Riemann hypothesis proof claimed with Claude: what Alpöge says he found
An Anthropic mathematician says he and Claude proved math's most famous open problem. We explain the claim, the Lean code and what still has to be checked.

Levent Alpöge, a mathematician who works at Anthropic, says he has found a proof of the Riemann hypothesis with help from the company’s AI model, Claude. The claim is not confirmed: so far there is only a short note on his website and a computer-checkable version of the proof, and mathematicians say it will take time to understand.
“I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis,” Alpöge wrote in a four-page note posted on Friday, 20 November. He says the same argument also settles the generalized Riemann hypothesis, a wider version of the problem.
What is the Riemann hypothesis?
In 1859 the German mathematician Bernhard Riemann made a guess about where the zeros of a special function, the zeta function, sit. If he was right, it tells us how prime numbers (2, 3, 5, 7, 11 and so on) are spread out among all numbers. Nobody has been able to prove it in 167 years.
It is one of the seven Millennium Prize Problems of the Clay Mathematics Institute, which offers $1 million for a solution. A huge number of other results in number theory start with the words “assuming the Riemann hypothesis is true”, so a proof would turn many of those into settled facts.
What did Alpöge actually release?
Not a normal paper. He released the proof written in Lean, a programming language in which a computer checks every step of an argument. According to his note, quoted by AFP, a full written version would be extremely long, he has not yet mastered all the mathematics in it, and he plans to write it up properly.
That is unusual. Normally a mathematician writes the human argument first and, sometimes, someone formalises it in Lean later. Here it is the other way round.
Andrew Sutherland, a mathematician at MIT, told Scientific American he had compiled the Lean proof himself and called it “a human triumph, and a day for celebration”, while saying that understanding it “is going to take some time”. Experts still have to confirm that the hypothesis was stated correctly in the code: a Lean proof is only as good as the statement it proves.
What other mathematicians said
Brian Conrey, one of the best-known specialists on the problem, told AFP it was “a monumental moment in human history”, though he said he had only seen Alpöge’s posted remarks and had asked for a copy. He added that he did not think any human was about to find this, and expected simpler proofs to follow.
Hector Pasten of the Pontifical Catholic University of Chile told Scientific American that, if correct, the proof could work like a Rosetta Stone linking branches of mathematics. Carlo Pagano of Concordia University said it could open new ways of thinking in number theory, again on the condition that it holds up.
Nestor Guillen, a visiting professor at New York University, praised Alpöge for saying openly that he does not fully understand the proof, which he said showed AI working more like a tool.
Why caution is still needed
Anthropic had not commented when AFP asked. Alpöge said he consulted chief executive Dario Amodei before announcing and admitted the timing, a day after Anthropic’s stock market debut, could be seen cynically.
There is history here too. In August, Anthropic said an unreleased Claude model had improved an old result connected to the hypothesis, while stressing it had not proved it. Alpöge was also part of a credit dispute with OpenAI over fluid-equation work about two months ago; OpenAI denied wrongdoing.
The Clay prize also has rules: a solution must be published in a peer-reviewed journal and survive two years of scrutiny.
Our take: what it means for Pakistan
If the proof holds, it is the strongest sign yet that AI models can do original research, not just answer homework. For Pakistani maths and computer science students, Lean is worth a look: it is free, and checking proofs by machine is becoming part of how serious mathematics is done.
What it does not mean is that AI has replaced mathematicians. Even here, the world is now waiting for humans to read, check and explain what the machine helped produce. Watch for a written paper and for Lean experts confirming the statement in the code.



