Major publishers from the United States and Europe filed a class-action lawsuit against Meta and founder Mark Zuckerberg on May 5 (local time), alleging copyright infringement over the use of books and academic papers to train the company’s AI model Llama.
According to Reuters and The Washington Post, five publishers — Elsevier, Cengage, Macmillan Publishers, Hachette Book Group, and McGraw Hill — along with author Scott Turow, filed the suit in federal court in Manhattan.
The plaintiffs allege that Meta illegally copied millions of copyrighted works, including textbooks, scientific papers, and novels, and used them without permission to train large language models (LLMs). They described the practice as “one of the largest copyright infringements in history.”
The lawsuit also claims Meta removed copyright notices and copyright management information from works used in AI training. The plaintiffs further alleged that Zuckerberg “personally approved and actively encouraged” the infringement.
They are seeking unspecified monetary damages and court approval to represent a broader class of copyright holders.
Maria Pallante, head of the Association of American Publishers, said in a statement that Meta’s large-scale infringement does not serve the public interest and warned that AI cannot develop responsibly if tech companies prioritize piracy sites over scholarship and creativity.
A Meta spokesperson countered that AI has significantly improved innovation, productivity, and creativity for individuals and businesses, adding that courts have appropriately recognized that training AI on copyrighted material can qualify as fair use.
Major AI companies including OpenAI, Anthropic, and Meta are currently facing multiple lawsuits alleging unauthorized use of copyrighted works as training data. The companies generally argue that such usage falls under fair use protections.
Meanwhile, Anthropic agreed last September to pay $1.5 billion to settle claims brought by authors who alleged the company used books without permission to train its AI model Claude.
