A proof in the form of code
Lean is a proof assistant: every step is written in a formal language and checked mechanically by a computer. That makes this claim different from a normal announcement. There is, in principle, something concrete to test, and the test does not require trusting the author's prose.
It also splits verification into separate questions, and only the first has any reported progress.
The first question is whether the code compiles: whether Lean accepts every step as following from the previous ones. Scientific American reports that MIT mathematician Andrew Sutherland independently compiled Alpöge's result in Lean and called it "an historic moment."
The second question is whether the theorem at the top of that code is actually the Riemann Hypothesis. A formalization can be internally perfect and still prove a weaker statement, a subtly different statement, or a statement with a hidden assumption in its definitions — and a statement match alone would not settle whether the proof relies on extra assumptions, axioms or admitted steps. Checking those things is human work, and on November 20 nobody quoted in the coverage had yet reported completing it. Understanding the argument is a further task again: Alpöge says he intends to work on understanding and explaining it, and his intuition is that it will be massively simplified.
Until those checks are done, the accurate headline is not "Riemann solved." It is "a Lean project that its author says proves Riemann has been posted, and it compiles."
The author is candid about the gaps
The note does not hide its own oddity. He writes that he is presenting a complete argument while his mastery of parts of it — automorphic forms on higher-rank groups, by his own account — is limited. He says the timeline Claude assembled for the work runs from October 6 to November 16, 2026, ending with "GRH" on the final date. He describes guiding the search toward a setting analogous to Stepanov's method, which in the note's framing resolved Riemann over function fields, and then watching Claude outpace him, in his words: "Claude flew, and danced, and sparkled near that finish line."
He also addresses the timing directly. The note appeared on his personal site the day after Anthropic's stock-market debut. Alpöge writes that the timing is "horrible," that the result came together late on a Sunday night, and that he messaged chief executive Dario Amodei to ask about announcing it. That contact does not by itself establish company approval or coordination. The disclosure is there in the note for readers to weigh.
A second claim arrived the same day
TechCrunch reports that Stanford professor Ravi Vakil posted a draft by Stanford postdoc Alessandro Masullo claiming a proof of the Generalized Riemann Hypothesis, described as entirely human work with no AI, the result of twelve years of effort. Vakil's note says the plan had been to polish the draft for a year; Alpöge's announcement made waiting untenable.
The two claims are separate, and nothing in the coverage establishes that their statements and assumptions match. Both now sit where Alpöge's does: awaiting expert review.
What to watch next
Three checks are outstanding. First, an independent statement check: a recognized expert confirming that the Lean theorem is the Riemann Hypothesis as standardly stated, and that the proof does not rest on extra assumptions, axioms or admitted steps. Second, replication of the compilation by other groups, on the full 14-part archive. Third, any human-readable account of the argument's main ideas, which Alpöge has promised to work toward.
If the first check fails, the story ends there, whatever the code proves. If it passes, the remaining question changes from whether the result is true to whether anyone can understand why.
Sources
- [1] Levent Alpöge, personal note (primary source), November 20, 2026Read source
- [2] Scientific American, "Anthropic mathematician claims to prove Riemann hypothesis, most famous problem in all of math," November 20, 2026Read source
- [3] TechCrunch, "AI claimed a solution to the 167 year-old Riemann Hypothesis — and a human did, too," November 20, 2026Read source
- [4] Phys.org (AP), "Anthropic researcher says Claude solved math's greatest riddle," November 20, 2026Read source