Technology & AI: Judge approves $1.5 billion copyright settlement between Anthropic and authors
Settlement resolves claims that Anthropic trained its Claude chatbot on hundreds of thousands of pirated books, with roughly $3,000 per title; future works excluded.
SAN FRANCISCO — A federal judge on Thursday approved a $1.5 billion settlement between artificial-intelligence company Anthropic and authors who allege nearly 465,000 books were pirated to train chatbots. The agreement would pay authors and publishers about $3,000 for each book covered and does not apply to future works.
U.S. District Judge William Alsup granted preliminary approval in San Francisco federal court after the sides addressed his concerns about how claims would be processed and how notices would reach eligible authors and publishers. About 465,000 books are on the list of works the authors say were pirated by Anthropic in training its Claude chatbot.
Alsup said, "This is a fair settlement," though he noted that distributing it to all parties would be "complicated." He had previously set a Sept. 22 deadline for submitting a claims form for his review ahead of Thursday's hearing. The judge raised concerns about two groups connected to the case — the Authors Guild and the Association of American Publishers — potentially pressuring authors to sign off without fully understanding their rights.
Attorneys for the authors argued in a Monday filing that the settlement would yield a high claims rate, respect existing contracts, and be consistent with due process and the court's guidance. The Authors Guild said the settlement "marks a milestone in authors’ fights against AI companies’ theft of their works. It sends a clear signal to AI companies that infringement of authors’ rights comes at a steep price and will undoubtedly push AI companies towards acquiring the books they want legally, through licensing."
Anthropic said it is pleased with the preliminary approval and will focus on developing safe AI systems that help people and organizations extend their capabilities, advance scientific discovery, and solve complex problems. The company added that, as it has maintained, the court’s June ruling that AI training constitutes transformative fair use remains intact and that the settlement simply resolves narrow claims about how certain materials were obtained.
The settlement comes after a mixed ruling in June in which Alsup found that training AI chatbots on copyrighted books wasn’t illegal but that Anthropic wrongfully acquired millions of books through pirate websites to help improve its Claude model. Several authors, including bestselling novelist Andrea Bartz, supported the agreement in court declarations and said the case sends a message that authors’ rights must be protected and that AI companies should license works when possible.
Industry and authors groups welcomed the decision. Maria Pallante, president and CEO of the Association of American Publishers, called the settlement a major step in holding AI developers accountable for reckless and unabashed infringement, while noting that many AI developers have trained on authors’ works using sources from infringing sites. The Authors Guild and its members have argued that the settlement should push the industry toward licensing rather than infringement.
Andrea Bartz, one of the authors who filed suit, said in a court declaration that she strongly supports the settlement and will work to explain its significance to fellow writers, adding that authors and publishers are sending a clear message to AI companies that infringement of IP rights will carry a price.
Alsup also announced in court that he plans to step down from the bench by the end of the year. The longtime jurist, nominated by President Bill Clinton in 1999, has presided over a number of high-profile tech and IP cases in the San Francisco federal court system.