Five major publishing houses and novelist Scott Turow allege Meta used pirated books and journal articles without permission to build its Llama AI models
James Manning - PA Images / Getty Images
Five major publishers and best-selling novelist Scott Turow filed a class-action copyright infringement lawsuit against Meta $META and its CEO Mark Zuckerberg on Tuesday, alleging the company pirated millions of books and journal articles to train its Llama artificial intelligence models.
Named as plaintiffs in the complaint are publishers Hachette, Macmillan, McGraw Hill, Elsevier, and Cengage, with the case brought before a federal court in Manhattan. According to the suit, the company's engineers obtained pirated books and journal articles by routing through Anna's Archive — a search engine that indexes piracy repositories such as LibGen and Sci-Hub — to acquire unlicensed copies for use in training. The complaint further accuses Zuckerberg of direct involvement, alleging he "personally authorized and actively encouraged the infringement," The New York Times reported.
The works allegedly used in training range from textbooks and scientific articles to novels, including "The Fifth Season" by N.K. Jemisin and "The Wild Robot" by Peter Brown. In terms of relief, the plaintiffs want the court to certify a wider class of copyright holders and are asking for monetary damages, though no specific dollar figure has been named.
The plaintiffs also point to Llama's own outputs as evidence. In one example cited in the filing, the chatbot was prompted to write a travel guide mimicking the voice of author Becky Lomax; it did so convincingly, and when pressed to explain its familiarity with her style, the model responded: "While I don't have personal interactions with Becky Lomax, I've been trained on a vast amount of text data, including her published works," according to The Times.
Meta disputed the allegations. "AI is powering transformative innovations, productivity and creativity for individuals and companies, and courts have rightly found that training AI on copyrighted material can qualify as fair use," a Meta spokesperson said in a statement. "We will fight this lawsuit aggressively."
Central to the plaintiffs' case is the claim that tools like Llama threaten the economic survival of authors and publishers; the complaint warns that AI-generated titles are arriving on Amazon $AMZN "in volumes that materially displace human-authored works." Maria Pallante, president of the Association of American Publishers, said in a statement that "tech companies prioritize pirate sites over scholarship and imagination."
Tuesday's lawsuit is one of the latest in a long line of copyright battles between AI companies and rights holders, spanning authors, news outlets, visual artists, and publishers. Meta, OpenAI, Anthropic, and others have all faced infringement claims over their use of copyrighted material in AI training. Courts have so far reached conflicting conclusions on whether such use qualifies as fair use under copyright law.
Anthropic, which counts Amazon and Google $GOOGL among its backers, reached what is believed to be the first major settlement of its kind in this area of litigation, with the company agreeing to pay $1.5 billion to a group of authors who had accused it of piracy in a class-action suit, according to reuters.com.
Join 500,000+ readers who start their day with Quartz.
By subscribing, you agree to our Terms of Service and Privacy Policy.
