SAN FRANCISCO: A U.S. judge ruled in Meta’s favor on Wednesday in a lawsuit where authors accused the tech company of violating copyright laws by training its Llama AI model on their books.
Judge Vince Chhabria stated that Meta’s use of the copyrighted materials was “transformative” enough to be considered fair use under U.S. copyright law. However, he also noted that the authors could have made stronger arguments about the potential market harm caused by AI tools.
The court acknowledged that while using books to train AI may be transformative, it could still damage the authors’ market by enabling the creation of competing content without permission or payment.
This decision marks Meta’s second legal win in a week, following another ruling where a San Francisco judge sided with AI company Anthropic in a similar copyright case. Both judgments strengthen the argument that AI training can fall under fair use.
Still, Judge Chhabria clarified that this decision doesn’t prove Meta’s use of the materials was fully lawful. He explained that the authors simply failed to present the right legal arguments and evidence to support their claims.
Authors like Sarah Silverman and Junot Diaz had filed the case, claiming Meta used pirated copies of their works to train its open-source Llama AI. Meta welcomed the court’s ruling, saying fair use protections are vital for building innovative AI tools.