Company Meta it won one of the key processes related to copyright and the use of data for training generative artificial intelligence. Judge of the U.S. Federal District Court Vince Chabbria dismissed a class-action lawsuit by a group of 13 authors, including writers Sarah Silverman and Ta-Nehisi Coateswho have accused the company of illegally using their works when teaching the language model Llama.
The court recognized that the actions of Meta are within the framework of the doctrine fair use — the principle of American copyright, which allows limited use of protected materials without the consent of the copyright holder, if this pursues transformative goals.
The judge noted that while "it is generally illegal to copy protected works without permission," in this case the plaintiffs couldn't prove any damage to the market or a substantial risk of such damage. The decision states that the arguments presented by the authors were "unconvincing" and "contradictory".
Chabbria clarified that Meta used books not in their original form, and for the purpose of transformation and generation of new content that falls under the protection of fair use. He stressed that the company has refuted accusations that its practice creates a significant market-like effect.
Meta said it welcomed the court's decision, noting in an official statement::
"Open AI models are the foundation of innovation and productivity. The legal concept of fair use plays a key role in the development of these technologies."
Despite this, the judge pointed out that questions about the market harm caused by the use of copyrighted texts in AI training remain open. In particular, he criticized Meta's position that banning the use of protected data without payment would seriously harm the public interest.
"Meta claims that this will stop the development of generative models. This is absurd, " Chabbria said.
The judge also clarified that the decision made it is not universal:
"This is not a class action, and the decision concerns only the 13 authors who participated in the trial. It does not mean that the use of Meta copyrighted works to train AI in general is legitimate."
The text of the resolution also contains a reminder that one of the claims remains unresolved — we are talking about a possible illegal distribution of protected texts via torrent networkswhich is also blamed on Meta.
Against the background of this case, a parallel lawsuit against the company continues. Anthropic, developer of the Claude model. This week, another federal court found that using books to teach AI could also comply with the fair use doctrine, but noted that the case should be considered in court, since the company could download millions of pirated copies of works.
"If Anthropic later purchased copies of books previously downloaded from the Internet, this does not release it from liability, although it may affect the amount of compensation," the judge said in an accompanying decision.
