OpenAI’s feud with mathematicians is only escalating

Details

The mathematical community's conflict with AI labs escalated sharply this week as 25 Fields Medal laureates co-signed an open letter asserting that AI labs are threatening the integrity of their field. The three specific harms cited: rushed proof announcements with no proper writeup, missing citations for prior work, and an OpenAI proof that remains unverified by the mathematical community.\n\nThe immediate catalyst was NYU professor Tristan Buckmaster's accusation that OpenAI pressured him not to credit an Anthropic collaborator who had contributed to solving an important mathematical problem. OpenAI also withdrew its sponsorship of a CalTech math event after facing researcher criticism. Other mathematicians are now paranoid that their Codex usage was ingested into OpenAI training data, effectively contributing to results that may be published without their credit.\n\nThe structural concern goes beyond any single dispute. Frontier labs can spend tens of millions of dollars in LLM inference to pursue proofs faster than independent researchers, creating an asymmetric race where secrecy becomes strategically rational. The letter warns this will erode the culture of open, cumulative knowledge-building that mathematics depends on — and that similar dynamics are coming for every other creative and scientific field.\n\nThe letter follows the Leiden Declaration in June, which offered a set of recommendations for mathematicians, institutions, and policymakers on navigating AI's role in research. With no resolution in sight, the conflict between frontier labs and research communities is likely to intensify.