A US federal judge has ruled in favor of Meta Platforms in a closely watched copyright lawsuit, concluding that the tech giant’s use of authors’ books to train its AI systems falls under “fair use.” The decision marks a pivotal moment in the ongoing debate over whether companies can legally use copyrighted content to develop generative AI tools, and comes just days after a similar ruling in favor of Anthropic. Despite disappointment from the authors involved, the ruling may now serve as a benchmark for other pending lawsuits across the country.
Court finds Meta’s AI training lawful under fair use
Judge Vince Chhabria of the U.S. District Court issued the summary judgment in Meta’s favor on Wednesday, stating that the authors who brought the lawsuit had failed to demonstrate that Meta’s actions had harmed the commercial value of their books. “The Court has no choice but to grant summary judgment to Meta,” he wrote in his opinion, as reported by Wired.
The lawsuit, originally filed in 2023 by a group of 13 authors including Sarah Silverman and Ta-Nehisi Coates, accused Meta of training its large language models (LLMs) on their copyrighted books without permission. They claimed this amounted to intellectual property theft on an unprecedented scale.
However, Judge Vince Chhabria focused his legal reasoning on a key principle of copyright law: whether the use substantially harms the market for the original work. He determined that the plaintiffs had not shown concrete evidence that Meta’s use of their books had reduced their sales or earnings. This market-harm test ultimately guided his decision to dismiss the case.
Contrasting legal strategies and industry-wide implications
While some past rulings have relied more heavily on the “transformative use” argument—whereby the new work must add something substantially new—Judge Vince Chhabria opted to center his judgment around economic impact. He explicitly noted that copying a work doesn’t automatically constitute infringement unless it diminishes the original’s commercial value.
This perspective differs slightly from the approach taken by Judge William Alsup earlier in the week in a case involving Anthropic, where he concluded that the AI company’s training was transformative enough to be protected under fair use. These diverging angles may shape legal strategies in similar cases yet to come.
Jacob Noti-Victor, a law professor at Cardozo School of Law, noted that the court’s rejection of “market dilution” as a standalone argument could influence future litigation. “We haven’t seen the last of this novel market dilution theory,” he said.
Still, Judge Vince Chhabria left the door open for future lawsuits. “In many circumstances it will be illegal to copy copyright-protected works to train generative AI models without permission,” he clarified. That means tech companies may eventually need to negotiate licensing deals with rights holders to avoid legal exposure.
Mixed reactions as legal battles continue
Meta welcomed the ruling, calling it a step forward for open-source AI development. Company spokesperson Thomas Richards said, “Fair use of copyright material is a vital legal framework for building transformative technology.”
On the other side, attorneys for the plaintiffs expressed disappointment. In a statement, law firm Boies Schiller Flexner said, “Despite the undisputed record of Meta’s historically unprecedented pirating of copyrighted works, the court ruled in Meta’s favour. We respectfully disagree.”
The legal debate is far from settled. Microsoft, for instance, is facing a separate lawsuit over claims it used over 200,000 pirated books to train its Megatron model. Meanwhile, Getty Images recently dropped key claims in its UK lawsuit against Stability AI, signaling the complexity and variability of these cases across jurisdictions.
While courts in the US have so far leaned toward shielding AI companies under the fair use doctrine, legal experts say future rulings may hinge on more precise facts—such as how the data is stored, reused, or monetized. As the legal framework evolves, the line between innovation and infringement continues to blur.
