Midjourney has filed a motion to compel Disney, Universal, and Warner Bros. to disclose their internal generative AI usage — potentially forcing the studios suing it for copyright infringement to reveal whether they rely on the same technology they claim caused them harm.

The motion, filed March 28, 2026, in Disney Enterprises Inc. v. Midjourney Inc. (U.S. District Court, Central District of California), appears in court records. Disney and Universal sued jointly in June 2025; Warner Bros. followed in September. The cases have been consolidated for discovery — the phase where both sides exchange documents and evidence.

The Discovery Dispute

Midjourney wants to overturn a restriction placed by Magistrate Judge Joel Richlin in June 2026. That ruling limited what the studios had to disclose: only generative AI usage that produced images or video the public actually saw. TechCrunch reported on July 4, 2026 that Midjourney now seeks to remove that barrier and compel disclosure of internal, unpublished AI work — tools used in production pipelines that touched copyrighted characters but never reached audiences.

Midjourney's legal defense rests on fair use: the argument that training image-generation models on copyrighted material is lawful, not infringement. Expanding discovery to cover studios' own internal AI practices fits that strategy. If Disney, Universal, and Warner Bros. are training their own models on proprietary intellectual property or running generative tools on characters they own, that fact could shape how a court weighs fair use — especially the "market harm" factor, which asks whether the defendant's use harms the copyright holder economically. It might also bear on what industry practice actually is.

David Singer, lead counsel for the studios, called Midjourney's requests a "fishing expedition," per Variety's reporting. That pushback is routine in broad discovery disputes, but Judge Richlin's original decision to narrow the scope suggests the court saw merit in the studios' position — at least to a point. Midjourney is now asking him to reverse course.

What Is Actually at Stake

This discovery fight matters beyond legal procedure. If Midjourney wins, it would require three of the largest media companies to release internal documentation on AI adoption that they have not disclosed publicly. Studios use machine-learning tools in visual effects, audio localization, marketing, and post-production — but which systems, which vendors, and which characters appear in those workflows remain confidential. Court-ordered disclosure of that detail would be unprecedented.

There is a strategic calculation here beyond mere defense. A defendant who can show a jury that the plaintiffs are themselves using substantially similar technology — even internally — builds a stronger fair use argument than one relying on abstract principle alone. Whether such evidence becomes admissible is the judge's call, but getting it into the case record through discovery is the first step.

The studios face their own logical tension. They argue that Midjourney's training on their copyrighted characters caused them economic injury. But if those same studios are running AI tools internally on the same characters without licensing or paying the underlying rights holders — the writers, artists, and performers whose likenesses appear in the source material — the question of who owes compensation to whom becomes less clear. Courts have not yet defined what fair use means when applied to AI training at scale, and discovery records showing studio practice on both sides could shape that definition.

For now, Judge Richlin holds the decision. His June order drew a line at public-facing output; Midjourney is asking him to erase it. Whenever he rules, the outcome will establish a precedent for how much latitude courts grant AI defendants in investigating the internal technology practices of the corporations suing them.