As generated AI tools continue to spread at a rapid pace, the way content creators’ litigation focuses on training for these systems is just as quickly as possible. While two rulings this week favor anthropomorphism and meta-elements and insist on using copyrighted books to train large language models (LLMS), they also focused on unresolved issues, including the use of pirated material and whether the emerging technology requires a new legal framework. Uncertainty remains about AI companies’ performance in future litigation.
“Both cases are positive,” notes: Brandon Butler, executive director of CREATE, is a coalition focused on balancing copyrights, he told Observer. “But these are the rulings of the district court, so there are more steps on the way, and there are many other cases.”
Judges debate on the “change nature” of generative AI
On June 23, a federal judge ruled personification in a lawsuit filed last year by a group of authors who claimed that the company’s Claude models received copyrighted books training without permission or compensation. Judge William Alsup found that human use was protected by the doctrine of “fair use” on the grounds of how companies use the material.
“Like any reader who aspires to be a writer, Anthropic’s LLM is trained to work without competing forward, copying or replacing them, but turning and creating something different,” Alsup wrote in his decision. However, he criticized Anthropic’s decision to download millions of copyrighted books from the Pirates’ website. A separate trial scheduled for December will determine whether the company owes losses.
Anthropic said in a statement that it was satisfied with the court’s recognition of its transformative use, saying the ruling was “consistent with the purpose of copyright to achieve creativity and promote scientific progress”.
In 2024, Meta was also sued by a group of writers, including comedian Sarah Silverman and writer Ta-Nehisi Coates. Despite some warnings, on June 25, Judge Vince Chhabria’s ruling was tied to the tech giant. Chhabria found that Meta used copyrighted books to train its fairly used camel model, but he noted that the plaintiff made flawed arguments that failed to show that Meta’s actions hurt the author’s market.
Chambria also criticized Judge Alsup’s earlier ruling that “focusing on the transformative nature of generative AI while abandoning concerns about the harm it may cause to the engineering it is trained”, suggesting that market impacts become increasingly important in future fair use domination. He warned that Generative AI has the potential to “sink the market” through endless images, songs, articles and books, rather than efforts created by humans, rather than incentives that inspire people to create “old-fashioned ways.”
Meta said it welcomed the decision. “The open source AI model provides transformative innovation, productivity and creativity for individuals and companies,” the company said in a statement. “The rational use of copyright materials is an important legal framework for establishing such transformative technologies.”
By raising questions about market dilution, Judge Chambria’s decision could impact an increasing number of lawsuits involving a wide range of AI companies sued by writers, news publishers, film studios and artists. “In cases involving uses like Meta, it seems that plaintiffs will usually win, at least in better cases on the record of the market impact these cases have on the defendant’s use,” Chhabria noted.
Butler said that due to the direct competitive threat posed by AI tools, some sectors (such as press and publishing) may have stronger arguments in this regard. “I do suspect that as these cases continue, other plaintiffs will use the theory to see if other judges agree,” he said.
However, for now, there is no doubt that the recent rulings supporting anthropomorphism and meta represent early victories for tech companies. “Of course, this is not the end of the story,” Butler said. “It’s a very, very positive start.”
