A federal judge ruled that Meta‘s unauthorized use of copyrighted works from Sarah Silverman and other authors to train generative AI models is a “fair use,” but warned that the practice may in many circumstances be illegal.
The ruling from U.S. District Judge Vince Chhabria is the second this week to determine that a major AI firm’s use of copyrighted works to train models falls within the “fair use” standard, a legal defense in copyright infringement cases.
Read the Sarah Silverman-Meta AI ruling.
Earlier this week, another federal judge ruled in favor of Anthropic over its use of copyrighted books to train its Claude model. But the judge ordered a trial on the question of whether the company still was liable for infringement as it had downloaded millions of pirated books in digital form off the internet.
Silverman and other authors, including Christopher Golden, Rachel Louise Snyder, Junot Díaz, Andrew Sean Greer and Richard Kadrey, sued Meta in 2023, claiming that the company never approached them about licensing their works.
The judge, though, found Meta’s use of their books to train its Llama large language model was “highly transformative.”
In his ruling granting summary judgement to Meta, Chhabria emphasized that his determination was specific to the circumstances of this case. He pointed to the way that the authors presented their argument that Llama diluted the market for their works, one of the factors that courts consider in determining whether something is “fair use.”
The judge wrote, “Meta introduced evidence that its copying hasn’t caused market harm. The plaintiffs presented no empirical evidence to the contrary—no evidence that the copying has already caused market harm, and no evidence that the copying is likely to cause market harm in the future. All the plaintiffs presented is speculation, and speculation is insufficient to raise a genuine issue of fact and defeat summary judgment.”
But the judge painted a picture of the future where AI firms will have to license content from authors, artists and other creators. Chhabria wrote that the “upshot is that in many circumstances it will be illegal to copy copyright-protected works to train generative AI models without permission. Which means that the companies, to avoid liability for copyright infringement, will generally need to pay copyright holders for the right to use their materials.” The judge even rejected as “ridiculous” the arguments that court copyright rulings could stufle the growth of AI technology.
He wrote, “These products are expected to generate billions, even trillions, of dollars for the companies that are developing them. If using copyrighted works to train the models is as necessary as the companies say, they will figure out a way to compensate copyright holders for it.”
The judge has not yet ruled on another claim from the authors — that Meta unlawfully distributed their works during a torrenting process.
The judge wrote that “as should now be clear, this ruling does not stand for the proposition that Meta’s use of copyrighted materials to train its language models is lawful. 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.”
The post Judge Rules For Meta In AI Lawsuit Brought By Sarah Silverman And Other Authors, But Warns Of Illegally Using Copyrighted Works In Training Models appeared first on Deadline.
