In a landmark settlement, Anthropic, a leading artificial intelligence company, has agreed to pay $1.5 billion to a group of authors and publishers after a judge ruled it had illegally downloaded and stored millions of copyrighted books.
The settlement is largest payout in the history of U.S. copyright cases. Anthropic will pay $3,000 per work to 500,000 authors.
The agreement is a turning point in a continuing battle between A.I. companies and copyright holders that spans more than 40 lawsuits across the country. Experts say the agreement could pave the way for more tech companies to pay rights holders through court decisions and settlements or through licensing fees.
“This settlement sends a powerful message to A.I. companies and creators alike that taking copyrighted works from these pirate websites is wrong,” said Justin A. Nelson, a lawyer for the authors who brought the lawsuit against Anthropic.
The agreement is reminiscent of the early 2000s, when courts ruled that file-sharing services like Napster and Grokster infringed on rights holders by allowing copyrighted songs, movies and other material to be shared free on the internet. “This is the A.I. industry’s Napster moment,” said Cecilia Ziniti, an intellectual property lawyer who is now chief executive of the artificial intelligence start-up GC AI.
The settlement came after a ruling in June by Judge William Alsup for the U.S. District Court for the Northern District of California. In a summary judgment, the judge sided with Anthropic, maker of the online chatbot Claude, in significant ways. Most notably, he ruled that when Anthropic acquired copyrighted books legally, the law allowed the company to train A.I. technologies using the books because this transformed them into something new.
“The training use was a fair use,” he wrote. “The technology at issue was among the most transformative many of us will see in our lifetimes.”
But he also found that Anthropic illegally acquired millions of books through online “shadow libraries” like Library Genesis and Pirate Library Mirror that many tech companies have used to supplement the huge amounts of digital text needed to train A.I. technologies. When Anthropic downloaded these libraries, the judge ruled, its executives knew they contained pirated books.
Anthropic could have purchased the books from many sellers, the judge said, but instead preferred to “steal” them to avoid what the company’s chief executive, Dario Amodei, called “legal/practice/business slog” in court documents. Companies and individuals who willfully infringe on copyright can face significantly higher damages — up to $150,000 per work — than those who are not aware they are breaking the law.
As part of the settlement, Anthropic said that it did not use any pirated works to build A.I. technologies that were publicly released. And anyone still has the right to sue Anthropic if they believe that the company’s technologies are reproducing their works without proper approval. Anthropic also agreed to delete the pirated works it downloaded and stored.
After the judge ruled the authors had cause to take Anthropic to trial over the pirated books, the two sides decided to settle.
“Anthropic saw the writing on the wall that there was a substantial amount of liability,” said Chad Hummel, a trial lawyer with the law firm McKool Smith, who was not involved in the case.
Anthropic did not immediately respond to a request for comment.
Judge Alsup still needs to approve the settlement. Another case involving A.I. reached a settlement last month, after two actors accused the A.I. voice company ElevenLabs of mimicking their voices. The settlement in that case, which was filed in a Delaware court, has not yet been made public.
In 2024, three book authors, Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, sued Anthropic, accusing the start-up of illegally using their work to train its A.I. models. The suit is among the four dozen cases that copyright holders have been brought against A.I. companies. Some have been dismissed by the courts.
Companies like Anthropic, OpenAI, Meta and Microsoft need enormous amounts of digital data, some of which is copyrighted, to build its A.I. models. The companies have long claimed that they are in their legal right not paying for the content because the material is public and they are not reproducing the material in its entirety. But authors, publishers, musicians and other artists have challenged this stance.
The New York Times has sued OpenAI and its partner, Microsoft, for copyright infringement of news content related to A.I. systems. OpenAI and Microsoft have denied those claims. (One of the law firms representing the authors in their case against Anthropic is also representing The Times in its case.)
Some A.I. companies have already signed agreements with news organizations and other copyright holders to license their material. OpenAI signed licensing deals with news organizations including Axel Springer, Condé Nast, News Corp and The Washington Post. In May, the tech giant Amazon signed an licensing agreement with The Times.
After The Times and several other organizations filed lawsuits over the use of copyrighted material in A.I. technologies, Anthropic tried a different approach by acquiring books “for legal reasons,” according to court documents. In early 2024, the company hired Tom Turvey, former head of the Google Books project, which created digital copies of millions of copyrighted works by scanning physical books from libraries.
The Google Books project also faced a copyright lawsuit, which was eventually decided in favor of Google.
After contacting major publishers about possibly licensing their books, Mr. Turvey bought physical books in bulk from distributors and retailers, according to court documents. He then hired outside organizations to dissemble the books, scan them and create digital copies that could be used to train the company’s AI. technologies.
Judge Alsup ruled that this approach was fair use under the law. But he also found the company’s previous approach — downloading and storing books from shadow libraries like Library Genesis and Pirate Library Mirror — was illegal.
The settlement in the Anthropic case does not set legal precedent, because the case did not go to trial. But it could influence other cases and the behavior of other A.I. companies, Mr. Hummel said.
During a deposition, a founder of Anthropic, Ben Mann, testified that he also downloaded the Library Genesis data set when he was working for OpenAI in 2019 and assumed this was “fair use” of the material.
In another case brought by book authors against Meta, court documents show that Meta employees used the Library Genesis database. One employee called it “a data set we know to be pirated,” according to the documents.
Like Judge Alsup, the judge in the Meta case also ruled that when books are legally obtained, using them to train A.I. models was fair use under the law.
But in another Delaware case, where the media organization Thomson Reuters sued the A.I. start-up Ross Intelligence, a judge found that the start-up’s use of copyrighted material to train its technologies was not considered fair use. The case has now been appealed.
Cade Metz is a Times reporter who writes about artificial intelligence, driverless cars, robotics, virtual reality and other emerging areas of technology.
The post Anthropic Agrees to Pay $1.5 Billion to Settle Lawsuit With Book Authors appeared first on New York Times.
