Artists are lawyering up against AI slop, and some are even winning

Kirk Wallace Johnson went looking for his name in a searchable dataset The Atlantic published, tracking works used to train AI. He found it. Books he'd spent five to six years researching and writing, including The Feather Thief, had been pirated and fed to a chatbot. His reaction, he says, was a mix of anger, worry, and a desire for payback against companies that got "galactically wealthy" off other people's work. He's not alone. Dozens of writers, musicians, and visual artists have taken their grievances to court over the last few years, with mixed results but at least one genuinely landmark outcome.

The legal assault started early. Illustrator Sarah Andersen, along with Karla Ortiz, Kelly McKernan, and others, sued Stability, Midjourney, DeviantArt, and Runway AI back in January 2023, just months after Stable Diffusion and Midjourney launched and before ChatGPT had even become a household name. That case is still grinding through the courts. Since then the list of defendants has grown to include Meta, Google, and music generator Suno, with plaintiffs ranging from novelist Andrea Bartz to musician Sam Kogon, who's suing Google over its Lyria AI music tool not on copyright grounds but for allegedly violating its own YouTube terms of service.

Kogon's case is a useful reminder that these fights aren't identical. His lawyers argue Google misused its Content ID system and YouTube data to train Lyria and ProducerAI, while Google leans on broad language in its terms of service that lets it "reproduce, distribute, [and] prepare derivative works." Entertainment lawyer Krystle Delgado, who's separately representing musicians against Suno and Udio, points out that YouTube's terms grant an irrevocable, perpetual license the moment you upload — a take-it-or-leave-it deal most users never think twice about.

The biggest domino fell in Bartz v. Anthropic, where a judge ruled the company had violated copyright by training Claude on pirated ebooks. Anthropic agreed to destroy that pirated trove and pay $1.5 billion, the largest settlement in a copyright case to date. But the same judge, William Alsup, found that scanning millions of legally purchased secondhand books for training was fair use, calling it "quintessentially transformative" — a distinction Bartz says she strongly disagrees with, noting that even libraries can't just scan and redistribute physical books. Meanwhile, a similar suit against Meta from Richard Kadrey, Sarah Silverman, and others saw most claims dismissed for lack of proven market harm, though narrower copyright claims are still alive.

What unites these plaintiffs isn't just the money, though $1.5 billion certainly gets attention. It's a shared sense that the people building these tools don't understand or respect what they're consuming. Andersen called it "reducing my life's work to an algorithm." Johnson argues the real threat isn't AI writing the next Godfather — it's AI producing enough mediocre books and films to crowd out the working writers and screenwriters who currently make a living in that middle tier. Delgado, for her part, thinks momentum is shifting: judges, and public opinion, seem to be leaning toward the artists. Whether that translates into lasting guardrails, or just a slower, quieter version of the same disruption, is still an open question.