Twenty-five mathematicians signed an open letter against OpenAI

Twenty-five mathematicians, among them Fields medallists, have signed an open letter objecting to how AI labs work. The objection is not to what the models can do, but to how results are announced.

According to the letter, labs rush to declare solutions to famous problems, leaving no room for verification, a proper writeup or citation of earlier work. In the signatories' words the pace leaves "no time for a proper writeup, the isolation of new methods and ideas, and citing relevant previous work of others".

Two incidents brought it into the open

The first was on 8 September. Tristan Buckmaster of New York University said OpenAI had pressured him not to credit an Anthropic researcher on the solution to a mathematical problem.

The second came this week. OpenAI withdrew its sponsorship of a mathematics event at CalTech after criticism from researchers.

Behind both sits OpenAI's announcement about the Navier-Stokes equations. The company published the claimed result on 8 September 2026 and described the work behind it as a marathon weekend of inference. Mathematicians say the result still has not been verified.

The timing is disputed too. About twelve hours before that announcement, Buckmaster said his work on the Euler equations had been leaked to OpenAI days before and may have fed the process. More detail: Fortune and Quanta Magazine.

What the letter is really worried about

The signatories are not shutting models out. Their worry is a broken chain of transmission. Without mathematicians in the process, the letter says, "AI-conceived ideas would never become fully alive and the crucial human transmission chain between mathematicians would be lost".

In mathematics a result is worth something not only because it is true, but because someone else can build on it. A proof nobody has read and understood is a dead result, however correct.

Source: TechCrunch.