The Daily Byte
Breaking
Welcome to The Daily Byte — real stories from across Pakistan.
Saturday, 21 November 2026 · Pakistan

The Daily Byte

Stories from Pakistan that deserve to be heard
Minor World

What Is the Riemann Hypothesis? Maths' Greatest Puzzle

By The Daily Byte ·21 Nov 2026 ·10 min read
Blackboard covered in chalk mathematical equations

A mathematician working at the artificial intelligence company Anthropic says he has proved the Riemann hypothesis, the most famous unsolved problem in mathematics. Levent Alpöge announced the claim on Friday 20 November 2026, saying he had found the proof working with Anthropic's AI system Claude. If the claim survives checking, it would settle a question that has resisted every mathematician for 167 years and earn a one-million-dollar prize. But for now it is a claim, not a confirmed result: no independent expert has yet verified the proof, and Alpöge himself has said the full human-readable argument does not exist yet.

So what is the Riemann hypothesis, why has it mattered for so long, and what would it mean if it turned out to be true? This explainer sets out the problem in plain terms, what exactly has been claimed, and what happens next.

What the Riemann hypothesis actually says

The hypothesis is about prime numbers: the numbers 2, 3, 5, 7, 11, 13 and so on, which can only be divided by one and themselves. Primes are the atoms of arithmetic, because every whole number can be built by multiplying primes together. Yet the primes do not fall into any neat pattern. They thin out as numbers get bigger, but they do so irregularly, and nobody has ever found a formula that predicts exactly where the next prime will appear.

In 1859 the German mathematician Bernhard Riemann published a short paper that connected the primes to a piece of mathematics called the Riemann zeta function. The zeta function is a formula that takes in a number and gives out another number, and mathematicians study the places where its output is zero. Most of those places, called trivial zeros, sit at the negative even numbers. The interesting ones, the non-trivial zeros, are scattered across a band of the complex plane. Riemann guessed that every single one of them lies on one precise vertical line, the line where the real part of the number equals one half. That guess is the Riemann hypothesis.

It sounds abstract, but it is really a claim about hidden order. Riemann's paper showed that the pattern of the primes is controlled by the positions of those zeros. If all the zeros really do sit on that one line, the primes are distributed as evenly as they possibly could be. If even one zero sits off the line, the distribution of primes is messier than mathematicians believe it to be.

Why prime numbers matter so much

Prime numbers are not just a curiosity for pure mathematicians. Much of modern digital security depends on them. The encryption that protects online banking, messaging and shopping relies on the fact that it is easy to multiply two large primes together but extremely hard to work backwards and find the two primes from the answer. A proof of the Riemann hypothesis would not suddenly break that encryption, and nobody is suggesting that it would. But it would deepen our understanding of how primes are spread out, and the mathematics of primes feeds directly into number theory, cryptography and computer science.

The hypothesis also sits at the centre of a web of other results. Hundreds of mathematical theorems have been proved on the assumption that the Riemann hypothesis is true. If it were proved false, much of that work would need to be rethought. If it is proved true, all of those results become solid. That is one reason mathematicians care so much: the hypothesis is a load-bearing wall in the building of modern mathematics.

Computers have checked enormous numbers of zeros, and every one tested so far has indeed fallen on the critical line. But checking examples, however many, is not a proof. A proof must show that no exception can ever exist, however far out the numbers go. That is what has defeated every attempt since 1859.

The million-dollar prize and 167 years of failure

In 2000 the Clay Mathematics Institute in the United States named the Riemann hypothesis one of its seven Millennium Prize Problems, each carrying a prize of one million dollars for a correct solution. It is the only one of the original seven that concerns the primes directly, and it is widely regarded as the hardest of the set. Only one Millennium problem has ever been solved: the Poincaré conjecture, proved by the Russian mathematician Grigori Perelman, who declined both the prize money and the Fields Medal.

Many of the greatest mathematicians in history have tried and failed. Riemann himself died in 1866, seven years after publishing his paper, without proving his guess. Since then the problem has survived every new technique thrown at it, from the analytic methods of the early twentieth century to the computer-assisted approaches of recent decades. That long record of failure is part of why a claimed proof is such enormous news, and also why mathematicians will insist on checking it with extreme care.

What was claimed on 20 November 2026

According to reports published on Friday, Alpöge wrote on his personal website: "I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis." He said the proof had been produced in Lean, a programming language that lets a computer check each step of a mathematical argument mechanically. He also said the proof covers the generalised Riemann hypothesis, a broader version of the problem that extends the same idea to a much wider family of mathematical functions related to the primes.

Several details in the announcement underline how unusual this claim is. Alpöge said he had consulted Anthropic's chief executive, Dario Amodei, before going public, and he acknowledged that the timing could be viewed cynically, coming a day after Anthropic's record-breaking debut on the stock market. The Wall Street Journal reported that the result was achieved late on Sunday night into Monday, days before the company went public. Alpöge also said the full written argument would be extremely long, and the Journal reported that the AI-written version runs to more than 250,000 pages. He has said he has not yet mastered all of the mathematics involved himself and plans to write the proof up properly.

That is not the first time Claude has made the news in connection with this problem. About a week earlier, Anthropic published research showing that an unreleased version of Claude had improved a long-standing lower bound for the fraction of zeta-function zeros known to satisfy the hypothesis, raising it from 41.6 per cent to 67.2 per cent. Two mathematicians at Anthropic studied and validated that result, and outside experts examined the paper. Alpöge had also been in the headlines in July after posting a Claude-generated approach to another famous problem, the Jacobian conjecture, helping to kick off what Scientific American called maths' unprecedented summer.

Notably, the same day also brought a separate claim: a Stanford postdoc, Alessandro Masullo, posted a proof of the generalised Riemann hypothesis written entirely by a human, shared by Stanford professor Ravi Vakil. The mathematics world therefore has two extraordinary claims to assess at once, produced by very different routes.

What is Lean, and why does it matter here?

Lean is an open-source programming language designed for writing mathematics in a form a computer can check line by line. When a proof is written in Lean, the software verifies every logical step, so a mistake anywhere in the chain is caught automatically. That makes a Lean proof much harder to get wrong than a traditional paper proof, which relies on human referees spotting errors.

But Lean has a catch, and it is an important one for this story. A Lean proof is only as good as the statement it proves. If the formal statement written in Lean does not exactly match the real Riemann hypothesis, the computer can check every step and still prove the wrong thing. Experts will therefore need to check two things: that the Lean code is correct, and that the statement it proves really is the Riemann hypothesis. That second check requires human mathematical judgement, and it is one reason the claim cannot be settled by a computer alone.

There is also the question of readability. A proof of more than 250,000 pages that no human has fully read is very different from the kind of proof mathematicians are used to, where the ideas can be explained, taught and built upon. Alpöge has said he intends to produce a human-intelligible explanation, and mathematicians quoted in the first reports said it will be a while before the community feels it has a proper handle on the argument.

What happens next

The claim now enters the slow machinery of mathematical verification. Other mathematicians are already working through the Lean code, and preliminary informal responses reported on Friday suggested the proof may indeed be correct, but that is a long way from a settled verdict. A formal check of a proof this large could take months. A human-readable version, and peer review of it, could take much longer.

For the Clay Mathematics Institute's million-dollar prize, the bar is specific: the proof must be published in a refereed mathematics journal of worldwide repute, and it must be generally accepted by the mathematics community two years after publication. Neither condition is close to being met. The prize rules exist precisely because mathematics has seen claimed proofs of famous problems before, and most of them have turned out to contain an error.

It is also worth remembering that the claim is exactly that: a claim. The reports on Friday were careful to say the proof has not been independently verified and has not yet been reviewed by outside experts. Alpöge's own wording, "I believe I have", is the language of a mathematician who knows the difference between finding an argument and having it accepted.

What it would mean if the proof holds

If the proof survives, it would be one of the great events in the history of mathematics. A 167-year-old question would be closed, hundreds of conditional results would become unconditional, and the Clay prize would be won. It would also be a landmark moment for artificial intelligence in science: the most famous problem in mathematics, solved with an AI system as a collaborator. The fact that a human-written proof of the generalised hypothesis appeared on the same day only adds to the sense that something remarkable is happening in the field.

But the more cautious lesson is that mathematics moves at its own pace. A claim announced on a website on a Friday in November is the beginning of a process, not the end of one. The zeros of the zeta function have kept their secret since 1859. Whether they have finally given it up will be decided not by headlines, but by the slow, exacting work of checking every step.