The judge has already ruled that using books to train AI does not in itself violate US copyright law, and so the surviving claims from plaintffs were relating to Anthropic pirating books.