Researcher claims to have used Claude AI to prove the Riemann hypothesis
A researcher at Anthropic claims to have used the chatbot to solve a 160-year-old math problem, though the proof remains unverified.

The short version
- The Riemann hypothesis concerns the distribution of prime numbers and has been unsolved since 1859.
- The researcher used Claude to produce a version of the proof in Lean, a programming language for verifying math.
- The claim involves one of the seven Millennium Prize Problems, which carry a $1 million reward.
- The proof has not been peer-reviewed and requires two years of scrutiny to be officially accepted.
Why it matters
If verified, this proof would confirm the pattern of prime numbers and have significant implications for fields like cryptography. It also marks a potential milestone in using AI to solve high-level theoretical problems that have stumped humans for centuries.
The 160-year-old mystery
A researcher at the artificial intelligence company Anthropic has announced a claim that could reshape the world of pure mathematics. He asserts that he used the company's Claude chatbot to solve the Riemann hypothesis, a problem that has remained unsolved for more than 160 years. First proposed in 1859 by German mathematician Bernhard Riemann, the hypothesis is widely considered the most important unsolved problem in the field.
The hypothesis focuses on the distribution of prime numbers and how frequently they occur along the number line. Specifically, it suggests that every non-trivial complex-number solution to the zeta function has a real component of 1/2. If this is true, all these solutions would align on a single straight line known as the critical line when plotted on a complex plane.
Proving this theory would provide a definitive way to calculate the distance between a prime number p and its successor, showing it is approximately the square root of p times the natural logarithm of p. Because prime numbers are fundamental to many systems, a confirmed proof could have far-reaching implications for various fields, including cryptography. Riemann spent seven years working on the problem before his death in 1866 at the age of 39.
AI and the Lean language
The researcher, Levent Alpöge, released a four-page document on Friday detailing his work with Claude. He claims to have discovered a proof for the Riemann hypothesis as well as a broader version known as the generalized Riemann hypothesis. To achieve this, he utilized Claude to generate a version of the proof in Lean, which is a programming language designed to allow computers to verify mathematical arguments.
Alpöge noted that a full written explanation of the proof would be incredibly long and admitted that he has not yet mastered all the mathematical fields involved in the work. For now, he is only releasing the version produced in Lean. He intends to produce a properly organized written version in the future to make the complex logic more accessible to the human community.
The use of Lean is significant because it provides a way to check the validity of a proof through code. By using an AI to generate the logic in this specific language, Alpöge is attempting to bridge the gap between human intuition and machine-verifiable certainty. However, the proof remains in a preliminary state and has not yet been reviewed by outside experts.
Expert reactions and scrutiny
The announcement has sparked significant interest and some skepticism among the mathematical community. Brian Conrey, a leading expert on the Riemann hypothesis, expressed surprise at the news. He noted that he had not yet seen a manuscript of the proof and had only seen the material posted on Alpöge's website. Conrey described the situation as a monumental moment but emphasized that the proof looks seriously complicated.
Conrey suggested that while the AI-assisted proof might be complex, there will likely be simpler ways to achieve the same result. He also noted that he did not see the claim coming at all. Despite the excitement, the scientific community remains cautious because the proof has not been independently verified by other mathematicians or institutions.
The researcher acknowledged that the timing of his announcement might raise suspicions, though he did not provide a specific reason for the timing. He did, however, mention that he consulted Anthropic CEO Dario Amodei before making the public announcement. Anthropic has not yet provided a formal response to requests for comment regarding the researcher's use of their chatbot.
The path to verification
To be officially recognized as a solution, the proof must meet the strict criteria set by the Clay Mathematics Institute. This institute designated the Riemann hypothesis as one of seven Millennium Prize Problems in 2000, offering a $1 million prize for each one solved. The rules require that a solution be published in a peer-reviewed journal and undergo at least two years of expert scrutiny.
This process ensures that any claim of a solution is thoroughly vetted by the global mathematical community before it can be accepted as fact. Because the Riemann hypothesis is so significant, the verification process is expected to be rigorous and lengthy. Until then, the claim remains an unverified assertion rather than a confirmed mathematical fact.
This story follows a similar trend in the intersection of AI and mathematics. Only two months ago, OpenAI announced that an AI model had solved another Millennium Prize Problem involving equations that model fluid motion. While that proof appeared correct, it also required comprehensive review. The current claim by Alpöge highlights a growing trend of using large language models to tackle some of humanity's most difficult theoretical questions.
What to watch
- Whether Alpöge releases a full written manuscript of the proof for peer review.
- Whether independent mathematicians can verify the Lean code provided by the researcher.
- How Anthropic responds to the claims regarding the use of Claude for this proof.
Try it
Look up the 'Millennium Prize Problems' to see the other six unsolved math problems that offer a $1 million reward.
Sources and links
- ReportingYnetnews: the article we sourced
First reported by Ynetnews. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.
Liked it? Say why
Add a line for other readers.
No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.