Anthropic will pay $1.5 billion to settle a lawsuit brought by a group of authors alleging that the AI company illegally pirated their copyrighted books to use in training its Claude AI models. The settlement was announced Aug. 29, as the parties in the lawsuit filed a motion with the 9th US Circuit Court of Appeals indicating they had reached an agreement.
"This landmark settlement far surpasses any other known copyright recovery. It is the first of its kind in the AI era," Justin Nelson, lawyer for the authors, told CNET. "It will provide meaningful compensation for each class work and sets a precedent requiring AI companies to pay copyright owners. This settlement sends a powerful message to AI companies and creators alike that taking copyrighted works from these pirate websites is wrong."
The settlement still needs to be approved by the court, which it could do at a hearing on Monday, Sept. 8. Authors in the class could receive approximately $3,000 per pirated work, according to their attorneys' estimates. They expect the case will include at least 500,000 works, with Anthropic paying an additional $3,000 for any materials added to the case.
"In June, the District Court issued a landmark ruling on AI development and copyright law, finding that Anthropic's approach to training AI models constitutes fair use. Today's settlement, if approved, will resolve the plaintiffs' remaining legacy claims," Aparna Sridhar, Anthropic's deputy general counsel, told CNET. "We remain committed to developing safe AI systems that help people and organizations extend their capabilities, advance scientific discovery, and solve complex problems."
This settlement is the latest update in a string of legal moves and rulings between the AI company and authors. Earlier this summer, US Senior District Court Judge William Alsup ruled Anthropic's use of the copyrighted materials was justifiable as fair use -- a concept in copyright law that allows people to use copyrighted content without the rights holder's permission for specific purposes, like education. The ruling was the first time a court sided with an AI company and said its use of copyrighted material qualified as fair use, though Alsup said this may not always be true in future cases.
Two days after Anthropic's victory, Meta won a similar case under fair use.
Read more: We're All Copyright Owners. Why You Need to Care About AI and Copyright
Alsup's ruling also revealed that Anthropic systematically acquired and destroyed thousands of used books to scan them into a private, digitized library for AI training. It was this claim that was recommended for a secondary, separate trial that Anthropic has decided to settle out of court.
In class action suits, the terms of a settlement need to be reviewed and approved by the court. The settlement means both groups "avoid the cost, delay and uncertainty associated with further litigating the case," Christian Mammen, an intellectual property lawyer and San Francisco office managing partner at Womble Bond Dickinson, told CNET.
"Anthropic can move forward with its business without being the first major AI platform to have one of these copyright cases go to trial," Mammen said. "And the plaintiffs can likely receive the benefit of any financial or non-financial settlement terms sooner. If the case were litigated through trial and appeal, it could last another two years or more."
Don't miss any of our unbiased tech content and lab-based reviews. Add CNET as a preferred Google source.
Will the settlement affect other copyright disputes?
Copyright cases like these highlight the tension between creators and AI developers. AI companies have been pushing hard for fair use exceptions as they gobble up huge swaths of data to train their models and don't want to pay or wait to license them. Without legislation guiding how companies can develop and train AI, court cases like these have become important in shaping the future of the products people use daily.
"The terms of this settlement will likely become a data point or benchmark for future negotiations and, possibly, settlements in other AI copyright cases," said Mammen. Every case is different and needs to be weighed on its merits, he added, but it still could be influential.
Watch this: The Hidden Impact of the AI Data Center Boom 05:13
There are still big questions about how copyright law should be applied in the age of AI. Just like how we saw Alsup's Anthropic analysis referenced in Meta's case, each case helps build precedent that guides the legal guardrails and green lights around this technology. The settlement will bring this specific case to an end, but it doesn't give any clarity to the underlying legal dilemmas that AI raises.
"This remaining uncertainty in the law could open the door to a further round of litigation," Mammen said, "involving different plaintiffs and different defendants, with similar legal issues but different facts."
For more, check out our guide to understanding copyright in the age of AI.