Meta sued by five publishers

Books are the point here — not just AI in the abstract. Five major publishers and novelist Scott Turow sued Meta and Mark Zuckerberg on May 5 in Manhattan federal court, saying Meta built Llama on a giant pile of copyrighted books and journal articles it never licensed. The stakes are simple: if that theory wins, one of the industry’s core habits — vacuum up text first, argue fair use later — gets a lot riskier. And this case matters because it brings heavyweight publishers directly into a fight that had mostly been framed around authors and artists. (money.usnews.com) ### Who sued Meta? The plaintiffs are Elsevier, Cengage, Hachette Book Group, Macmillan, McGraw Hill, and Scott Turow, who filed a proposed class action against Meta and Zuckerberg in the Southern District of New York. They want to represent a broader class of copyright owners, which means this is not just about a handful of books — it is aimed at the whole pipeline Meta used to train Llama. (money.usnews.com) ### What are they accusing Meta of? Basically, piracy at training scale. The complaint says Meta copied millions of protected works without permission and used them to train Llama to answer prompts, generate summaries, and in some cases reproduce text from the underlying works. The plaintiffs(money.usnews.com)ally costs money to access. (money.usnews.com) ### Why is Zuckerberg named personally? That is one of the sharpest parts of the suit. The publishers say Zuckerberg did not just sit above the process as CEO — they allege he personally authorized and encouraged the use of pirated material for AI training. Naming him directly is a way to argue this was a conscious business decision from the top, not a messy technical choice made somewhere deep in the company. (nytimes.com) ### What is the big number everyone is noticing? The complaint points to more than 267 terabytes of allegedly pirated books and articles used in Meta’s data pipeline. That number matters because it makes the scale feel real. We are not talking about a few disputed excerpts. We are talking about a warehouse-sized corpus of literature, textbooks, and journal content. (culture.org) ### What does Meta say back? Meta is leaning on the same defense much of the AI industry has used: training on copyrighted material can be fair use because the model is creating something new and transformative rather than republishing the books themselves. Meta has said it will fight the lawsuit aggressively. That means the case (culture.org)and when is it just copying at industrial scale? (money.usnews.com) ### Why are publishers stepping in now? Because this fight has moved beyond individual creators. Authors, artists, and news organizations have already sued AI companies including Meta, OpenAI, and Anthropic over training data. But publishers bring a different kind of weight — bigger catalogs, clearer licensing markets, and a stronger argument that there was an obvious legal way to get this material if Meta had wanted to pay. (money.usnews.com) ### Why does this case matter more than one lawsuit? The catch is that courts are still split on the core issue. Reuters notes early rulings in AI copyright cases have diverged, which means no clean rule has emerged yet. So this suit is not just another complaint on the pile — it is part of the process that may decide whether AI companies can keep treating copyrighted text as free raw material unless a court stops them. (money.usnews.com) ### Bottom line This is the book industry telling Meta that “move fast and break things” does not work when the thing being broken is the licensing market for books. If the publishers can prove deliberate, large-scale copying — and tie it directly to Zuckerberg — the legal and financial pressure on AI training practices goes up fast. (wtop.com)