In a landmark decision, a federal court has ruled that Anthropic, an Amazon-backed AI company, is largely protected by the fair use doctrine when using millions of books to train its artificial intelligence system, Claude. The ruling, penned by U.S. District Judge William Alsup, marks the first judicial opinion to tackle the contentious issue of AI training on copyrighted material and is poised to influence similar lawsuits, including those brought by Disney and Universal against AI image generator Midjourney. While the decision offers a lifeline to AI developers, it also underscores potential liabilities, with Anthropic facing a trial over illegally downloading seven million books, a misstep that could cost the company hundreds of millions in damages.
The core question in these cases revolves around fair use, the legal principle allowing limited use of copyrighted material without permission. Judge Alsup’s ruling found Anthropic’s use of books to train Claude to be transformative, likening it to a person reading books to learn how to write. “Everyone reads texts, then writes new texts,” Alsup wrote in the court order. “To make anyone pay specifically for the use of a book each time they read it… would be unthinkable.” The court emphasized that Anthropic’s AI does not replicate or replace the original works but creates something new, a finding that bolsters the argument that AI training on copyrighted datasets is permissible under fair use.
However, the victory is not absolute. The court ruled that Anthropic’s initial act of illegally downloading seven million books to build its training library was not protected by fair use. The company’s attempt to retroactively purchase copies of the books does not absolve it of liability, Alsup noted, setting the stage for a trial that could result in significant damages. With statutory damages for willful copyright infringement reaching up to $150,000 per work, Anthropic could face a financial hit in the hundreds of millions, a precedent that could ripple through the industry.
The ruling has sparked unease among major studios like Disney and Universal, who are suing Midjourney for allegedly using thousands of their films to train its image generator. Unlike Anthropic, which implemented guardrails to prevent infringing outputs, Midjourney has been accused of producing near-exact replicas of film frames, such as Thanos from Avengers: Infinity War or scenes from Top Gun: Maverick. These outputs, detailed in the studios’ complaints, suggest Midjourney may have directly copied copyrighted works, a practice Alsup’s ruling deemed “inherently, irredeemably infringing.” The studios argue that Midjourney’s ability to replicate their films’ animation styles—from Shrek to The Lego Movie—proves unauthorized use in training.
The ruling also draws parallels to Google’s Book Search, where limits on text snippets prevented copyright violations. Alsup noted that if Anthropic’s outputs were infringing, the authors’ case would be stronger, but no such claims were made. For Disney and Universal, however, Midjourney’s outputs could be their trump card, potentially leading to massive payouts if piracy or direct infringement is proven.
As AI technology reshapes industries, Alsup described it as “among the most transformative many of us will see in our lifetimes.” Yet, the balance between innovation and intellectual property rights remains precarious, with this ruling serving as both a shield for AI developers and a warning of the costly consequences of overstepping legal boundaries.
Please follow us on Facebook and X for more news, tips, and reviews. Need cord cutting tech support? Join our Cord Cutting Tech Support Facebook Group for help. You can find Luke on X HERE.