Anthropic is staring down another lawsuit. This one’s for $75 million, and it comes from authors who say the company pirated their copyrighted books to train Claude.
The fight between writers and AI companies isn’t close to over.
The lawsuit claims Anthropic grabbed books from well-known shadow libraries — sites that host copyrighted material without permission. The authors say the company never asked for a license and never offered payment before feeding the books into its training pipeline.
Here’s the legal distinction that matters. A previous ruling found that training AI on legally acquired books counts as fair use. But downloading pirated copies? That’s a separate act of infringement. The piracy angle is the battleground.
The plaintiffs argue existing settlements don’t come close. Copyright law allows up to $150,000 per willfully infringed work. They think the per-book payouts in earlier deals were a joke relative to the scale of what happened.
This isn’t Anthropic’s only legal headache. A class action filed in June targets its Claude Max subscription plans — alleging advertised usage boosts collapsed under hidden caps. That one covers the $100 and $200 tiers and seeks refunds.
The copyright case has much higher financial stakes. Anthropic already settled a class action for roughly $1.5 billion, paying authors around $3,000 each for about 500,000 pirated books. Some authors opted out of that deal to keep the right to sue individually. This $75 million lawsuit is exactly that play — seeking far bigger per-work damages.
Anthropic can afford the fight. The company is valued in the hundreds of billions. But repeated legal battles could reshape how AI firms source training data and market their subscription products.
