Anthropic ordered to pay $1.5 billion to authors in landmark AI copyright case

Q: How will the $3,000 per eligible work be distributed to authors and what proof must authors provide to claim payment?

The settlement provides an estimated $3,000 in damages per eligible work, covering 482,460 qualifying books, to be paid out from the Settlement Fund; the judge said any remaining funds may be redistributed to Settlement Class Members if feasible. The article notes that more than 91 percent of authors and publishers covered by the settlement have claimed their share. The article does not specify what proof authors must provide to claim payment.

Q: Does the settlement require Anthropic to delete the original pirated book files and any derivative training data, and how is compliance enforced?

The settlement requires Anthropic to destroy the original pirated book files and any derivative data it illegally obtained from the piracy websites. The article does not describe how compliance with that destruction requirement is enforced.

Q: Will authors who have publishing or rights agreements with their work receive the full $3,000 or will those agreements affect payouts?

The court estimated $3,000 per eligible work in damages paid to the authors and explicitly did so "without taking into account any publishing or rights agreements the authors have with their work," indicating those agreements do not reduce the stated $3,000 per stolen book payment. The settlement covers 482,460 qualifying books and the judge anticipated redistributing any remaining funds to settlement class members if feasible.

Q: How does this settlement affect other AI companies that used Books3, LibGen, or similar shadow libraries for training data?

The settlement sets a major legal precedent indicating that AI firms can be held accountable for using pirated books in training. It reinforces that large-scale scraping of copyrighted material can lead to damages and requires companies to destroy illegally obtained files and any derivatives, giving other AI companies a clearer path to liability if they are discovered using such sources.

Anthropic has agreed to a $1.5 billion copyright settlement, under which it must pay $3,000 per stolen book back to the authors of those books.

VIEW GALLERY - 2

Jak Connor

Tech and Science Editor

Published Jul 22, 2026 8:10 AM CDT

1 minute & 15 seconds read time

TL;DR: A judge approved Anthropic's $1.5 billion class-action settlement over training models on over 7 million pirated books, covering 482,460 qualifying works at about $3,000 per book and requiring destruction of illegally obtained files and derivatives, setting a major precedent for AI copyright accountability.

Voice: Jak ConnorSpeed

0:00 / 2:42

A California judge has approved Anthropic's $1.5 billion copyright settlement with authors, marking the largest copyright payout in U.S. history and a case that will likely be used as a precedent in future copyright cases involving AI. The settlement resolves a class-action lawsuit over Anthropic's use of pirated books to train its models without the consent of the authors.

Anthropic ordered to pay $1.5 billion to authors in landmark AI copyright case 2

VIEW GALLERY - 2 IMAGES

The case, which began in 2024, alleged Anthropic trained its AI on millions of copyrighted works sourced from sites like Books3 and LibGen without permission from the authors. Anthropic downloaded more than 7 million pirated books from shadow libraries, which were then used to train popular models such as Claude. The judge called the settlement "fair," as it covers 482,460 qualifying books downloaded by Anthropic, which the court estimated equates to $3,000 per eligible work in damages paid to the authors, without taking into account any publishing or rights agreements the authors have with their work.

This settlement sets a major precedent for the AI industry, reinforcing that large-scale data scraping of copyrighted material, especially since Anthropic not only has to pay out those it has wronged, but also destroy the original files, along with any derivatives, it illegally obtained from the piracy websites. Other AI firms now face a clearer path for being held accountable if it's discovered they are using copyrighted data, particularly books from authors.

Read more: Meta accused of downloading torrents of 81.7TB of pirated books to train its Llama AI models

"If any funds remain in the Settlement Fund after all Valid Claims are paid, the Parties anticipate a redistribution of the remaining funds to Settlement Class Members unless it is economically infeasible to do so," said ruling judge Martínez-Olguín

With AI development accelerating, this ruling is undoubtedly going to influence how AI companies handle data ethics and intellectual property. What remains to be seen is whether this settlement will prompt more lawsuits, change within AI companies, or be simply a speeding ticket.

Frequently Asked Questions

TweakBot answers common questions about this news using TweakTown's own coverage from this page and related content from our archive. Tap a question to reveal the answer, or type your own below.

Question #1

How will the $3,000 per eligible work be distributed to authors and what proof must authors provide to claim payment?

Question #2

Does the settlement require Anthropic to delete the original pirated book files and any derivative training data, and how is compliance enforced?

Question #3

Will authors who have publishing or rights agreements with their work receive the full $3,000 or will those agreements affect payouts?

Question #4

How does this settlement affect other AI companies that used Books3, LibGen, or similar shadow libraries for training data?

Have a question not listed here? Ask below and TweakBot will answer it.

"We are pleased that more than 91 percent of authors and publishers covered by the settlement have claimed their share of the payment, and we're looking forward to bringing this matter to a close," said Anthropic's deputy general counsel, Aparna Sridhar to Ars Technica