Back to News

A claimed Riemann hypothesis proof leaves RSA, banks and Bitcoin untouched

Anthropic's Levent Alpöge says he proved the Riemann hypothesis with Claude. Even if the Lean proof checks out, it does not weaken RSA or the encryption behind payments and crypto. The prize is $1 million, and the real encryption risk is still quantum computing.


A claimed Riemann hypothesis proof leaves RSA, banks and Bitcoin untouched

Anthropic mathematician Levent Alpöge said on Nov. 20 that he has proved the Riemann hypothesis with the help of Claude, in a statement on his website. The claim is not yet verified. Even if it holds, it does not weaken the encryption that protects bank transfers, card payments or crypto wallets.

Alpöge released the proof as Lean code, a language that checks each logical step by machine, New Scientist reported. He has not posted a human-readable version and said an unpublished draft ran past 250,000 pages. He also says the method proves the Generalized Riemann Hypothesis.

"Let's see if it compiles first," Kevin Buzzard of Imperial College London told New Scientist. No independent check had been reported as of Friday afternoon, Crypto Briefing reported.

What the Riemann hypothesis says

Bernhard Riemann stated the conjecture in 1859. It says every nontrivial zero of the zeta function has real part 1/2, which fixes how far the count of primes can stray from its average, according to the Clay Mathematics Institute. Clay says the first 10 trillion zeros have already been checked.

Why RSA and bank encryption are unaffected

RSA, the oldest widely used public-key scheme, is safe only as long as factoring a large number into its two primes stays hard. The Riemann hypothesis speaks to how many primes there are below a bound. It does not tell you which two primes make up a given key.

Cryptographers and number theorists have also treated the hypothesis as true for decades. Any attack that needed it could have been built and run already. "We can simply accept it as true now," reads a long-standing answer on Cryptography Stack Exchange. One answer on Mathematics Stack Exchange calls the claimed link to cryptography "greatly exaggerated and overhyped". Algorithms whose speed was proved only under the hypothesis, such as an RH-conditional factoring method, run no faster than elliptic-curve factoring, which needs no such assumption.

Elliptic-curve signatures, used in most web traffic and in Bitcoin and Ethereum wallets, rest on a different hard problem, and the hypothesis says nothing about it.

The cryptography risk banks are spending on

The threat to today's encryption is quantum computing. NIST published its first three post-quantum standards, FIPS 203, 204 and 205, on Aug. 13, 2024, and told system administrators to start moving to them immediately, the agency said. A Riemann proof does not change that timeline or its budget.

Where a proof could matter

The one real channel is the method. If the tools behind the proof give new structure on primes, someone could later turn them into a faster factoring algorithm. That would be a separate result, published and checked in the open, and the post-quantum migration already under way would cover most exposed systems.

The money directly at stake is small. Clay's Millennium Prize pays $1 million, and its rules require publication in a qualifying outlet, two years to pass after publication, and general acceptance by mathematicians before Clay considers a solution.

The larger market story is Anthropic itself. The company listed on Nov. 19, and Alpöge wrote that he held the announcement until after the listing, New Scientist reported. A verified proof would be the strongest public evidence yet of what frontier AI models can do in research, and investors are likely to price it into AI stocks, not bank stocks.

What to watch

  • Next few days: whether Lean experts confirm the code compiles and that its formal statement matches Riemann's.
  • Next few weeks: a human-readable write-up, and comment from number theorists on the method.
  • Nov. 2028 at the earliest: if a paper appears in a qualifying outlet this month, the two-year wait means Clay could not take it up before then.