The studios sued Midjourney for training AI on copyrighted characters. Midjourney says the studios are doing the same thing and wants a judge to make them prove it.
Midjourney, the AI image generator, has asked a federal court to force Disney, Warner Bros. Discovery and Universal to hand over their internal AI training materials, arguing the studios are doing the very thing they sued Midjourney for in a 2025 copyright fight over characters like Superman and Batman.
The motion, filed as Document 112 in the Central District of California, escalates a discovery fight that has become the real battleground in the case. Midjourney wants the studios to produce their AI business plans, research reports, training datasets, model weights (the numerical parameters that shape how an AI model generates output), and board-level AI presentations.
"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," Midjourney attorney Bobby Ghajar wrote, according to Variety's reporting on the filing.
Fair use is the legal doctrine that can permit using copyrighted material in certain contexts. Unclean hands is a defense arguing a plaintiff accusing someone of misconduct should not itself be engaged in similar misconduct. Midjourney's theory is straightforward: if its training on publicly available images counts as copyright infringement, then the studios' own AI training practices deserve the same scrutiny, and proving that requires access to internal workflows rather than just consumer-facing products.
The studios have a legitimate interest in protecting proprietary AI workflows, board materials and trade secrets. They sued Midjourney last year alleging the image generator could produce copyrighted characters on demand. Midjourney has argued throughout that training on publicly available images is fair use.
In mid-June 2026, a magistrate judge narrowed what the studios must produce, according to Engadget's coverage of the procedural history, letting them withhold most internal AI information and limiting disclosure to consumer-facing AI applications. Midjourney's new motion asks the district judge to overturn that protective order and expand the scope of what the studios must hand over.
The studios' lawsuit has become a double-edged discovery instrument. By opening a copyright fight over AI training, the plaintiffs gave Midjourney a hook to demand their adversaries' internal AI data. The district judge's decision on Document 112 will not resolve the underlying copyright question, but it will set what evidence either side can actually use at trial.
That ruling could also shape how much AI training information courts can force companies to disclose in future copyright cases, an industry-wide question that reaches beyond these two parties. The Warner Bros. case has been consolidated with the Disney lead matter, and both will turn on whatever scope the district judge sets.
The motion is pending. The next concrete milestone is the district judge's order on Document 112.