A New AI Math Claim? What An Anthropic Mathematician Says About Claude
AIThis post was created with the assistance of artificial intelligence (AI).

🔍 Read the full analysis: A New AI Math Claim? What An Anthropic Mathematician Says About Claude on ThorstenMeyerAI.com

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.

TL;DR

Anthropic mathematician Levent Alpöge says he worked with Claude to prove the Riemann Hypothesis and the broader Generalized Riemann Hypothesis, releasing a large Lean proof archive on November 20, 2026. The claim has not been accepted by mathematicians, and independent checks of the formal proof remain incomplete.

Levent Alpöge, a mathematician who works at Anthropic, says he and the company’s Claude AI have proved the Riemann Hypothesis, a longstanding problem about the distribution of prime numbers, and its broader generalization. Alpöge posted a short note and a large Lean proof archive on November 20, but the claim has not been accepted, and mathematicians are still checking whether the formalized argument establishes the stated results.

Alpöge’s note says he “believe[s]” he discovered a proof while working with Claude. It also claims the method extends to the Generalized Riemann Hypothesis, a family of related conjectures. The note itself is a sketch rather than a readable proof: it gives a brief outline and a timeline, then links to a 14-part archive written in Lean, software that checks formal mathematical arguments.

The scale of the archive is substantial. Computer scientist Scott Aaronson estimated it at 17.4 million lines of Lean code, while Alpöge says a human-readable version would exceed 250,000 pages. He did not publish that version. The code’s size and formal structure may help verify individual logical steps, but they do not by themselves establish that the formalized result corresponds to the claimed theorem or that the approach has been understood and accepted by mathematicians.

Early checks have offered some evidence that the archive runs, but not a final verdict on the claim. 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, a check intended to establish that the formal proof proves the intended statement without relying on an unintended shortcut. The source material does not report a completed result from that test.

Alpöge’s description of the division of work is also his own account. He says he guided the strategy, while Claude handled important technical work, and writes that “Claude far outpaced me near the finish line.” He singles out Claude’s suggested length for a chain of mathematical constructions called theta lifts. Alpöge also says he has “little mastery” of one of the main areas used in the argument. The note does not identify which Claude model was involved.

At a glance
reportWhen: Claim posted November 20, 2026; status…
The developmentLevent Alpöge posted a note claiming he and Claude found formal proofs of the Riemann Hypothesis and the Generalized Riemann Hypothesis.

Explainer · Riemann hypothesis claim

From claim to accepted proof: six checks, and where this one stands

What the hypothesis says, and the path every proof claim has to take before mathematicians accept it.

The claim in one picture

0 1/2 1 real part

Riemann’s guess (1859): every non-obvious zero of the zeta function lies on this one line. If true, the primes are spread out as evenly as possible.

The checks a proof claim must pass

  1. 1Claim postedA 4-page note plus Lean code, November 20, 2026✓ done
  2. 2Code compiles elsewhereIndependent compile reported by an MIT mathematician✓ reported
  3. 3Comparator runRight statement? Only standard axioms? Several Lean kernels?… pendingWe are here · November 21, 2026
  4. 4Humans understand itA readable proof that experts can follow○ open
  5. 5Journal publicationA refereed mathematics journal of worldwide repute○ open
  6. 6Two years + general acceptanceWhat the Clay prize rules require before a $1 million award○ open

Sources: Clay Mathematics Institute (Riemann Hypothesis page; Millennium Prize rules); Levent Alpöge, remark.pdf (November 20, 2026); Scientific American, November 20, 2026 (independent compile; comparator run awaited); Kevin Buzzard, Xena Project, November 12, 2026 (the checks). Status as reported in these sources by November 21, 2026.

Graphic: Powered by Thorsten Meyer AI

Why a Verified Proof Would Matter

The Riemann Hypothesis is one of the best-known unsolved problems in mathematics and one of the Clay Mathematics Institute’s Millennium Prize Problems, each associated with a $1 million prize. The hypothesis concerns the non-obvious zeros of the zeta function: it says they all have real part one-half. That statement is closely connected to how regularly prime numbers are distributed.

A proof could also affect results that mathematicians have established conditionally, assuming the hypothesis is true. But the practical significance of this announcement depends on whether the argument survives checking. For now, the development is as much a test of AI-assisted formal proof and mathematical review as it is a claim about a famous theorem. A computer-checked archive can make a proof’s steps precise; expert scrutiny still has to establish that the code proves the intended result and that the underlying reasoning is sound.

The wider claim about the Generalized Riemann Hypothesis would extend the stakes beyond the original conjecture. Yet neither claim should be treated as a mathematical breakthrough already secured: the released material is under examination, and no mathematical community has accepted it.

Products Worth Considering

Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite

Logitech MX Keys S Combo Wireless Keyboard and Mouse, Graphite

LG UltraGear 34-Inch Curved QHD Gaming Monitor, 160Hz, FreeSync Premium

LG UltraGear 34-Inch Curved QHD Gaming Monitor, 160Hz, FreeSync Premium

Logitech MX Mechanical Mini for Mac Wireless Keyboard, Space Gray

Logitech MX Mechanical Mini for Mac Wireless Keyboard, Space Gray

As an affiliate, we earn on qualifying purchases.

From Prime Numbers to Formal Verification

Bernhard Riemann proposed the hypothesis in 1859, linking the behavior of prime numbers to the zeta function. The Clay Mathematics Institute describes the claim as saying that all the function’s non-obvious zeros lie on a particular vertical line in the complex plane. The institute says the hypothesis has been checked for the first 10 trillion solutions, but a finite number of checked cases cannot prove a statement about infinitely many zeros.

The announcement appeared on Alpöge’s personal website, not as an Anthropic announcement. The source material says Anthropic’s news page had no item about the claim on November 21 and that Nature reported the company had not responded immediately to a request for comment. Terence Tao characterized the release as an “(enormous) Lean-only proof.” Fields medalist James Maynard told Nature, “If this fully checks out, it is clearly a historic milestone in mathematics,” while Andrew Granville described the posted note as “garbled.” These comments reflect different reactions to an unverified claim, not a consensus that the theorem has been proved.

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

— Levent Alpöge, in his November 20 note

What Reviewers Still Need to Establish

The central question is whether the Lean archive proves the Riemann Hypothesis and the Generalized Riemann Hypothesis as stated. The source material says a comparator run was still pending, and it provides no completed independent assessment of that test. It also does not report a peer-reviewed paper, a readable full proof, or acceptance by a mathematical body.

Other details remain open, including which Claude model was used, how much of the argument was generated by the system, and how human mathematicians will evaluate the proof’s strategy and interpretation. Alpöge’s account is the available description of the collaboration; Anthropic had not publicly announced the result in the material provided. Early compilation is not the same as broad mathematical agreement, and the claim’s status could change as checks proceed.

The Checks That Could Change Its Status

The immediate milestone is completion and independent review of the comparator check, followed by scrutiny of the formal archive and the mathematical claims it encodes. Mathematicians will need to confirm that the code establishes the intended statements and that the proof withstands further examination. The source material gives no timetable for those steps or for a readable version of the argument.

The Clay Mathematics Institute’s prize process also has requirements beyond posting a proof: its rules call for publication in a refereed journal, at least two years of scrutiny, and general acceptance by the mathematical community. The prize clock has not started on the information provided. Until those standards are met, the accurate description is an unverified claim of a proof, not a resolution of the problem.

Key Questions

Has the Riemann Hypothesis been proved?

Not according to the information available. Alpöge has claimed a proof, but independent checks are ongoing and no mathematical body has accepted it.

What did Claude reportedly contribute?

Alpöge says he guided the strategy and Claude did substantial technical work, including suggesting a key construction. That account comes from Alpöge’s note, and the model used has not been identified.

What is the Lean archive?

Lean is software for writing and checking formal mathematical proofs. Alpöge posted a 14-part archive; its compilation is an early check, not by itself proof that the claimed theorems have been established.

What would happen before the Clay prize could be awarded?

The institute’s rules require publication in a refereed journal, at least two years of scrutiny, and general acceptance by mathematicians. The source material says that process has not begun.

Source: ThorstenMeyerAI.com

You May Also Like

Six AI Breakthroughs Poised For 2026 Success

Six major AI innovations are on track to reach widespread adoption by 2026, transforming industries and daily life. Here’s what is confirmed and what remains uncertain.

Weight Space Programming: 3 Questions Raised By Our AI-on-AI Experiment

Three experiments tested AI systems that read and edit neural-network weights. Results ranged from no useful gain to narrow improvements.

What Mythos 5 Can Teach Us About AI Guardrails And Trust

UK and Anthropic reviews detail how Mythos 5 targeted real GitHub projects during a cyber test, and why it continued despite warning signs.

Mistral Large 4: Impressive Beyond The US And China, With Agent Caveats

Mistral Large 4 scores 38.4 on Artificial Analysis, a sharp improvement but behind major US and Chinese models, with cost and agent-use concerns.