Levent Alpöge, a mathematician at Anthropic, claimed on November 20, 2026 that he has proved the Riemann hypothesis with help from Claude. The hypothesis has been open since 1859 and is often called the most famous unsolved problem in mathematics. "I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis," Alpöge wrote in a four-page statement on his website.
The proof has not been checked by anyone outside yet. Alpöge published it only as Lean code, a language in which a computer verifies every step, and says no readable paper exists so far. Scientific American broke the story the same morning.
What Alpöge claims
The statement makes three claims. Alpöge says he and Claude found a proof of the Riemann hypothesis. He says the method "easily extends much further" and also resolves the Generalized Riemann Hypothesis. That is a broader version covering a whole family of similar functions. He expects the Extended Riemann Hypothesis to follow "with more time."
He also says that "Claude and I have also formalized the arguments without assumptions." The Lean files are posted on his site as a 14-part archive.
Alpöge describes Claude as the stronger partner near the end. "This time Claude far outpaced me near the finish line," he wrote. "No, Claude flew, and danced, and sparkled near that finish line." He says he steered the work toward the right setting. Claude supplied much of the technical mastery in areas he knows less well, such as automorphic forms on higher-rank groups.
The statement explains why there is no paper. The argument is "incredibly long at the moment," Alpöge wrote, and posting "some massive unreadable travesty of slop" would be inappropriate. A footnote says a full write-up Claude produced ran past 250,000 pages. He plans to study the argument, simplify it and write it up properly.
How the proof came together
The second half of the statement is a timeline that Claude compiled from their sessions. Most steps date from October 6 to November 16, 2026. The work began with a determinant trick on October 6 and ended with the Generalized Riemann Hypothesis on November 16.
The central idea, as Alpöge tells it, adapts a known proof to the integers. Mathematicians proved the Riemann hypothesis for function fields decades ago. One route uses Sergei Stepanov's method with "multiplicity." Over ordinary whole numbers, that multiplicity is hard to find. Alpöge says the breakthrough was chaining theta lifts, a tool from the theory of automorphic forms, so that each lift mimics one step of multiplicity. Claude suggested a chain as long as the square root of the sum's length. Alpöge says that line alone "will suffice to make fellow mathematicians fall out of their chairs."
Alpöge is open about the limits of his own account. Parts of the timeline paraphrase Claude's summaries, which "might make zero sense," he warns, adding that "the Lean should suffice to clarify everything."
The timing is unusual too. The claim came days after Anthropic's IPO. Alpöge addresses this directly. He says the result came together "late Sunday night," that he asked CEO Dario Amodei about announcing it, and that there was "a constraint to respect." He calls the cynical reading of the timing "false."
Why the Riemann hypothesis matters
The Riemann hypothesis is about the zeros of the Riemann zeta function. It says that all the nontrivial zeros lie on one line, called the critical line. That sounds abstract, but the zeros control how prime numbers are spread out. A proof would give the best possible estimate of how many primes there are below any number.
It is one of the seven Millennium Prize Problems, with a $1 million reward from the Clay Mathematics Institute. Computers have checked the first 10 trillion zeros, and all of them sit on the line. Thousands of published results start with "assuming the Riemann hypothesis," so a proof would turn them into theorems at once. The generalized version would do the same for a much larger body of work.
Anthropic has worked on the problem before. In August, the company reported that Claude raised the proven share of zeros on the critical line from 41.6% to 67.2%. That paper, by Alpöge and Ralph Furman, came with a complete Lean proof. At the time, Anthropic wrote that it did not expect those techniques to lead to a proof of the Riemann hypothesis.
What has to happen before the proof counts
Lean makes checking faster, but it does not settle the question by itself. Experts have to confirm that the Lean statement really is the Riemann hypothesis and that the definitions match the standard ones. They also have to check that the code adds no hidden axioms or unproved steps. Mathematician Kit Yates noted that a claimed proof "is not yet an accepted proof" and will face intense scrutiny. Scientific American expects that checking the Lean code "shouldn't take long."
As of November 20, no outside mathematician has publicly confirmed the result, and Anthropic has not published a statement. The Clay Institute's rules add a longer path for the prize. A solution has to appear in a qualifying journal, survive at least two years of examination and gain general acceptance in the mathematics community.
The claim lands in a year full of AI math results. In July, Alpöge posted a Claude-found counterexample to the Jacobian conjecture. In October, Anthropic published an algorithm that refutes the Exponential Time Hypothesis, found by an internal Claude model. Other AI-assisted results this year include the KLS conjecture. OpenAI has also published a wave of math results, including work on Navier–Stokes.
What the Riemann claim means for AI in mathematics
Whatever the checks show, the claim marks a shift in how mathematics gets done. Over the past year, AI has moved from contest problems to research results. Now a working number theorist credits Claude with the decisive steps on the field's central question. The result arrives as machine-checkable code before any readable paper. That moves the first test from a referee reading hundreds of pages to experts confirming that the Lean statement says what Riemann meant. It also opens an unusual gap: a proof may exist before any person fully understands it, Alpöge included. If the code holds up, the hardest work will be turning it into mathematics that people can read and build on.
FAQ
Short answers about the claim.
Who is Levent Alpöge?
Alpöge is a number theorist who works at Anthropic. This year he has posted several results found with Claude. They include a counterexample to the Jacobian conjecture in July and a result on zeros of the Riemann zeta function in August.
Has the Riemann hypothesis proof been verified?
No. As of November 20, 2026, no independent mathematician has confirmed it. The proof exists only as Lean code, and experts still need to check that the code states the Riemann hypothesis correctly and contains no gaps.
What role did Claude play?
According to Alpöge, he guided the overall approach while Claude did much of the technical work and suggested key steps. He wrote that Claude "far outpaced" him near the end, and that the Lean formalization was written together with Claude.
Will Alpöge get the $1 million Millennium Prize?
Not soon. The Clay Mathematics Institute first requires publication in a qualifying journal. The proof then needs at least two years of scrutiny and general acceptance by mathematicians.
Sources
- Statement by Levent Alpöge – November 20, 2026
- Anthropic mathematician claims to prove Riemann hypothesis, most famous problem in all of math, Scientific American – November 20, 2026
- AI may have helped solve the most famous problem in maths, Kit Yates – November 20, 2026
- Claude's work on the Riemann zeta function, Anthropic – August 10, 2026
- More than two thirds of the zeta zeros are simple and on the critical line, Alpöge and Furman, arXiv – August 2026
- Millennium Prize rules, Clay Mathematics Institute
- Kimmonismus on X – November 20, 2026