[AI Summary]: Brandon Butler of Re:Create Coalition argues that the Bartz v. Anthropic settlement should not be interpreted as establishing that AI training requires licensing. He emphasizes that Judge Alsup had already ruled AI training to be “spectacularly transformative” fair use, and the settlement primarily concerned Anthropic’s data sourcing from “shadow libraries” rather than the fair use question itself. Butler notes that copyright’s statutory damages of $750-$150,000 per work (potentially reaching 500,000 works) likely drove the settlement, and that another federal court (Judge Chhabria in Kadrey v. Meta) had recently ruled oppositely on the shadow library question, suggesting the plaintiffs settled because their position wasn’t guaranteed on appeal.
- Author: Brandon Butler
- Year: 2025
- Organization: Re:Create Coalition
- Type: Blog post/Legal commentary