science
What We Know About the Claimed Riemann Hypothesis AI Proof

A proof posted Friday runs 17 million lines of code and claims to settle a question that has stood unresolved for 167 years, according to Al Jazeera. No outside mathematician has yet read the full argument.
Levent Alpoge, a member of Anthropic's technical staff, wrote on his personal website that he believes he has "discovered a proof of the Riemann Hypothesis" working with Claude, the company's AI model, Al Jazeera reported. He specified the result covers the generalised Riemann hypothesis, a broader version of the original 1859 conjecture.
What did Alpoge actually publish?
Alpoge released a four-page explanation rather than a conventional journal paper, Al Jazeera reported. The underlying argument was written in Lean, a programming language and proof assistant that checks mathematical reasoning line by line. Alpoge credited the AI model directly for the late-stage work, writing that "Claude far outpaced me near the finish line," according to the outlet.
What is the Riemann hypothesis, and why does it matter?
First proposed by German mathematician Bernhard Riemann in 1859, the hypothesis describes the distribution of prime numbers — whole numbers divisible only by one and themselves. Primes appear scattered with no obvious pattern, but Riemann's conjecture holds that a hidden order governs them, encoded in an object called the Riemann zeta function, Abhishek Saha, a mathematics professor at Queen Mary University of London, told Al Jazeera. The problem is one of seven Clay Mathematics Institute Millennium Prize Problems, each carrying a $1m award, per the outlet.
Why hasn't anyone confirmed it yet?
The scale of the proof is the bottleneck. At 17 million lines, no single mathematician has read through it end to end, Al Jazeera reported.
"At the minute, we don't even know why the theorem is true," Kevin Buzzard, professor of pure mathematics at Imperial College London, told Al Jazeera.
Lean's step-by-step verification means the code can be machine-checked for logical consistency even before humans fully grasp the mathematical intuition behind it — a gap Buzzard's comment underscores.
What happens next?
Al Jazeera's report notes independent experts are now working through the argument, but verification could take weeks. The process typically involves mathematicians tracing the Lean code section by section, checking that each step follows the rules of the proof assistant and that the overall logical chain has no gaps.
Timeline
- 1859 — Bernhard Riemann proposes the hypothesis describing the hidden order of prime numbers.
- 2000 — The Clay Mathematics Institute names it one of seven Millennium Prize Problems, each worth $1m.
- Friday, Nov. 21, 2026 — Levent Alpoge posts his four-page claim and 17-million-line Lean proof online.
- Now — Independent mathematicians, including Kevin Buzzard and Abhishek Saha, begin reviewing the argument; no full read-through has been completed.
Al Jazeera's original report is available here.
Questions
Has the Riemann hypothesis definitely been proven?
No. Anthropic researcher Levent Alpoge has posted a claimed proof, but independent mathematicians have not yet verified the 17-million-line argument, Al Jazeera reported.
What role did AI play in the claimed proof?
Alpoge said he worked with Anthropic's Claude model and that Claude 'far outpaced' him in the final stages of the work, according to Al Jazeera.
Why does the Riemann hypothesis matter?
It describes a hidden pattern governing how prime numbers are distributed and is one of the Clay Mathematics Institute's seven Millennium Prize Problems, each worth $1m, per Al Jazeera.