Anthropic Faces $75M Authors Lawsuit Over Book Piracy to Train Claude
More than 100 authors filed a $75 million lawsuit against Anthropic, alleging the company pirated their books to train Claude. Here is what the case means for authors who write with AI.

More than 100 authors and rights holders filed a federal lawsuit against Anthropic on June 17, 2026, seeking over $75 million in damages. They claim the company downloaded more than 500 copyrighted books, shared them through BitTorrent, and used them to train its Claude language models without permission. The case, filed in the Northern District of California, is the sharpest legal challenge yet to how AI companies source their training data.
What the lawsuit claims
The complaint, Shakespeare et al. v. Anthropic PBC, alleges that Anthropic built a dataset of pirated books to train Claude. According to the filing, the company did not merely scrape publicly visible text. Counsel for the plaintiffs say Anthropic obtained complete book files through peer-to-peer networks, including BitTorrent, and retained them in a centralized repository. The plaintiffs are seeking statutory damages of $150,000 per infringed work, which for the 500-plus titles at issue pushes the total demand past $75 million. The New York Post first reported the damages figure in its coverage on July 4.
The lead plaintiffs include published fiction and nonfiction authors. The suit is structured as an opt-out action: these writers chose not to join the earlier class settlement between Anthropic and authors represented in the Bartz case, and they are now pursuing their claims independently.
How we got here
This is not Anthropic's first courtroom fight over books. In a parallel class action, Bartz et al. v. Anthropic PBC, authors alleged the company trained Claude on nearly half a million pirated books. In 2025, U.S. District Judge William Alsup ruled that Anthropic's use of books for training qualified as fair use, but the company crossed a line by building what the judge called a "central library" of pirated works and sharing them internally. A proposed $1.5 billion settlement in that case drew sharp criticism from Judge Alsup, who told the parties in a September 2025 hearing that he wanted to ensure authors did not "get the shaft" and scheduled further review.
The new Shakespeare suit picks up where the Bartz settlement left off. Authors who saw the settlement terms and walked away are now pressing their own case with a narrower set of works and a more direct claim: the books were taken, copied, and distributed.
What it means if you write with AI
Three things matter for working authors:
The training-data question is not settled. Judge Alsup found fair use for the act of training, but the method of acquisition matters. If a company downloaded books from torrent sites and kept them in a shared library, that looks different in court than scraping publicly posted text. The distinction will shape how AI companies build their next training sets.
Opt-out lawsuits are multiplying. The Shakespeare case signals that more authors are willing to litigate rather than settle. If courts let these opt-out claims proceed, the liability exposure for AI companies grows fast. That could change what models are available, what they are trained on, and what they cost.
Disclosure rules keep tightening. Amazon KDP now asks authors to declare AI use. Kobo rejects low-quality AI-generated books at scale. Platforms respond to the legal climate, and a large judgment against a model provider would accelerate that. The books that get through are the ones an author shapes in their own voice, not clean AI drafts shipped without revision.
For authors who draft with AI, the practical habit is the same one platforms already reward: outline your own structure, revise every chapter, and stand behind the final text. Our guides on how to edit and humanize AI-generated book content and how to publish an AI book on Amazon KDP cover the current requirements in detail.
PageWriter Studio is built for exactly that workflow. The AI generates outlines, drafts, and full chapters, and you steer every revision with a style profile tuned to your writing. New court rulings may shift what training data powers the models you use, but the author's role at the steering wheel does not change. If you want to try that approach on your next title, you can start a free trial and keep control from the first page.
Ready to write your book with AI?
Turn your idea into a finished, publishable book. Start your 5-day free trial today, no installation needed.
Start your free trialMore news

UK Magazine Fiction Feast Swaps Freelance Writers for AI-Generated Stories
Bauer Media's Fiction Feast paused writer commissions and started publishing AI-generated short stories in-house. Here is what happened and what it means for authors.

The AI Writing Tell 'It's Not X, It's Y' and How Authors Can Avoid It
The Atlantic traces the most recognizable AI writing tic, 'it's not X, it's Y.' Here is why models keep using it and how authors can keep this AI writing pattern out of their books.

Hachette, Elsevier, Cengage and Scott Turow Sue Google Over Gemini AI Training
Hachette, Cengage, Elsevier and author Scott Turow filed a class action against Google, alleging Gemini was trained on copyrighted books without permission. Here is what it means for authors.
