Staff Reports
Five major publishers and bestselling author Scott Turow have fired a legal salvo at Meta and its CEO, accusing the company of stealing millions of books and articles to feed its Llama AI models. The complaint, filed this week in the Southern District of New York, paints a picture of deliberate choices and discarded licensing plans — and it names Mark Zuckerberg as a central figure in the decision-making. If true, this is not a legal gray area; it looks like a business choice with big consequences for authors and the law.
What the lawsuit says
The publishers say Meta pulled in roughly 267 terabytes of copyrighted books, textbooks, and journal articles — the kind of haul that would make any pirate site blush — and used that material to train Llama. The complaint alleges Meta obtained the content through torrenting from shadow libraries and unauthorized web scraping, then stripped copyright-management information to hide the source. Plaintiffs also claim Meta considered paying for licenses and even eyed a dataset budget of up to $200 million, but abruptly stopped that path after the matter was escalated to CEO Mark Zuckerberg.
Allegations at a glance
The suit is a coordinated move by Elsevier, Cengage, Hachette, Macmillan, McGraw Hill and Scott Turow. They seek class certification, damages, and injunctions to stop further use and to force destruction of infringing copies. The core accusation: Meta willfully ignored licensing markets and built an “infinite substitution machine” that can spit out near‑verbatim text, replacement chapters, and textbook knockoffs — all of which eat the publishers’ business and the authors’ paychecks.
Why this case matters
This isn’t just another tech lawsuit. It could reshape how courts treat AI training on copyrighted works. There is precedent that leans one way in some cases, but judges have repeatedly said outcomes depend on facts like scale, intent, and market harm. Here, the publishers put facts on the table — alleged internal talks about licensing, a claimed order to stop paying, and the huge dataset size. If the court finds the allegations true, it could force Big Tech to change how it builds models or to pay for the content that powers them.
Meta’s defense and the legal road ahead
Meta’s answer is predictable: training AI can be fair use, and the company says it will fight. That defense has won in prior disputes, but the new complaint leans heavily on alleged market harm and deliberate concealment. Expect motions to dismiss, fights over discovery, and a battle for internal documents that could prove whether licensing was abandoned at the top. For creators and readers alike, the outcome matters: either the new AI economy pays its way, or it gets a free pass to hollow out creative markets. If you care about rule of law and paying people for their work, that should make you pay attention.
