Anthropic is set to begin compensating a group of authors and publishers who accused the AI company of copyright violations. A federal judge has given the green light to a significant $1.5 billion settlement from a class action copyright suit against Anthropic, as reported by various outlets.
Judge William Alsup from the U.S. District Court for the Northern District of California had previously granted preliminary approval for the settlement last year, determining that Anthropic had unlawfully downloaded and cached millions of copyrighted books.
Since then, Judge Alsup has retired, and Judge Araceli Martinez-Olguin approved the settlement on Monday.
The settlement will distribute $3,000 for each work across approximately 500,000 titles, to be apportioned among the authors and publishers who hold the rights. Though this settlement is considered unprecedented in the landscape of U.S. copyright law, many creators still perceive it as an unsatisfactory outcome.
The issue arises from how the legal matters were resolved. Judge Alsup leaned in favor of Anthropic on a pivotal point. He ruled that utilizing copyrighted text to train AI can be classified as fair use— a verdict viewed as significant for the AI sector. However, this ruling didn’t address the methods by which Anthropic acquired the books. The company had sourced its training materials from two avenues: books it legitimately purchased and scanned, and those it illegally downloaded from sites like Library Genesis and Pirate Library Mirror. Judge Alsup deemed the latter method illegal and mentioned that this piracy issue could potentially go to trial. Anthropic opted for settlement soon after, likely to avoid the uncertainties of a trial and possible jury-awarded damages.
While the conclusion of this case offers closure for this particular dispute, it does not resolve broader industry legal questions, as Judge Alsup’s ruling stems from a single district court and Anthropic’s decision to settle prevents the case from progressing to an appeals court for a binding precedent.
Other judges retain the authority to reach their own decisions based on differing facts, and this dynamic is unfolding in various ongoing cases. Numerous copyright lawsuits continue to challenge companies such as Google, Meta, Midjourney, and OpenAI regarding the legality of training AI models on copyrighted materials. Recently, a new class action was filed against Google by a consortium of publishers and authors, including Hachette, Cengage, and Elsevier, along with author Scott Turow, accusing the tech giant of employing their copyrighted works to develop its AI platform, Gemini.
