Two philosophers argue on Tao's blog that OpenAI produced an answer to Navier-Stokes, not a proof.

Terry Tao hosted a September 12 guest post by Silvia De Toffoli and Eamon Duede attacking two assumptions behind the coverage: that AI solved a mathematical problem, and that mathematics is only about solving problems. Their split is between logical validity and intelligible understanding, granting that "a Lean formalization meets these standards exactly" while arguing mathematicians want something a machine-checked certificate doesn't supply. The recommendation isn't defensive: treat AI as a technology for advancing mathematics' human purposes rather than a competitor to beat. Separately, computational biologist Lior Pachter published a direct rebuttal to the 25-signatory Fields medalist declaration, which makes that document a two-sided argument instead of a consensus statement.