Tristan Buckmaster, a mathematics professor at New York University, teamed up with Levent Alpöge, a mathematician from Anthropic, to make initial headway on the Millennium Prize Problem concerning the existence and smoothness of the Navier-Stokes equations. They employed Codex and Claude AI models in their research. However, before their findings could be made public, OpenAI launched parallel research and published a complete proof of the problem, asserting that it was achieved by a new generation of models. This research, completed within a week, utilized 300 billion output tokens and incurred a cost of around $22.5 million.

Buckmaster accused OpenAI of pursuing the same specialized research path after becoming aware of their progress. He also criticized OpenAI for being ambiguous about the start time of their research and the extent of human involvement in their communications. OpenAI even went so far as to request the removal of Alpöge's name from the research and threatened to damage his career.

In response, OpenAI stated that its research commenced on September 1, motivated by rumors, and without having directly viewed the other party's work. They also claimed that their proof differed significantly from the other party's findings. Nevertheless, OpenAI conceded that it could not entirely dismiss the possibility that data generated by the other party using its products might have contributed to enhancing the model.

This incident has ignited a widespread debate over the role of AI in mathematical research and OpenAI's profit-driven motives. Buckmaster aims to reveal pertinent information to enable the public to grasp the full sequence of events.