Meta Wins in Court: AI Training on Protected Books Falls Under Fair Use | WhatsApp Meta AI | Meta AI image generator | Meta Facebook login | Turtles AI

A U.S. federal judge, Vince Chhabria, has issued a summary judgment in favor of Meta in a lawsuit filed by 13 authors, ruling that the use of their books to train AI models falls within fair use, but does not establish general lawfulness.

Key Points:

• Summary judgment: Chhabria ordered a non-jury dismissal in favor of Meta due to the plaintiffs’ lack of evidence and argument.
• Transformative use: The training was found to be transformative, the models do not entirely reproduce the works.
• No evidence of harm: The authors failed to demonstrate market impact or dilution, which is essential in fair use.
• Limited applicability: The decision does not ratify the systematic legality of the use of protected material, but applies to the specific facts of the case.

A California federal judge, Vince Chhabria, has dismissed a lawsuit filed by 13 authors — including Sarah Silverman and Ta-Nehisi Coates — who accused Meta of using their copyrighted books to train the Llama language model. In his summary judgment ruling, Chhabria repeatedly noted the inadequacy of the plaintiffs’ arguments and documentation, saying that “Plaintiffs have advanced the wrong arguments.” The court found the training to be “transformative” use, as the models generate new content and do not restrict the full reproduction of the original texts. It also found that the authors failed to demonstrate to any significant extent either economic harm or market dilution — key elements in the fourth factor of fair use. While the ruling is a technical victory for Meta, the judge noted that it is only relevant in its own context, leaving the door open for future lawsuits based on more specific facts.

This ruling comes a few days after a similar decision in favor of Anthropic, which had also invoked fair use for training AI models on books, but had received criticism for using pirated material. In that case too, the Court distinguished between purchased material and illegal content, declaring that the use of legally acquired books was protected, while legal questions about pirated texts remain open.

The judges in both cases highlighted the importance of market damage, introduced the new theory of "dilution", or the possibility that AI produces competing content that reduces interest in original works. This concept risks redefining future arguments on fair use in the AI ​​field.

Similar lawsuits continue to rain down in the US judicial landscape: the New York Times has sued OpenAI and Microsoft for the use of newspaper articles; Disney and Universal have sued Midjourney for using protected films and TV programs; Microsoft itself has been sued by a group of authors for allegedly adopting pirated texts in its Megatron model. The current decisions outline a line of interpretation favorable to fair use conditioned by concrete contexts and factual convolutions, but none provides generalized legislative coverage for indiscriminate training.

The Chhabria decision represents an important step in the legal debate on the treatment of protected works in the AI ​​era, focusing on solid evidence and the demonstration of real market bias.