Sébastien Bubeck, the OpenAI researcher at the center of the Navier-Stokes credit dispute, has given his account of the private talks with the two mathematicians who claim OpenAI mistreated them. Bubeck denies trying to strip one of them of credit for their own work and apologizes for a remark he made during a call.
Bubeck set out his version of the events in a post on X, where he explained that he first contacted Levent Alpöge to coordinate their releases. He claims he came in with the best possible intentions and was surprised to learn that the pair had solved the Euler problem, not Navier-Stokes.
After the call, the two sides brainstormed options, including making Tristan Buckmaster the lead author on a rewrite of OpenAI's Navier-Stokes proof. Bubeck says he suggested it would be simpler if Alpöge was not an Anthropic employee, as it would be inappropriate for an Anthropic employee to author OpenAI's work.
Bubeck claims he never asked to remove Alpöge from authorship of his own work and that the pair had admitted using internal Anthropic models on their Euler proof. He offered the pair access to OpenAI's internal model to try to close the gap from Euler to Navier-Stokes.
The call turned hostile, with Bubeck claiming he was met with a litany of slander, including a threat to go to the press with unfounded accusations if OpenAI announced its result. He apologized for asking why anyone would risk their career over such accusations, and retracted the remark on the spot.
OpenAI chief executive Sam Altman backed Bubeck in his own post, saying that Bubeck and everyone else acted with integrity and generosity throughout. Altman confirmed that OpenAI tried to solve the Navier-Stokes problem after rumors emerged that Anthropic's models had solved a Millennium problem.
Buckmaster's account of the events tells a different story, with him claiming that Bubeck twice asserted that he wanted Alpöge removed from authorship. The dispute remains unresolved, with no independent process having tested the competing accounts.
The question of whether OpenAI accessed the pair's user data through its coding tool, Codex, remains unclear. OpenAI says no specific user data was accessed, but cannot rule out that de-identified data derived from their usage helped improve its models.
Questo articolo è stato scritto con l'assistenza dell'IA.
News Factory APP - notizie agentiche per potenziare il tuo SEO e AEO.