OpenAI Claims Navier-Stokes Breakthrough, but Accusations of Academic Scooping Follow
OpenAI announced Tuesday that an unreleased AI model solved the Navier-Stokes Millennium Prize problem in 88 hours using roughly 10,000 AI agents, but the announcement immediately drew accusations of scooping a rival research team.
👁 114 reads
OpenAI announced on Tuesday that one of its unreleased models solved the Navier-Stokes Millennium Prize problem - a $1 million, nearly 90-year-old challenge about fluid movement - in just 88 hours, deploying a swarm of roughly 10,000 AI agents on the task. The company called the result a “milestone,” but the announcement landed in the middle of an unfolding dispute over how OpenAI came to pursue the problem in the first place.
- An unreleased OpenAI model solved the Navier-Stokes fluid-dynamics problem - worth a $1 million prize - in 88 hours using a swarm of roughly 10,000 AI agents
- NYU professor Tristan Buckmaster says OpenAI became aware of his related work, then pressured him to drop coauthor Levent Alpöge and credit OpenAI's model instead
- OpenAI flatly denied accessing user data, but the episode has prompted allegations of scooping and violations of academic norms
One day before OpenAI’s blog post, New York University mathematics professor Tristan Buckmaster published findings on a closely related problem, co-authored with Levent Alpöge, a researcher at Anthropic - though Alpöge was acting independently, not on behalf of his employer. Buckmaster said he reached out to OpenAI after learning the company had become aware of his team’s progress. The exchange, he said, turned hostile quickly. An OpenAI researcher told him that going public “would ruin your career,” and when Buckmaster pressed further, the researcher reportedly replied: “If you don’t want me to be nice, then I don’t have to be nice.” OpenAI also urged Buckmaster to publish the work crediting its internal model and to drop Alpöge as a coauthor.
Buckmaster said he asked OpenAI directly whether it had accessed his sessions on Codex, a tool he had used while working on the problem. He described OpenAI’s responses as increasingly evasive. In its blog post, OpenAI denied any such access: “We (the researchers and the agents) did not see any of their work through any means until they released it publicly - in particular, no specific user data was accessed in order to solve this problem.”
Abhishek Saha, a mathematics professor at Queen Mary University of London, described OpenAI’s conduct as “the kind of things that mathematicians will generally not do.”
Practitioner takeaway: For developers and creators who use OpenAI tools like Codex to explore research or prototype ideas, this episode is a concrete reminder that work done inside a commercial AI platform is not the same as work kept in a private notebook. Even if OpenAI’s denial holds, the controversy shows that companies can observe usage patterns, and the line between platform provider and research competitor is not fixed. Anyone working on novel or competitively sensitive problems should think carefully about which tools they use and what those tools’ operators can see.
Google · Your sourcesQuantslant higher in your Google - one clickAdd to preferred sources →
More on this from QUANTSLANT
benchmarks·Nano Banana 2.1: Google Cuts Price in Half - Launch Day TestGoogle released Nano Banana 2.1 with 50% lower pricing, 8:1 panoramas, and better mask editing. We tested it on launch day across 6 benchmarks.