Tristan Buckmaster's own statement: the Euler write-up 'can only be described as AI slop', and the credit belongs to Córdoba and Martínez-Zoroa
Buckmaster (NYU) posted a signed statement alongside three results he released with Levent Alpöge (Anthropic) on 2026-09-07: finite-time blowup with smooth forcing for incompressible porous media, Boussinesq, and 3D incompressible Euler, plus an unreleased hypo-dissipative Navier–Stokes result whose Lean verification had not finished. He states the program was neither theirs nor proposed by an LLM, credits Diego Córdoba and Luis Martínez-Zoroa, and says Martínez-Zoroa deserves a Fields Medal. Concrete timeline: slow progress for most of a year, blowup results on August 15, Lean verification on August 22, and he calls the first LLM-generated proof Alpöge sent him 'the most horrendous I have ever read'. He used Claude and Codex (mainly GPT-5.6 Sol), with Astra only for write-ups and auditing, and apologises for presentation quality under outside pressure.
↳ Follow the thread