OpenAI withdraws three mathematical results
A sign error in OpenAI's paper on Weil classes on split abelian eightfolds invalidated a key cancellation argument, taking down that result and two papers built on it, including a claimed proof of the rational Hodge conjecture for products of K3 surfaces. Fourteen other manuscripts were revised with proof repairs, and only about 42% of the 719 top-line results are formalized in Lean so far.