A federal court in San Francisco has ruled that AI company Anthropic’s use of copyrighted books to train its large language model, Claude, qualifies as fair use under U.S. law. However, the court found that the company’s storage of millions of pirated books constitutes a copyright violation, setting the stage for a December trial to determine potential damages.
U.S. District Judge William Alsup stated that the use of works by authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson in training Anthropic’s AI system was “exceedingly transformative” and aligned with the goals of copyright law. “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,” he wrote in the ruling.
However, Judge Alsup criticized Anthropic for retaining over 7 million pirated books in what he described as a “central library.” He determined that this storage was not protected under fair use and infringed on the rights of the authors—even if not all the works were directly used in training the model.
The authors, who filed a proposed class action lawsuit last year, alleged that Anthropic used pirated copies of their works without consent or compensation. The company is backed by major tech firms including Amazon and Alphabet.
In response to the ruling, an Anthropic spokesperson said, “We are pleased that the court recognized our AI training as transformative and consistent with copyright’s purpose in enabling creativity and fostering scientific progress.”
This ruling marks the first major U.S. decision to directly address how fair use applies to generative AI—an argument central to ongoing legal battles involving companies like OpenAI, Meta, and Microsoft.
While AI developers assert that training on copyrighted content enables innovation and the creation of transformative works, rights holders argue that such practices erode their ability to earn income from original content. Judge Alsup’s decision acknowledges both viewpoints: validating the legality of transformative AI training while condemning the unauthorized retention of pirated materials.
Anthropic could face statutory damages of up to $150,000 per work for willful copyright infringement in the upcoming trial. The outcome is expected to set a significant precedent for how AI companies access and utilize protected content in training generative models.
Recent Random Post:
