
Independently written programs accepted the argument in Levent Alpöge’s proof files, according to tests described by MIT’s Andrew Sutherland and reported by Nauka w Polsce on 21 November. Researchers separately compared the theorem in the files with the standard statement of the Riemann hypothesis. That distinction matters: every step can follow from the one before it and still establish a slightly different theorem from the one the author intended.
On 20 November, Alpöge released a four-page note and a set of files containing his argument in the formal language Lean. He said he had found a proof with the help of the AI model Claude. The files let other mathematicians start testing that day whether the conclusion followed from the stated assumptions. Suppose those assumptions were just a little more restrictive than the standard ones: a valid proof would then cover only some of the intended cases. That is a reason to compare the statements, not a suggestion that Alpöge made such an error.
There was a practical benefit to releasing the files early: mathematicians got an argument they could test without waiting for a paper. Nauka w Polsce reports that the test results are public and others can repeat the checks. Repeating them serves a different purpose from mathematical peer review. A program can follow all the deductions without identifying the idea that made them possible or showing how to use it on another problem. Alpöge himself said he wanted to understand the reasoning better before writing it up.
Sources
- Nauka w Polsce · Fundacja PAP — Źródło: Nauka w Polsce (Fundacja PAP) (https://naukawpolsce.pl/) · Source date: 2026-11-21
Independent editorial content. WUA is not affiliated with any political party. 2026-11-21