About 300 Authors Sue Meta Solo Over AI Training on Books

About 300 authors sued Meta on Oct. 19, and every one of them is suing alone. Each wants statutory damages of up to $150,000 per infringed work over books Meta allegedly used to train its Llama and Muse AI models, Bloomberg Law reports.

That’s a direct challenge to the class-action model that has shaped AI copyright fights so far. If you’ve published a book, it’s also the clearest sign yet that you have more than one way to respond.

What happened

The suits were filed in the U.S. District Court for the Northern District of California. Publishers Lunch counted 231 authors across four complaints, led by Jacqueline Winspear, Daniel José Older, Robert Sheard and Tosca Reno. A fifth suit filed the same day pushed the total to 292 authors, according to the case tracker Chat GPT Is Eating the World. The plaintiffs range from novelists to journalists.

The claims go further than piracy. The Winspear complaint says Meta trained Llama on shadow-library collections including Books3, LibGen and Z-Library. It then describes “Project Vellum,” a scanning operation the authors say began around mid-2025. Meta allegedly bought physical books in bulk, scanned them, ran the pages through text recognition, and fed the results into Muse Spark and Muse Glimmer, released in April and August 2026. The authors argue that buying a book doesn’t give you the right to copy it.

Winspear, who writes the Maisie Dobbs mysteries, leads that complaint. One footnote corrects a Copyright Office record that lists her debut as “Maisie Debbs.”

Meta didn’t immediately respond to Bloomberg Law. AI companies, Meta included, have argued that training is fair use because it turns copies into a new technology.

Why it matters for working writers

Until now, most authors had one realistic option: join a class action and take a share of whatever it paid. The individual route lets each writer control their own case. The complaint says the plaintiffs chose it to avoid class settlements that resolve claims for far less than they’re worth.

Two firms are behind the filings: Elizabeth Brannen of Stris & Maher and Kyle Roche of Freedman Normand Friedland. Roche told Publishers Weekly his firm represents more than 2,200 authors. All of them signed up through ClaimsHero, a platform that recruited writers to opt out of Anthropic’s settlement and sue on their own. Brannen compared Meta’s scanning to a teacher who “got one copy of the book and ripped the spine off” to hand out copies.

This isn’t only a bestseller problem. The Winspear plaintiffs include authors of calculus textbooks, cheese-making guides and software-testing manuals. If you’ve turned expertise into a nonfiction book as part of a freelance writing career, the same datasets may hold your work.

Class action or solo suit: the money math

The numbers explain the split. Anthropic’s $1.5 billion class settlement works out to roughly $3,000 for each author who opted in, Bloomberg Law reports. A single work’s statutory damages can reach $150,000.

That ceiling isn’t a promise. In Kadrey v. Meta, an earlier proposed class action, the judge found the authors hadn’t shown that Llama harmed the market for their books. A jury can award well below the maximum. And each solo case has to be argued on its own.

So what should you do if you suspect your book trained a model? Start by checking whether your titles appear in the shadow libraries these complaints name: Books3, LibGen, Z-Library and Anna’s Archive. Gather your copyright registrations, since the Winspear filing attaches each plaintiff’s registered works as an exhibit. Then read any representation agreement closely before you sign it. Neither ClaimsHero’s terms nor the firms’ fee arrangements have been reported.

Talk to a lawyer before you choose a route.

What to watch next

The filings aren’t finished. Bloomberg Law says the firms plan suits for close to 1,000 more authors in the coming months. Roche told Publishers Weekly he expects up to 1,000 individual claims by year-end. Those may be the same pipeline counted two ways, or they may not.

Watch the judge assignment, too. The tracker expects the cases to go to Judge Vince Chhabria, though that hasn’t been confirmed.

The real test is Project Vellum. In Anthropic’s case, Judge William Alsup found AI training was fair use but singled out pirated books, Publishers Weekly notes. These suits ask whether buying a paper copy first makes the copying lawful, and that question now hangs over Meta’s Muse models.

Meta’s first response in court will show how hard it plans to fight the scanning claim.