OpenAI's Navier–Stokes claim: what the proof says (and what it doesn't)

Process map by Brian. 50 steps, 2 layers. Tags: science, math, physics, ai, navier-stokes.

The Clay problem: Clay asks: can a smooth 3D fluid flow break down in finite time?
The Clay problem: Which version of the problem?
The Clay problem: Still unsolved. This proof does not cover it.
OpenAI's agents: How the agents got there
OpenAI's agents: State the result: a smooth force can make a fluid blow up
OpenAI's agents: Build the blow-up: a vortex that contracts
OpenAI's agents: Keep the force smooth by cancelling the errors
The fluid: Velocity grows without bound at a finite time while energy stays finite
OpenAI's agents: It holds for every positive viscosity, not just an idealized fluid
OpenAI's agents: A periodic version too (Corollary 10.6)
Lean proof checker: Translate the whole argument into Lean
Mathematicians and reviewers: Experts read the 166-page manuscript
Mathematicians and reviewers: Independently verified and accepted?
The Clay problem: Clay's own process: publication, a two-year wait, broad acceptance
Mathematicians and reviewers: Where things stand today: claimed, not yet independently verified