In what legal experts are calling the first copyright ruling favoring the artificial intelligence (AI) industry, a federal judge on Tuesday sided with Anthropic in a closely-watched case.
U.S. District Judge William Alsup of Northern California ruled Anthropic’s use of legally purchased books to train Claude, its AI model, did not violate federal copyright law. Anthropic’s use of copyrighted books to train its language learning model (LLM) was “quintessentially transformative” and did not violate “fair use” doctrine under copyright law, Alsup decided in a case considered pivotal in how judges respond to AI copyright cases.
“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,” Alsup wrote in his decision. “Authors’ complaint is no different than it would be if they complained that training schoolchildren to write well would result in an explosion of competing works.”
The ruling stems from a case filed last year by writers Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson after Anthropic used copies of their books to train Claude AI models. The authors claimed the company’s practices amounted to “large-scale theft.”
It was not a complete vindication for Anthropic, however. Alsup said the company may have broken the law when it separately downloaded millions of pirated books and said it will face a separate trial in December over the issue.
Alsup’s decision also does not address whether the outputs of an AI model infringe copyrights, which is at issue in related cases.
Court documents showed Anthropic employees were concerned about the legality of using pirate sites to access books. The company later shifted its approach and hired a former Google executive in charge of Google Books, a searchable library of digitized books that successfully weathered years of copyright battles.
Anthropic applauded the decision.
“We are pleased that the Court recognized that using ‘works to train LLMs (language learning models) was transformative — spectacularly so,” Anthropic said in a statement. “Consistent with copyright’s purpose in enabling creativity and fostering scientific progress, ‘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.'”
On Wednesday, Meta Platforms Inc. won yet another AI copyright lawsuit filed by several authors when a judge ruled the company’s use of the authors’ copyrighted works to train AI models was “fair use.”
“We are now 2-for-2 on it being highly transformative and thus, fair use to use copyrighted materials without permission to train an AI system (LLM),” Randy McCarthy, who heads the IP Law Group at national law firm Hall Estill, said in an email. “In the Kadrey v. Meta case, [federal] Judge Vince Chhabria used a slightly different approach to evaluating the fair use factors than Judge Alsup did in the Anthropic case, but both arrived at the same ultimate result.”
