As we previously blogged, Anthropic has filed a consent motion with the Ninth Circuit asking to stay its appeal of a federal district court ruling on class certification, because the company agreed to settle with the plaintiffs — journalists and book authors – who filed suit in 2024 alleging that Anthropic used “hundreds of thousands of copyrighted books” from unauthorized sources to train Anthropic’s Claude AI chatbot.
In June, the district court issued a mixed order on fair use in connection with the training of Large Language Models (LLMs) for use in generative AI (GenAI) tools like Claude and ChatGPT.
The court found that “the purpose and character of using copyrighted works to train LLMs to generate new text was quintessentially transformative” and thus allowed under the doctrine of fair use.
Said the court,
Like any reader aspiring to be a writer, Anthropic’s LLMs trained upon works not to race ahead and replicate or supplant them — but to turn a hard corner and create something different. If this training process reasonably required making copies within the LLM or otherwise, those copies were engaged in a transformative use.
However, the court also found that building a library of pirated copies of copyrighted works in order to train an LLM was not protected as fair use.
As the New York Times described it,
Anthropic had illegally acquired millions of books through online 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.
Now, the plaintiffs have filed a “Supplemental Brief in Support of Motion for Preliminary Approval of Class Settlement” with the court, including the plan of distribution for the settlement.
According to the brief,
The Court should grant preliminary approval of this historic $1.5 billion Settlement. The Settlement provides an extraordinary monetary recovery to the class, releases only certain past claims, and requires the permanent destruction of the two allegedly pirated datasets.
As the brief notes,
Class actions often must navigate agency problems, with representative plaintiffs and counsel advocating for absent class members who have no direct voice and whose interests largely must be surmised. This class action is different: absent class members whose rights are being litigated are not only able to speak, but to participate actively. …From the moment the Court certified the class and first included publishers in the case, they have been in an all-out sprint: mobilizing to work with Class Counsel to assemble the Works List, enlisting and preparing publisher witnesses to be poised for an historic trial, and now helping to deliver this groundbreaking settlement for the benefit of the entire Class.
According to the New York Times, Anthropic will pay $3,000 per work to the 500,000 authors and publishers whose works were copied without authorization.
According to the brief, notice of the settlement and the procedure for authors and publishers to claim their shares
will be distributed via first-class mail and e-mail. In addition, a robust outreach plan exists through social media, publication in print, publication online, and targeted outreach by several well-known membership organizations in the United States and other countries. … Moreover, due to the publicity this case has attracted, Plaintiffs expect that the case will continue to generate both press and commentary among interested groups.
According to the Times,
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 case is just one of at least 40 in the United States involving claims by copyright and trademark owners against GenAI companies.
Just like the haiku above, we like to keep our posts short and sweet. Hopefully, you found this bite-sized information helpful. If you would like more information, please do not hesitate to contact us here: https://aeonlaw.com/contact-us/.
