Post by @vesting_tv

Swipe to see the $1 million problem OpenAI says its AI solved, and the mathematician calling it theft. OpenAI says one of its internal models produced a proof for a problem tied to the Navier-Stokes equations, part of the Clay Institute's Millennium Prize Problems, each of which carries a $1 million reward and has stood open for decades. The company says the model resolved it over a single weekend using around 10,000 AI agents running in parallel, at a computing cost reportedly in the millions. The claim is disputed. NYU mathematician Tristan Buckmaster says he and Levent Alpöge had already made major progress on the same approach, and that word of it reached OpenAI before they could publish, with the company finishing the job off their direction. Buckmaster has also questioned whether private drafts stored in OpenAI's own tools could have fed the model, and says he didn't get a clear answer on the training question. It's a live disagreement, not a settled result, and OpenAI has claimed math breakthroughs before that didn't hold up, including a GPT-5 post last year that was pulled after the "solutions" turned out to already exist in the literature. Alpöge works at Anthropic, the company behind Claude, which wrote this. Source: Scientific American and Entrepreneur. Subscribe to the Vesting daily for more stories like this: vesting.beehiiv.com/subscribe #OpenAI #AI #Mathematics #MillenniumPrize #technews