Judge Skeptical of Meta’s Claims It Did Not Violate Copyright Law in AI Lawsuit

The lawsuit was filed by authors who claim Meta used their work to train its large language model without informing them.

A pedestrian walks in front of the Meta logo at the Facebook headquarters in Menlo Park, Calif., on Oct. 28, 2021. Justin Sullivan/Getty Images

A federal judge in San Francisco on April 1 expressed skepticism over social media giant Meta’s argument that it did not violate copyright law when it used the works of 13 authors to train its artificial intelligence (AI) models without their permission.

District Judge Vince Chhabria in California questioned lawyers for both parties over Meta’s request for a ruling that it made “fair use” of books by Junot Diaz, comedian Sarah Silverman, and others to train its large language model known as Llama.

Chhabria said he did not think this was the case.

“You have companies using copyright-protected material to create a product that is capable of producing an infinite number of competing products,” Chhabria told Meta’s attorneys.

“You are dramatically changing, you might even say obliterating, the market for that person’s work, and you’re saying that you don’t even have to pay a license to that person.

“I just don’t understand how that can be fair use.”

Chhabria’s comments stem from a copyright infringement lawsuit filed by comedian Sarah Silverman and two authors against Meta in 2023.

The suit argued that the company used pirated versions of their books to train Llama, which was released by Meta in February 2023, without first seeking their permission or offering compensation.

Part of the legal challenge—allegations that text generated by Llama infringed their copyrights—was initially dismissed by Chhabria in November 2023, though he allowed Silverman and the other authors to amend most of their claims.

They refiled their lawsuit in February of this year, and it has since expanded to cover 13 authors, including award-winning writer Ta-Nehisi Coates.

It alleges that Meta pirated their books through online digital repositories of pirated books called “shadow libraries” in order to help train Llama.

“Plaintiffs and Class members did not consent to the use of their copyrighted books as training material in any version of Llama, despite there being a vibrant market for content for AI training data—a market within which Meta participates,” they wrote in the lawsuit.

“Indeed, while Meta internally discussed licensing copyrighted books for training data and reaching out to a variety of publishers from the very beginning of its development of Llama, it decided to cut corners and save time and money by using free online ’shadow libraries’ to source this highly valuable content.”

Meta Argues Use of Books Was ‘Transformative’

Meta has not denied that it used the authors’ works but says it was protected by the legal doctrine of “fair use,” which allows for the unauthorized use of copyrighted material under specific circumstances, and has asked for the lawsuit to be thrown out.

Specifically, it has argued its use of their books was transformative and helped train Llama to “serve as a personal tutor on nearly any subject, assist with creative ideation, and help users to generate business reports, translate conversations, analyze data, write code, and compose poems or letters to friends.”

Chhabria acknowledged that Meta’s use may have been transformative, but said it still may not have been fair.

“This seems like a highly unusual case in the sense that though the copying is for a highly transformative purpose, the copying has the high likelihood of leading to the flooding of the markets for the copyrighted works,” the judge said.

Kannon Shanmugam, a lawyer for Meta, said copyright owners are not entitled to “protection from competition in the marketplace of ideas.”

“But if I’m going to steal things from the marketplace of ideas in order to develop my own ideas, that’s copyright infringement, right?” Chhabria responded.

Chhabria also told David Boies, an attorney representing the authors, that the legal challenge may have failed to adequately address the potential market impacts of Meta’s conduct.

The judge also asked Boies for evidence that Llama’s creations would affect the market for the authors’ books specifically.

“It seems like you’re asking me to speculate that the market for Sarah Silverman’s memoir will be affected by the billions of things that Llama will ultimately be capable of producing,” Chhabria said. “And it’s just not obvious to me that that’s the case.”

Llama has been downloaded an average of 1 million times per day since its release, according to Meta. The tech giant released its most recent version of the Chat GPT rival at the end of last year.

The Epoch Times has contacted a Meta spokesperson for comment.

The Associated Press contributed to this report