Edward Baxter

A class action copyright infringement lawsuit brought by U.S. authors against the AI company Anthropic has reached settlement, avoiding a trial set to begin in December. The class of plaintiff-authors alleged in the suit that Anthropic used millions of pirated books without authorization to train its popular Claude AI assistant. This case has had an unusual path to settlement, following a split opinion in June by Northern California District Judge, William Alsup. Alsup held that Anthropic's use of copyrighted works for AI training constituted fair use, but determined the company nonetheless violated copyright law by maintaining pirated books in a "central library" for use far beyond training purposes. This liability exposure, which could have potentially led to billions or even trillions of dollars in penalties assessed against Anthropic if the pending trial did not go its way, was likely the primary factor which drove Anthropic toward settlement.

The settlement's broader impact on pending AI copyright litigation against other major defendants such as OpenAI, Microsoft, and Meta remains uncertain, as this landscape of copyright law remains largely unsettled. Just two days following Judge Alsup’s ruling, Northern California District Judge Vince Chhabria issued a somewhat contrasting opinion in a similar authors' lawsuit against Meta, which suggested that Meta's fair use defense held water. However, the judge suggested that the defense could fail if that suit’s plaintiffs adjusted their arguments to indicate AI models’ potential to flood the market with reproductions of the authors’ works. With dozens of AI copyright cases pending, the unpredictability surrounding these novel legal questions may either incentivize additional settlements or encourage defendants to hold out for potentially favorable precedential rulings.

As authors, attorneys, executives, and judges alike continue to navigate this new copyright landscape, time will tell whether more AI companies follow in Anthropic’s footsteps and seek dispute resolution before trial.