A 167-Year-Old Puzzle Just Made Headlines Again
Few math problems have survived as long as the Riemann Hypothesis. Bernhard Riemann proposed it in 1859, and generations of mathematicians have chased it since.
On Friday, November 20, 2026, the chase got a new twist. Levent Alpöge, described in news reports as a scientist at Anthropic, said he had found a proof with help from the company’s Claude AI model.
What Was Actually Claimed
Alpöge posted his work publicly, and mathematicians were quick to react. Physicist Brian Greene wrote on social media that the announcement would be astounding if it holds up.
The important words there are “if true.” A claim like this is not a settled result until other mathematicians have checked it.
Some coverage describes the claim as a proof of the Riemann Hypothesis. Other reports, including a blog post by computer scientist Scott Aaronson, refer to the Generalized Riemann Hypothesis. According to one report, the final entry in the posted material is dated November 16 and carries the label “GRH.”
Why the Riemann Hypothesis Matters
The hypothesis is about a function Riemann studied called the zeta function. It predicts where certain special values of that function, known as the nontrivial zeros, must sit.
That might sound abstract, but it connects to something familiar: prime numbers. The zeros control how evenly primes are spread across the number line. Proving the hypothesis would sharpen what mathematicians know about that pattern.
The problem is also one of the seven Millennium Prize Problems named by the Clay Mathematics Institute, each carrying a $1 million prize. The Clay Institute has its own rules for awarding the prize, which include publication and a waiting period, so a public announcement alone does not trigger anything.
The Role of Claude
One report on the announcement says Alpöge credits Claude with a specific contribution involving a chain of theta lifts. Theta lifts are a technique from advanced number theory, and the detail suggests the AI did more than format or summarize.
Aaronson’s blog post says the work produced 17.4 million lines of Lean. Lean is a proof assistant, a programming language in which a computer can check every logical step of a mathematical argument. That is an enormous amount of code, and it is worth noting that Aaronson reported the figure while reacting to the news.
If a proof really is written in Lean and the software accepts it, the checking becomes mechanical. The harder questions are whether the statement being proved matches the actual hypothesis and whether the underlying definitions were set up correctly. Human experts will want to examine exactly that.
A Human Claim Entered the Picture Too
The story got more complicated within hours. According to Slashdot, a statement circulated saying that mathematician Alessandro Masullo had put forward a proposed proof of the Riemann Hypothesis without AI several months earlier.
TechCrunch covered the same twist under a headline noting that an AI and a human had both claimed a solution. In the comments under Aaronson’s post, one reader wondered whether a human had already done it with a far shorter write-up. That was a reader’s guess, not a verified fact.
The takeaway is that two separate claims now exist. Neither has been confirmed by the wider mathematical community as of November 21, 2026.
Why Mathematicians Are Cautious
Claimed proofs of the Riemann Hypothesis have appeared many times over the years. Slashdot has even tracked them, with a story as far back as 2008 about another claimed proof.
Most of those claims fell apart under scrutiny. That history is why careful readers treat any new announcement as a starting point for review rather than a conclusion.
What makes this announcement different is the tooling. A machine-checkable proof, if it exists in complete form, gives reviewers something concrete to test. It also raises a new question about how mathematics should credit and verify work that an AI helped produce.
Market and Public Reaction
The news spread fast. A commenter on Aaronson’s blog claimed Anthropic’s stock market value dropped 12-14% that day, but that figure came from a reader comment and I could not confirm it from a reliable financial source, so treat it with caution.
What is clear is that the story ran across many outlets within hours, from tech sites to international news agencies. Interest in the Riemann Hypothesis rarely gets this kind of spotlight.
How to Follow This Story Without Getting Misled
If you want to keep up, a few simple habits will help. They work for any big science claim, not just this one.
- Look for peer review. A posted PDF is not the same as a paper accepted by a journal.
- Watch for expert responses. Number theorists who study the zeta function are best placed to judge the argument.
- Check the exact statement. The Riemann Hypothesis and the Generalized Riemann Hypothesis are related but not identical claims.
- Be wary of single-source numbers. Figures from comment sections or social posts need a second source.
What Happens Next
In the coming days, expect mathematicians to read Alpöge’s posted material closely and to test whether the Lean code really captures the hypothesis. They will also compare it with the separate human claim that surfaced the same day.
If the proof survives that review, the next step would be formal publication, followed by the Clay Institute’s own evaluation process. If it does not, the episode will still be a useful test of how AI-assisted mathematics gets checked in the open.
For now, the safest reading is this: a bold claim has been made, the first reactions have been loud, and the verification work has only just begun.



