OpenAI Claims AI Model Solves Long-Standing Math Puzzle in 88 Hours

OpenAI announced on Tuesday that an unreleased artificial‑intelligence model had produced a solution to a mathematical puzzle that has resisted proof for generations. The company said the model, developed internally and not yet available to the public, generated a novel proof that satisfies the rigorous standards of the mathematical community, marking what OpenAI described as a “milestone in AI‑driven discovery.” The announcement referenced the problem’s long history in the field of number theory, noting that prior attempts by human researchers had yielded only partial results.

The claim sparked a dispute with a group of independent researchers who had been working on the same problem, with both parties contesting credit for the breakthrough. OpenAI’s statement emphasized that the model’s output was derived from its own training data, while the rival team argued that earlier collaborative efforts had laid essential groundwork. Observers noted that the episode highlights emerging tensions over intellectual ownership as AI systems become more capable of contributing to advanced scientific research.