Meta has encountered obstacles in its efforts to license books for training its generative AI models, according to recent court filings in the ongoing Kadrey v. Meta Platforms lawsuit.
The filings indicate that Meta paused discussions with publishers due to difficulties in securing necessary rights and low publisher engagement. Meta’s outreach to publishers reportedly received limited responses, leading to the company’s decision to explore alternative approaches.
A Meta executive, Sy Choudhury, revealed that publishers of fiction books often lacked the rights to license content, necessitating extensive negotiations with authors. This, along with logistical setbacks, prompted Meta to pause certain book licensing efforts in April 2023.
Meta’s licensing endeavors for AI development have previously encountered similar challenges. The company had unsuccessfully attempted to license 3D worlds for research purposes, ultimately opting to develop its own solution.
The plaintiffs in the lawsuit, including renowned authors Sarah Silverman and Ta-Nehisi Coates, have expanded their allegations against Meta. They claim that the company cross-checked pirated books with licensed books to evaluate licensing agreements and used “shadow libraries” containing pirated e-books to train its AI models, including the Llama series.
The plaintiffs’ complaint alleges that Meta obtained some of these libraries through torrenting, which involves the simultaneous uploading of files for distribution. This practice is alleged to constitute copyright infringement.
Original source: Read the full article on TechCrunch
