Anthropic Mathematician Claims Proof of Riemann Hypothesis Using AI
Levent Alpoge asserts he has solved a 167-year-old math problem with the help of Claude, though independent experts are still verifying the complex Lean code.
By Daniela Kelly ·

Levent Alpoge, a mathematician at the technology company Anthropic, has announced that he believes he has proven the Riemann hypothesis. This conjecture, first proposed by Bernhard Riemann in 1859, is widely considered the most significant unsolved problem in pure mathematics. Alpoge made the claim on Friday on his personal website, stating that he discovered the proof while working with Claude, an artificial intelligence model developed by his employer.
The Riemann hypothesis concerns the distribution of prime numbers. While these numbers appear random, the hypothesis suggests they follow a deep, hidden order encoded in the Riemann zeta function. If Alpoge's claim is verified, it would settle a question that has challenged mathematicians for 167 years. The problem is one of the seven Clay Mathematics Institute Millennium Prize Problems, each carrying a one million dollar award.
Alpoge did not publish a traditional academic paper. Instead, he released the argument in Lean, a programming language designed to verify mathematical proofs step by step. He noted that Claude played a major role in the process, writing that the AI far outpaced him near the finish line. The proof covers the generalised Riemann hypothesis, which extends the original conjecture to a broader family of functions.
Independent experts are currently working to verify the claim, but no human has yet read the entire argument. Kevin Buzzard, a professor of pure mathematics at Imperial College London, explained that the proof consists of 17 million lines of Lean code. This volume of code could translate to tens of thousands of pages of standard mathematics. Buzzard noted that it is unclear how much of the work represents new ideas and how much is standard material that the machine had to develop.
Buzzard described the current state of the proof as millions of lines of code that are difficult for humans to interpret immediately. He said that experts around the world are translating parts of the code back into English. He expects a better understanding of the proof to emerge within a week or two as humans analyze the Lean code.
The importance of the Riemann hypothesis extends beyond pure mathematics. Abhishek Saha, a professor at Queen Mary University of London, explained that hundreds of other mathematical results have been proved assuming the hypothesis is true. Buzzard added that the result has implications for internet security. Some algorithms used in cryptography are known to work if the generalised Riemann hypothesis is true, allowing cryptographers to operate with greater confidence.
Saha said the claimed proof has overturned his expectations. He had believed that a proof would require substantial new theories developed over decades. Instead, he described the proof as taking a complicated tour through existing mathematical infrastructure to create a direct path to the solution. He argued that the credit should go to Alpoge for directing the AI effectively, suggesting the mathematician deserves a Fields Medal if the proof holds up.
Buzzard rejected the idea that AI has made human mathematicians redundant. He compared AI to the telescope in astronomy, a tool that allows for clearer observation of complex fields. While AI lacks taste and must be directed by humans, it is a powerful new instrument. Buzzard noted that the current breakthroughs are concentrated in tech companies with access to advanced models, but this may change as more powerful models become available to the public.
The past year has seen significant AI driven advances in mathematics. Previous AI assisted results showed that at least 67.25 percent of a specific set of mathematical values follow the pattern predicted by the Riemann hypothesis. The new claim, if correct, would extend this confirmation to all such values. Saha stated that AI is creating a revolution in the field, changing how research is conducted and emphasizing the importance of asking the right questions and directing AI effectively.
Buzzard acknowledged that some mathematicians feel unease about the rapid changes in the field. He noted that people are grieving the loss of traditional methods and the impact on professional egos. However, he maintained that mathematics is infinitely large and that AI provides a powerful tool to explore it. He concluded that the rumors of mathematics dying have been greatly exaggerated.






