deniz.in

Markets

Weather

Loading weather

· via The Verge

Anthropic mathematician claims Claude-assisted proof of the Riemann hypothesis

Levent Alpöge says he and Claude proved the Riemann hypothesis and formalized the claim in Lean, but it awaits independent verification and years of scrutiny under Millennium Prize rules.

Anthropic mathematician claims Claude-assisted proof of the Riemann hypothesis

Levent Alpöge, a mathematician at Anthropic, says he has proved the Riemann hypothesis — one of the most famous unsolved problems in mathematics — working together with Claude, the company's AI model. According to The Verge, the claimed proof has not been independently verified, and Anthropic has neither commented on it nor involved itself in the announcement.

A proof posted to a personal website

The claim currently exists only as a document on Alpöge's own site, written up in Lean, a proof assistant that allows arguments to be checked mechanically. In an accompanying statement, Alpöge describes Claude as central to the work: he says the technique he developed with the model stretches beyond the original problem to the Generalized Riemann Hypothesis, and that he expects it to reach the Extended Riemann Hypothesis with more time. He also says the arguments have been fully formalized with no assumptions left in place, and reflects that witnessing the progress felt less like an ending than the start of something far larger for science.

The timing is notable. The Verge reports that Alpöge messaged Anthropic chief executive Dario Amodei about the result late on a Sunday night, and that his write-up suggests he held publication until after the company's IPO on Thursday.

There is an awkward wrinkle in the company's own record, too. In August, Anthropic acknowledged it was working on the Riemann hypothesis with Claude but said at the time that it did not expect the model's techniques to lead to a proof. The Verge says it approached Anthropic for comment and had not received a reply.

A prize with a two-year waiting period

The Riemann hypothesis, first posed in 1859, concerns the distribution of prime numbers and the role played by the zeros of a particular function. It is one of seven Millennium Prize Problems overseen by the Clay Mathematics Institute, each carrying a $1 million award.

Recognition would not come quickly even if the proof holds. Under the institute's rules, a proposed solution must be published, withstand at least two years of scrutiny, and win general acceptance in the global mathematics community before it can be considered for the prize. In the roughly 25 years since the prizes were established, only one problem — the Poincaré conjecture, concerning three-dimensional spheres — has been officially solved.

AI's accelerating push into mathematics

The claim lands amid a turbulent stretch for AI in the field. In September, OpenAI said it had cracked the Navier-Stokes problem, another Millennium Prize problem concerning fluid flow. The Verge reports that OpenAI took on that problem after hearing rumors that rival researchers were closing in on a solution, and that Alpöge had been working on it with Tristan Buckmaster of New York University. Buckmaster has said OpenAI's result followed the same route he and Alpöge had been exploring with OpenAI's Codex and Claude.

OpenAI reportedly spent millions of dollars and deployed 10,000 agents on the effort. The result is widely considered credible but has received no formal recognition, and the project has been dogged by allegations that unpublished work from mathematicians who had shared their research with its AI tools was used along the way — claims the company has strongly denied.

Alpöge himself claimed another significant result in July, saying he had settled the Jacobian conjecture using Claude. The Verge notes that the deluge of findings has left mathematicians scrambling to digest the material and reconsider their place in a rapidly shifting landscape.

Why it matters

If Alpöge's proof survives review, it would resolve a problem that has stood for more than 165 years and establish AI as a genuine research collaborator at the very top of mathematics rather than a curiosity. Even before any verification, the pattern is striking: within a matter of months, AI-assisted efforts have claimed two separate Millennium Prize problems plus at least one other major conjecture. The Lean formalization matters as well, because machine-checkable arguments can shorten the distance between a dramatic claim and confidence in it — something this particular claim will badly need. And the circumstances around it, from a publication timed after an IPO to disputes over whose unpublished ideas fed earlier results, show that the hardest questions raised by AI in mathematics are social and commercial as much as technical.

  • #anthropic
  • #claude
  • #mathematics
  • #riemann-hypothesis
  • #lean

Related posts