A federal judge sides with Anthropic in lawsuit over training AI on books without authors’ permission

Federal judge William Alsup ruled that it was legal for Anthropic to train its AI models on published books without the authors’ permission. This marks the first time that the courts have given credence to AI companies’ claim that fair use doctrine can absolve AI companies from fault when they use copyrighted materials to train LLMs.

This decision comes as a blow to authors, artists, and publishers who have brought dozens of lawsuits against companies like OpenAI, Meta, Midjourney, Google, and more. While the ruling is not a guarantee that other judges will follow Judge Alsup’s lead, it lays the foundations for a precedent that would side with tech companies over creatives.

These lawsuits often depend on how a judge interprets fair use doctrine, a notoriously finicky carve out of copyright law that hasn’t been updated since 1976 — a time before the internet, let alone the concept of generative AI training sets.

Fair use rulings take into account what the work is being used for (parody and education can be viable), whether or not it’s being reproduced for commercial gain (you can write Star Wars fan fiction, but you can’t sell it), and how transformative a derivative work is from the original.

Companies like Meta have made similar fair use arguments in defense of training on copyrighted works, though before this week’s decision, it was less clear how the courts would sway.

In this particular case of Bartz v. Anthropic, the group of plaintiff authors also brought into question the manner in which Anthropic attained and stored their works. According to the lawsuit, Anthropic sought to create a “central library” of “all the books in the world” to keep “forever.” But millions of these copyrighted books were downloaded for free from pirate sites, which is unambiguously illegal.

While the judge granted that Anthropic’s training of these materials was a fair use, the court will hold a trial about the nature of the “central library.”

“We will have a trial on the pirated copies used to create Anthropic’s central library and the resulting damages,” Judge Alsup wrote in the decision. “That Anthropic later bought a copy of a book it earlier stole off the internet will not absolve it of liability for theft but it may affect the extent of statutory damages.”