OpenAI faces fraud claim in Navier-Stokes dispute

The dispute centers on OpenAI's claimed AI solution to the Navier-Stokes equations, a Millennium Problem. Buckmaster and his co-author Alpöge say they had entered a similar approach into OpenAI's systems, which may have been used for training without consent. OpenAI employees argue the chance is low, and CEO Sam Altman backed the team, calling the accusations unfounded. However, OpenAI's official acknowledgment that it cannot rule out data usage has intensified concerns.

The case highlights how AI labs can rapidly mobilize resources based on rumors, as OpenAI did after hearing Anthropic's models might have solved the problem. This raises questions about the fairness of racing to publish results that build on others' private work. For the research community, the takeaway is that feeding data into AI platforms could lead to being beaten by one's own findings. Terence Tao's warning underscores a potential shift away from open sharing, which could damage long-term scientific progress. The outcome depends on whether any evidence emerges that OpenAI trained on the researchers' submissions, which remains publicly unknown.