Mathematical formulas written on a chalkboard (illustrative photo)

A mathematician at the artificial intelligence company Anthropic has claimed to have proved the Riemann hypothesis — the most famous unsolved problem in mathematics — working with the company’s AI chatbot Claude, and independent checkers say the proof is correct.

Key Points

  • Riemann hypothesis proof — Anthropic mathematician Levent Alpöge announced on Friday that he believes he has, “working with Claude, discovered a proof of the Riemann Hypothesis”.
  • The proof was released as about 17 million lines of code in Lean, a programming language that lets computers check mathematical arguments line by line.
  • On Saturday, the Lean Focused Research Organization and Mathlib Initiative said they had finished verifying the code: “We have full confidence that Alpöge’s proof of the Riemann Hypothesis is correct.”
  • Alpöge says the method also resolves the broader Generalised Riemann Hypothesis, which underpins much of modern number theory.
  • Hours after his announcement, a second, entirely human-written proof was posted by a Stanford mathematician — so two rival claims now await formal review.
A mathematics classroom with formulas on the chalkboard (illustrative photo)

The Riemann hypothesis, first proposed by the German mathematician Bernhard Riemann in 1859, concerns the distribution of prime numbers — the numbers divisible only by one and themselves — as they grow towards infinity. It is widely regarded as the most important open question in pure mathematics and is one of the seven Millennium Prize Problems selected by the Clay Mathematics Institute.

Alpöge, a mathematician on Anthropic’s technical staff, posted his announcement on his personal website on Friday in a four-page note. “I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis,” he wrote. He said the full written argument would be extremely long, that he had not yet mastered all of the mathematics involved, and that he planned to write it up properly when he could simplify it. For now he released only the Lean version — code that a computer can verify, but that few humans can read.

Verification — and a rival claim

The announcement came a day after Anthropic’s record-breaking debut on the stock market, and Alpöge said he had consulted chief executive Dario Amodei before going public, acknowledging that the timing could be viewed cynically. Mathematicians reacted with astonishment — and caution. “What he released is garbled,” said Andrew Granville, a number theorist at the University of Montreal, who argued Alpöge should have prepared a clearer write-up first. But James Maynard of Oxford University said: “I suspect that the proof will turn out to be correct,” even in its current form as “incomprehensible code”. A rival claim also emerged on Friday, when Stanford professor Ravi Vakil posted a preprint by Alessandro Maria Masullo describing an entirely human-written proof. The Masullo preprint will have to be checked in the same way.

Then, on Saturday, the community of mathematicians and programmers who maintain the Lean proof-checking software finished verifying the 17 million lines of code and issued their ruling. The Lean Focused Research Organization and Mathlib Initiative said in a joint statement: “We have full confidence that Alpöge’s proof of the Riemann Hypothesis is correct.”

What happens next

Computer verification is a powerful signal, but it is not the final word. The Clay Mathematics Institute still lists the Riemann hypothesis as unsolved, and formal recognition requires the proof to be published and scrutinised by the mathematical community — a process that could take months or years, especially for an argument of this size. The rival Masullo preprint will have to be checked in the same way. If either proof stands, it would be a spectacular result in the bicentenary year of Riemann’s birth, and a landmark moment for AI in science. For now, the claim is verified by machine, doubted by some experts on presentation, and awaiting the slow, human work of understanding what has actually been proved.

For background on Anthropic’s record stock market debut, see our report on Anthropic Raises Record $81 Billion in IPO Priced at $83 as Shares Debut on Nasdaq, and on AI Data Centres to Double Global Power Use by 2030 and Push Emissions 70% Higher, IEA Warns. More in our Technology section.

Sources: Scientific American, “Anthropic mathematician proves Riemann hypothesis, most famous problem in all of math” (https://www.scientificamerican.com/article/anthropic-mathematician-claims-to-prove-riemann-hypothesis-most-famous-problem-in-all-of-math/); Nature, “Mathematicians astounded by AI and human-written proofs of iconic Riemann hypothesis” (https://www.nature.com/articles/d41586-026-03812-7)

Leave a Reply

Your email address will not be published. Required fields are marked *