On Monday, a federal judge approved a $1.5 billion settlement between Anthropic and a class of book authors and publishers — the largest copyright class-action payout in U.S. history. The suit, filed in 2024, accused Anthropic of downloading more than 7 million books from pirate websites to train its Claude AI chatbot. Judge William Alsup had earlier ruled that training AI on legally-acquired books is fair use. The piracy, he found, is not protected. The deal pays roughly $3,000 per book across approximately 482,000 titles. More than 91% of eligible authors and publishers claimed their share; the judge overruled 54 objections.

1. Authors Who Said No (The Authors Guild, John Carreyrou)

Most authors accepted the deal. The ones who didn't say Anthropic got out cheap.

$3,000 per book is a fraction of what the law allows. Copyright law lets courts award up to $150,000 per work for willful infringement — at that maximum, liability across 482,000 titles could have exceeded $72 billion. Objecting authors argued the settlement undervalued their claims, overcompensated the attorneys, and wrongly excluded some copyright holders. The judge overruled all of them, saying the objections were "not grounded in a realistic assessment of the overall risks and rewards of a trial."

Not every author accepted — one filed suit against the whole industry. John Carreyrou — author of Bad Blood, the Theranos exposé — opted out of the class settlement and filed a copyright lawsuit in December 2025 against Anthropic, OpenAI, Google, Meta, xAI, and Perplexity. He's suing the whole industry, not just Anthropic. Some publishers are fighting separately too.

The amount "feels paltry," says the Authors Guild. The Guild urged authors to file claims while arguing that what Anthropic did was theft at scale. Its position: the deal sends a signal that piracy is costly, but doesn't set a fair price for how the industry used creative work.

2. Anthropic Says It Won (Deputy General Counsel Aparna Sridhar)

For Anthropic, the settlement confirms that training AI on books is legal — and that the company can fight large copyright suits and survive them.

Training on books is now court-tested. Anthropic Deputy General Counsel Aparna Sridhar said the company settled "after the court's landmark ruling that training AI on books is fair use under copyright law." For AI companies watching this case, that's the headline: a federal judge looked at their core practice and called it legal.

$1.5 billion is the most ever paid in a copyright case. Lead plaintiffs' attorney Justin Nelson called it "the largest known copyright recovery in history." Anthropic says 91% of eligible authors claiming their share shows most considered the deal fair.

Training on books you own is legal; downloading pirate copies isn't. That's the line Anthropic wants the industry to take from this case. The company admits to building a piracy library — but says that's the narrower wrong. AI training itself can continue at scale; the problem was where the books came from.

3. Courts Are Still Split on the Same Facts (Universal Music Group, Scott Turow)

The ruling isn't binding, courts are already disagreeing, and Anthropic faces more lawsuits using the same evidence.

The Alsup fair-use ruling binds no other court. Because Anthropic settled before the case reached an appeals court, Judge William Alsup's ruling that training on books is fair use was never reviewed at a higher level. Other judges can — and do — ignore it.

A different court reached a different result on the piracy question. In Kadrey v. Meta, a separate federal court found the manner of acquisition from pirate libraries "non-dispositive" — the training was transformative, and how they got the books didn't change that outcome. Judge Alsup's Anthropic ruling excluded piracy from fair use; the Kadrey court said piracy-sourcing wasn't determinative. Two courts, same pirate library, different rules for AI companies.

A fresh $3 billion suit against Anthropic reuses the same evidence. On January 28, 2026, Universal Music Group, Concord Music Group, and ABKCO sued Anthropic for $3 billion. They allege the company downloaded more than 20,000 copyrighted songs from the same pirate repositories — and they built their case using evidence from the book settlement. More than 70 AI copyright lawsuits remain active. The most recent came July 10: Hachette, Cengage Learning, Elsevier, and novelist Scott Turow suing Google over Gemini training data.

Where This Lands

Objecting authors and the Authors Guild argue that $3,000 per book doesn't compensate for what Anthropic built, and that the fair use signal from Judge Alsup's ruling now favors the industry. Anthropic says it paid the most any company has ever paid in a copyright case and that the training itself is legally validated. Legal watchers say neither side should declare victory: the Alsup ruling is one district court's opinion, courts are reaching different conclusions on the same facts, and the music, news, and visual-art industries haven't settled anything with anyone. Monday's ruling closed one case — Anthropic's past conduct with books, nothing else.

Sources