The fight over who pays for the material that trains AI is now running on two tracks at the same time. A music publisher filed billion-dollar lawsuits, a German children’s book publisher sued over a character that ChatGPT reproduced, and Hollywood signed its first licensing deal with an AI company. Those who can, license. Those who cannot, litigate.
The AI copyright lawsuits are getting bigger
Round Hill Music is suing Anthropic and Suno over the alleged use of lyrics from more than 500 copyrighted songs to train their AI models, seeking over $1 billion in damages. The publisher says it’s not looking to settle, and it’s pushing for jury trials. That’s the real clash here.
No backroom meetings, no shady deals, no agreements by handshakes. Copyright issues are real, and they have to be addressed properly.
Across the Atlantic, German publisher Carlsen Verlag, author Marc-Uwe Kling, and illustrator Astrid Henn sued OpenAI over the children’s book “Das NEINhorn.” They allege ChatGPT generated text and illustrations nearly identical to their copyrighted book, including unauthorized print templates.
I’m not personally familiar with the author’s work, nor with the proper legal terms, but if ChatGPT provides near-identical graphics and all the templates needed for publishing, I believe that indeed could be a problem, because for me it sounds like plagiarism.
Hollywood picked the other path
While the music and book worlds litigate, Hollywood chose to sign. ByteDance and the Motion Picture Association reached a memorandum of understanding, the first deal of its kind between the studios’ main lobby and an AI company. The agreement adds copyright guardrails to ByteDance’s Seedance video and Seedream image generators, so in this way, Seedance can train its models to generate better videos, and Hollywood can be sure the work will be done properly, without content piracy.
The deal closes a dispute that started about six months ago, when an AI-made clip of Tom Cruise fighting Brad Pitt went viral. But instead of shutting down its tools as OpenAI did with Sora under similar pressure, ByteDance chose to negotiate and keep building.
Why this matters for everyone
Generative AI’s progress is so impressive, especially in the past few months, that it has become a real threat to the established status quo. I have seen the trailers of some of the Chinese AI shows, and fair enough, I had a hard time staying convinced that I’d call them AI-generated if I hadn’t known it beforehand. So no, this isn’t a niche legal squabble.
The outcome decides the economics of everything built on top of content, from music to books to video. If the lawsuits win, AI companies face huge costs and may have to retrain. If the licensing path wins, the content world gets a new revenue stream, and the tools keep getting better and better.
For the average person, the question is simply about cost and creativity. If training on protected content gets expensive or blocked, the price of AI tools could rise, and the range of what they can produce could narrow. If licensing works, the creators who fed the machines may finally get paid, and the rest of us get better, more reliable tools. The barrier in front of creativity could start to fall.
The copyright war has two paths, and this week showed both in action. The next year will reveal which one wins.
