Meta's AI: Built on Stolen Books, Publishers Allege

Major publishers including Elsevier, Cengage, Hachette, Macmillan and McGraw Hill sued Meta Platforms in Manhattan federal court on Tuesday, alleging the company infringed copyrights by using their books and journal articles to train its Llama artificial intelligence model.

The proposed class-action complaint, which also includes author Scott Turow, alleges Meta copied millions of works without permission to train large language models that generate responses to human prompts. The plaintiffs are seeking unspecified monetary damages and court permission to represent a broader class of copyright owners.

Meta said training AI on copyrighted material can qualify as fair use and said it intends to fight the lawsuit.

The case adds to a growing wave of disputes between creators and technology companies over whether using copyrighted material for AI training is permissible. Similar lawsuits have been filed against OpenAI and Anthropic.

sources has reported that the pending cases are expected to hinge on whether AI systems make fair use of copyrighted material by using it to create new, transformative content, and that judges have issued differing rulings on the issue.