AI · SEMICONDUCTOR · FACT CHECK
SKD WIREMostly True
OpenAI says its AI has produced proofs for three difficult problems spanning mathematics, computer science and physics, described as challenges that have remained open for decades. The company's claims, along with a review framework said to include Nobel laureates, are now facing formal scrutiny from the academic community, though independent verification is still in progress and the physics case cannot yet be assessed because the underlying material has not been released.
According to OpenAI's announcement, the AI addressed one problem each in mathematics, computer science and physics, all framed by the company as decades-old unsolved challenges. The company has also put forward a verification structure for the work, with participation from Nobel laureates, establishing the basis for academic review of the results.
The mathematics and computer science claims rest on primary sources that establish the problems' existence and their decades-long history. However, the headline framing that the AI "proved" the problems reflects OpenAI's own assertion rather than an established fact at this stage.
The physics claim stands on different footing: the source material has not been made public, leaving reviewers unable to examine the work at all. Independent academic verification of the announcement as a whole is underway but not complete, meaning the AI's results cannot yet be treated as confirmed by the research community.
OpenAI's central assertions — that the announcement occurred, that it concerns three named difficult problems, and that a review process involving Nobel laureates exists — are supported by primary documentation. What remains open is whether the proofs themselves hold up, a question that only the ongoing academic review can answer. On the facts as they stand, the claim is Mostly True.