US court formally approves Anthropic's nearly $1.5 billion settlement

2026-07-21

A US federal court on Tuesday approved a settlement under which Anthropic will pay about $1.5 billion (about 240 billion yen) to US writers and others in a class action over copyright infringement during AI development.

Settlement and record scale

It is the first major US copyright case involving AI developers to reach a settlement. According to the court, it is the largest settlement ever in a copyright class action. A federal court in California on the same day gave final approval, saying the deal was fair.

Only pirated books at issue

The parties announced the settlement in September 2025 and had been awaiting court review and approval. Anthropic's legal representative said claims had been filed by more than 90% of the writers and publishers covered by the deal, and said, 'We look forward to bringing this matter to a close.'

Writers sued the company in 2024, saying their books had been used without permission to train AI. The district court found copyright infringement over about 7 million books that the company had downloaded for free from pirate sites. For about 500,000 qualifying books, it will pay $3,000 per book to authors and publishers.

By contrast, in 2025 the court ruled that Anthropic's use of legally purchased books for AI training fell under fair use. The settlement payments are limited to downloads of pirated books. Under US copyright law, fair use can allow use of a copyrighted work without infringement in cases such as when it is used for a different purpose from the original.

The company also faces a copyright lawsuit from a major music publisher over the use of song lyrics. In US copyright cases involving generative AI, disputes including one between OpenAI and The New York Times are also ongoing.

Enjoyed this article? Share it with your network!