ADVERTISEMENT

Technology

Sarah Silverman Lawyers Get Judge’s Harsh Rebuke in Meta AI Case

Published: 

(Bloomberg Law) -- The legal team bringing a proposed class action accusing Meta Platforms Inc. of training its AI model on copyrighted books of authors including comedian Sarah Silverman is failing to adequately represent its clients, a San Francisco federal judge said at a Friday hearing.

“You and your team have barely been litigating the case, that’s obvious,” Judge Vince Chhabria told the authors’ attorney Joseph Saveri of the Saveri Law Firm LLP.

“I will not certify a class represented by this legal team,” Chhabria said at the virtual case management hearing. “There is no way that I would find adequacy of representation based on the representation that I’ve seen take place thus far.”

The judge said the attorneys representing Silverman and other authors including Pulitzer Prize-winning novelist Andrew Sean Greer should bring in other law firms that have “the resources and the wherewithal” to litigate the case “properly.”

Saveri and his co-counsel Matthew Butterick didn’t immediately return requests for comment about the hearing.

Chhabria’s admonishment came in response to the plaintiffs’ attorneys’ request to extend the deadline for factual discovery in the case by 97 days. The deadline had been set for the end of the month, and Chhabria said he would extended it by 14 days to account for a few remaining depositions.

Meta’s Internal Documents

The authors first filed the class action against Meta in the US District Court for the Northern District of California last year, alleging the company took their copyrighted books without permission to train its open source AI model called LLaMA. Chhabria dismissed parts of suit, but the authors’ core claim of copyright infringement is proceeding.

Meta will argue that its copying of the book is permitted by copyright law’s fair use exception, which allows unauthorized copying under certain circumstances.

Saveri and other attorneys at his firm argued that they needed additional time to obtain internal documents from Meta to identify whether the company knew it was training its AI model on ebooks taken from pirated databases. They also argued that Meta has produced tens of thousands of technical documents that the plaintiffs’ experts must analyze.

But Chhabria said the discovery process had been dragging on for too long, based largely on failures by the plaintiffs’ attorneys to raise discovery disputes in a timely manner.

The judge said he was in a “bind” because on the one hand, the discovery problems are the fault of the plaintiffs. But on the other hand, the question of copyright law and AI training is an “important societal issue” and the court has a duty to the proposed class members, Chhabria said.

Cooley LLP, Cleary Gottlieb Steen & Hamilton LLP, and Lex Lumina PLLC represents Meta.

The case is Kadrey v. Meta Platforms Inc., N.D. Cal., No. 3:23-cv-03417, 9/20/24.

To contact the reporter on this story: Isaiah Poritz in San Francisco at iporitz@bloombergindustry.com

To contact the editor responsible for this story: Stephanie Gleason at sgleason@bloombergindustry.com

©2024 Bloomberg L.P.