Meta Platforms Inc., the parent company of Facebook, is embroiled in a legal battle over its use of copyrighted materials to train its artificial intelligence models. This lawsuit, part of a broader wave against Big Tech, alleges that Meta utilized books without the consent of the authors to develop its AI technologies.
The litigation, spearheaded by law firms Lieff Cabraser Heimann & Bernstein and Cowan, DeBaets, Abrahams & Sheppard, represents author Christopher Farnsworth. Filed in the U.S. District Court for the Northern District of California, the complaint accuses Meta of copyright infringement, claiming the company sourced “hundreds of thousands” of copyrighted books from an unauthorized online collection to construct its “Llama” large language model.
Meta’s journey into AI began with the launch of its LLM family, initially titled LLaMA, in February 2023, as part of its competitive strategy against OpenAI’s ChatGPT. The company continued to evolve its AI capabilities, releasing “Llama 2” for commercial use in July 2023 and “Llama 3” in April 2024, aimed at enhancing its AI assistant “Meta AI.”
The lawsuit asserts that Meta accessed nearly 200,000 books from “Books3,” a repository of works extracted by developer Shawn Presser from the illicit site Bibliotik. This collection is part of “The Pile,” a dataset designed to train LLMs, hosted by EleutherAI. The complaint references a February 2023 research paper where Meta allegedly acknowledged using Books3 data for its AI training.
Meta has not yet responded to the claims, and the defendant’s legal representation remains undeclared. The case highlights ongoing tensions in AI development, as tech companies operate under the ethos of rapid advancement, often facing legal repercussions after integration into the market.
This isn’t Meta’s first encounter with such accusations. In July 2023, a group of writers, including Sarah Silverman, filed a lawsuit against both Meta and OpenAI, citing similar copyright violations. Additionally, Mark Zuckerberg, Meta’s CEO, is slated for deposition in this class action.
Similar allegations have been directed at AI startup Anthropic, accused in an August lawsuit of improperly using Books3 for its “Claude” LLM series.
Industry experts like Mike Palmisciano suggest that disputes over copyright in AI will persist until regulatory frameworks or judicial decisions establish clear guidelines. He notes that the ongoing legal challenges may eventually lead to a Supreme Court ruling on the application of fair use in AI, determining whether large-scale data ingestion is legally transformative.
In the meantime, companies may resort to settlements and licensing agreements, as this strategy appears to be a significant financial consideration for AI platforms like OpenAI, which are navigating the complexities of content licensing alongside technological development.
