META WINS AI COPYRIGHT CASE, BUT QUESTIONS LINGER

A federal judge in California has ruled in favor of Meta Platforms, dismissing a copyright infringement lawsuit filed by 13 authors, including Sarah Silverman, Ta-Nehisi Coates, and Jacqueline Woodson.

- 

The authors accused Meta of using their books without permission to train its Llama AI model, claiming the practice violated their intellectual property rights. U.S. District Judge Vince Chhabria granted Meta’s motion for summary judgment, finding that the authors failed to prove Llama’s outputs infringed their copyrights or caused market harm.

The lawsuit, one of several targeting AI companies, argued that Meta illegally sourced books from “shadow libraries” like Bibliotik and Library Genesis, online repositories of pirated works. The authors sought damages, alleging that Llama’s ability to generate text relied on their creative content without consent or compensation. Chhabria acknowledged that training AI with copyrighted material is often illegal but ruled that the plaintiffs’ arguments missed the mark, focusing on weak claims about Llama itself being an infringing derivative work rather than proving market damage.

Meta celebrated the decision, calling fair use a “vital legal framework” for AI innovation. The authors’ lawyers disagreed, insisting that Meta’s use of pirated works was unlawful and vowing to explore further legal steps. The ruling, limited to these 13 authors, leaves the door open for future cases, with Chhabria hinting that stronger arguments could succeed.

THE FAIR USE DEBATE