OpenAI claims to have made a major breakthrough in mathematics, stating that artificial intelligence has solved the Navier-Stokes problem, one of the seven great mathematical problems of the millennium, for which there is a one million dollar prize.
The company says that approximately 10,000 AI agents worked simultaneously on the problem and arrived at a solution in 88 hours, as reported by The Guardian.
However, the announcement was quickly followed by controversy. Mathematician Tristan Buckmaster, a professor at New York University, claimed that OpenAI accelerated its research after learning that he and another researcher from Anthropic, OpenAI's rival, were about to announce their own progress in solving the problem.
Buckmaster also raised questions about the fact that their ongoing work was stored in Codex, the OpenAI model used for writing computer programs, which could have made the materials accessible to the company's team.
What is the Navier-Stokes problem
The Navier-Stokes problem refers to the equations used to describe the motion of fluids, such as water and air. One of the central questions is whether these equations can "blow up" under certain conditions, causing the fluid velocity to become mathematically infinite.
OpenAI claims that its demonstration shows that such a phenomenon can occur under certain conditions. The solution was then verified by GPT-6 Astra in about 17 hours.
The Navier-Stokes problem is one of the seven "Millennium Problems," established by the Clay Mathematics Institute as some of the most important unsolved mathematical problems. Solving any of them is rewarded with a million dollars.
10,000 AI agents worked simultaneously
OpenAI says it used an internal system for this project that is more powerful than the newest GPT-6 Astra model. Approximately 10,000 AI agents, capable of autonomously performing complex tasks, worked in parallel on the problem.
According to the company, the entire process took 88 hours.
"A major goal of our work is to enable scientists to advance research and technology that benefit all of humanity. We believe it is important to inform the world about the pace of AI progress and what to expect from future models," wrote OpenAI researchers on the company's blog.
Sebastien Bubeck, an OpenAI researcher, described the result as "a spectacular culmination of the journey we have seen in the past 12 months."
A mathematician accuses OpenAI of having access to his research
Tristan Buckmaster says he doesn't know exactly what the OpenAI system did to reach the result and whether his research was used.
"I don't know what their model did or how. I don't know if our data was used. I'm not accusing anyone of anything," the mathematician clarified.
At a press conference, Sebastien Bubeck denied that OpenAI used Buckmaster's and his colleague's research or accessed the materials stored on the company's servers.
However, OpenAI stated that it cannot completely rule out the possibility that data from the use of its products by the two researchers contributed to improving its models.
AI delves deeper into mathematics
The result announced by OpenAI comes at a time when artificial intelligence models, supported by massive computing power, are beginning to play an increasingly important role in mathematical research.
In May, OpenAI announced progress in solving another mathematical problem formulated about 80 years ago. Google DeepMind also announced significant results obtained by its models in the field of mathematics.
OpenAI specified that it does not intend to claim the one million dollar prize awarded by the Clay Mathematics Institute for solving the Navier-Stokes problem.
The announcement also provides an opportunity for the company to highlight the positive applications of artificial intelligence, at a time when the development of increasingly autonomous systems raises concerns about their safety.
In July, OpenAI revealed that a group of AI agents managed to breach Hugging Face's infrastructure during a cybersecurity test. A similar incident was later reported in the case of Anthropic.
Such episodes have fueled calls for limiting the development of artificial intelligence, including appeals from some American senators to permanently ban the development of "superintelligence" systems capable of surpassing humans in all cognitive tasks.