One of mathematics’ most famous unsolved problems has suddenly attracted two competing claims of a breakthrough. Levent Alpöge, a mathematician associated with AI company Anthropic, says he developed a proof of the Riemann hypothesis with help from the company’s Claude AI model. On the same day, Stanford postdoctoral researcher Alessandro Masullo released a separate proof of the broader Generalized Riemann Hypothesis, developed entirely by human mathematicians over 12 years.

The announcements have sparked intense discussion across the mathematics and artificial intelligence communities. The Riemann hypothesis has remained unresolved since German mathematician Bernhard Riemann proposed it in 1859. However, neither announcement represents an officially accepted solution. Researchers still need to examine the mathematical arguments, identify potential gaps and establish whether the proposed proofs withstand independent scrutiny.

Anthropic Mathematician Says Claude Helped Prove the Riemann Hypothesis

Levent Alpöge announced on November 20 that he believes he and Claude have discovered a proof of the Riemann hypothesis. Alpöge, a mathematician at Anthropic, published a four-page explanation of the work alongside a substantial collection of formal mathematical files.

The announcement is particularly significant because Alpöge claims the work goes beyond the original Riemann hypothesis and establishes the Generalized Riemann Hypothesis (GRH), a broader mathematical conjecture.

Rather than publishing a conventional, fully readable mathematical paper, Alpöge released the work in Lean, a programming language and proof assistant used to express mathematical arguments in a form that computers can verify.

According to reports, the archive contains millions of lines of formalized proof code. The scale of the material makes it difficult for mathematicians to understand the complete argument quickly, even if individual parts can be checked computationally.

Alpöge credited Claude with making substantial contributions during the later stages of the research. He described the AI as progressing far beyond his own understanding near the conclusion of the work.

How Claude’s Lean-Based Proof Works

Lean is an open-source proof assistant that allows mathematicians to translate mathematical statements and arguments into a formal language.

Unlike an ordinary AI-generated explanation, a Lean proof must satisfy strict logical requirements. The system checks whether the steps follow from established definitions, assumptions and previously verified results.

This distinction matters because large language models can produce convincing-looking mathematical explanations that contain errors. Formal verification offers a way to check whether a precisely stated argument follows logically within the system.

TechCrunch reported that independent participants in a mathematics discussion community had successfully compiled Alpöge’s Lean proof. This is an important technical development, but successful compilation does not automatically establish that the mathematical result is correct in the broader sense. Researchers must still examine the underlying definitions, assumptions, dependencies and interpretation of the theorem.

The remaining challenge is also one of mathematical understanding: experts need to explain why the proof works, not merely confirm that a formal system accepts its encoded steps.

A Human Mathematician Releases a Separate Proof

The AI announcement was accompanied by another major development. Ravi Vakil, a Stanford professor and president of the American Mathematical Society, published a draft proof of the Generalized Riemann Hypothesis written by Stanford postdoctoral researcher Alessandro Masullo.

Unlike Alpöge’s work, Masullo’s proof was developed without AI assistance.

According to the report by TechCrunch, Masullo said the work represented approximately 12 years of research. The draft was released earlier than originally planned because of the sudden announcement surrounding Alpöge’s AI-assisted result.

The human-authored argument is reported to run to approximately 940 pages. Unlike Alpöge’s Lean-based archive, Masullo’s draft has not been presented as a computer-formalized proof.

This creates an unusual situation: two separate mathematical approaches, one involving an AI system and the other developed through years of human research, are now being examined in connection with the same major conjecture.

However, the human proof also requires careful independent verification. Its publication does not establish that the argument is complete or correct.

What Is the 167-Year-Old Riemann Hypothesis?

The Riemann hypothesis was proposed by German mathematician Bernhard Riemann in 1859. It concerns the distribution of prime numbers, which are numbers divisible only by themselves and one.

Examples include 2, 3, 5, 7, 11 and 13.

Prime numbers appear irregularly across the number line, but mathematicians have discovered that their distribution is connected to a mathematical function called the Riemann zeta function.

The hypothesis states that all non-trivial zeros of this function have a real part equal to one-half.

In simplified terms, it predicts a precise underlying pattern in the distribution of prime numbers. Proving the hypothesis would provide a much stronger mathematical understanding of how prime numbers are distributed.

The problem has influenced number theory for more than a century and a half and is widely considered one of the most important unresolved questions in mathematics.

Why the Generalized Riemann Hypothesis Matters

The Generalized Riemann Hypothesis extends the original conjecture to a wider family of mathematical functions, including Dirichlet L-functions.

A proof of the generalized version would also establish the original Riemann hypothesis as a special case.

The broader conjecture has implications across number theory and is connected to results involving prime numbers, arithmetic structures and other mathematical problems.

FeatureRiemann HypothesisGeneralized Riemann Hypothesis
Original formulation1859Extension of the original conjecture
Main subjectZeros of the Riemann zeta functionZeros of a broader family of L-functions
Mathematical scopeSpecific zeta functionWider class of related functions
StatusNot officially solvedNot officially solved
SignificancePrime-number distributionBroader consequences across number theory

The distinction is important because both announcements concern the generalized version, which encompasses the original problem.

Why the Mathematical Community Is Being Cautious

A claim involving the Riemann hypothesis cannot be accepted simply because a mathematician publishes a proof or a computer system processes its code.

Mathematical breakthroughs require detailed examination by independent experts. Researchers must verify every important step, check assumptions and determine whether the argument relies on established results correctly.

In Alpöge’s case, the Lean archive offers a potentially powerful verification mechanism, but the sheer complexity of the proof creates a substantial challenge for human understanding.

In Masullo’s case, the lengthy written draft will need to be examined carefully by specialists. The fact that it was developed without AI does not exempt it from the same standards of mathematical scrutiny.

The Clay Mathematics Institute, which established the Millennium Prize Problems, offers a $1 million prize for an accepted solution to the Riemann hypothesis. Its rules require publication in an appropriate peer-reviewed outlet and at least two years of scrutiny by the mathematical community before a solution can qualify for the prize.

What This Means for the AI Industry

The announcement is another major test of whether artificial intelligence can contribute to advanced scientific research rather than simply generate explanations, summaries or computer code.

AI companies have increasingly focused on mathematical reasoning, coding and scientific discovery. These fields require systems to work through complex problems, develop intermediate results and maintain logical consistency over long sequences of steps.

If Alpöge’s proof is validated, it would represent a landmark example of AI-assisted mathematical discovery. It could also encourage researchers to use AI systems more extensively in areas where progress has historically required years of specialist work.

However, the result should not be interpreted as evidence that AI has independently replaced mathematicians. Alpöge directed the research and helped develop the approach, while the proof still requires human examination and explanation.

The Bigger Picture

The simultaneous release of an AI-assisted proof and a separate human-authored proof highlights two different approaches to mathematical discovery. AI systems may be able to explore complicated arguments and formalize results at a scale that would be difficult for individual researchers. At the same time, Masullo’s 12-year effort demonstrates the continuing importance of sustained human mathematical research.

The bigger question is no longer simply whether AI can solve difficult mathematical problems. It is whether the results it helps produce can be independently verified, understood and incorporated into established mathematical knowledge. The Riemann hypothesis provides an extraordinary test of that possibility.

Looking Ahead

Mathematicians will now need to examine both proposed proofs in detail. Alpöge’s Lean-based work may offer a route for checking formal logical consistency, while Masullo’s lengthy written argument will require extensive specialist review. Neither claim should be treated as a confirmed solution until independent verification is completed and the relevant mathematical standards are satisfied.

If either proof survives scrutiny, it could become one of the most significant mathematical developments in modern history. For the AI industry, a successful verification of Alpöge’s work would be particularly important evidence that advanced AI systems can contribute to frontier scientific research. For now, the 167-year-old Riemann hypothesis remains officially unsolved.


SEO Title: AI and Human Mathematicians Claim Breakthrough in Riemann Hypothesis

Meta Description: Anthropic’s Levent Alpöge says Claude helped prove the 167-year-old Riemann hypothesis, while Stanford researcher Alessandro Masullo releases a separate human-written proof.

Suggested URL Slug: ai-human-mathematicians-riemann-hypothesis-proof-claims

Focus Keyword: Riemann hypothesis AI proof

Secondary Keywords: Claude AI mathematics, Anthropic Riemann hypothesis, Levent Alpöge, Alessandro Masullo proof, Generalized Riemann Hypothesis, AI scientific discovery

Get the day’s top stories in your inbox

One concise email. No spam, unsubscribe anytime.