Court grants final approval to $1.5 billion Anthropic copyright settlement

What happened

On 20 July 2026, Judge Araceli Martinez-Olguin of the U.S. District Court for the Northern District of California granted final approval to a $1.5 billion class-action settlement resolving claims that Anthropic downloaded copyrighted books from unauthorized online libraries (Library Genesis and Pirate Library Mirror) and used them in developing its Claude AI models. The settlement fund covers 482,460 works on the court-approved Works List, with an estimated gross allocation of about $3,100 per eligible claimed work; the court also cut the attorneys' fee award from the requested $187.5 million to about $101.6 million. Anthropic resolved the claims without admitting wrongdoing. The claim deadline (30 March 2026) has passed, and payments cannot begin until the judgment becomes effective after the appeal period or any appeals are resolved.

Why it may matter to you

This is the largest known U.S. copyright settlement, and it is specifically about how training data was acquired, not about whether training itself is lawful. The case drew a sharp line: a 2025 ruling in the same case held that training on lawfully purchased books was fair use, but that acquiring and retaining pirated copies was not — and the piracy exposure is what produced the $1.5 billion figure. For authors, the settlement's 50/50 author-publisher split rule (and full payment where rights reverted before 10 August 2022) is the practical detail that now determines who gets paid.

What is not yet known

Distribution is not finished: claim notices sent by 4 September 2026 surfaced many competing claims between authors, publishers and agencies, and funds tied to contested works are held back while splits are worked out. Works appearing only in the separate Books3 dataset, or scanned books not on the Works List, are not necessarily covered or compensated. The settlement covers Anthropic's conduct only through 25 August 2025 and leaves output-based and future-infringement claims outside its scope.