Meta halts book licensing for AI training amid copyright disputes
What's the story
Meta Platforms, the parent company of Facebook, has reportedly halted its efforts to secure book licensing deals for AI model training.
The move comes after a copyright infringement lawsuit was filed against the tech giant.
The case, Kadrey v. Meta Platforms, is one of many that have seen AI firms and IP owners locked in legal battles over copyright issues.
Licensing challenges
Meta's struggle to secure book licensing deals
The court documents feature excerpts from depositions of Meta employees, indicating that some staff members thought negotiating AI training data licenses for books could be impractical.
Sy Choudhury, who leads Meta's AI partnership initiatives, said their outreach to different publishers received "very slow uptake in engagement and interest."
He added they had created a long list from initially scouring the internet of top publishers but didn't receive contact and feedback from many of their cold call outreaches to try to establish contact.
Past efforts
Previous attempts to license AI-related content
The court transcripts show that Meta stopped some AI-related book licensing efforts in early April 2023 over "timing" and other logistical issues. Choudhury revealed that some publishers, particularly those of fiction books, did not even have the rights to the content Meta was considering licensing.
Allegations
Meta accused of using pirated books for AI training
The plaintiffs in the case, including bestselling authors Sarah Silverman and Ta-Nehisi Coates, have amended their complaint multiple times since it was filed in 2023.
The most recent iteration claims Meta cross-referenced some pirated books with copyrighted ones available for license to determine if it was worth pursuing a licensing deal with a publisher.
The complaint also alleges Meta used "shadow libraries" with pirated e-books to train several of its AI models, including the popular Llama series of "open" models.