Mathematicians challenge OpenAI over use of unpublished work

Mathematicians are pressing OpenAI for clarity on the data behind its models’ recent mathematical results, following a dispute over whether unpublished work informed those outputs. According to The Verge, a second mathematician has come forward with accusations following an earlier, contentious disagreement over the same issue days prior.

The complaint centers on transparency. The researcher characterized OpenAI’s conduct as unethical and, in the researcher’s own word, “dishonest,” according to the report, and said the company has not adequately disclosed the origin of the material its systems appear to draw on. The dispute follows an earlier clash in which another mathematician raised similar concerns about whether OpenAI’s models had benefited from work that had not yet been published.

The Verge frames the pattern as a recurring question for OpenAI, whose models have produced increasingly notable mathematical results in recent months. The report does not specify which unpublished materials are alleged to have been used, nor does it detail OpenAI’s response to the newer claim.

The story was also circulated on Hacker News, where it drew further discussion, per the aggregation.

As of this filing, the underlying data pipeline behind OpenAI’s mathematical capabilities remains undisclosed by the company, and the researchers’ specific allegations have not been independently verified beyond the accounts reported by The Verge.