A U.S. federal judge has granted final approval to a $1.5 billion class action settlement between artificial intelligence company Anthropic and hundreds of thousands of authors and copyright holders whose books were used to train the company's large language models, Qazinform News Agency correspondent reports.
In a ruling issued Monday, U.S. District Judge Araceli Martínez-Olguín said the agreement is "fair, reasonable, and adequate."
The settlement covers copyright owners of books included on a court-approved list of works that Anthropic downloaded from the online libraries LibGen and PiLiMi. The fund is non-reversionary, meaning the money will remain available for eligible class members rather than returning to the company.
The court said the $1.5 billion settlement provides meaningful compensation given the risks of continued litigation. Eligible works are expected to receive about $3,000 each before fees and costs, roughly four times the minimum statutory damages typically awarded for ordinary copyright infringement. More than 91% of eligible works had already been claimed by April 2026, while only 350 valid opt-out requests were submitted.
Under the agreement, Anthropic must destroy the original pirated copies of books it obtained from LibGen and PiLiMi, subject to legal preservation requirements. The settlement does not prevent authors from bringing future claims over AI-generated outputs or future conduct by the company.
The judge also reduced the attorneys' fee award sought by class counsel. Instead of the requested $187.5 million, the court approved about $101.6 million, or nearly 6.8% of the settlement fund, along with reimbursement of about $2.6 million in litigation expenses and $15,000 service awards for each of the three class representatives.
Earlier, Qazinform News Agency reported that the US lifted restrictions on Anthropic’s frontier AI models.