Anthropic Pays $1.5B as Artists Press AI Copyright Lawsuits
Get the Tech newsletter
Daily tech — startups, AI labs, chips, the launches that shape the next decade. Free.
- The Atlantic published a searchable dataset of works used to train AI, enabling authors like Kirk Wallace Johnson to discover his books had been scraped without consent, prompting him to join the Susman Godfrey suit against Anthropic.
- Anthropic settled for $1.5 billion — the largest copyright settlement ever — and agreed to destroy its trove of pirated ebooks, though Judge William Alsup ruled that legally acquired books used under "Project Panama" qualified as fair use as "quintessentially transformative."
- Sarah Andersen and other visual artists filed a class action against Stability, Midjourney, DeviantArt, and Runway AI in January 2023, just months after Stable Diffusion and Midjourney's debut, in a case still working through the courts.
- Independent musician Sam Kogon is leading a suit against Google's Lyria AI music engine on terms-of-service grounds, arguing that YouTube's "irrevocable perpetual license" can't justify training AI on content uploaded before the technology existed.
- Kadrey v. Meta saw initial claims dismissed for lack of market harm evidence, but narrower copyright claims around the use of pirated materials continue, with the original complaint warning AI could prevent "the next Agatha Christie from getting noticed."
- Lead plaintiff Andrea Bartz called the Anthropic settlement "the first major win for creatives against an AI company" but disputed the fair use ruling, noting that "even a library can't buy a physical copy of a book, scan it, and start lending it out as an ebook."
- Artists warn that AI's biggest threat isn't to top-tier creators but to working-class writers, musicians, and illustrators producing "mediocre" work that sustains their careers — the very volume that's most easily displaced by AI output.
Why it matters: The courtroom pendulum is shifting against AI companies: Anthropic's $1.5 billion settlement and rising public support suggest courts are losing patience with unconsented training data, even as fair use rulings still shield legally sourced material. The biggest losers aren't the best-selling authors in headlines — it's the working-class creators whose mid-tier work sustains their livelihoods, now threatened by an unending tide of AI-generated content.
