Copyright Alliance guest author Bill Rosenblatt analyzes the class-action copyright infringement lawsuit filed by author and journalist Julian Sancton against AI developer Anthropic. The analysis points out potential weaknesses in the plaintiffs' case, which alleges that Anthropic’s AI model Claude was illegally trained on their copyrighted works. Rosenblatt argues that the complaint lacks direct evidence that the plaintiffs' specific works were actually included in the training dataset, relying instead on circumstantial evidence like the AI’s ability to summarize their books.

The article further explores the significant legal hurdle of the “fair use” doctrine, a key defense for AI companies in such lawsuits. It suggests that using copyrighted works for the technical purpose of training an AI model could be considered a transformative, non-expressive use, and thus fair use. The analysis highlights the difficulty plaintiffs may face in demonstrating actual market harm and concludes that the case underscores complex legal questions at the intersection of copyright law and generative AI.