Court filings reveal that Anthropic, a prominent AI firm, has agreed to pay at least $1.5 billion to resolve a copyright infringement complaint brought by a group of authors who claimed the platform had unlawfully utilized stolen copies of their books to train large-language models.
According to the authors’ attorney, Justin Nelson, if the historic deal is accepted, it will be the biggest copyright recovery ever made public, surpassing both individual copyright cases that were litigated to a final verdict and any previous copyright class action settlement.
About 500,000 published works were at the core of the complaint, which was filed in a California federal court last year. In a note to the case’s court, Nelson stated that the proposed settlement amounts to a gross recovery of $3,000 per work.
This outcome is truly astounding, Nelson continued.
By downloading and “commercially exploiting” books that it purportedly obtained from pirate websites like Library Genesis and Pirate Library Mirror, Anthrophic was accused by the plaintiffs of “committing large-scale copyright infringement.”
According to U.S. copyright law, Anthropic had claimed that what it was doing qualified as “fair use.” The federal court assigned to the case decided in late June that Anthropic’s acts qualified as fair use due to the fact that the final product was “exceedingly transformative.”
However, there were important asterisks in Judge William Alsup’s decision. According to him, downloading publications that have been pirated does not qualify as fair usage.
According to Aparna Sridhar, Anthropic’s deputy general counsel, “the District Court issued a landmark ruling on AI development and copyright law in June, finding that Anthropic’s approach to training AI models constitutes fair use.”
“The plaintiffs’ outstanding legacy claims will be resolved if today’s settlement is authorized. Our dedication to creating safe AI systems that support human and organizational growth, scientific advancement, and the resolution of challenging issues is unwavering.” Sridhar added.
Three authors, Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson, first filed the complaint. Journalists Graeber and Johnson have both published factual works, while Bartz is both a journalist and a writer.
The settlement may influence the course of further ongoing legal disputes between published writers and AI platforms.
OpenAI has been sued by almost 20 best-selling writers, including John Grisham, George R.R. Martin, and Jodi Picoult, who claim that the company used their works to train ChatGPT and other tools, committing “systematic theft on a mass scale.”
According to Nelson, Anthropic committed to provide $300 million to the settlement fund over the course of four installments, with the first payment being due five business days after the court approved the terms.
The proposed $1.5 billion compensation is the “minimum size” of the settlement, according to Nelson’s note to Alsup. “Anthropic will pay an extra $3,000 per work that Anthropic adds to the Works List above 500,000 works if the Works List eventually exceeds 500,000 works,” he stated.
