In brief
- Anthropic mathematician Levent Alpöge says Claude helped him prove the Riemann hypothesis, along with a broader generalization.
- He says he notified CEO Dario Amodei on Sunday but held the announcement until Friday, following Thursday’s IPO.
- A Lean proof archive is publicly available, though independent verification and any prize recognition remain outstanding.
An Anthropic mathematician says that, working alongside Claude, he has cracked the Riemann hypothesis, a 167-year-old problem carrying a $1 million prize.
His claim emerged Friday, just one day after the AI company’s stock-market debut, and has yet to be independently verified.
“I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis,” Levent Alpöge wrote in a four-page note. He says the approach also settles the broader Generalized Riemann Hypothesis, which extends the problem to other classes of mathematical functions.
Alpöge says he and Claude formalized their argument in Lean, software that checks mathematical proofs step by step. His note links to 14 downloadable components of the proof archive, but making the work available is not the same as acceptance by the mathematical community.
Alpöge says he messaged Anthropic CEO Dario Amodei about a possible announcement “late Sunday night,” November 15. He then consulted company leadership and said there was a constraint to honor, without specifying exactly what it entailed.
“The precise timing is horrible,” he acknowledged. Alpöge pushed back on a cynical reading of the sequence, though he stopped short of saying the company’s IPO was the cause of the delay.
Anthropic began trading on Nasdaq on Thursday, November 19, after raising $81.34 billion in its offering.
The “Mathocalypse”
Alpöge was a central figure in September’s AI mathematics disputes. OpenAI claimed it had solved the Navier-Stokes Millennium Prize problem while he and NYU mathematician Tristan Buckmaster were pursuing related work on fluid equations, triggering a disagreement over who deserved credit for the research.
OpenAI followed in October with 722 manuscripts generated by an unreleased model, organized into 372 families of results. Only 162 contained Lean-verified main results, and mathematicians called for stronger evidence rather than taking the company’s sweeping claims at face value.
Some mathematicians worry about a coming “mathocalypse,” in which AI solves prized problems before researchers can learn from working through them. Toronto mathematician Daniel Litt cautioned that, absent adaptation, “there’s just no more math in 50 years.” At an MIT panel this week, Buckmaster argued that mathematics still depends on human understanding.
What a proof would mean
First proposed by Bernhard Riemann in 1859, the hypothesis predicts where certain zeros of the Riemann zeta function lie. The question carries weight because those zeros are tied to the distribution of prime numbers, whose patterns underpin a vast body of mathematical research.
A proof would rank among the most significant mathematical achievements, given that experts have pursued it for well over a century.
Lean can confirm that formal deductions follow its rules. Mathematicians would still need to reproduce the archive, verify that the formalized theorem matches the original conjecture, and scrutinize the claimed result before treating it as settled.
Anthropic has previously reported progress on the zeta function without asserting a solution to the full hypothesis. Alpöge’s new claim extends well beyond that earlier work.
The Clay Mathematics Institute counts the Riemann hypothesis among its Millennium Prize Problems, each carrying a $1 million award. Under the institute’s rules, a proposed solution must be published in a qualifying venue, withstand at least two years of scrutiny, and achieve general acceptance before the prize can be awarded.


