The Global Fever

Tech

Anthropic Riemann Hypothesis claim: what is known and what is not

An Anthropic researcher says he and Claude have a proof of the Riemann Hypothesis. Mathematicians have not yet verified it, and the Clay Institute still lists the problem as unsolved.

Wasif Mumtaz

Founder and Editor · November 21, 2026 · 6 min read

Blackboard covered in chalk mathematical formulas, diagrams and equations
Photo: Magnific

An Anthropic researcher says he has found a proof of the Riemann Hypothesis, one of the best-known open problems in mathematics, working with the company's Claude AI. The Anthropic Riemann Hypothesis claim was made on Friday, November 20, 2026, and as of Saturday no independent confirmation had been reported. It matters because the problem has stood unsolved since 1859 and carries a $1 million prize, and because the claim arrived one day after Anthropic's record stock market debut.

What the Anthropic researcher claimed

The researcher is Levent Alpöge, whom the Associated Press describes as a member of Anthropic's technical staff. He posted a four-page note on his website on Friday. In it he wrote, "I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis." The wording is a statement of belief, and the claim has not been presented as a finished, reviewed result.

AFP and Scientific American report that Alpöge also says the same method settles the Generalized Riemann Hypothesis, a broader version of the problem. Scientific American notes that its own headline and body describe the claim a little differently on this point, so the exact scope is worth watching as more details appear.

He has released the proof only as a Lean file for now. Lean is a programming language for writing arguments that a computer can check step by step. The AP reports that the Lean version runs to about 17 million lines of code, and that Alpöge says the AI work behind it produced more than 250,000 pages. AFP adds that a full written argument would be extremely long, that Alpöge says he has not yet mastered all of the underlying mathematics, and that he plans to write it up properly.

According to the AP, Alpöge says he found the proof late on Sunday night, before Anthropic's stock market debut, and announced it the day after the listing.

Why the Riemann Hypothesis is such a big target

The Riemann Hypothesis dates to 1859 and concerns the zeros of the Riemann zeta function. The Clay Mathematics Institute named it one of seven Millennium Prize Problems in 2000, with a $1 million reward for a solution. Because it is so well known, any claim about it draws immediate scrutiny from specialists.

Anthropic has worked on the problem before. In August, the company said an unreleased version of Claude had improved a long-standing result related to the Riemann Hypothesis, while stressing that it had not proved the hypothesis itself. TechCrunch reported at the time that two of Anthropic's in-house mathematicians confirmed that earlier result and that it was formalized in Lean.

Alpöge has also been in the news this year. Scientific American notes that in July he tweeted a Claude-generated solution to the Jacobian conjecture, another well-known open problem.

What mathematicians have said so far

The AP reported no independent confirmation. Alex Kontorovich of Rutgers told the AP that Claude checked the Lean code and judged the proof to be true. He added that a computer is a stricter reviewer than a human, but said he does not yet understand the proof himself.

Andrew Sutherland of MIT told Scientific American that he independently compiled the result in Lean, and he called the moment historic. The magazine noted that the mathematical details still need study, and that experts must confirm both that the conjecture is coded correctly and that the proof holds.

Other mathematicians quoted by Scientific American were conditional. Hector Pasten of the Pontifical Catholic University of Chile said that if the proof is correct and understood, it would connect branches of mathematics that were only partly linked. Carlo Pagano of Concordia University said its new ideas could open research directions if the result holds.

AFP reported a more cautious picture. It said the claim had not been independently verified or reviewed by outside experts, and that Brian Conrey, a leading specialist on the problem, had seen only Alpöge's posted remarks and not a manuscript. Conrey called the announcement a monumental moment, a reaction given before any such review.

The Clay Institute held back. Hours after the announcement it still listed the problem as unsolved, and its president, Martin Bridson, told the AP that assessing such a claim takes time. He called the announcement of historic significance but said the road from here to real understanding would be long and difficult.

How the Riemann Hypothesis proof would be verified

The accounts differ on how much outside checking has happened, but they agree on the central gap. Compiling a Lean file shows that the machine accepts the proof as written. Whether the statement it proves is the Riemann Hypothesis, and whether the argument holds up under expert reading, is the part mathematicians still have to settle.

Kontorovich noted that a proof in Lean usually stands in for a formal paper, because the machine checks each step. In this case there is no readable paper yet. Alpöge says he plans to write one, and Bridson's comments suggest that understanding the argument will take far longer than checking it.

The prize itself is a separate track. AFP reports that the Clay Institute's Millennium Prize requires publication in a peer-reviewed journal and two years of scrutiny. Even if the proof survives review, no prize decision is likely for some time.

Timing around Anthropic's IPO and earlier AI math disputes

The announcement came the day after Anthropic's Nasdaq debut, which the AP says raised a record $81 billion at a $1.5 trillion valuation. Alpöge said he consulted chief executive Dario Amodei before announcing and acknowledged that the timing could be viewed cynically. Anthropic did not respond to requests for comment from the AP or AFP. For background on the listing, see our report on the Anthropic IPO record raise and on the global AI stock selloff around the debut.

Other AI mathematics claims this year have also been contested. AFP reports that in September OpenAI said an AI model had solved another Millennium Prize Problem, on fluid equations, and that mathematicians say the proof appears correct but has not been fully vetted. Scientific American notes that NYU mathematician Tristan Buckmaster suggested OpenAI had taken his and Alpöge's work, while AFP says OpenAI denied wrongdoing.

What to watch next

Four developments will show how this resolves: a statement from the Clay Institute, checks by outside mathematicians on whether the Lean statement is correct, a written paper from Alpöge, and any formal response from Anthropic. Each could confirm, narrow or undercut the claim.

Until then, the accurate description is a claim with a machine-checkable file attached. It is not an established proof, and this article will be most useful to readers who treat it that way.

Disclosure: The Global Fever uses AI tools, including Anthropic's Claude, to help research and write articles.

Frequently asked questions

Did Anthropic prove the Riemann Hypothesis?

Not as an established result. One Anthropic researcher, Levent Alpöge, says he believes he has a proof made with Claude. No independent confirmation had been reported by November 21, and the Clay Institute still listed the problem as unsolved.

Who is Levent Alpöge?

The Associated Press describes him as a member of Anthropic's technical staff. Scientific American says he tweeted a Claude-generated solution to the Jacobian conjecture in July.

What is Lean, and does it settle the question?

Lean is a programming language for writing mathematical arguments that a computer checks step by step. A proof that compiles is strong evidence the logic is sound. Experts still need to confirm that the statement it proves matches the Riemann Hypothesis.

Would this win the $1 million Millennium Prize?

Not soon. AFP reports the Clay Institute requires publication in a peer-reviewed journal and two years of scrutiny before a prize can be awarded.

Is the claim connected to Anthropic's IPO?

The claim came the day after the debut. Alpöge said he found the proof before the listing, consulted Dario Amodei, and accepted that the timing could look cynical. The sources reviewed show the timing but no further link.

ShareXLinkedInWhatsApp

Sources

Related stories

Tech

Nvidia Stock Plunge: Chip Selloff Deepens as SEC Probe Reports Add Pressure

The Nvidia stock plunge of about 14% on Nov. 20 pulled chip and AI infrastructure shares sharply lower and pushed the S&P 500 into a correction, with reports of an SEC inquiry adding pressure.

Tech

Java4Shell and AI-Found Vulnerabilities: Why Tech Outages Are Raising Questions

A Java flaw found with AI help, called Java4Shell, landed during a run of bank and NSA outages. Here is what is confirmed, what is disputed and what to patch.

Tech

Alibaba Warns US Chip Rule Could Disrupt Data Center Expansion Plans

Alibaba's CEO said the new U.S. rule on advanced chips could slow its cloud growth and disrupt data center expansion, while Tencent said it expects little impact.