A federal judge has granted final approval to Anthropic’s US$1.5 billion (approximately GBP 1.2 billion) settlement of the largest copyright class-action lawsuit ever certified, closing a chapter of the AI company’s legal history while leaving the industry’s core fair-use question unresolved.
US District Judge Araceli Martinez-Olguin signed off on the settlement on July 20. She succeeded William Alsup, who had granted preliminary approval last September before retiring. Alsup had previously ruled that Anthropic’s training on copyrighted text qualified as fair use — a landmark finding for the AI industry — but also that the company had obtained some of its training materials illegally.
How Anthropic built its library
Anthropic assembled its training corpus from two sources. Books it purchased and scanned were deemed legally obtained. But the company also downloaded millions of works from pirate sites including Library Genesis and the Pirate Library Mirror. Alsup found the second method illegal. Rather than risk a jury trial on the piracy question, Anthropic settled.
The settlement covers approximately 500,000 works, with authors and publishers set to receive roughly US$3,100 (approximately GBP 2,450) per work. Only 350 class members opted out, a figure that both sides characterized as validation of the agreement’s terms. The payout includes over US$100 million (approximately GBP 79 million) in attorney fees.
“The US$1.5 billion settlement provides substantial benefits to the class in light of the novel claims asserted,” Martinez-Olguin wrote in her judgment. “Success at trial was not assured, and a loss would have left the class with no recourse.”
A precedent that is not a precedent
Despite being the largest copyright settlement in US history, the resolution does not settle the legal question that the entire AI industry is watching. Because Anthropic settled, the case will never reach an appeals court, meaning Alsup’s fair-use ruling stands as a single district court decision with no binding weight on other judges.
Other courts remain free to reach their own conclusions. Dozens of similar lawsuits remain active against Google, Meta, Midjourney, and OpenAI. Just last week, a group of publishers including Hachette, Cengage, and Elsevier, along with author Scott Turow, filed a class-action suit against Google over its training of Gemini. A separate group of authors secured a settlement from OpenAI last year, but the terms were confidential.
What it means for the industry
For Anthropic, the approval removes a significant legal overhang as it expands its enterprise business. The company recently reported 41 percent share of enterprise AI spending among its peer group, ahead of OpenAI’s 39.5 percent, according to Ramp transaction data. Removing the copyright cloud could accelerate enterprise adoption.
But the broader question — whether training AI models on publicly available copyrighted text is fair use — remains unanswered at the appellate level. Until a case reaches the Supreme Court, every AI company operates under a patchwork of district-level rulings and settlements, none of which carry the weight of law beyond their own jurisdiction.
Sources: Anthropic’s landmark $1.5B copyright settlement is approved (TechCrunch, July 2026); Anthropic’s $1.5B copyright settlement approved; only 350 authors opted out (Ars Technica, July 2026); Anthropic to pay $1.5 billion copyright settlement to authors, publishers (Courthouse News, July 2026)