Tristan Buckmaster, an NYU mathematician, publicly accused OpenAI in early September 2026 of leveraging unpublished methods from his Codex sessions—developed with Anthropic collaborator Levent Alpöge on fluid dynamics problems related to the Navier-Stokes equations—after learning his identity, while pressuring him over authorship credit in a tense call with researcher Sébastien Bubeck. OpenAI categorically denied any influence from his prompts or training data, stating its internal large language model produced a separate proof with minimal human input, and emphasized the results addressed distinct equation variants. The episode highlights competitive pressures among AI labs racing to demonstrate advanced mathematical reasoning capabilities, alongside researcher concerns about data privacy, intellectual property leakage via AI coding tools, and ethical credit allocation. Buckmaster has since avoided sharing original ideas with OpenAI platforms but has not filed suit as of mid-October, leaving any legal timeline dependent on further verification of the proofs or new disclosures.