OpenAI's machine-checked proof does not match the proof it published.
Three mathematicians posted a paper on Tuesday about OpenAI's claimed Navier-Stokes breakthrough. They say the Lean proof the machine verified does not correspond to the written proof of fluid breakdown. They also argue faithful translation into Lean is harder than any computable problem, including the halting problem.
Why it mattersLean verification is the industry's answer to the question of who checks AI mathematics. If the checked thing is not the claimed thing, that answer stops working.
▶ Related video: Did AI Really Solve Navier–Stokes? Show Me The Proof!
Read in the app — free, in 9 languages
Related stories
Nvidia found one training step changes just 1% of a model.2026-10-07OpenAI posted 722 maths papers from one prompt to one agent.
2026-10-07The Erdős problems site stopped taking proofs after an AI flood.
2026-10-07Twenty tries at the same task cut the best agent's score by a third.
2026-10-06A model learned to write without the method that trains every AI.
2026-10-06