
An Anthropic mathematician says he has used the company’s Claude chatbot to prove the Riemann hypothesis. No one has confirmed it. Levent Alpöge announced the claim on Friday, according to the BBC, a day after Anthropic made its US stock market debut.
If the proof holds up, the BBC reports, it would be among the most significant in history. That “if” is doing a lot of work. Here’s what has been claimed, who has responded, and why a result like this can’t be judged in a day.
What is the Riemann hypothesis, and why does it matter?
It’s a number theory problem, which is the branch of maths that studies whole numbers and especially primes. The BBC calls it nearly 170 years old and widely considered the single most famous problem in maths. Tom’s Hardware says Bernhard Riemann proposed it in 1859.
In plain terms, the hypothesis says prime numbers aren’t random. They follow a determined order along the number line, according to the BBC. Prime numbers are the bedrock of higher maths, the BBC adds, so breakthroughs tied to them have broad impact.
The Clay Mathematics Institute, quoted by Tom’s Hardware, puts it more formally. It says the hypothesis asserts that all interesting solutions of the equation ζ(s) = 0 lie on a certain vertical straight line. That has been checked for the first 10,000,000,000,000 solutions. A proof for every solution, it says, “would shed light on many of the mysteries surrounding the distribution of prime numbers.”
Tom’s Hardware also quotes Scientific American calling the problem the “North Star for the entire field of number theory.” It adds that the problem was once considered so insurmountable that experts didn’t bother working on it.
What does this mean for everyday users and businesses?
For now, nothing practical. The claim is unverified, and neither report describes any product, service or security impact. We won’t guess at one.
The more immediate stake is trust. AI companies are announcing big results in maths and science, and readers, students and business buyers are being asked to believe them. How those claims get checked tells you how much weight to give the next one.
The backstory, in order
The timeline below uses only dates and sequences given in the BBC and Tom’s Hardware reports. Neither gives exact calendar dates for most events.
1859: The problem is posed
Riemann proposes the hypothesis, per Tom’s Hardware. It then sits unsolved for roughly 167 years.
September 2026: OpenAI makes its own claim
OpenAI said in September that employees using a non-public version of its AI tools solved part of a notoriously difficult problem in 88 hours, the BBC reports. The problem was part of the Navier-Stokes equations, which concern how fluids move.
More than 10,000 AI agents, meaning bots that operate somewhat autonomously, were put to the task as an AI training exercise, per the BBC. The BBC says the proof has yet to be confirmed as accurate. It also set off criticism in academic and higher maths communities.
Tom’s Hardware describes the same episode as a breakthrough that was “somewhat overshadowed by a plagiarism controversy.” The BBC doesn’t mention plagiarism, and the two outlets frame the result differently. We can’t resolve that from the notes.
Later in 2026: Mathematicians push back
A group of maths scholars said in a letter that AI companies boasting of major solutions to show how sophisticated their products had become was “detrimental to the science of mathematics,” the BBC reports. They also wrote: “We are witnessing a general threat to intellectual work, with misalignment between the outcome of the use of AI and its initial purpose.”
The BBC says OpenAI later released what it claimed were solutions, proofs or major progress on 372 more problems. It also says Anthropic has claimed other progress in maths and physics.
About six weeks of work, then a message on Sunday
Alpöge said that over about six weeks he used Claude to build a detailed proof, according to the BBC. The BBC says it was found using a non-public version of Claude.
He messaged Anthropic CEO Dario Amodei on Sunday, the BBC reports, asking directly how he could announce the discovery.
Thursday and Friday: IPO, then the claim
Anthropic made its US stock market debut on Thursday, per the BBC’s account of the timing. Tom’s Hardware says the company floated on the Nasdaq that same week. Alpöge made his announcement on Friday.
What Alpöge actually claimed
His wording is careful. “I believe I have, working with Claude, discovered a proof of the Riemann Hypothesis,” he wrote, as quoted by Tom’s Hardware. That’s a belief, not a declaration of a settled result.
He also believes the method could easily extend to the Generalized Riemann Hypothesis and the Extended Riemann Hypothesis, per Tom’s Hardware. Those are broader versions of the problem. That part is his expectation, not something anyone has tested.
He wrote that “an astounding amount of mathematics is present in the argument, far more than I had imagined would be necessary as a student.”
He also addressed the timing head-on. “The precise timing is horrible,” he said, according to the BBC. “If I could have, I’d have finished this all much earlier.” He added: “I am clearly unilaterally responsible for feeling I just could not live if I waited even longer. There is an obvious cynical view, to do with the timing, but it is false.”
That’s his account. The BBC reports it without independent confirmation of his reasons, and we have none either.
Why a Riemann hypothesis proof is so hard to verify
The BBC says the complete proof was put forward in Lean, a computer language used for confirming maths proofs. The file is over 250,000 lines long.
That’s a striking detail. Lean is designed to check proofs mechanically. Neither report says whether the file has been checked, or by whom. We’d treat that as an open question.
Our take (analysis): a quarter of a million lines is far more than one person can read over a weekend. Even with a machine-checkable format, experts will likely want to understand what the proof says and how it’s built. Understanding takes people, and people take time.
Mathematician Tao said on Friday that verification is still some way off, according to the BBC. “I imagine the digestion process will take some time, though in this case at least we have a human author to communicate with to assist with this task,” he said.
The BBC points out the contrast. No such human was obviously part of the major AI-assisted proofs earlier this year. Having an author who can answer questions could help reviewers, though neither report says anyone has started a formal review.
What’s confirmed, and what isn’t
- Confirmed by both outlets: Alpöge, an Anthropic employee and mathematician, has claimed a proof made with Claude.
- Not confirmed: that the proof is correct. Tom’s Hardware’s headline calls it “allegedly discovered but remains unconfirmed.”
- Slightly different in the reports: the BBC places the announcement on Friday, a day after the IPO. Tom’s Hardware says Alpöge “wrote in a paper this week.” They aren’t necessarily in conflict, but we can’t reconcile them further.
- Not in the notes: any statement from Anthropic itself, or any review by other mathematicians beyond Tao’s comment.
What happens next
Our take (analysis, not reporting): the next signal to watch is whether other mathematicians publicly engage with the argument. Tao’s comment suggests that process has barely begun. A claim like this typically earns trust slowly, if it earns it at all.
The earlier Navier-Stokes claim is a useful reference point. By the BBC’s account, it still hasn’t been confirmed, and it drew sharp criticism about how AI results are announced. A proof with a human author and a formal Lean file may be easier to scrutinize. That’s a possibility, not a prediction.
The timing next to Anthropic’s stock debut will keep drawing questions, whatever the maths turns out to be. Alpöge rejects the cynical reading. Readers can hold two thoughts at once: the explanation is his own, and the proof is what matters.
Practical takeaway: treat this as a claim under review. Until independent mathematicians report back, the honest label is “unverified.” We’ll add a dated update here if that changes.
Sources
- Anthropic may have solved major maths problem Reimann hypothesis – BBC News (bbc.co.uk)
- Anthropic mathematician claims to have solved 167-year-old math problem with Claude — Riemann hypothesis proof allegedly discovered but remains unconfirmed | Tom's Hardware (tomshardware.com)
This article was researched and drafted with AI assistance and reviewed against the linked sources by the Nexonic editorial team.
Join the conversation