According to the complaint, the authors’ books were part of a dataset called “The Pile” that Nvidia received from the open-source AI research organization EleutherAI.
Nvidia, Databricks sued by authors over the unapproved use of copyrighted material in training AI