Did Claude Really Solve the Riemann Hypothesis? What Anthropic’s AI-Assisted Proof Actually Shows

Did Claude Really Solve the Riemann Hypothesis? What Anthropic’s AI-Assisted Proof Actually Shows
Share

An Anthropic mathematician claims he used Claude to solve the Riemann Hypothesis, a 167-year-old problem that has challenged generations of mathematicians. The proposed proof contains millions of lines of computer-checkable code, but independent experts have not yet confirmed that the famous problem is solved.

Anthropic’s Claude AI may have helped solve one of the most difficult problems in mathematics, according to an extraordinary announcement from mathematician Levent Alpöge on November 20, 2026.

Alpöge, a number theorist working at Anthropic, says he collaborated with Claude to develop a proof of the Riemann Hypothesis, a mathematical conjecture first proposed in 1859. The problem concerns the distribution of prime numbers and has resisted proof for more than 160 years.

The claim has attracted attention because the Riemann Hypothesis is widely considered one of the most important unsolved problems in mathematics. It is also one of the Clay Mathematics Institute’s seven Millennium Prize Problems, which carry a $1 million award for an accepted solution.

However, the mathematical community has not yet accepted Alpöge’s result.

According to Nature’s November 20 report, leading mathematicians are examining the proposed proof, which Alpöge released primarily as code written in the Lean proof assistant rather than as a conventional mathematical paper.

The announcement raises two important questions: Has Claude genuinely helped solve a problem that defeated generations of mathematicians, and how can researchers verify a proof so large that its creator says he does not yet fully understand every part?

Anthropic Researcher Levent Alpöge Claims a Historic Breakthrough

Levent Alpöge announced on November 20 that he believes he and Claude have developed a proof of the Riemann Hypothesis.

Alpöge is a mathematician specializing in number theory and is associated with Harvard University. He currently works at Anthropic, the company behind Claude.

In a four-page statement, Alpöge explained that he had been working with Claude on the problem and had reached a result that surprised even him.

He credited the model with producing a substantial amount of mathematics during their collaboration.

According to AFP’s report on the announcement, Alpöge also said the proposed argument establishes the Generalized Riemann Hypothesis, a broader mathematical conjecture.

That would make the result considerably more significant than a proof of the original hypothesis alone.

However, Alpöge presented the finding as a belief that he had discovered a proof, rather than an independently established mathematical result.

The distinction matters because even highly respected mathematicians can make mistakes in arguments involving difficult and unfamiliar mathematical ideas.

The announcement begins the process of verification rather than ending it.

What Is the Riemann Hypothesis?

The Riemann Hypothesis is a mathematical conjecture about the hidden patterns behind prime numbers.

Prime numbers are whole numbers greater than one that can be divided evenly only by one and themselves. Examples include 2, 3, 5, 7, 11 and 13.

These numbers are fundamental to mathematics because every whole number greater than one can be expressed as a product of prime numbers.

Although mathematicians understand many properties of primes, their distribution becomes increasingly difficult to predict as numbers grow larger.

In 1859, German mathematician Bernhard Riemann discovered an important connection between prime numbers and a mathematical function now called the Riemann zeta function.

The hypothesis concerns the locations of certain points where this function equals zero.

Riemann proposed that all its nontrivial zeros have a real part equal to one-half.

In simpler terms, the hypothesis predicts that these zeros all lie along one particular vertical line in the complex-number plane.

Proving this statement would establish important results about how accurately mathematicians can describe the distribution of prime numbers.

Despite enormous progress in number theory, nobody has produced a universally accepted proof since Riemann introduced the conjecture.

Why the Riemann Hypothesis Has Remained Unsolved for 167 Years

The difficulty of the Riemann Hypothesis comes from the relationship between seemingly irregular prime numbers and the mathematical structure of the zeta function.

Mathematicians have verified that enormous numbers of nontrivial zeros lie on the predicted line.

However, checking individual cases does not prove that every possible zero satisfies the hypothesis.

A complete proof must establish that the statement holds universally, including for infinitely many cases that cannot be tested one by one.

Over the past century, mathematicians have developed powerful methods to study the zeta function and related mathematical objects.

These advances have produced important results in number theory, even without resolving the original conjecture.

Many mathematical theorems also depend on the Riemann Hypothesis being true.

If the hypothesis were proved, some results that currently rely on an unproven assumption could become unconditional.

That is one reason the problem has remained so important.

The challenge is not finding more numerical evidence supporting the hypothesis. It is discovering a mathematical argument that establishes the result without exceptions.

Claude’s Proposed Proof Contains 17 Million Lines of Lean Code

Claude's Proposed Proof Contains 17 Million Lines of Lean Code

One of the most unusual aspects of Alpöge’s announcement is the form of the proposed proof.

Instead of releasing a traditional mathematical paper containing a readable sequence of arguments, he published a large collection of code written for Lean, a formal proof assistant.

According to the Associated Press report, the formalization contains approximately 17 million lines of Lean code.

Alpöge also described the broader collaboration as producing more than 250,000 pages of mathematical material.

These figures show the extraordinary scale of the work, but they do not establish its correctness.

A very long mathematical argument can still contain an error, and the amount of code does not measure the importance of a proof.

The reason Lean matters is that it allows mathematical arguments to be expressed in a language that a computer can check according to precise logical rules.

Rather than relying entirely on a person reading every step, researchers can use Lean to verify whether a formal argument follows from its definitions, assumptions and previously established results.

However, a successful computer check is only one part of establishing that the intended mathematical problem has been solved.

ALSO READ: Anthropic Launches Claude Fable 5.2 and Mythos 5.2 With Faster Performance

Why Mathematicians Are Still Checking Claude’s Riemann Hypothesis Proof

The biggest obstacle is not simply the size of the proposed proof.

Mathematicians need to establish that the formal argument corresponds to the original mathematical conjecture and that its logical foundations are sound.

That requires several kinds of examination.

Researchers must inspect the theorem statement to confirm that it accurately represents the Riemann Hypothesis.

They must also examine the definitions used throughout the proof, particularly those involving the zeta function and its zeros.

Another concern is whether the code depends on unproved assumptions or incomplete sections.

A Lean project can contain declarations that assume a mathematical statement without proving it. If an essential part of the argument depends on such a declaration, the result may not establish the claimed theorem.

Independent researchers also need to reproduce the verification using the published code and appropriate software environment.

Finally, mathematicians want to understand the central ideas behind the proof.

A computer-verified argument can establish logical validity without immediately providing a clear explanation of why the result is true.

That distinction is particularly important for a proof that could transform several areas of mathematics.

ALSO READ: AI Can Now Mathematically Verify Software Security. Here’s Why That Matters

Fields Medal Winner James Maynard Says the Claim Could Be Historic

The announcement has attracted cautious interest from leading mathematicians.

James Maynard, a number theorist at the University of Oxford and a Fields Medal winner, told Nature that a fully verified proof would represent a historic milestone.

Maynard also expressed optimism that the argument might ultimately prove correct.

However, he emphasized that researchers would need time to examine the Lean verification and understand the mathematical reasoning.

His reaction illustrates the difference between taking the claim seriously and declaring the problem solved.

An established mathematician can consider a proposed proof promising without treating it as confirmed.

The complexity of Alpöge’s work also creates a challenge for researchers who want to explain the result in conventional mathematical language.

A proof may be logically valid but remain difficult for humans to understand.

If the claim survives verification, mathematicians may spend considerable time identifying its main ideas and finding shorter or clearer explanations.

Some Experts Are More Skeptical About the Announcement

Not every mathematician has welcomed the way the result was presented.

Andrew Granville, a number theorist at the University of Montreal, criticized the clarity of Alpöge’s initial written explanation.

Speaking to Nature, Granville argued that the researcher should have prepared a more understandable account before making such a major announcement.

The criticism reflects a long-standing expectation in mathematics that researchers should explain their arguments clearly enough for other experts to evaluate them.

A large computer-verified codebase can provide evidence of formal correctness, but it may not communicate the underlying mathematical ideas effectively.

Other researchers have been more optimistic.

Tristan Buckmaster, a mathematician at New York University who has worked with Alpöge on separate research, praised the techniques he developed for using language models in mathematical work.

The differing reactions do not establish whether the proof is correct.

They show that mathematicians are assessing both the technical claim and the unusual method used to present it.

Claude May Have Helped Prove the Generalized Riemann Hypothesis Too

Alpöge’s announcement goes beyond the original Riemann Hypothesis.

He says the argument also establishes the Generalized Riemann Hypothesis, commonly abbreviated as GRH.

The generalized version extends the original idea to a broader family of mathematical functions known as Dirichlet L-functions.

These functions play an important role in understanding how prime numbers appear in different arithmetic progressions.

For example, mathematicians study patterns among primes that leave particular remainders when divided by a fixed number.

The Generalized Riemann Hypothesis would provide stronger information about several such patterns.

If Alpöge’s argument genuinely proves the generalized statement, it would establish a broader collection of results than a proof of the original Riemann Hypothesis alone.

However, this part of the announcement requires independent verification as well.

Researchers must check the exact formal statements and determine whether they establish the full generalized conjecture.

Until that work is complete, both results remain claims rather than accepted mathematical theorems.

Could Claude Win the $1 Million Millennium Prize?

The Riemann Hypothesis is one of seven Millennium Prize Problems selected by the Clay Mathematics Institute in 2000.

The institute established a $1 million prize for an accepted solution to each problem.

However, Alpöge’s announcement does not automatically qualify for the award.

The institute requires a proposed solution to satisfy formal publication and acceptance conditions.

According to the Clay Mathematics Institute’s Millennium Prize rules, a solution must be published in an appropriate refereed venue and remain subject to scrutiny for at least two years before the institute considers awarding the prize.

That means even a successful independent check of the Lean code would not immediately settle the prize question.

The mathematical community would need time to evaluate the result and establish its acceptance.

There is also a distinction between the role of Claude and the person responsible for submitting the mathematical work.

The use of an AI system does not by itself determine prize eligibility or how any award would be attributed.

Those questions would depend on the institute’s rules and its eventual assessment of a qualifying solution.

For now, no $1 million prize has been awarded for Alpöge’s claim.

What Did Claude Actually Contribute to the Proof?

Alpöge has credited Claude with making a substantial contribution to the proposed argument.

However, the precise division of work between the mathematician and the AI model has not been independently established.

AI systems can assist mathematical research in several ways.

They can suggest possible approaches, generate formal proof steps, identify relevant theorems and help translate mathematical reasoning into a proof assistant’s language.

They can also produce incorrect arguments that require human review.

Alpöge’s account suggests a collaboration in which Claude contributed mathematical ideas and formal material while he directed the research and evaluated the results.

This is different from establishing that Claude independently solved the problem without human involvement.

The distinction matters because complex mathematical discoveries often involve choices about which ideas to investigate, how to formulate the problem and how to interpret intermediate results.

If the proof is verified, researchers will likely examine how the collaboration worked and which parts of the process were most important.

That could provide valuable evidence about the ability of language models to contribute to original scientific research.

Why This Could Be a Bigger AI Breakthrough Than a Benchmark Record

AI companies frequently announce improvements on mathematics benchmarks.

These evaluations generally measure how well a model solves problems for which correct answers or solutions are already known.

A verified proof of the Riemann Hypothesis would be different.

It would mean an AI-assisted research process contributed to resolving a major mathematical question that had remained open for generations.

That would provide evidence of AI contributing to new mathematical knowledge rather than simply reproducing or extending established solutions.

However, the significance would depend on what researchers learn about Claude’s actual contribution.

If the system generated important new mathematical ideas, that would be particularly notable.

If it mainly helped formalize reasoning developed by a human mathematician, the achievement would still demonstrate the usefulness of AI tools but support a different conclusion about their capabilities.

The distinction should be established through evidence rather than assumptions about the model’s intelligence.

A verified result would also raise questions about how mathematicians should review future discoveries involving enormous amounts of computer-generated reasoning.

Anthropic’s Announcement Comes Just After Its Record IPO

Anthropic's Announcement Comes Just After Its Record IPO

The timing of the claim has attracted attention because it came immediately after Anthropic’s public stock market debut.

The company completed its record-breaking IPO on November 19, 2026, one day before Alpöge announced the proposed proof.

According to AFP, Alpöge said he consulted Anthropic CEO Dario Amodei before making the announcement.

He also acknowledged that some observers might question the timing.

However, the coincidence does not establish that the claim was released to influence investors or the company’s valuation.

Nature reported that Alpöge rejected a cynical interpretation of the announcement’s timing.

The relevant scientific question remains whether the mathematical argument is valid.

A company’s financial position does not prove or disprove a theorem.

Independent verification is especially important in this situation because a confirmed result could have major implications for both mathematics and the reputation of AI-assisted research.

What Happens If the Riemann Hypothesis Proof Is Correct?

If mathematicians confirm Alpöge’s result, it would resolve one of the longest-standing problems in modern mathematics.

The proof could establish important results in number theory and remove assumptions from theorems that currently depend on the Riemann Hypothesis.

A confirmed proof of the generalized version could have even broader consequences.

Researchers would then need to study the argument and identify which mathematical techniques made the breakthrough possible.

Some of those techniques could potentially be useful in other areas of mathematics.

The result would also provide an important case study for AI-assisted scientific discovery.

Researchers could examine how Claude contributed to the work and whether similar methods might help solve other difficult problems.

However, it would be premature to claim that the proof would immediately transform encryption, banking systems or everyday computing.

Although prime numbers are important in cryptography, the Riemann Hypothesis is not a secret formula that automatically breaks modern encryption.

Its direct technological consequences would depend on the mathematical results and methods that emerge from the verified proof.

What If Experts Find an Error?

If mathematicians discover a flaw, the Riemann Hypothesis would remain unsolved.

That would not necessarily mean the entire research effort was worthless.

A proposed proof can contain useful ideas, partial results or new techniques even when its central claim fails.

Researchers might identify a gap that can be repaired or discover that the argument establishes a weaker but still valuable theorem.

Alternatively, the problem could involve an incorrect formal statement or an unsupported assumption that undermines the main result.

In either case, the outcome would provide information about the strengths and limitations of AI-assisted mathematical research.

It could also help developers improve methods for generating, checking and explaining formal proofs.

The important point is that scientific verification must determine the outcome, regardless of how impressive the announcement appears.

Claude’s Riemann Hypothesis Claim Could Change Mathematics, If It Survives Review

Anthropic mathematician Levent Alpöge’s announcement is one of the most ambitious claims yet involving artificial intelligence and original mathematical research.

The proposed proof targets a problem that has challenged mathematicians since 1859 and could establish results extending far beyond the original conjecture.

Claude’s reported contribution also raises important questions about how AI systems might help researchers discover new mathematics.

However, the difference between a proposed proof and an accepted theorem remains essential.

The Lean code must be independently checked, the mathematical statements must be confirmed, and researchers must determine whether the argument genuinely resolves the problem.

If the proof survives that scrutiny, it could become a historic achievement for mathematics and AI-assisted research. Until then, the Riemann Hypothesis remains an open problem, and Claude’s reported breakthrough remains an extraordinary claim awaiting confirmation.