Key Highlights
- An Anthropic employee says he worked with Claude on a proof of the Riemann hypothesis that is not independently verified.
- The claim is published in Lean, a proof-checking language, with workings passing 250,000 pages and no prose explanation yet.
- The Riemann hypothesis, formulated in 1859, is a Millennium Prize problem carrying a USD 1 million reward.
- The president of the London Mathematical Society called the announcement a historic development, while experts said it faces intense scrutiny.
The Claim
An Anthropic employee, himself a mathematician, says he worked with the company's Claude model to develop a proof of the Riemann hypothesis, a problem about prime numbers. If confirmed, it is described as the most significant advance yet in the use of AI in mathematics. The alleged proof has not been independently verified.
The employee described the moment in a post on his website on Friday as era-defining. He wrote that it is natural to feel the result marks an end, and that for him it is the loss of a principal guiding light, but that what he saw during the course of progress amazed him and he believes it is the dawn of something far greater across science.
How It Was Presented
The employee has not released a formal account of the solution with a prose explanation. Instead, he published the findings in Lean, a specialist programming language used for checking and expressing mathematical proofs. He said his AI-assisted workings passed 250,000 pages and described posting them as inappropriate, calling the alternative an unreadable mass of low-quality output.
He credited Claude as the main driver of the final push, writing that he and Claude have enjoyed working together across a wide range of topics in recent months, but that this time Claude far outpaced him near the finish line. He said he is now in an extraordinary situation made possible by this age, presenting a complete argument without having mastered the wider body of mathematics around it.
Background on the Problem
The Riemann hypothesis has confounded mathematicians since the German mathematician Bernhard Riemann formulated it in 1859, making it a 167-year-old problem. It is a central problem in number theory, the study of the properties of whole numbers. It is one of the problems set 26 years ago as a Millennium Prize challenge, with a USD 1 million reward attached to each.
Reaction From Experts
The president of the London Mathematical Society called the announcement a historic development. He said it opens the way to new levels of human understanding of mathematics and its applications.
Other experts said assessing the proof takes time. A professor of mathematical biology at the University of Bath said an announcement of this magnitude faces intense scrutiny, adding that if the argument survives, the consequences are large both for mathematics and for understanding of what AI can do.
AI and Mathematics Context
The claim follows a series of AI-assisted advances in mathematics. In September, OpenAI announced that it had developed a solution to another Millennium Prize problem, the Navier-Stokes problem. The Anthropic employee had been working on that challenge with a New York University mathematician, and the pair claimed that data from their work had potentially informed OpenAI because they had used its tools. OpenAI chief executive Sam Altman rejected what he called "unfounded accusations of plagiarism."
Last month, OpenAI unveiled hundreds of papers, including what it said were proofs of longstanding problems. Experts said the set mixed impressive work with papers that contained errors or were unreadable. The Association for Human Mathematics, a group pledging to protect the field against the "threat of artificial intelligence," urged mathematicians to stop working with OpenAI. The company defended its actions and pledged USD 5 million to a mathematics-focused foundation.






Please wait processing your request...