Meta Wins Lawsuit Over AI Training on Books: Copyright Not Infringed

A US federal judge has ruled in favor of Meta , dismissing a copyright infringement lawsuit filed by a group of authors who accused the company of violating copyright law by training the artificial intelligence “Llama” on their creations without permission.

However, San Francisco District Court Judge Vince Chhabria left the door open for similar lawsuits, noting that the ruling "does not say that Meta's use of copyrighted materials to train its language models is legal. It only finds that the plaintiffs have made incorrect arguments and failed to present evidence to support their correct ones."

A few days earlier, the same court had similarly ruled that the company Anthropic had not broken the law by training its chatbot “Claude” with copyrighted books, because the process had been sufficiently “transformative” (even though it will be prosecuted for illegally acquiring those books from pirated websites). Judge Chhabria criticized this view, saying that too much emphasis had been placed on the transformative aspect of AI, while ignoring the potential harm to the market for original works , and suggested that such harm could constitute a valid legal argument.

“As transformative as training (generative AI) might be, it is difficult to imagine that it would be fair use to use copyrighted books to develop a tool that would allow billions or trillions of dollars of revenue to be made, while also enabling the creation of a potentially endless stream of competing works that could significantly harm the market for those books,” the judge said in his ruling.

Massive amounts of data are needed to train the large language models that power generative AI. Musicians, book authors, visual artists, and news organizations have sued multiple AI companies that have used their data without permission or payment.