# Anthropic mathematician says he proved the Riemann hypothesis with Claude

> **TL;DR:** Levent Alpöge, a mathematician at Anthropic, has published a proof of the Riemann hypothesis built with Claude, written in Lean and posted on his personal website There is no independent verification yet, and Anthropic has not commented The Clay Institute requires a proof to be published and withstand scrutiny for at least two years before it considers the 1 million USD prize The author says the method can be extended to the Generalized Riemann Hypothesis OpenAI previously claimed to have solved the Navier-Stokes problem, and that claim is still disputed

- Original: https://migovi.com/en/gia-thuyet-riemann-claude-anthropic-alpoge/
- Author: [duongchautantuan](https://migovi.com/en/author/duongchautantuan/)
- Published: 2026-11-21T02:42:09+07:00
- Categories: AI, News
- Source: [The Verge](https://www.theverge.com/ai-artificial-intelligence/1010051/anthropic-riemann-hypothesis-millennium-prize-problem)

---

The proof of the Riemann hypothesis so far exists only on the author's personal website, but if it survives scrutiny, it would be AI's biggest milestone in mathematics.

## Where the proof comes from

Levent Alpöge, a mathematician at Anthropic, says he worked with Claude to prove the Riemann hypothesis, a problem that some of the world's best mathematicians have chased for decades without success. Proposed in 1859, the hypothesis concerns how prime numbers are distributed and the role played by the zeros of a particular function, which is why it is still considered one of the most important unsolved problems. It is also one of the seven Millennium Prize Problems, the set of open challenges regarded as the hardest in mathematics.

In the introduction on his website, Alpöge writes that he believes he found the proof together with Claude and that the method extends easily to the Generalized Riemann Hypothesis. He expects that, given more time, the Extended Riemann Hypothesis could be solved as well. He and Claude also formalized the arguments without any assumptions, and the proof is presented in Lean. By his own account, he messaged Anthropic CEO Dario Amodei about the result on Sunday evening, and the text suggests he waited until after the company's IPO on Thursday to publish.

![Image illustrating the Riemann hypothesis proof announced by an Anthropic mathematician with Claude](https://migovi.com/wp-content/uploads/2026/11/gia-thuyet-riemann-claude-anthropic-alpoge-2.jpg)

## Why this isn't settled yet

For now, the proof exists only on Alpöge's website, and no one has independently verified it. Anthropic did not comment on or get involved in the announcement, and The Verge's request for comment did not get an immediate response. Notably, when Anthropic said in August that it was trying to solve the Riemann hypothesis with Claude, the company stated clearly that it did not expect the techniques Claude used to lead to a proof, so this announcement calls for even closer scrutiny.

As for the prize, the Clay Mathematics Institute, which administers the Millennium Prizes with 1 million USD for each problem, has not recognized the proof. Under the institute's rules, a proposal must be published and withstand scrutiny for at least two years, during which it must be "widely accepted by the global mathematical community," before it can be considered for the award. The numbers show how hard that is: more than 25 years after the prizes were created, only the Poincaré conjecture, a topology problem about three-dimensional spheres, has been officially solved.

## AI's race in mathematics

![Image related to AI and the Riemann hypothesis in mathematics research](https://migovi.com/wp-content/uploads/2026/11/gia-thuyet-riemann-claude-anthropic-alpoge-3.jpg)

The Riemann hypothesis is not the first Millennium Problem to get AI involvement. In September, OpenAI claimed to have solved the Navier-Stokes problem, which concerns the flow of fluids, and said it pursued the problem after hearing that rival research teams were closing in on a solution. Alpöge was working on that very problem at the time with Tristan Buckmaster, a mathematics professor at New York University, and Buckmaster said OpenAI had produced a solution along the same lines the two were pursuing.

OpenAI spent millions of USD and deployed 10,000 agents on the effort. The result is generally considered credible, though not formally recognized, but the achievement was overshadowed by allegations about its research practices, including the use of unpublished work from mathematicians who had shared their research with the company's AI tools. OpenAI strongly denied this, but it has not stopped the backlash. While the spotlight has mostly been on OpenAI, Alpöge says he used Claude in July to solve the Jacobian conjecture, and this rapid string of results has left mathematicians scrambling to digest it while finding their own place in a mathematical landscape that is changing very fast.

**MIGOVI VIEWPOINT**

Migovi thinks this should be treated as a notable but unsettled claim. The proof has no independent verification, Anthropic has not commented, and the company itself said in August that it did not expect Claude's approach to lead to this proof. Alpöge formalizing the arguments in Lean is a plus, since it lets experts check each step instead of simply taking the author's word. For readers in Vietnam, the sensible course is to wait for responses from the mathematical community and the Clay Institute before concluding that AI has done something humans could not, while still following the story, since this direction could change how mathematics research is done in the coming years.

Source: [The Verge](https://www.theverge.com/ai-artificial-intelligence/1010051/anthropic-riemann-hypothesis-millennium-prize-problem)
