- Get link
- X
- Other Apps
Featured Post
By Mesum Mukhtar
LearnDailyAI
-
- Get link
- X
- Other Apps
On Friday 20 November 2026, a mathematician at Anthropic announced a claim that could be one of the biggest moments in the history of mathematics. Levent Alpoge said he believes he and Claude have found a proof of the Riemann Hypothesis, the 167 year old problem about prime numbers that has resisted every attempt to solve it since 1859. If the claim is true, it would settle the most famous open question in pure mathematics.
Here is the part that matters most, and it must come first. The claim has not been independently verified, and the proof has not yet been reviewed by outside experts. That is the position reported by phys.org, and it is the position this post keeps throughout. This is a claim, not a confirmed result, and no one should call the problem solved until independent mathematicians have checked the work and agreed.
That warning does not make the story small. The claim comes from a serious mathematician, it is backed by computer checked code, and leading mathematicians are already starting the long work of verification.
Watch the official video
This video is Anthropic's own upload from the official Anthropic YouTube channel, press play and watch without leaving this page. It shows Claude solving a puzzle game, a clear view of the same careful reasoning behind the maths work in this post.
Claude Opus 4.5 solves a puzzle game
Watch Claude Opus 4.5 solves a puzzle game on YouTube
Video: official Anthropic upload, © Anthropic, embedded from the official Anthropic YouTube channel.
What exactly was announced
Levent Alpoge is a mathematician at Anthropic, the company behind Claude. On 20 November 2026 he announced that he believes he and Claude have proved the Riemann Hypothesis, first posed by German mathematician Bernhard Riemann in 1859. The announcement was reported by phys.org and by The Hindu.
Alpoge says the proof also covers the Generalized Riemann Hypothesis, a stronger version of the problem. He released the proof in Lean, a programming language that lets computers check mathematical arguments step by step, as a four page note plus about 17 million lines of Lean code in 14 compressed files. The full written argument would run to more than 250,000 pages and was not published.
Alpoge was open about the limits of his own understanding. He said he has not yet mastered all of the mathematics involved and plans to write it up properly. A proof of this size needs many experts, each checking a different part, before the community can agree.
The announcement came the day after Anthropic's record stock market debut on 19 November 2026, and Alpoge consulted Anthropic chief executive Dario Amodei before announcing. Alpoge acknowledged the timing could be viewed cynically, according to phys.org. The key work came together late on Sunday night 15 November into the early morning of 16 November.
What is the Riemann Hypothesis
To understand why this claim matters, start with prime numbers. A prime number is a whole number greater than 1 that can be divided evenly only by 1 and by itself, for example 2, 3, 5, 7 and 11. Primes are the building blocks of whole numbers, because every whole number can be built by multiplying primes together.
Primes do not appear in a simple pattern. They get rarer as numbers get larger, but they never stop. The Riemann Hypothesis is a precise claim about the pattern behind prime numbers, and it is widely regarded as the most important open question in pure mathematics. Riemann looked at a mathematical function called the zeta function, and he said that all of its interesting zeros lie on one single straight line.
Mathematicians have tested the idea on enormous numbers of cases with computers, and it has held up every time, but testing cases is not the same as proving it for every case. A proof must show the claim is true in every case, with no exception possible, and that is why the problem has stayed open for 167 years. It is one of seven Millennium Prize Problems chosen by the Clay Mathematics Institute in 2000, each carrying a 1 million dollar prize, according to phys.org.
Lean formal verification is the part that makes this claim different from a claim on paper alone. Lean is a programming language that lets a computer check a mathematical argument step by step, so a mistake in logic can be caught by the machine. That does not prove the argument is correct, but it does give independent checkers a concrete place to start.
The Generalized Riemann Hypothesis is a stronger version of the same idea. It extends the claim to a wider family of related functions, so proving it would settle even more than the original problem alone.
What is verified and what is not
The most useful way to read this story is to separate what is confirmed from what is still a claim. Two separate proof checking systems have accepted the result, and MIT mathematician Andrew Sutherland confirmed the code passes several rigorous checks, according to The Hindu. That is encouraging, but it is not the same as independent verification by the wider mathematics community.
Here is the whole picture side by side:
| Claim | Status | Evidence |
|---|---|---|
| Claude and Alpoge have proved the Riemann Hypothesis | Claimed, not verified | Announcement on 20 November 2026, no independent review yet |
| The proof also covers the Generalized Riemann Hypothesis | Claimed, not verified | Stated by Alpoge, same proof files, same review still needed |
| Lean code and a four page note were released | Confirmed | Note and about 17 million lines of Lean code in 14 compressed files |
| Two proof checking systems accepted the result | Confirmed | Reported by The Hindu, with checks confirmed by Andrew Sutherland |
| Leading mathematicians are now trying to verify the work | Confirmed | Assessment by Terence Tao and comment by James Maynard |
| The 1 million dollar prize is won | Not yet | Prize needs peer review first, claim must pass review |
Verdict: The claim is serious, the computer checked code is a real step forward, and leading mathematicians are treating it with respect. But the proof is not verified, the prize is not won, and the problem is not solved until independent experts have checked the work and agreed. Treat this as a claim to follow, not a result to celebrate yet.
Terence Tao, a Fields Medal winner, assessed the work in the early hours after the announcement. His preliminary analysis shows the proof draws on modern research in the Langlands Program and arithmetic geometry, and the mathematics community is now trying to verify it, which will take some time. That assessment was reported by 36Kr. Tao did not declare the proof correct.
James Maynard, Regius Professor of Mathematics at the University of Oxford and a 2022 Fields Medal winner, put the stakes clearly. If this fully checks out, it is clearly a historic milestone in mathematics, establishing one of the great mysteries of the subject. That quote, reported by IBM, is careful in exactly the right way. The words if this fully checks out carry the whole story.
The 1 million dollar prize also needs care. The Clay Mathematics Institute offers 1 million dollars for each Millennium Prize Problem, and the claim must pass peer review first before any prize can be considered. No prize has been awarded, and no official body has declared the problem solved.
Why mathematicians are excited and cautious
The excitement is easy to understand. Mathematicians have waited 167 years for this proof, and a computer checked argument at this scale would be a new kind of breakthrough, not just for maths but for what AI can do in science. If Claude really helped to build a proof that humans alone could not finish, it would change how research is done.
The caution is just as easy to understand, and it comes from experience. Big claims in mathematics have failed before, sometimes after months of careful review, because a single hidden gap can undo a long argument. A proof that would run to more than 250,000 pages in full written form cannot be checked in a day, and even its author says he has not yet mastered all of the mathematics involved.
There is also a rival story unfolding at the same time. A rival human proof appeared hours later, when Stanford's Ravi Vakil hurried out Alessandro Masullo's all human draft of the Generalized Riemann Hypothesis, called The Music of the Prime, after months of review by Brian Conrad. That story was reported by The Context. Two claims arriving on the same day does not make either one true.
The context around Anthropic adds to both the excitement and the caution. In August 2026 Anthropic said an unreleased Claude improved a long standing result related to the Riemann Hypothesis, raising the share of zeros proved from 41.6 percent to 67.2 percent. In October 2026 Anthropic released hundreds of major proofs at once.
Earlier, a Claude Fable model found a counterexample to the 87 year old Jacobian conjecture. Those steps were reported by The Hindu and by The Wall Street Journal.
For readers who follow Claude closely, this is the same company and the same model family covered in our Claude review. It is also worth reading Gemini Can Now Control Your Mac to see how the other leading AI labs are racing in the same direction, and our post on Muse showing people the wrong private chats as a reminder that powerful AI systems still need careful checking.
What happens next
Verification is now the whole story, and it will take time. Mathematicians around the world will read the four page note, run the Lean code, test the steps, and look for gaps, and that work will take weeks and months, not hours. Terence Tao said the community is now trying to verify the work, which will take some time, and that is the honest timetable.
Alpoge's own background helps explain why the claim is being taken seriously. He holds an undergraduate degree from Harvard University, a doctorate in mathematics from Princeton University, and he was a Junior Fellow at the Harvard Society of Fellows. That background was reported by 36Kr. A strong background does not make a proof correct, but it does mean the claim comes from someone who knows the field deeply.
Here are the four things to follow as the verification unfolds:
- The Lean code checks: watch whether more independent teams run the code and report the same result, or find an error in a step.
- The human readable write up: watch whether Alpoge publishes a proper written version that other mathematicians can read and judge.
- The expert verdicts: watch what leading number theorists say after they have had time to study the argument in depth.
- The rival draft: watch how the all human draft by Alessandro Masullo is received.
If the proof survives, the effects would reach far beyond a prize. A confirmed proof would reshape number theory, give new tools to every mathematician who works with primes, and show that AI can help solve problems that have resisted human effort for more than a century.
If the proof fails, the story still matters. A failed attempt at this scale would still teach the community a great deal, and it would show exactly where AI reasoning breaks down on the hardest problems. Either way, the careful review now under way is the right process, and the worst mistake would be to declare a winner before the checking is done.
Try it yourself
- Learn the primes: write out the numbers from 1 to 50 and circle every prime, then look for a pattern in where they appear.
- Read the note: open the four page note linked in the sources below and read the first page slowly.
- Follow the verification: check back in a week and search for what Terence Tao and other mathematicians have said since the announcement.
Quick quiz
1. Who announced the claim, and on what date?
Answer: Levent Alpoge of Anthropic, on Friday 20 November 2026.
2. How much is the prize for each Millennium Prize Problem, and what must happen before it can be considered?
Answer: 1 million dollars, and the claim must pass peer review first.
3. What did Anthropic say Claude improved in August 2026?
Answer: The share of zeros proved in a long standing result related to the Riemann Hypothesis, from 41.6 percent to 67.2 percent.
Your turn
Do you think AI will solve the Riemann Hypothesis before humans do alone? Tell us in the comments, share this post with a friend who loves maths, and follow @learningdailyai on Instagram for the next chapter, explained simply, every day.
All logos and videos in this post are official Anthropic media, © Anthropic, used here for editorial coverage. The prime number illustrations are original illustrations, not official photographs.



Comments
Post a Comment