An event has occurred that will go down in the history of not only cryptocurrencies but all of mathematics. On July 20, Anthropic researcher Levent Alpoge announced that the Claude Fable 5 language model found a counterexample to the Jacobian conjecture — one of the most famous unsolved problems in algebraic geometry. This is not just another benchmark, but a full-fledged breakthrough that overturns our understanding of AI capabilities.

What is the Jacobian conjecture and why is it so important?

Formulated by German mathematician Ott-Heinrich Keller in 1939, this conjecture has long been considered a cornerstone. Simply put, it states that if we have a polynomial function of several variables that is locally invertible at every point (i.e., its derivative is non-zero), then it is also globally invertible. In other words, if a function behaves "well" everywhere, the original data can always be uniquely reconstructed from the result. Over 87 years, there have been many erroneous proofs, but none correct.

How Claude Fable 5 broke the rule

Claude Fable 5 constructed a specific formula that passes the local test at every point, yet maps three different sets of numbers to the same result. This means the function cannot be inverted — the conjecture is false. Alpoge published links to Wolfram Alpha, where anyone can personally verify the calculations. Notably, the question about the conjecture was posed by a friend of the researcher, and Claude itself worked on the problem during the FIFA World Cup final.

Reaction from the scientific community and context

Mathematician Qiaochu Yuan called this "the most famous open problem solved by a language model." He emphasized the irony: the Jacobian conjecture is notoriously known for attracting false proofs. For example, Beniamino Segre published three incorrect proofs, and Igor Shafarevich used it as a proven theorem. Jared Duker Lichtman called the result "outstanding," recalling that a special case of the conjecture was the topic of Ethan Zhang's dissertation — the future author of a breakthrough in prime number theory.

My analysis: This is not just another AI victory over a mathematical puzzle. The main difference between this result and previous claims by OpenAI and Google DeepMind is that it is absolutely verifiable. Unlike solutions to Erdős problems, which often require complex verification through formal systems, a counterexample to the Jacobian conjecture is simply a formula that can be plugged in and checked. This undermines trust in "closed" claims by AI companies and sets a new standard for transparency.