More than 100 authors just filed a lawsuit against Anthropic seeking over $75 million in damages, alleging the AI company pirated their books to train its Claude large language models. The suit, filed on June 17 in a California district court, claims Anthropic downloaded more than 500 copyrighted works from shadow libraries like Library Genesis and Pirate Library Mirror using BitTorrent.
The piracy playbook
The authors behind this new lawsuit opted out of a prior class-action settlement reached in August 2025. That earlier deal, which covered approximately 482,000 pirated works, resulted in a $1.5 billion payout. Quick math puts that at roughly $3,000 per pirated work, which, for context, is less than many authors receive as an advance on a single book.
The original legal saga started in August 2024, when authors Andrea Bartz and others first took Anthropic to court. A June 2025 judicial ruling introduced an important distinction: training AI on legally acquired books qualifies as fair use. Downloading and storing pirated copies of those books does not.
The court found that roughly 7 million pirated books were stored as part of training operations, and the unauthorized copying and centralized storage of that content constitutes copyright infringement. The new lawsuit leans heavily into this framing, focusing on the act of piracy and storage rather than relitigating the fair use question around AI training itself.
