What happened
Midjourney is pushing three Hollywood studios to disclose how they use generative AI internally. The move comes during a legal dispute where Disney, Universal, and Warner Bros. are suing the AI startup for alleged copyright infringement, as reported by TechCrunch AI. Midjourney argues that if studios are also training models on unlicensed content for internal tasks like storyboarding, it would support its own fair use defence. The current filing seeks to overturn a judge's ruling that limited disclosure to only 'consumer-facing' AI outputs.
How the room's reading it
Legal analysts see this as a classic discovery battle tactic. Midjourney is trying to make the process painful for the plaintiffs and expose potential hypocrisy. On X, many developers and AI artists are framing this as studios wanting to have it both ways — using AI for their own gain while suing the toolmakers. The studios' counsel, meanwhile, has publicly called the move a 'fishing expedition' designed to distract from the central copyright claims. The consensus among legal watchers is that this filing isn't just about fair use — it's a strategic attempt to put the entire industry's AI practices on trial, not just Midjourney's.
Sailfish's take
We see this as more than just a legal spat. It's a warning shot for every builder relying on models with opaque training data. 'Fair use' is a lawyer's argument, not a durable product strategy, especially when your potential customers hold the copyrights. We've shipped enough products to know that supply chain risk is real — and for AI, the data is the supply chain. This case highlights a simple truth: if you can't explain where your model's intelligence came from, you're building on shaky ground. We'd avoid any model that can't provide a clear statement on data provenance. The legal bills will always outweigh the performance gains.