Mathematicians are asking OpenAI to prove that it did not use unpublished research in solving the Navier-Stokes equations and to demonstrate how it accessed the information. The debate centers on separate work by Tristan Buckmaster and Levent Alpöge, and on information about this work being conveyed to OpenAI on September 3. OpenAI, meanwhile, says it launched the initiative on September 1, did not use the researchers’ prompts or proofs, and did not access private data. However, the company does not entirely rule out anonymized usage data contributing to model development.
Why it matters
The significance of this debate lies in the questions it raises about which channels artificial intelligence systems use to access scientific knowledge and how their relationship with unpublished work can be monitored. The differing accounts of the date when the initiative began and when the researchers’ information reached the company mean that not only the technical result but also the transparency of the process must be examined. For mathematicians, the open question is how it can be proven that specific prompts and proofs were not used; for OpenAI, it is how the boundaries of anonymized usage data’s role in model development should be defined. This distinction has become decisive for those seeking to assess whether researchers’ work was processed indirectly, without their knowledge.
Background
OpenAI is not a new name in the FikirPilot archive: we have published 29 news reports mentioning the name in the last 90 days; the most recent is dated September 11, 2026.