US Federal Court Sides with Anthropic in Landmark AI Copyright Case
- firstcounsel.ai
- Jul 3, 2025
- 2 min read
United States · Los Angeles · July 3, 2025
In a significant ruling for the AI industry, U.S. District Judge William Alsup of the Northern District of California granted partial summary judgment to Anthropic PBC in the case of Bartz v. Anthropic (Case No. C 24-05417 WHA), finding that training AI models on copyrighted books without permission constitutes fair use under Section 107 of the Copyright Act.
The Fair Use Victory
Judge Alsup ruled that Anthropic's use of copyrighted works to train its Claude AI models was "exceedingly transformative" and justified under fair use doctrine. The court found that the training process was analogous to a person reading books to learn writing skills, noting that "everyone reads texts, too, then writes new texts" without owing additional compensation beyond the initial purchase.
Crucially, the plaintiffs—authors Andrea Bartz, Charles Graeber, and Kirk Wallace Johnson—never alleged that Claude's outputs directly reproduced their copyrighted works. This distinction proved pivotal, as the court emphasized that the transformative nature of AI training distinguished it from mere copying. Judge Alsup wrote that "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."
The Piracy Problem
However, Anthropic's victory was incomplete. Judge Alsup sharply criticized the company's acquisition of over seven million pirated books from sites like LibGen and Books3, rejecting Anthropic's argument that fair use justified downloading copyrighted works from "pirate sites." The court will proceed to trial in December on damages related to these pirated copies, potentially including willful infringement penalties up to $150,000 per work.
"That Anthropic later bought a copy of a book it earlier stole off the internet will not absolve it of liability for the theft but it may affect the extent of statutory damages," Judge Alsup wrote, emphasizing that companies cannot retroactively legitimize piracy through subsequent purchases.
Fair Use in the Digital Age
This decision highlights the tension between copyright law's 1976 framework and modern AI technology. Fair use doctrine traditionally considers four factors: purpose and character of use, nature of the copyrighted work, amount used, and market impact. Judge Alsup's analysis suggests courts may view AI training as sufficiently transformative to qualify for protection, mirroring arguments made by Meta, OpenAI, and other tech companies in similar pending litigation.
Future Implications and Appeal Prospects
While this ruling provides precedent favoring AI companies, it's not binding on other courts. Copyright lawyer John Strand of Wolf Greenfield called the decision from a "well-respected" judge "very significant," noting that "dozens of other cases involving similar questions" are pending throughout the U.S. Given the proliferation of AI copyright cases, Strand expects "the primary question of whether training LLMs on copyrighted materials is fair use likely will be addressed by the US Supreme Court."
The ruling establishes that AI companies cannot simply "bless themselves" by claiming research purposes while using pirated content—they must still respect copyright acquisition rights even when training constitutes fair use. This nuanced approach may become the template for balancing innovation with creators' rights in the evolving AI landscape.
LINKS
Comments