Meta Platforms Inc. was unable to escape legal claims that it violated copyright law by removing copyright notices attached to millions of books used to train its generative AI model called Llama.

Judge Vince Chhabria Friday ruled that a proposed class of authors, including Sarah Silverman and journalist Ta-Nehisi Coates, can advance claims that Meta violated the Digital Millennium Copyright Act by removing their books’ copyright notices in an effort to conceal infringement.

The ruling comes after the authors produced internal documents from Meta showing that top executives including CEO Mark Zuckerberg approved AI training on pirated books and the …