25 Fields Medalists warn AI threatens mathematical knowledge ecosystem

September 12, 2026 at 11:51 AM2 min

This digest was compiled by AI from multiple sources — links to the originals are below.

Twenty-five Fields Medalists signed an open letter warning that AI's involvement in mathematics threatens scientific knowledge and the mathematical community. The letter argues AI labs prioritize demonstrating capabilities over conceptual understanding, disrupting the human knowledge ecosystem. The warning follows OpenAI's claim of solving a Navier-Stokes smoothness problem, which a New York University professor suspects used his and an Anthropic colleague's work without credit.

Key Facts

Twenty-five Fields Medalists signed an open letter warning that AI's role in mathematics threatens scientific knowledge and the mathematical community.
The letter states that AI labs aim to demonstrate capabilities by solving known problems, while mathematicians seek conceptual understanding and new ideas.
OpenAI announced this week it solved the smoothness existence problem for Navier-Stokes equations, with internal verification taking 17 hours.
A New York University mathematics professor suspects OpenAI used work belonging to him and a colleague at Anthropic without credit.

The Open Letter

Twenty-five Fields Medalists signed an open letter stating that AI's participation in mathematical science creates a threat to scientific knowledge and the mathematical community. The letter argues that the goals of AI laboratories and the mathematical community diverge: developers see solving known problems as a way to demonstrate AI capabilities, while mathematicians view AI as a means to gain new knowledge and methods, with the primary goal being conceptual understanding and new ideas. Previously, solving known problems involved lengthy study of new ideas and methods, with results discussed, simplified, and connected to prior research before entering textbooks and other fields. With AI, this human knowledge ecosystem is breaking down: results are published hastily without proper formatting, indication of new methods, or links to other scientists' work.

Authorship and Plagiarism Concerns

New questions arise about authorship and plagiarism: for AI-generated ideas to be fully realized, participation of mathematical community members is required to develop and integrate these ideas into the existing knowledge system. AI can indeed accelerate mathematical research, but the actual benefit to humanity will depend on the people who manage this technology. This week OpenAI reported solving the smoothness existence problem for Navier-Stokes equations; verification and formalization of the result took the company's internal model an additional 17 hours. Subsequently, a mathematics professor at New York University suspected OpenAI of using work belonging to him and his colleague who works at Anthropic. OpenAI could neither refute the scientists' suspicions nor offer them a compromise solution.