US judge vindicates Anthropic in copyright case

Image Source: Unsplash

A US judge has ruled that AI firm Anthropic did not violate copyright law by using published books to train its language models, marking a major legal win for the tech industry and a setback for authors and publishers.

According to TechCrunch, Federal Judge William Alsup determined that the company's use of copyrighted texts qualifies as "fair use" under US copyright law, making it the first court decision to validate AI developers' claims that training AI models on such materials can be legally protected. 

The decision could influence dozens of similar lawsuits brought by writers and artists against companies like OpenAI, Meta, and Google.

However, the court stopped short of clearing Anthropic entirely. 

Judge Alsup said a trial will still be held to examine the company's alleged creation of a massive "central library" of pirated books. 

According to the lawsuit, Anthropic downloaded millions of copyrighted books from illegal sources to create a permanent training archive. "We will have a trial on the pirated copies used to create Anthropic's central library and the resulting damages," the judge wrote.

The plaintiffs in Bartz v Anthropic argue that while AI companies may eventually buy legal copies, initial use of pirated versions causes direct harm and cannot be retroactively excused. 

The judge agreed that purchasing copies later "will not absolve it of liability for theft," though it could impact the scale of damages.

This ruling underscores the growing legal and ethical debate around how tech companies acquire and use content for AI training. Fair use — a flexible doctrine that weighs purpose, transformation, and commercial impact — has become a key legal battleground, particularly as the law has not been significantly updated since 1976.

A trial date has yet to be announced.