The Anthropic copyright settlement worth $1.5 billion has cleared its final legal hurdle, with Judge Araceli Martinez-Olguin of the U.S. District Court for the Northern District of California signing off on Monday, formally closing a lawsuit that exposed one of the AI industry’s least flattering habits: training on books lifted from piracy sites.
The case, formally captioned Bartz v. Anthropic, was originally filed in 2024 by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, who accused Anthropic of using unauthorised copies of their books while developing its Claude AI models. Judge William Alsup granted preliminary approval in September 2025 before retiring at the end of that year, leaving Martinez-Olguin to deliver the final ruling.
What the $1.5bn Anthropic Copyright Settlement Actually Delivers
Under the terms of the agreement, rightsholders will receive $3,000 per work across an estimated 500,000 eligible works, divided between the authors and publishers who hold rights to them. Anthropic told the court that claims had been filed for more than 91% of the eligible works covered by the settlement, a participation rate that points to an organised and well-informed rights-holder community.
Plaintiffs’ attorney Justin Nelson described it as the largest publicly reported copyright recovery in history. The attorneys’ fees tell their own story: Judge Martinez-Olguin awarded counsel more than $101 million of the $187.5 million they had requested, according to Publishers Lunch. The judge overruled objections to the settlement’s overall size, stating that complaints were ‘not grounded in a realistic assessment of the overall risks and rewards of a trial.’
Many authors and creators are still reluctant to call any of this a victory. The reason lies in the structure of the case. Alsup sided with Anthropic on the central legal question, ruling that training an AI model on copyrighted text constitutes fair use. That conclusion drew wide attention across the technology and publishing industries. But it did not absolve how Anthropic built its training library in the first place.
Anthropic had assembled that library from two sources: books it purchased and scanned, which Alsup found acceptable, and books it downloaded from piracy sites including Library Genesis and Pirate Library Mirror, which he found illegal on its own terms. Alsup ruled the piracy question could proceed to trial; Anthropic settled shortly afterwards to avoid whatever damages a jury might have awarded.
A Precedent Gap That Leaves the Industry Unsettled
The settlement closes the case without settling the law. Alsup’s fair-use ruling was a single district court decision, and Anthropic’s choice to pay rather than appeal means it will never reach a circuit court and acquire binding authority. Judges presiding over other AI copyright cases remain free to reach different conclusions on different facts.
Those cases are multiplying. On 10 July 2026, a group of publishers and authors including Hachette, Cengage, Elsevier, author Scott Turow, and S.C.R.I.B.E. filed a class action against Google in the U.S. District Court for the Southern District of New York (Case No. 1:26-cv-05870). The Hachette v. Google complaint accuses Google of reproducing unauthorised copies under the Copyright Act and also raises claims for the removal of copyright management information, a charge that goes beyond what was at stake in the Anthropic case.
The complaint alleges that Google copied works not only from Google Books and Google Play Books but also from unauthorised web scrapes and known pirate sources, and from content sitting behind paywalls. Hachette Book Group said the publishers had originally intended to pursue their claims as intervenors in an existing case, known as In re Google Generative AI Copyright Litigation, before filing the new separate suit to preserve claims falling outside that putative class. That earlier litigation was itself brought by illustrators and writers in 2023, as Publishers Weekly reported at the time.
The Anthropic settlement has now produced the largest disclosed copyright payout in US legal history. What it has not produced is a rule. With Google, Meta, Midjourney, and OpenAI all still facing active litigation, the fair-use question Alsup tried to answer will get another hearing in another courtroom, almost certainly before another judge who is not bound by his conclusions.
