Anthropic to Pay $1.5 Billion to Book Authors: Record Settlement and a Win for AI Labs

Anthropic must pay book authors $1.5 billion, the largest copyright settlement in the history of US class actions. The money is not for training AI. It is for downloading about 482,460 works from pirate databases.

Why the record settlement turned out to be a win for AI labs

The record payout punishes Anthropic for how it obtained the books, not for training a model on them. The company downloaded works from pirate libraries, and it is paying for that violation. The court left standing the principle that training AI on legally purchased texts is permitted. For the whole industry, this is the key signal: training on copyrighted content is not equated with theft if the data was obtained legally. That is why a $1.5 billion loss doubles as a strategic win for AI labs.

• Settlement amount: $1.5 billion, a record for a copyright class action
• Basis: downloading about 482,460 works from pirate databases
• On average, that works out to roughly $3100 per book
• The case was handled by Judge William Alsup, Northern District of California
• The court ruled that training AI on legally acquired books is fair use

What exactly is Anthropic paying for?

Anthropic is paying for piracy, not for training its models. As The Decoder notes, the $1.5 billion compensates authors for the download of about 482,460 of their books from illegal databases. In essence, it is payment for obtaining the texts unlawfully, not for how they were used afterward. Had the company bought the same books legally, then by the court's logic there would have been no legal claims against the model's training. The distinction is subtle, but it determines who is liable and for what.

What Judge Alsup decided

In June 2025, Judge William Alsup ruled that training AI on legally obtained books falls under the fair use doctrine. He called this use "transformative," meaning it creates a fundamentally new product rather than copying the original. That distinction turned a record-setting loss in dollar terms into a win on the merits: Anthropic was punished for the source of its data, not for the training technology itself.

"Using the books for training was fair use and exceedingly transformative," from

Judge William Alsup's ruling.

Why the sum is a record

At $1.5 billion, this is the largest copyright settlement in the history of US class actions. The record comes down to arithmetic: nearly 482,460 works multiplied by compensation of around $3000 each yields a total of $1.5 billion. Under US copyright law, the penalty for a single willful infringement can reach $150,000 per work, so in the worst case the payout could have been many times larger. The settlement let Anthropic cap that risk.

What this means

The precedent sets the rules for the entire industry: models can be trained on protected content, but the data must be acquired legally. By paying $1.5 billion, AI labs got an expensive but unambiguous guideline: pay for sources, rather than fear lawsuits over the act of training models itself.

Frequently asked questions

What is Anthropic paying $1.5 billion for: training AI?

No. The payout covers the download of about 482,460 books from pirate databases, that is, the illegal way the texts were obtained. The court did not treat training the model on this data as a separate violation.

Did the court rule that training AI on books is legal?

Yes, provided the source is legal. Judge William Alsup ruled that training AI on legally acquired books is "transformative" use and falls under the fair use doctrine.