If that becomes the norm, the incentive flips toward secrecy — a direct threat to the culture of open research that mathematics has long run on.

The Leiden Declaration came first

This is not the field's first collective response. The letter follows the Leiden Declaration, released in June by a working group of mathematicians, which also examined how LLM-generated proofs will change mathematical work and set out recommendations for mathematicians, institutions and policymakers.

As TechCrunch observes, mathematicians are leaning on an argument familiar from software engineering and other fields being reshaped by AI tools: the value of the discipline lies not just in proofs and who gets credit, but in the intellectual superstructure that trains students, surfaces new questions and integrates ideas into broader human knowledge.

Why it matters

The letter explicitly frames mathematics as an early warning rather than an isolated case. The issues the field now faces, the signatories write, resemble those confronting other scientific and creative professions, and point to a question "all of humanity might face": how to make sure that, as AI changes the way work is done, we do not lose sight of what that work was meant to achieve in the first place.

For AI labs, the dispute is also a concrete reputational problem. OpenAI has drawn public criticism from some of the most decorated mathematicians alive, withdrawn a sponsorship amid that backlash, and put forward a proof that the community it apparently aimed to impress has not verified. And for other disciplines, the mathematicians' attribution problem — who did the thinking, and who gets the credit — reads like a preview of conflicts heading their way.