New court filings in an ongoing AI copyright lawsuit against Meta support previous reports that the company had "paused" discussions with book publishers about licensing deals to train its generative AI models.

The filings relate to Kadrey v. Meta Platforms, one of several cases moving through the U.S. legal system that pit AI companies against authors and intellectual property holders. AI firms, including Meta, have largely defended their use of copyrighted materials under the "fair use" doctrine, while copyright holders strongly dispute this claim.

Documents submitted on Friday include partial transcripts from depositions of Meta employees, taken by attorneys representing the plaintiffs. These transcripts suggest that some Meta staff viewed book licensing for AI training as an impractical strategy.

According to one transcript, Sy Choudhury, head of Meta’s AI partnerships, stated that outreach to publishers saw “very slow uptake in engagement and interest.”

“I don’t recall the entire list, but I remember we compiled a long list of top publishers after researching online,” Choudhury said, according to the transcript. “However, many of our cold-call attempts to establish contact didn’t receive responses or feedback.”

Choudhury added that while a few publishers did engage in discussions, “not many” showed interest.

The court documents indicate that Meta paused some AI-related book licensing initiatives in early April 2023 due to timing and logistical challenges. Choudhury noted that many fiction publishers, in particular, lacked the rights to license content Meta was considering using.

“In the fiction category, we quickly realized from our business development team that many publishers we were in talks with did not actually have the rights to license their content,” Choudhury said in the transcript. “Engaging with all their authors would have taken significant time.”

Choudhury also acknowledged that Meta had previously halted licensing efforts in other AI-related areas.

“I’m aware of other licensing efforts—for example, we attempted to license 3D worlds from game engine and game manufacturers for our AI research team,” he said. “As with fiction and textbook data, we saw little engagement and ultimately chose to develop our own solution.”

The lawsuit, brought by bestselling authors including Sarah Silverman and Ta-Nehisi Coates, has undergone multiple amendments since being filed in the U.S. District Court for the Northern District of California in 2023. The latest version alleges that Meta cross-referenced pirated books with copyrighted ones available for licensing to decide whether securing a deal with publishers was worthwhile.

Additionally, the plaintiffs claim Meta used “shadow libraries” containing pirated e-books to train AI models, including its Llama series. The complaint suggests Meta may have obtained these libraries via torrenting, a file-sharing method that requires users to both download and upload content—an act plaintiffs argue constitutes copyright infringement.