Riemann hypothesis has puzzled scientists for 167 years. Has AI solved it?

Screenshot 2026-11-22 091359
AI Generated Image
  • +
  • -
A mathematician working at the US company Anthropic has drawn the attention of the scientific community after announcing that, with the help of its AI model Claude, he had arrived at a proof of the Riemann hypothesis, one of the most complex mathematical problems that has eluded scientists since 1859. At the same time, another proof, developed by a human researcher without the help of AI, has emerged, and scientists are awaiting the results of peer review that will determine whether either discovery holds up.اضافة اعلان

The mathematician Levent Alpöge, who works at the US company Anthropic, announced that he believes he has proved the Riemann hypothesis, one of the most prominent unsolved problems in mathematics, in collaboration with the AI model Claude.

Alpöge revealed his results in a four-page document published on 20 November 2026, explaining that he had arrived at a proof of the question that has puzzled generations of mathematicians for 167 years.

In his announcement, he said: "I believe that, in collaboration with Claude, I have arrived at a proof of the Riemann hypothesis."

However, this announcement does not mean the question is definitively settled. The proof still needs rigorous scientific review to confirm its validity, and how far the researcher's results match the original hypothesis.

According to the journal Nature, verifying the proof could represent a historic achievement in mathematics, especially since the hypothesis relates to understanding the distribution of prime numbers, one of the central themes of number theory.

What is the Riemann hypothesis that has puzzled scientists?
The Riemann hypothesis dates back to 1859, when the German mathematician Bernhard Riemann put forward a mathematical idea for understanding the patterns that govern the distribution of prime numbers.

Prime numbers are whole numbers greater than one that are divisible only by themselves and by one, such as 2, 3, 5, 7 and 11.

These numbers are the building blocks of positive whole numbers, since every whole number greater than one can be expressed as a product of prime numbers.

Despite the importance of prime numbers, their appearance among natural numbers does not follow a simple pattern that makes it easy to determine the position of the next prime number.

The Riemann hypothesis therefore proposes that there is a precise mathematical structure governing the deviations in the distribution of these numbers from their expected average distribution.

Riemann relied on a mathematical function known as the "Riemann zeta function", represented by the Greek letter zeta, which is used to study the properties of prime numbers and their mathematical relationships.

He found that the distribution of prime numbers is closely linked to the points at which the value of the function becomes zero, known as the "zeros of the zeta function".

His hypothesis states that all the non-trivial zeros of this function lie on a specific line in the complex plane, where the real part of the number equals one-half, known as the "critical line".

Although computers have been used to check enormous numbers of these zeros, there has never been an accepted mathematical proof showing that the hypothesis holds in every case without exception.

17 million lines of code to prove the hypothesis
In developing the proof, Alpöge relied on the Claude model, alongside the Lean system, an open-source tool that allows mathematical proofs to be written and checked for logical validity using a computer.

According to a report published by TechCrunch, the researcher submitted around 17 million lines of code written in the Lean language, which he said contain a formal proof of the hypothesis, and even extend to a broader version known as the "generalised Riemann hypothesis".

This technique makes it possible to check the logical steps of a proof automatically, rather than relying exclusively on traditional human review.
The report noted that a number of specialists, including the mathematician Andrew Sutherland of the Massachusetts Institute of Technology, were able to confirm that the code translated and ran successfully within the verification system.

However, successfully running the code is not enough on its own to declare the problem solved conclusively. Experts must confirm that the proof does in fact establish the required mathematical statement, and that it does not rely on unjustified assumptions or on definitions that change the meaning of the result.

Researchers also face another challenge, which is understanding the proof and explaining it in clear mathematical language that specialists can review and absorb.

A divided mathematical community
The announcement has drawn mixed reactions in the scientific community, ranging from optimism about the possibility of an unprecedented breakthrough to scepticism about whether the evidence published so far is sufficient.

Number theorist James Maynard, a Fields Medal winner and professor at the University of Oxford, said that if the proof is fully confirmed it would be a historic achievement in mathematics.

By contrast, mathematician Andrew Granville of the University of Montreal in Canada criticised the way the results were presented, arguing that the document published by Alpöge lacks the clarity needed to review a discovery of this size.

Quanta magazine noted that the use of automated verification systems opens a debate about the difference between proving that a mathematical result is logically correct and understanding the deeper reasons that make it true.

Some researchers argue that a mathematical proof is not only about establishing a result, but must also offer an explanation that helps scientists build new knowledge and develop other theories.

A human proof that rivals AI
Alongside Alpöge's announcement, the mathematical community saw another development, after mathematician Ravi Vakil, professor at Stanford University and president of the American Mathematical Society, published a draft proof of the generalised Riemann hypothesis prepared by researcher Alessandro Massolo without the help of AI.

Vakil explained that Massolo had worked on his research for 12 years, and that mathematician Brian Conrad had reviewed parts of it over several months, providing comments that are being incorporated.

The plan had been to continue developing and reviewing the proof before publishing it, but rapid developments led the researchers to make the draft available early.

Vakil described the work as entirely human, confirming that AI was not used in its development.

However, in an update published on 20 November, he clarified that a number of researchers had informed him of possible problems in the proof, identified with the help of various AI models.

He stressed that these objections need to be addressed and reviewed before the results can be accepted scientifically.

The human proof is therefore also still under evaluation, with no final confirmation of its validity.

Why does proving the Riemann hypothesis matter?
The importance of the Riemann hypothesis extends beyond number theory, as it is linked to many mathematical results and estimates that rely on an understanding of the distribution of prime numbers.

Proving it could settle mathematical results that are conditional on its truth, and open the way to developing new tools and methods in different areas of mathematics.

Prime numbers are also linked to important applications in computer science and digital cryptography, including the systems used to protect financial transactions and electronic communications.

However, proving the Riemann hypothesis does not necessarily mean it would be possible to predict every new prime number directly, or to break the encryption systems in use today, since these applications depend on a range of mathematical problems and algorithms.

The main value of a possible proof lies in providing a more precise understanding of the mathematical structure that governs the distribution of prime numbers.

A million-dollar prize awaits a solution
Given the difficulty and importance of the Riemann hypothesis, the Clay Mathematics Institute included it in 2000 on its list of the "Millennium Prize Problems", which comprises some of the world's most prominent mathematical challenges.

The institute set aside a prize of one million dollars for the solution of each problem, subject to its verification and acceptance criteria.

The Riemann hypothesis remains listed among the unsolved problems on the institute's official website, despite the two recent announcements.

This means the scientific community has not yet reached final acceptance of either proof, and that receiving the prize requires meeting the institute's approved verification procedures.

Is AI entering a new era in mathematics?
These developments come at a time when AI models are increasingly contributing to scientific research, particularly in mathematics, programming and the analysis of complex problems.

Alpöge described his experience with Claude as a journey full of challenges and confusion, noting that the model showed capabilities that exceeded his expectations during the final stages of developing the proof.

He also acknowledged that he published the results before he had been able to absorb all their mathematical details, pledging to work on providing a clearer explanation for researchers.

On the other hand, the emergence of Massolo's proof highlights the continuing role of human scientists in tackling complex mathematical problems, whether independently or in collaboration with modern technologies.

While the scientific community awaits the results of the review of the two proofs, the Riemann hypothesis remains one of the biggest open questions in mathematics.

If either proof is confirmed, it could mark a turning point in the history of mathematics, not only because it would solve a problem that eluded scientists for 167 years, but also for what it may reveal about the future of the relationship between artificial intelligence and scientific discovery.