Twenty-five Fields Medalists have signed an open letter arguing that AI labs are threatening mathematical research. The letter follows NYU professor Tristan Buckmaster accusing OpenAI of pressuring him not to credit a collaborator who works for Anthropic for solving an important math problem, and questioning whether the company used their work with Codex to produce its own proof.
On Thursday, OpenAI withdrew its sponsorship of a math event at CalTech after criticism from researchers. The signatories argue that AI-conceived solutions must be understood and communicated by the math community, and that the rush to announce proofs without proper writeups raises severe attribution and plagiarism concerns.
The letter follows the Leiden Declaration from June and warns that if frontier labs can spend millions using LLMs to beat researchers to a proof, the culture of open research will be threatened. Mathematicians who use Codex are now wondering if their usage was fed into OpenAI's models.
