Book authors file mass copyright suits against Meta

Two law firms, Stris & Maher and Freedman Normand Friedland, filed five near-identical copyright suits against Meta in the Northern District of California on 19 October, on behalf of 292 book authors in total, according to the litigation tracker Chat GPT Is Eating the World. The tracker reported that the firms told Bloomberg Law they plan to sue Meta for 1,000 plaintiffs “in the next few months”. The lead names include Jacqueline Winspear, author of the Maisie Dobbs series, and Daniel José Older.

The filings are individual actions, not a proposed class action. The Older complaint, which was read for this entry, says the plaintiffs elected not to proceed as a class because the Copyright Act entitles them to individualised statutory damages decided by a jury, and because they wished to avoid class settlements it characterises as resolving claims “for pennies on the dollar”. The firms previously represented authors who opted out of the Anthropic class settlement. The tracker’s author called the approach permissible, closer to a mass-tort filing than a typical copyright case, and questioned why five suits were filed rather than one and whether every plaintiff would sit for deposition.

The Older complaint alleges two separate wrongs. First, that Meta downloaded books from “shadow libraries” by torrenting, which also uploads, to train its Llama models. Second, that from around mid-2025 Meta ran a book-scanning operation, “Project Vellum”, buying physical books, scanning them and feeding the text into the Muse family, including Muse Spark and Muse Glimmer. The complaint calls that an independent infringement, since buying a copy does not carry the right to reproduce it. The tracker noted that this runs against the Bartz v. Anthropic ruling, in which Judge William Alsup held scanning purchased print books to be fair use, and that the cases are expected to be assigned to Judge Vince Chhabria, who presides over the related suits against Meta. Meta had not responded publicly, and the claims are allegations.

The tracker counted copyright suits against AI companies at 150 when the first of these was filed. If the 1,000-plaintiff plan is carried out, it will test whether individual filings can be run at a scale that class actions normally handle, and it puts the physical-scanning question before a second judge; the tracker says Alsup is the only federal judge to have ruled on it.