Anthropic to Pay Authors $1.5B in Landmark AI Copyright Settlement

Anthropic has agreed to pay $1.5 billion to settle lawsuits brought by authors and publishers who accused the company of using millions of copyrighted books to train its Claude model. Specifically, Anthropic allegedly downloaded material from free pirated book-sharing websites and incorporated this content into model training without authorisation, with senior executives reportedly aware that pirated works were being used.

Under the settlement, the total payout will amount to $1.5 billion to affected authors and publishers, described as the largest copyright-related payout in U.S. history, with an average of roughly $3,000 per book though the actual figure per author will be significantly lower after publishers take their share.

The four installments are: $300 million within five days of the court’s preliminary approval, $300 million within five days of the court’s final approval, $450 million (including interest) within twelve months of preliminary approval, and $450 million within one year of preliminary approval. The court previously ruled that Anthropic’s use of such data for model training constituted “fair use,” a position the company maintains while agreeing to compensate to resolve outstanding claims.

While this settlement resolves the dispute with authors and publishers, a separate case will examine allegations that Anthropic executives knowingly allowed the use of pirated materials, with a trial scheduled for December 2025.