AI claimed a solution to the 167 year-old Riemann Hypothesis — and a human did, too

Glass Epoch DeskOriginal digestSource: TechCrunch AI

AI-assisted digest, checked against the original source before publishing. How we work

An Anthropic mathematician says he has a Lean-verified proof of the 167-year-old Riemann Hypothesis — while a competing human solution lands the same week.
AI claimed a solution to the 167 year-old Riemann Hypothesis — and a human did, too
Photo: Brett Jordan · free stock license

Levent Alpöge, a mathematician employed at Anthropic, has announced what he describes as a Lean-compilable solution to the Riemann Hypothesis, the conjecture first posed in 1859 that posits a hidden structure beneath the apparent randomness of prime numbers. It is one of the Millennium Prize problems, and the mathematics community is only beginning to pick through the claim. Complicating the moment: a competing human solution emerged alongside it.

Alpöge flagged the awkwardness of the moment directly in the essay announcing his findings. "The precise timing is horrible," he wrote, explaining that he had messaged Anthropic CEO Dario Amodei on Sunday night to ask how best to publicize the result — the same stretch in which Anthropic became the first publicly-traded frontier AI lab. His account of the work reads as much as a collaboration log as a research announcement: a "stream-of-consciousness jaunt" through the argument with Claude, in which the model "flew, and danced, and sparkled near that finish line."

Experts are not taking the claim on faith. On a Zulip chat, mathematicians have been working through the technical core, with Chris Birbeck, a number theorist at the University of East Anglia, describing what appears to be an automorphy-lifting argument that leans on Galois representations attached to regular algebraic, polarized cuspidal automorphic representations of GL_n over CM fields, plus base change, Jacquet–Shalika, Clozel's purity, Bernstein–Zelevinsky, the main theorem of CM, and Deligne–Ribet. Because the proof is formalized in Lean — an open-source language built to machine-verify mathematical arguments — most observers in the thread appear confident the formalized result will hold even as they puzzle over the underlying reasoning.

Alpöge acknowledged releasing the work without a truly human-intelligible explanation, an omission he seemed to feel guilty about. His fixation on the problem since childhood made him say he "could not live" if he delayed publication any further, though he has committed to producing an explanation humans can actually follow and teach one another.

What to watch next is the gap between machine-checkable and human-understandable. The community still needs to translate the Lean-verified argument into prose that mathematicians can defend, extend, and teach — and to reconcile it with the competing human proof that arrived in the same week. Until then, the result sits in an unusual state: formally verified, publicly contested, and largely unexplained.

Key facts

  • • Alpöge works at Anthropic and messaged CEO Dario Amodei about announcing the proof.
  • • The Riemann Hypothesis dates to 1859 and is a Millennium Prize problem.
  • • The claimed proof is formalized in Lean, an open-source proof-verification language.
  • • Anthropic became the first publicly-traded frontier AI lab the same week.
  • • A competing human solution to the same problem emerged alongside the AI-assisted one.

Why it matters

This is the clearest test yet of AI as a research partner: a frontier-lab employee used Claude to reach a machine-verified result on a problem that has stood for 167 years. For builders, the interesting constraint is the verification gap — Lean can confirm an argument long before humans can explain it, so expect formal verification tooling and human-readable proof generation to become a real produ

Spotted an error? Send a correction.

Get the day's AI news in one email

The Daily Epoch · 60-second read · unsubscribe anytime

Related