Fields Medal winners blast OpenAI over AI math proofs

The open letter did not emerge in isolation. It follows the Leiden Declaration, released in June by a working group of mathematicians, which likewise examined how LLM proofs will reshape their field. Both documents share a concern that the speed at which AI labs announce solutions leaves no room for the slower, collaborative work that validates new methods and credits prior contributions.

The specific incidents this week gave those concerns a sharper edge. NYU professor Tristan Buckmaster accused OpenAI of pressuring him not to credit a collaborator who works for Anthropic, and questioned whether the company used their Codex work to produce its own proof over a marathon weekend. OpenAI then withdrew its CalTech math event sponsorship after university researchers criticized it. The letter's signatories argue that if solutions arrive without proper writeups or citation, the human transmission chain between mathematicians — the superstructure that trains students and generates new questions — could erode.

The stakes may extend beyond mathematics. The letter itself says the issues the math community faces are similar to those other scientific and creative professions are facing, indicating issues all of humanity might face as AI changes how work is done. Whether the field preserves open research or shifts toward secrecy may hinge on whether frontier labs and the mathematicians they rely on can agree on norms for credit and verification — and on whether institutions and policymakers take up the recommendations the Leiden Declaration set out.