Databricks can't seem to shake authors' copyright claim that could result in 'extraordinary' damages

Databricks faces ongoing copyright infringement lawsuit from bestselling authors claiming its DBRX language model was trained using a dataset containing pirated versions of approximately 196,000 books. A federal judge rejected Databricks' motion to dismiss, allowing the suit to proceed. The core dispute involves whether DBRX inherited infringing training data from the RedPajama dataset used by Databricks' acquired Mosaic team.