Five major publishing houses (Elsevier, Cengage, Hachette, Macmillan, McGraw Hill) and author Scott Turow filed a putative class action against Meta and CEO Mark Zuckerberg on May 5, 2026, alleging Meta torrented over 267 terabytes of copyrighted material from pirate sites to train its Llama LLMs and stripped copyright management information to conceal the sourcing. The complaint is deliberately structured to address weaknesses in prior failed AI copyright suits, specifically the absence of market-harm evidence, by alleging Llama outputs (full papers, textbook chapters, study guides) functionally substitute for plaintiffs' works and that Meta bypassed established licensing markets. The case is the first brought by institutional publishers rather than individual authors, and its outcome could materially shape whether AI developers must pay for training data licensing, altering the cost structure for open-weight model development at scale.
279 other tracked developments touching the same claims.