Midjourney wants discovery. Specifically, it wants Warner Bros., Disney, and Universal to hand over everything about how they use AI — business plans, training datasets, model weights, boardroom presentations. The whole stack.
This is the sequel to a lawsuit the studios filed last year, accusing Midjourney of copyright infringement for generating images of Superman, Batman, and other copyrighted characters. Midjourney’s defense? Training on publicly available images is fair use — and the studios do the exact same thing.
A magistrate judge already let the studios withhold most of their AI info, saying they only need to disclose consumer-facing AI applications. Midjourney is now asking a federal court to overturn that ruling.
The argument is simple. “If Plaintiffs are doing the very thing they seek to punish, that evidence goes to the heart of Midjourney’s fair use and unclean hands defenses,” wrote Midjourney’s attorney Bobby Ghajar.
Hard to argue with that logic. If Disney has been training its own models on copyrighted material while suing someone else for doing it, that’s not just hypocritical — it’s relevant to the case.
The judge’s decision could set a precedent for what AI companies can demand in discovery going forward. This one matters beyond just Midjourney.
