i
News
News · 2026-11-20

Anthropic mathematician claims Claude helped prove Riemann hypothesis

Dataist Dataist

A mathematician at Anthropic and Harvard University, Levent Alpöge, says he has proved the Riemann hypothesis with Claude. One of mathematics’ most important unsolved problems and a Millennium Prize Problem, it is backed, he says, by millions of lines of Lean code that check the proof. Alpöge has not yet presented a full scientific paper: he claims the proof created with Claude runs to more than 250,000 pages. Other specialists now need to verify the result.

Cover: Anthropic mathematician claims Claude helped prove Riemann hypothesis

What Alpöge claims

The Riemann hypothesis was formulated in 1859 by German mathematician Bernhard Riemann. It describes a pattern that may lie behind the apparently random distribution of prime numbers along the number line. The problem concerns how prime numbers of a given size are distributed, and has guided research in number theory for about 150 years. Many mathematicians consider it the most important unsolved problem in their field.

Alpöge announced the purported proof in a statement. He wrote that he worked on it with Claude. The mathematician also noted that the Riemann hypothesis has influenced more than number theory: its ideas are reflected in physics as well.

The claim has not yet been independently verified. Along with it, Alpöge published millions of lines of Lean code, which he says can be used to check the result. He has not yet presented a formal scientific paper; according to Alpöge, the proof runs to more than 250,000 pages.

AI in mathematics

If specialists confirm the result, it would show how AI can help solve mathematical problems that people have worked on for decades. In September, OpenAI said its advanced model had helped solve the Navier–Stokes problem. The claim sparked debate over who should receive credit for the solution.

AI’s growing role prompts mixed feelings among mathematicians, from excitement and curiosity to sadness. Alpöge acknowledged that such a discovery could feel like the end of an important chapter and the loss of a central guiding problem. But what he saw during the work convinced him that science is on the brink of major changes.

Dataist's take: Alpöge’s claim remains unverified. The published code and the specialists’ future review matter more here than the claim itself. If the proof withstands independent scrutiny, it would be a notable example of collaboration between a mathematician and AI; for now, it is too early to say the hypothesis has been solved.

Daily AI news

Every day we pick what actually matters in AI and explain it plainly — no hype, no filler. Subscribe if you want to follow where the industry is going.

Only what matters — every day

Follow on X