Anthropic researcher claims Claude helped solve 167-year-old math puzzle
A mathematician working at Anthropic says he used the Claude AI model to find a proof for the famous Riemann hypothesis.
A mathematician at Anthropic claims to have found a proof for the Riemann hypothesisRiemann hypothesisA famous mathematical conjecture about where the Riemann zeta function equals zero, linked to prime numbers. Open in glossary by working with the company’s artificial intelligence model, Claude. The claim arrived during the same week Anthropic completed its initial public offeringInitial public offeringThe first sale of stock by a private company to the public on an exchange. Open in glossary on the Nasdaq.
A 167-year-old mystery
Proposed by Bernhard Riemann in 1859, the Riemann hypothesis is widely considered the most famous open challenge in number theoryNumber theoryA branch of pure mathematics devoted primarily to the study of integers and arithmetic properties. Open in glossary. The problem concerns the distribution of prime numbers and suggests that their patterns are determined rather than random.
According to the Clay Mathematics Institute, the hypothesis states that all non-trivial solutions to the Riemann zeta function equation fall along a specific vertical line. While computers have verified this pattern for the first 10 trillion solutions, a complete mathematical proof remained out of reach, leading many researchers to view the problem as insurmountable.
Using Claude to build a proof
Anthropic mathematician Levent Alpöge detailed the findings in a paper this week, stating that he and Claude discovered a valid proof. Alpöge indicated that the argument involves an immense amount of advanced mathematics. He also believes the same approach could apply to broader variations of the puzzle, including the Generalized Riemann Hypothesis and the Extended Riemann Hypothesis.
The announcement follows other recent attempts to apply artificial intelligence to difficult scientific questions, including earlier work linked to the Navier-Stokes problem by OpenAI.
Verification still pending
Independent researchers must now review and verify the proposed proof, a process that is expected to take time. Alpöge noted mixed feelings about the milestone, calling the problem a guiding light for researchers, but expressed optimism about how automated tools could accelerate scientific discoveries in the future.
How does this story land?
One tap per reader.
Quick poll
Will AI models become standard co-authors on major mathematical proofs?
One tap. No sign-up.
Ask this story
Still have a question?
Answers use only what is in this story. They are written by AI and can be wrong.