Claude, An Anthropic Mathematician, And The Riemann Hypothesis: A Claim Under Review
AIThis post was created with the assistance of artificial intelligence (AI).

TL;DR

Anthropic mathematician Levent Alpöge posted a note on November 20, 2026, saying he believes he and Claude proved the Riemann Hypothesis and its generalized version. The proof is an extensive Lean code archive that has not been accepted by mathematicians; an independent check of whether it proves the intended statement was still pending as of November 21.

Levent Alpöge, a mathematician who works at Anthropic, posted a note on November 20 saying he believes he and Claude proved the Riemann Hypothesis, a longstanding problem about the distribution of prime numbers. The claim has not been accepted as a proof: the released material is chiefly a large Lean code archive, and important independent checks remain open.

Before you orderOffer from Amazon

Get the latest gadgets delivered free with Prime

  • Fast, free delivery on millions of items
  • Prime Video, Amazon Music and more included
  • Member-only deals all year
Start your free Prime trial Free trial for eligible customers · Cancel anytime
As an affiliate, we earn on qualifying purchases.

Alpöge’s four-page note says the method also proves the Generalized Riemann Hypothesis, a broader family of related conjectures. It links to a Lean archive divided into 14 parts. Lean is software that checks formal mathematical arguments. Scott Aaronson put the archive at 17.4 million lines of code; Alpöge says a human-readable version would exceed 250,000 pages and did not publish it.

The note offers a sketch of ideas and a timeline, rather than a conventional proof that mathematicians can read and assess directly. Alpöge says he guided the strategy but that Claude did much of the technical work, writing that “Claude far outpaced me near the finish line.” He also says he has “little mastery” of one of the main areas used in the argument. The note does not identify the Claude model.

Early reports describe some checks, but not a complete validation. MIT mathematician Andrew Sutherland told Scientific American that the code compiled independently “with only the three standard axioms.” Kevin Buzzard wrote that a comparator run was still awaited. That check is intended to test whether the formalized code proves the stated result using only the permitted assumptions. Compilation alone does not settle whether the archive proves the intended theorem.

At a glance
reportWhen: Claim posted November 20, 2026; checks…
The developmentLevent Alpöge published a note claiming he and Claude proved the Riemann Hypothesis, alongside a 17.4 million-line Lean archive that experts have not yet fully checked.
Bar chart titled 20 steps in six weeks, eight of them on one day. Steps per date in the timeline of Alpöge's note: October 6, 4; October 8, 1; October 9, 1; October 14, 2; October 19, 2; October 20, 1; November 15, 8; November 16, 1, the last step, which reads GRH, highlighted
The 20-step timeline in Alpöge's note, which he says Claude compiled, by date. Source: Levent Alpöge, remark.pdf, pages 3 and 4, November 20, 2026. Chart: Thorsten Meyer AI.

A Possible Result With Major Stakes

If verified, a proof would resolve one of mathematics’ best-known open problems and have consequences for number theory. The Riemann Hypothesis concerns the zeros of the zeta function and predicts a precise pattern for the distribution of prime numbers. Its generalized version covers related functions and would extend the scope of the claim.

The Clay Mathematics Institute offers a $1 million prize for a solution to the Riemann Hypothesis, but posting a claim does not qualify it for an award. Clay’s rules require publication in a refereed journal, at least two years of scrutiny after publication, and general acceptance by the mathematical community. Those conditions have not been met.

The episode also puts formal proof software and AI-assisted mathematics under attention. A proof checker can verify that each encoded step follows under its rules; it does not, by itself, show that the formalization matches the intended mathematical claim. Experts must examine both the code’s foundations and the relationship between the code and the conjecture.

Products Worth Considering

LG 27GS60QC-B Ultragear 27-inch Curved Gaming Monitor (QHD, 180Hz, 1ms, 1000R)

LG 27GS60QC-B Ultragear 27-inch Curved Gaming Monitor (QHD, 180Hz, 1ms, 1000R)

Samsung Odyssey OLED G8 34-Inch Curved Ultra WQHD Gaming Monitor

Samsung Odyssey OLED G8 34-Inch Curved Ultra WQHD Gaming Monitor

As an affiliate, we earn on qualifying purchases.

From Riemann’s Conjecture to Lean

Bernhard Riemann proposed the hypothesis in 1859. In simplified terms, it says that every nontrivial zero of the zeta function has real part one-half. The conjecture is connected to how regularly prime numbers are distributed. It has remained unsolved and is one of the seven Millennium Prize Problems named by the Clay Mathematics Institute.

Lean is an interactive theorem prover: mathematical arguments can be encoded in a formal language and checked by its kernel. Such formalization can catch errors in a proof as represented in the system. But a very large archive may be difficult for people to understand, and reviewers still need to establish that the encoded statement is the Riemann Hypothesis and that the formal argument uses sound definitions and assumptions.

The first-day coverage included both signs of progress and substantial reservations. Terence Tao described the submission as an “(enormous) Lean-only proof.” Fields Medal recipient James Maynard told Nature, “If this fully checks out, it is clearly a historic milestone in mathematics.” Andrew Granville, by contrast, called the released note “garbled.” These comments address the material available at the time; they do not amount to a collective verdict on the full archive.

“I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis.”

— Levent Alpöge

The Proof Still Needs Review

As of November 21, no journal had published the result, and no mathematical body had accepted it. The reported independent compilation is one check, while the comparator run remained pending. The available sources do not establish that the full archive has passed that test or that independent mathematicians have verified its correspondence to the Riemann Hypothesis.

It is also unclear how long a human review would take, whether experts can produce a comprehensible account of the formal argument, or whether the claimed extension to the Generalized Riemann Hypothesis withstands scrutiny. Anthropic had not announced the claim on its news page when checked on November 21. Nature reported that the company had not immediately responded to its request for comment.

Until those questions are answered, the appropriate description is a proof claim under review. A large checked codebase is significant evidence to investigate, but neither the note nor the early reports establish that the conjecture has been solved.

Independent Checks Come First

The immediate milestone is the comparator run and further independent examination of the Lean archive. Reviewers will need to confirm that the code proves the intended statements without hidden assumptions, and mathematicians will need to understand and assess the argument. The sources provided do not give a date for those checks or a timetable for publication.

If the work survives those stages, journal review and broader mathematical scrutiny would follow. For the Clay prize, publication in a refereed journal, a period of at least two years, and general acceptance are required under the institute’s rules. Until then, the claimed proof remains an unverified development rather than a settled result.

Key Questions

Has the Riemann Hypothesis been proved?

Not according to the available evidence. Alpöge has claimed a proof, but mathematicians have not accepted it and key checks were still pending as of November 21, 2026.

What did Alpöge release?

He posted a four-page note and linked a 14-part Lean archive. He did not publish the human-readable version, which he says would exceed 250,000 pages.

What role did Claude play?

Alpöge says he guided the strategy and Claude carried out much of the technical work. The note does not name the Claude model, and the account has not been independently verified.

What is the comparator run?

It is a check intended to establish that the formal code proves the stated result using only the permitted assumptions. Kevin Buzzard said this check was still awaited.

Would the claim qualify for the Clay prize now?

No. Clay’s rules require refereed journal publication, at least two years of scrutiny, and general acceptance by mathematicians. Those requirements have not been satisfied.

Source: Thorsten Meyer AI

You May Also Like

Must-Have AI Automation Software Tools In 2026: Top 13

Discover the 13 leading AI automation software tools in 2026, their features, and what makes them essential for modern workflows.

A Reader’s Guide To Spotting And Fixing Tiring AI Text

A preprint comparing 14 language models with human writing identifies six patterns linked to reading difficulty and offers editing advice.

The 2028 Model Lab Endgame: How Six Becomes Two, Three, or Twelve

Forecasts for 2028 suggest the Western frontier AI labs could consolidate into two, three, or twelve major players, with significant implications for the industry.

AI Demystified: The Engine Room Behind Twelve Machines

An in-depth look at the core components behind twelve AI models, explaining how they work and why it matters for understanding AI technology today.