Experts say OpenAI's Navier-Stokes proof uses a different approach

The short version

• Scientific American reports that OpenAI's early September Navier-Stokes solution differs from the approach initial accounts described.[Scientific American]
• The proof was produced by an agent swarm of about 10,000 copies of an internal LLM, using roughly 88 hours of computing time.[Scientific American]
• Mathematician Tristan Buckmaster alleges OpenAI tried to bribe and threaten him, and these allegations remain unverified in the reporting.[Scientific American]
• Mathematician Camillo De Lellis says the dispute no longer rests on clearly verifiable mathematical facts.[Scientific American]

Key facts

• OpenAI posted a 166-page solution to the Navier-Stokes problem in early September.[Scientific American]
• Initial accounts, including from OpenAI, suggested the proof followed an approach devised by Spanish mathematicians Diego Córdoba and Luis Martínez-Zoroa.[Scientific American]
• Mathematician Gonzalo Cao-Labora says that on close examination the proof takes a different approach.[Scientific American]
• OpenAI used an agent swarm of about 10,000 copies of its math-adept LLM, which spent about 88 hours of computing time and 130 billion tokens constructing a blowup.[Scientific American]
• Tristan Buckmaster has alleged that OpenAI attempted to bribe and threaten him to prevent him from going public with insinuations of malfeasance.[Scientific American]

Sources

Outlet counts describe coverage, not independent confirmation. Reports may share a wire service or original source.

• OpenAI’s blockbuster Navier-Stokes proof isn’t what we thought it wasScientific American metered