A U.S. judge ruled in favor of Meta on Wednesday in a copyright lawsuit brought by a group of authors who claimed their works were unlawfully used to train the company’s Llama AI model without authorization.
District Judge Vince Chhabria, presiding in San Francisco, determined that Meta’s use of the copyrighted content was sufficiently “transformative” to qualify as “fair use” under U.S. copyright law. This marks the second major courtroom win in the same week for companies developing generative AI technologies.
However, Judge Chhabria acknowledged a potential vulnerability in the broader legal debate, noting that the plaintiffs could have presented a stronger case by arguing that AI trained on copyrighted materials creates tools that directly compete in the literary market.
“No matter how transformative the training may be,” he wrote, “it’s difficult to see how it qualifies as fair use if it results in a system capable of producing endless competing content while generating immense profits—especially if it harms the market for the original works.”
Large language models powering generative AI require enormous volumes of data for training. This demand has led to lawsuits from musicians, authors, visual artists, and news outlets who allege that AI companies are using their creative works without consent or compensation.
In response to the ruling, a Meta spokesperson said the company welcomed the decision. “Open-source AI models are driving innovation and creativity, and fair use is an essential legal principle that enables this progress,” the spokesperson noted.
The lawsuit against Meta involved allegations that pirated copies of books—including Sarah Silverman’s The Bedwetter and Junot Díaz’s The Brief Wondrous Life of Oscar Wao—were used in training Llama, Meta’s open-source generative AI.
Despite ruling in Meta’s favor, Judge Chhabria clarified that the decision did not validate the legality of using copyrighted material for AI training in general. “This ruling only reflects that the plaintiffs failed to make the appropriate legal arguments or support them with sufficient evidence,” he wrote.
Second Ruling Favors Anthropic
Earlier in the week, another San Francisco-based judge, William Alsup, sided with AI firm Anthropic in a similar case. Authors had accused Anthropic of using pirated and purchased books to train its Claude AI model, a competitor to ChatGPT.
Judge Alsup found the use of the materials to be “exceedingly transformative,” qualifying as fair use under the U.S. Copyright Act. He described AI training as one of the most transformative technologies of our era, likening it to how humans read books to learn.
The class-action lawsuit, brought by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, accused Anthropic of building its AI on unauthorized copies of their works.
While ruling in Anthropic’s favor, Judge Alsup rejected the company’s attempt to seek blanket immunity, stating that downloading millions of pirated books to build a permanent digital library went beyond what fair use protects.
