Riemann Hypothesis Proof Claimed With Claude and Lean

Chalkboard of equations illustrating the Riemann hypothesis proof claimed with AI

A mathematician at Anthropic says he has solved the Riemann hypothesis, and the machine-checkable file behind his Riemann hypothesis proof already compiles. Levent Alpöge published the result on Friday, the same week his employer became the first publicly traded frontier AI lab.

The timing was awkward, and he said so. In an essay announcing the work, he called it “horrible” and described messaging Anthropic CEO Dario Amodei late last Sunday to ask how to announce a Lean-compilable solution.

What the Riemann hypothesis says

Bernhard Riemann posed the problem in 1859. It predicts that prime numbers follow a hidden pattern, and it sits on the list of Millennium Prize problems, each worth $1 million.

Mathematicians have tested the claim against enormous amounts of data without finding a counterexample. A full proof has never existed until now, if this one holds up. Many results in number theory are written as “assuming the Riemann hypothesis is true,” so a confirmed proof would turn a long list of conditional results into settled ones.

Why Lean matters here

This is what separates the Riemann hypothesis proof from earlier claims, which relied on peer review alone.

Lean is an open-source programming language for expressing mathematics as code. If a proof compiles, the software has checked every logical step, so nobody has to trust the author or read thousands of pages first.

Several people in a public Lean discussion thread report that Alpöge’s file builds. MIT mathematician Andrew Sutherland is among those who confirmed it. Most participants appear confident the formal result will stand.

The caveat is comprehension. A compiled file proves the statement is true, but it doesn’t explain why. Alpöge has said he feels uneasy about releasing a proof without an explanation people can follow, and he has committed to working on one.

How the mathematicians are reacting

The discussion thread has become a live seminar. Chris Birkbeck, a number theorist at the University of East Anglia, posted that the key bound appears to come from an automorphy-lifting argument. He listed the machinery involved, including Galois representations attached to automorphic representations over CM fields, Deligne-Ribet, and results by Clozel, Jacquet-Shalika and Bernstein-Zelevinsky.

Roughly 20 readers answered that post with a mind-blown emoji. For non-specialists, the takeaway is simple: the argument draws on a wide stretch of modern number theory rather than a single trick.

A childhood obsession and a model that “danced”

Alpöge wrote that he has been fixated on the problem since childhood. He said he “could not live” if he waited to release the finding. He described the search as a stream-of-consciousness trip through confusion with Claude, which he said “flew, and danced, and sparkled” near the finish line.

That tone drew attention on its own. The proof came from heavy collaboration with an AI model, and the author works for the company that makes it.

Two more results the same week

Alpöge and NYU mathematician Tristan Buckmaster also released a proof related to the Navier-Stokes equations. Those equations helped start this summer’s race between AI labs and mathematicians over unsolved problems.

Separate from that, a human-readable proof of finite-time blowup for the unforced three-dimensional Euler equations appeared, co-written with Matei Coiculescu and formalized in Lean. The Navier-Stokes paper is readable, but it needs more work before it becomes a polished proof. The Riemann file, by comparison, is huge and exists only in Lean for now.

What happens next

Verification by compiler is nearly done. The slower job is human digestion: experts have to extract the strategy, check that the Lean statements match the real hypothesis, and turn the result into a paper people can teach from.

Alpöge is around to help with that, and has said he wants to, which makes the process easier than it would be for an anonymous upload. Another group has also posted a rival, human-written attempt, so the Riemann hypothesis proof may end up with more than one claimed route. Until a readable explanation exists, the claim stands as a verified file with its mathematics still being decoded.

Source: TechCrunch

Written by

Saksham Bhargava

Saksham Bhargava runs Alteroid and has been writing about consumer technology for over five years, covering iPhone, Android, Windows, AI tools and gaming. With an MCA and six years of hands-on web development experience, he tests the features he writes about on his own devices and focuses on clear, practical guides that solve real problems. Away from the screen, he enjoys travelling to places most tourists skip.