Did Claude Really Prove the Riemann Hypothesis? Terence Tao and Other Mathematicians Respond
Anthropic mathematician Levent Alpöge claims that, working with Claude, he found a method for proving the Riemann hypothesis and the generalized Riemann hypothesis, and released approximately 17 million lines of Lean formal proof code. Terence Tao and other mathematicians are following the development, but the mathematical community is still verifying the code and the proof strategy. Under the Clay Mathematics Institute’s rules, the work must also be formally published, withstand at least two years of scrutiny, and gain broad acceptance.
Why read this
SelectedAI score 82
The materials compile the announcement, the scale of the formalized code, and preliminary feedback from several mathematicians, helping distinguish between claims of a breakthrough and recognition by the academic community.
Topics
Body · AI translation
Did Claude Really Prove the Riemann Hypothesis? Fields Medalists Including Terence Tao and Deng Yu Respond Download the Zhidx App for an Exclusive First Look November 21, 2026, 07:12:50, from Beijing
Zhidx reported on November 21 that last night, Anthropic mathematician Levent Alpöge published a statement on his personal website claiming that, in collaboration with Claude, he had found a way to prove the Riemann hypothesis and its stronger version, the Generalized Riemann Hypothesis (GRH).
▲Statement published by Alpöge (Image source: his personal homepage)
This problem, which has confounded the mathematical community for 167 years, may be on the verge of a major breakthrough.
Terence Tao, the 2006 Fields Medalist, also published a post on his personal blog discussing the development. He said that preliminary analysis suggested the proof appeared to synthesize modern research from the Langlands program and arithmetic geometry. The mathematical community is currently trying to verify the proof’s correctness and identify its central ideas. He expects this process to take some time.
▲Tao’s post
Alex Kontorovich, a distinguished professor of mathematics and chair of the mathematics department at Rutgers University, exclaimed: “I’m actually more confident now that the proof may succeed. The Riemann hypothesis is being proved by using ‘all’ the proofs!” Álvaro Lozano-Robledo, a mathematics professor at the University of Connecticut, also reposted the news and expressed his astonishment with two exclamation marks.
Ayush Khaitan, a mathematics and AI researcher at Princeton University, said: “Just yesterday, I was discussing with collaborators how keeping humans in the loop remains the best way to use large language models to solve difficult mathematical problems.”
MIT mathematician Andrew Sutherland, after independently compiling the relevant Lean code, told American Scientist: “This truly is a historic moment. The world has changed… In my view, this is a victory for humanity and a day worth celebrating.”
Alpöge revealed that the written proof he generated with Claude exceeded 250,000 pages. Because it was too long and difficult to read, it was ultimately not made public. Instead, he released a four-page explanation and the Lean formal proof code, divided into 14 compressed files. The code contains approximately 17 million lines. Alpöge said it includes a complete formal proof of the Generalized Riemann Hypothesis, although the result still awaits further scrutiny by the mathematical community.
▲Anthropic mathematician Levent Alpöge (Image source: personal homepage)
Original statement:
Original code (01 through 14):
Alpöge holds a bachelor’s degree from Harvard University and a PhD in mathematics from Princeton University. He previously served as a Junior Fellow at the Harvard Society of Fellows and now works as a researcher at Anthropic.
In the statement, he admitted that the Riemann hypothesis had been his most important mathematical research goal since childhood and that Claude’s performance in this collaboration far exceeded his expectations. “As we approached the finish line, Claude far surpassed me. No, Claude was practically flying, dancing, and shining near the finish line.”
Alpöge said that he pointed Claude toward the key research direction, after which Claude proposed a series of mathematical ideas that astounded him. He expects the currently extremely lengthy proof to be substantially simplified in the future.
Notably, Alpöge had actually achieved this result just days before Anthropic went public. According to a Wall Street Journal report early this morning, he completed the crucial work late on November 15 and in the early hours of November 16 local time, then contacted Anthropic CEO Dario Amodei to ask whether he could make it public.
Alpöge acknowledged that the timing of the announcement was “extremely bad” and that there were internal company restrictions he needed to observe, but he ultimately decided to publish it, saying that he “simply couldn’t wait any longer.”
Just How Difficult Is the Riemann Hypothesis?
In 1859, German mathematician Bernhard Riemann proposed the conjecture, whose central question is whether all nontrivial zeros of the Riemann zeta function have a real part equal to 1/2.
This seemingly abstract question is closely related to the distribution of prime numbers among the natural numbers. Mathematicians have used computers to verify vast numbers of zeros, but checking a finite number cannot prove that the conjecture holds for infinitely many zeros.
The Generalized Riemann Hypothesis extends the question to a broader class of mathematical functions, and whether it holds affects many important results in number theory.
In 2000, the US-based Clay Mathematics Institute named the Riemann hypothesis one of the seven Millennium Prize Problems and established a prize of $1 million (approximately RMB 6.677 million).
Earlier, in August this year, Anthropic published Claude’s research findings on a related problem, raising the lower bound on the proportion of zeros lying on the critical line from 41.6% to 67.2%. At the time, however, there was still a substantial gap between that result and resolving the complete conjecture.
Interestingly, as early as February 2022, OpenAI co-founder and CEO Sam Altman celebrated on X that OpenAI had solved two International Mathematical Olympiad (IMO) problems, while half-jokingly asking: “When will the Riemann hypothesis be cracked?”
Now, however, it is Anthropic, its top competitor, that has taken the lead in claiming to have used AI to prove this century-old problem. Altman’s post from more than four years ago has also resurfaced, with one commenter joking: “Anthropic beat you to it, bro.”
After the announcement was released, Shunyu Yao, a recipient of Tsinghua University’s Special Scholarship and now a research scientist at Google DeepMind, reposted Alpöge’s statement and exclaimed: “Oh my God, the Riemann hypothesis?!”
The announcement also quickly drew attention from the mathematics community.
Mathematician Elliot Glazer said on X that his preliminary examination found that the formalized mathematical proposition in the Lean files released by Alpöge was indeed the generalized Riemann hypothesis.
Although the code was still compiling at the time and was expected to take several hours, he believed it had probably already passed checks beforehand. He also noted that a vulnerability in Lean itself could not theoretically be ruled out, but estimated the probability of that to be far below 1%.
Thomas Bloom, a mathematician at the University of Manchester in the United Kingdom, remarked that he did not know which was more astonishing: an AI claiming to have proved the Riemann hypothesis, or an AI claiming to have generated a logically coherent proof exceeding 250,000 pages.
A purported screenshot of a chat in which Deng Yu, a possible 2026 Fields Medal recipient, commented on the proof has also circulated online. In response to a related question, Sayaka replied: “I won’t even look at it until there is at least a 500-page version.” The remark presumably means that the 250,000-page original draft is too voluminous to be reviewed and that the author must first distill it into a condensed version of roughly 500 pages before it would be worth devoting effort to verification. Otherwise, it would be difficult to determine whether the proof is valid.
▲Image source: Internet
For now, Alpöge’s proof still awaits further scrutiny by the mathematics community. Under the rules of the Clay Mathematics Institute, a solution to a Millennium Prize Problem must be formally published, undergo at least two years of examination, and gain general acceptance from the international mathematics community before it is eligible for the prize.
Meanwhile, the recently listed Anthropic has suffered a sharp setback in the capital markets.
On November 20 Eastern Time, Anthropic shares closed at $88.91 (approximately RMB 593.7), down 8.50%, giving the company a total market capitalization of approximately $1.18 trillion (approximately RMB 7.87 trillion). The share price plunged nearly 13% intraday and at one point approached its offering price. By only its second trading day, nearly all the gains since its IPO had already been erased.
▲Anthropic share-price performance (Image source: Tencent Stock Selection)
“Special notice: The above content (including any videos, images, or audio) was uploaded and published by users of Dafeng Hao, a self-media platform operated by Phoenix New Media. This platform merely provides information-storage services.
Notice: The content above (including the videos, pictures and audios if any) is uploaded and posted by the user of Dafeng Hao, which is a social media platform and merely provides information storage space services.”
AI Trends