Twenty‑five leading mathematicians, each a Fields Medalist, signed an open letter this week warning that artificial‑intelligence laboratories are threatening their intellectual work by racing to produce AI‑generated proofs, neglecting proper attribution and fostering secrecy.
What the open letter says
The letter, released on 11 September 2026, argues that AI labs are "rushing to produce AI‑generated proofs" and that the speed of announcement leaves "no time for a proper write‑up, the isolation of new methods and ideas, and citing relevant previous work of others" — a concern the signatories say jeopardises the scholarly record. Each signatory has been awarded the Fields Medal, the most prestigious prize in mathematics.
“Often these solutions are 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, they wrote — and OpenAI’s proof remains unverified.”
The signatories stress that, as in all creative professions, the situation raises "severe attribution and plagiarism questions." The open letter is the first coordinated public statement from the mathematics community specifically targeting AI‑generated research.
Tristan Buckmaster’s accusation
NYU professor Tristan Buckmaster, whose own research appears in the same TechCrunch story, accused OpenAI of pressuring him not to credit a collaborator who works for Anthropic on a recent breakthrough. Buckmaster wrote that OpenAI "pressured him not to credit a collaborator who works for Anthropic for solving an important math problem" and wondered whether the company had used its Codex system to produce the proof over a marathon weekend of inference.
“NYU professor Tristan Buckmaster accused OpenAI of pressuring him not to credit a collaborator who works for Anthropic for solving an important math problem, and wondered if the company had used their work with Codex to produce its own groundbreaking proof over a marathon weekend of inference.”
The allegation adds a concrete example to the broader concerns expressed in the open letter, suggesting that the pressure to claim priority may extend to the omission of co‑authors.
OpenAI’s sponsorship pullback
On 10 September 2026, OpenAI withdrew its sponsorship of a mathematics event at the California Institute of Technology after criticism from researchers at the university. The move follows the same week’s public criticism and underscores the growing tension between AI labs and the academic community.
“On Thursday, OpenAI withdrew its sponsorship of a math event at CalTech after the company was criticized by researchers at the university.”
OpenAI, headquartered in San Francisco and led by CEO Sam Altman, employs roughly 4,500 people according to the latest Wikidata entry. Anthropic, another San Francisco‑based AI firm, is led by CEO Dario Amodei and employs about 2,500 staff. Both companies are central players in the race to develop advanced mathematical reasoning systems.
Why the letter matters now
The timing of the open letter coincides with two high‑profile disputes: Buckmaster’s accusation and OpenAI’s sponsorship withdrawal. Together they illustrate a shift from private concern to public confrontation. The letter’s 25 signatories represent a cross‑section of the discipline’s most influential voices, giving the warning weight that could influence funding decisions, conference policies and corporate research practices.
For AI labs, the criticism points to a trade‑off between speed and scholarly rigour. The letter notes that many AI‑generated proofs are announced before they can be independently verified, leaving the community with “unverified” results that may later be retracted or corrected. The lack of citation also threatens the traditional credit system that underpins academic careers.
What remains unknown
- How many AI‑generated proofs have been published without peer review, and what proportion of those remain unverified?
- Whether OpenAI or Anthropic will change their internal publication policies in response to the letter.
- The extent to which other AI labs beyond OpenAI and Anthropic are engaged in similar practices.
The open letter does not specify any concrete policy demands, leaving the next steps to be negotiated between the mathematics community and AI companies.
What could happen next
Mathematical societies and conference organisers may adopt stricter attribution guidelines for AI‑generated results. Funding agencies could require proof‑verification steps before accepting AI‑produced research. Meanwhile, AI labs might introduce internal review processes to ensure that any public claim of a new proof is vetted by external experts.
For now, the open letter serves as a formal record of the mathematicians’ concerns and a signal that the debate over AI‑generated mathematics is entering the public arena.