A claim that artificial intelligence helped solve one of mathematics’ most famous problems has triggered an extraordinary verification effort. On November 20, 2026, Anthropic mathematician Levent Alpöge announced that he and Claude had developed a proof of the Riemann hypothesis, a question open since 1859.

Has Claude really proved the Riemann hypothesis, and what do the independent checks establish? The early results are substantial, but the distinction between checking computer code and understanding a mathematical breakthrough matters.

What did Levent Alpöge announce?

Alpöge published a four-page note on his personal website on Friday, November 20. He said he believed he had discovered a proof with Claude, Anthropic’s AI system, and that the method also established the generalized Riemann hypothesis, an extension involving other mathematical functions.

Instead of a conventional mathematical paper explaining each step, he released an archive of roughly 17 million lines of Lean code. Lean is a proof assistant: its small verification kernel checks whether formally expressed arguments follow the specified rules.

Alpöge acknowledged that the argument is exceptionally long and said he intends to develop a clearer mathematical explanation. His announcement therefore raises two questions: whether the machine-readable proof checks out, and what the underlying reasoning actually reveals.

Have independent mathematicians verified the Lean proof?

Several independent technical checks have succeeded. Reporting by Quanta Magazine on November 20 identified successful compilations by Andrew Sutherland at MIT and Simon DeDeo at Carnegie Mellon University, including a check using a variant of Lean.

A public independent verification project also reported that the complete argument was accepted by two different proof-checking kernels and that its stated theorems matched separately specified mathematical statements. Those findings address critical concerns: whether the files can be checked independently and whether the program proves the proposition it claims to prove.

The scrutiny is more demanding than asking an AI chatbot to assess its own answer. Researchers must inspect possible hidden assumptions, definitions, dependencies and the behavior of the checking software. Different checks answer different questions, and technical success does not immediately produce a human-readable understanding of the result.

What is the Riemann hypothesis, in plain English?

Proposed by Bernhard Riemann in 1859, the hypothesis describes where the nontrivial zeros of the Riemann zeta function lie. It predicts that each has real part one-half.

The claim matters because the zeta function is closely linked to prime numbers. Mathematicians already understand the average distribution of primes; the Riemann hypothesis would provide exceptionally strong control over how their actual distribution differs from that average.

The generalized hypothesis extends similar ideas to a wider family of functions. Establishing both would have consequences throughout number theory, although it would not automatically turn every speculative application into a working technology.

Is the $1 million Millennium Prize now settled?

No prize has been awarded on the basis of this announcement. As of November 21, the Clay Mathematics Institute still lists the Riemann hypothesis among its unsolved Millennium Prize Problems.

Its published rules require a proposed solution to appear in a qualifying outlet, receive general acceptance among mathematicians and remain published for at least two years before the institute will consider it. A successful formal proof check is a major milestone; the prize procedure follows its own timetable.

What happens next?

The immediate work is to reproduce and audit the Lean checks, investigate the precise mathematical statements and turn the enormous formal argument into an explanation specialists can evaluate and build on.

There is another development to watch: Quanta reported that a separate, human-written draft addressing the generalized Riemann hypothesis by mathematician Alessandro Masullo was also made public on November 20. That manuscript faces its own independent scrutiny.

The strongest conclusion on November 21 is that Alpöge’s Claude-assisted claim has passed important independent computational tests. Whether it becomes an accepted, comprehensible landmark in mathematics depends on continued verification and expert examination.