Anthropic Mathematician Claims Riemann Proof in 17M Lines of Code

Levent Alpöge, a mathematician at the AI company Anthropic, announced he has discovered a proof for the Riemann hypothesis recorded in approximately 17 million lines of code.

Levent Alpöge, a mathematician at the AI company Anthropic, announced he has discovered a proof for the Riemann hypothesis with the assistance of the Claude model.

This 167-year-old mathematical problem, first formulated by Bernhard Riemann in 1859, concerns the distribution of prime numbers and remains one of the most significant unsolved challenges in the field, as reported by Seznam Zprávy. The announcement occurred during the same week that Anthropic debuted on the Nasdaq stock exchange, according to Seznam Zprávy.

Alpöge released a four-page outline and 14 archives containing source code in the Lean 4 programming language, according to Novinky. The digital files comprise 175,187 files amounting to 239 MB, which in readable form exceeds a quarter of a million pages, Deník N reported. The proof, which is recorded in approximately 17 million lines of code, also purports to solve the generalized variant of the hypothesis, Seznam Zprávy noted.

“Žádný text důkazu v podobě odborného článku srozumitelného lidským čtenářům však zatím nebyl zveřejněn,” AIWeekly said.

“Věřím, že jsem ve spolupráci s Claudem objevil důkaz Riemannovy hypotézy,” said Levent Alpöge, a mathematician at Anthropic, in a statement cited by Seznam Zprávy. He added, “Ale to, co jsem po cestě viděl, mne ohromilo natolik, že teď věřím v novou kapitolu napříč vědou. Je to úsvit něčeho mnohem většího.”

Anthropic Mathematician Claims Riemann Proof in 17M Lines of Code
Photo: Seznam Zprávy

Verification of the 17 million lines

The mathematical community has begun the rigorous process of auditing the massive codebase. Mathematician Andrew Sutherland at MIT has successfully compiled the proof in the Lean system, though he cautioned that the details require extensive further examination, according to Forbes Česko.

Experts are currently verifying the consistency of definitions, the structural purity of the argument, and translating the machine code into human-readable interpretation, including a comparison with a 940-page human manuscript by Alessandro Masullo from Stanford, according to Wired CZ.

While some initial automated checks suggest the solution is likely correct, the complexity of the code makes it difficult for human mathematicians to read directly, Seznam Zprávy noted. Some experts have already expressed skepticism regarding the clarity of the initial documentation. For instance, mathematician Andrew Granville described the four-page outline as “zmatečný” (confusing), according to Novinky.

Timeline for the Clay Mathematics Institute prize

The Clay Mathematics Institute, which manages the Millennium Prize Problems, requires a formal publication in a recognized journal before the proof can be officially considered. Once published, a two-year period follows during which the global mathematics community will review the work, followed by an evaluation by a special commission, Seznam Zprávy reported.

Martin Bridson, president of the institute, said that while the announcement is of historical significance, the path to full understanding will be “dlouhá a obtížná,” as reported by Deník N.

Questions about the Riemann proof

Is the Riemann hypothesis now officially solved?

No. While Alpöge has presented a proof, it remains subject to a multi-year audit by the international mathematical community. The Clay Mathematics Institute continues to list the problem as one of the unsolved millennium challenges, Deník N reported.

Anthropic Mathematician Claims Riemann Proof in 17M Lines of Code
Photo: Deník N

What happens if the proof is confirmed?

A confirmed proof could have profound implications for number theory and cryptography, potentially impacting how large numbers are factored, according to Seznam Zprávy. Furthermore, it would validate the use of AI as a tool for solving complex, long-standing scientific problems. Mathematician Carlo Pagano of Concordia University expects the new ideas in the procedure may open further research possibilities in number theory, as reported by Forbes Česko.

Born in 1992, Levent Alpöge is a mathematician specializing in number theory and arithmetic statistics who holds dual citizenship in the United States and Turkey. Before his current role at Anthropic, he was a junior fellow in the Harvard Society of Fellows and earned his PhD from Princeton University.

His career has been marked by significant mathematical contributions, including establishing bounds on integers expressible as the sum of two rational cubes and providing an alternative proof for the negative solution to Hilbert’s tenth problem over algebraic number fields. His influence in the field is such that the main-belt asteroid 25898 Alpöge is named in his honor.

Did Alpöge solve any other problems?

Yes. Along with colleague Tristan Buckmaster, Alpöge announced a potential solution to the “Navier-Stokes” problem, another millennium challenge, using an AI assistant, Seznam Zprávy reported. The Riemann hypothesis was originally formulated by German mathematician Bernhard Riemann (1826–1866) in 1859.