Meta wins copyright lawsuit over AI training, but judge signals door still open for future claims

Meta Platforms secured a legal win on June 25 in a copyright case involving the use of books to train its generative AI model, Llama. Thirteen authors, including comedian Sarah Silverman and novelist Ta-Nehisi Coates, had alleged that Meta infringed on their copyrights by using pirated versions of their works without consent or compensation. However, US District Judge Vince Chhabria ruled that the plaintiffs failed to present a compelling legal argument that Meta’s actions violated US copyright law.

The case, one of the first major legal battles testing the limits of fair use in the AI era, revolved around whether training AI models on copyrighted content without permission constitutes infringement. In his ruling, Judge Chhabria emphasised that the fair use doctrine protected Meta’s use of the books in this instance, noting that the plaintiffs could not prove that the company’s practices caused meaningful harm to the market for their works.

“On this record, Meta has defeated the plaintiffs’ half-hearted argument that its copying causes or threatens significant market harm,” Chhabria wrote. He acknowledged that using copyrighted materials without permission is “generally illegal” but criticised the authors’ legal strategy as poorly developed. “This ruling does not stand for the proposition that Meta’s use of copyrighted materials to train its language models is lawful,” Chhabria clarified. “It stands only for the proposition that these plaintiffs made the wrong arguments and failed to develop a record in support of the right one.”

Fair use doctrine shields Meta for now

Meta’s core legal defence relied on the fair use doctrine, a provision of US copyright law that permits limited use of copyrighted material without the rights holder’s permission for purposes such as criticism, commentary, or transformative uses. Meta argued that its AI training process was transformative, as it did not reproduce the original works but instead used them to teach the AI to generate new, original text.

“No one can use Llama to read Sarah Silverman’s description of her childhood or Junot Diaz’s story of a Dominican boy growing up in New Jersey,” Meta’s attorneys wrote in filings. The company also claimed that the authors’ works could not be substituted by AI-generated content and that there was no evidence of market harm. “After nearly two years of litigation, there still is no evidence that anyone has ever used Llama as a substitute for reading Plaintiffs’ books, or that they even could,” the lawyers stated.

While Chhabria acknowledged that Meta’s use may have economic implications, he noted that no significant harm was demonstrated in court. He did, however, criticise Meta’s assertion that banning the use of copyrighted works in AI training would “badly disserve” the public interest or stifle innovation. “This is nonsense,” Chhabria wrote, calling Meta’s claim exaggerated and speculative.

Ruling limited in scope, new lawsuits likely

Though Meta prevailed in this case, the ruling is not a sweeping endorsement of AI companies’ use of copyrighted materials. Judge Chhabria stressed that the decision applied only to the 13 plaintiffs and did not provide blanket legal protection for Meta’s practices.

“In the grand scheme of things, the consequences of this ruling are limited,” he wrote. “This is not a class action, so the ruling only affects the rights of these thirteen authors, not the countless others whose works Meta used to train its models.” Chhabria also left open the possibility of future legal challenges, suggesting that a more thoroughly developed case could reach a different outcome.

The ruling came just days after another case involving AI firm Anthropic, in which Judge William Alsup found that the company’s use of copyrighted books to train its Claude chatbot was also “transformative” and qualified for fair use. However, that case will still go to trial over allegations that Anthropic illegally downloaded pirated books from online sources.

Chhabria criticised Alsup’s decision for focusing too heavily on the transformative nature of AI without adequately weighing potential harm to content creators. He expressed concern about the broader implications of generative AI on creative markets, saying such tools could undermine the incentive to produce original work. “By training generative AI models with copyrighted works, companies are creating something that often will dramatically undermine the market for those works,” Chhabria noted.

More legal challenges loom for AI companies

The recent ruling does not close the book on legal scrutiny of AI training practices. The plaintiffs’ attorneys argued that Meta used pirated versions of books from “shadow libraries”, with internal Meta documents showing the decision was escalated to CEO Mark Zuckerberg due to its potential legal risk. A separate claim related to Meta’s alleged illegal distribution of copyrighted works via torrenting is still pending.

Meta has joined a growing list of tech companies, including OpenAI, Microsoft, and Anthropic, that face lawsuits from authors, news organisations, and other rights holders over the unauthorised use of copyrighted materials in AI training. As generative AI continues to disrupt content creation industries, more complex legal battles over the limits of fair use are expected to emerge.