AI NewsModels & agentsAnnouncement
Mathematician Levent Alpöge claims Riemann Hypothesis proof with Claude
A mathematician at Anthropic says he has a proof of the Riemann Hypothesis. He worked it out with Claude.

Why it mattersThe claim is not yet verified. The Lean files are on his site, so anyone can check the proof by machine.
Levent Alpöge is a mathematician at Anthropic. He posted a short note on his own site this week. He says he has a proof of the Riemann Hypothesis. He says he worked the proof out with Claude. The note lives at alpo.ge/remark.pdf. The Lean formalization sits beside it. It is in 14 parts, from alpo.ge/lean.tar.gz.part01 to part14.
The Riemann Hypothesis is a 150 year old problem in number theory. It is one of the seven Clay Mathematics Institute Millennium Prize problems. Each one carries a one million dollar prize.
Alpöge writes in the note: "I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis."
He says the method also proves the Generalized Riemann Hypothesis. He says it should extend further with more time. He writes that he and Claude have formalized the arguments without assumptions.
Who he is
Alpöge is a Junior Fellow at Harvard's Society of Fellows. His own site at alpo.ge lists him as working at Anthropic. His academic work is in number theory. He has published papers on Hilbert's tenth problem and integral points on curves. He names colleagues at Anthropic in the note. These include David Dohan, Ben Michel, Ralph Furman, Eric Price and Cora Vinther. He says he messaged Dario Amodei, Anthropic's CEO, before announcing. He wanted to check the timing.
What the note contains
The note is seven pages. It describes the key step. The step is a chain of theta lifts. The chain has a length of roughly the square root of X. Here X is the length of the von Mangoldt sum. Alpöge says the argument builds on Stepanov's method for the Riemann Hypothesis over function fields.
He writes: "The argument is incredibly long at the moment." He says the formalization is finished. He also gives a rough timeline. The first entries sit in early October 2026. The entry labelled "GRH" is dated 11/15/26.
No other mathematician has checked it
No other mathematician has verified the proof. Anthropic has published nothing about the claim on its blog or in its newsroom. The note is Alpöge's own announcement, on his own site. He writes that he is presenting a complete argument before he has fully mastered the mathematics around it.
The Hacker News thread about the paper carried 1,230 points in its first day. A GitHub user has mirrored the Lean files as a public repository. Any mathematician can clone it and read the formal proof.
For a team building software with model-written code, this is a working example of one method. Write the argument. Formalize it in a tool that checks every step by machine. Post the machine-checked files beside the note that reads in plain prose. Whether this proof holds is for the number theory community to decide. The pattern is now in front of a wider audience than it has ever had.
Source
The remark on alpo.ge, by Levent Alpöge.
This item was written by an AI system from the linked source. Reveneau is responsible for what it publishes.
Get AI News in your inbox
New developer tools, model and agent releases, and how teams are actually using them to release software. Short, and only when there is something worth reading.


