SIAgentNews.com
← All news

Anthropic mathematician claims to have solved the Riemann hypothesis with Claude

Levent Alpöge says he used Anthropic's AI to crack a famous prime number puzzle, potentially marking a major shift in how humans and machines solve math.

· First reported by The Verge

$1 millionMillennium Prize bounty for each solved problem

The short version

  • Alpöge claims his method also resolves the Generalized and Extended Riemann Hypotheses.
  • The solution was presented in Lean, a formal language for verifying mathematical proofs.
  • The Clay Mathematics Institute requires two years of scrutiny before awarding the $1 million prize.
  • Alpöge previously claimed to have solved the Jacobean conjecture using Claude in July.

Why it matters

If verified, this would be a landmark moment for AI's ability to perform high-level reasoning. It signals a shift toward a landscape where AI agents may soon handle the heavy lifting of complex mathematical proofs.

A claim of a major breakthrough

Levent Alpöge, a mathematician at Anthropic, has announced that he believes he has discovered a proof for the Riemann hypothesis. This specific problem is one of the seven Millennium Prize problems, which are considered some of the most difficult open challenges in the field of mathematics. The hypothesis, first proposed in 1859, focuses on the distribution of prime numbers and the behavior of the zeroes of a specific function.

Alpöge states that the AI model Claude was integral to developing this proof. He claims that the method he used is not limited to just this one problem, as it easily extends to the Generalized Riemann Hypothesis and the Extended Riemann Hypothesis. He even went as far as to formalize the arguments without any assumptions using a tool called Lean.

The announcement has come at a time of significant interest in AI's mathematical capabilities. Alpöge reportedly texted Anthropic CEO Dario Amodei about the solution late one Sunday night. However, he chose to wait until after the company's IPO on Thursday to make the public announcement. Anthropic has not officially commented on the specific claim or involved itself in the announcement at this time.

The path to verification

While the claim is significant, the solution has not yet been independently verified by the broader scientific community. Alpöge has published the proof on his own website, but it has not been accepted by the Clay Mathematics Institute. This institute is the body that administers the Millennium Prize and offers a $1 million bounty for each problem that is successfully solved.

The rules set by the institute are very strict regarding the validation of new proofs. A proposed solution must be published and then withstand at least two years of intense scrutiny by other experts. During this period, the solution must receive general acceptance within the global mathematics community before it can be officially recognized. This process ensures that the complexity of the math is fully understood by peers.

To date, only one of the seven Millennium Prize problems has been officially solved. That achievement was the Poincaré conjecture, which is a topological problem involving three-dimensional spheres. The remaining problems, including the Riemann hypothesis, remain some of the most formidable and elusive targets for mathematicians worldwide.

AI's growing role in math

This claim follows a series of recent events where AI has been used to tackle high-level mathematical concepts. In September, OpenAI claimed to have cracked the Navier-Stokes problem, which deals with the flow of fluids. That achievement was notable because OpenAI deployed 10,000 agents and spent millions of dollars to pursue the solution after hearing that rival researchers were close.

The Navier-Stokes result is considered credible by many, though it has faced criticism regarding research practices. There were allegations that the AI used unpublished work from mathematicians who had shared their research with the tools. OpenAI has strongly denied any wrongdoing, but the controversy has created a lot of noise around the use of AI in formal research.

Alpöge's work with Claude is part of a broader trend of mathematicians using AI to navigate a changing landscape. He previously stated in July that he solved the Jacobean conjecture using Claude as well. These frequent findings are forcing the mathematical community to figure out how to digest and categorize new discoveries made with machine assistance.

How the process works

The use of Claude in this process involves a collaborative effort between human intuition and machine processing. Alpöge worked on the problem alongside New York University professor Tristan Buckmaster. They utilized the AI to explore complex paths, similar to how Buckmaster and Alpöge previously worked with OpenAI's Codex and Claude on the Navier-Stokes problem.

By using Claude, the researchers can explore vast amounts of data and logical permutations much faster than a human could alone. The AI helps to identify patterns in prime numbers and the zeroes of functions, which are the core components of the Riemann hypothesis. This allows the human mathematician to focus on the overarching logic and the final structure of the proof.

The use of Lean is a critical technical step in this specific announcement. Lean is a formal proof language, which means it allows mathematicians to write out their arguments in a way that a computer can check for errors. By formalizing the arguments without assumptions, Alpöge is attempting to provide a level of certainty that is highly valued in the mathematical community.

What comes next for the field

The next step for Alpöge's work is the period of intense peer review. Because the Riemann hypothesis is so important, the mathematical community will need to scrutinize every step of the proof to ensure there are no logical gaps. If the proof holds up over the next two years, it could be one of the most significant mathematical achievements of the century.

This situation also highlights a growing tension between traditional research and AI-assisted discovery. Mathematicians are currently scrambling to understand how to integrate these new tools into their workflows. The rapid pace of these findings suggests that the landscape of mathematics is becoming profoundly different as AI becomes a standard collaborator.

For the public, this story marks a milestone in the capabilities of large language models. It moves the conversation beyond simple chat interactions and into the realm of solving problems that have stumped the smartest minds for many decades. Whether the proof is eventually accepted or not, the collaboration between Alpöge and Claude has already changed the conversation about what AI can do.

What to watch

  • Independent verification of the proof by the global mathematics community.
  • Whether the Clay Mathematics Institute will officially recognize the solution.

Sources and links

First reported by The Verge. 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.

More AI news