OpenAI is facing pushback from academic mathematicians over its strategy for publishing hundreds of AI-generated solutions to open mathematical problems. In August, the company convened approximately 40 mathematicians to consult on responsible release practices, but academics report their recommendations to publish formal peer-reviewed papers were ultimately ignored in favor of dropping raw results on GitHub.
OpenAI spokesperson Lindsay McCallum confirmed that an internal model trained in late August resolved over 100 open problems, including work on the Navier-Stokes Millennium Prize problem. The firm stated it is drawing on public recommendations from the Institute for Advanced Study’s advisory group to inform the release, though it has not finalized a specific timestamp.
The dispute highlights growing friction between academic research protocols and competitive pressures among frontier AI labs like OpenAI and Anthropic. Leading researchers allege that labs are prioritizing rapid public demonstrations to showcase capability ahead of potential initial public offerings at the expense of traditional scientific attribution.
Why it matters
Frontier models are rapidly demonstrating core scientific discovery capabilities, threatening to disrupt traditional academic peer-review processes.
Highlighting IP and attribution disputes between researchers and AI labs, emphasizing the need for clear rules on AI-assisted research credit.
Source: wired.com



