Midjourney Demands Studios Reveal Their Own AI Practices
Midjourney is trying to flip the script in its copyright battle with three of Hollywood’s biggest studios — by asking a federal judge to force Disney, Universal, and Warner Bros. Discovery to open up their own AI playbooks.
The image-generation company filed a motion asking U.S. District Judge John A. Kronstadt to overturn an earlier ruling that limited how much the studios have to disclose about their internal use of generative AI. If Midjourney gets its way, the studios could be compelled to hand over research reports, training data, model documentation, internal presentations, and even board-level discussions about AI — not just the material tied to their public-facing products.
How the Case Got Here
The lawsuit dates back to 2025, when Disney and Universal accused Midjourney of building a business on the back of their intellectual property. Their complaint pointed to the platform’s ability to generate recognizable versions of characters like Bart Simpson and Darth Vader. Warner Bros. Discovery joined the fight soon after, adding its own roster of characters — including Batman, Superman, and Scooby-Doo — to the list of allegedly infringed properties.
Midjourney’s defense rests on two pillars. First, it argues that training an AI model on publicly available images, even ones that depict copyrighted characters, can qualify as fair use. Second, it’s leaning on what’s known as an “unclean hands” argument — the idea that a plaintiff shouldn’t get to sue over conduct it’s quietly doing itself.
Also Read : Notta AI Review 2026: Is It Still Worth Your Money?
The Discovery Fight, Explained
Discovery — the pre-trial phase where both sides exchange evidence — has become the real battleground for now. In mid-June, Magistrate Judge Joel Richlin sided largely with the studios, ruling that they only had to turn over AI-related materials connected to consumer-facing tools and products. Anything happening behind the scenes — internal testing, storyboarding experiments, model research — was ruled out of scope.
Midjourney wasn’t willing to leave it there. Attorney Bobby Ghajar, representing the company, argued in the new filing that the ruling let the studios cherry-pick evidence that supports their case while shielding anything that might undercut it. According to Ghajar’s brief, if the studios are internally building AI tools using unlicensed material for tasks like storyboarding, that alone would show such practices are treated as normal across the industry — including by the very companies suing over it.
The studios aren’t budging. Their lead attorney, David Singer, has called the broader request a fishing expedition designed to pull attention away from the actual claims — that Midjourney’s tool reproduces their characters without permission. Singer has stressed that the studios aren’t trying to block AI development altogether; they’re asserting the same rights any copyright owner would claim against unauthorized copying, regardless of the technology involved.
Midjourney isn’t only after internal AI documents, either. The company also wants every prompt ever entered on its own platform by the studios’ teams, not only the specific prompts tied to the disputed images — a request the studios have pushed back on just as hard.
Why the Line Between “Internal” and “Consumer-Facing” Matters
The entire dispute now hinges on a fairly narrow-sounding distinction: does it matter how a studio uses AI before something reaches an audience?
Midjourney says yes. Internal tools used for early concept art, animation tests, or pitch decks may never be seen by the public, but they can still shape what eventually lands in a finished film or show — and they’d still reveal whether unlicensed training data is standard practice inside these companies. The studios counter that none of that internal experimentation has any bearing on whether Midjourney itself infringed their copyrights, and that expanding discovery that far would turn the case into an audit of their entire AI strategy.
A hearing on the matter is scheduled for August 17, 2026, and the outcome could determine how much of Hollywood’s AI activity becomes part of the public record.
The Bigger Picture in Hollywood
This case isn’t happening in isolation. Around the same time, Disney, Paramount Skydance, and the Motion Picture Association sent cease-and-desist letters to ByteDance over an AI tool capable of generating realistic clips of copyrighted characters — a sign that studios are fighting AI-generated likenesses on multiple fronts at once, not just in court against Midjourney.
Meanwhile, Hollywood’s labor side has already staked out its own position on AI. SAG-AFTRA ratified a new four-year contract in June 2026 that restricts when an AI-generated performer can replace a human actor, requiring that it add meaningful value rather than simply cut costs. Separately, the Directors Guild of America reached a four-year framework agreement covering AI use in directing and production work.
Taken together, these moves show studios walking a fine line: pushing hard against unauthorized use of their characters by outside AI companies, while simultaneously exploring — and in some cases quietly adopting — generative AI tools inside their own production pipelines.
What’s at Stake
If Judge Kronstadt sides with Midjourney, the studios could be forced to open up internal AI records that they’d almost certainly prefer to keep private, even under confidentiality protections. That outcome would hand Midjourney fresh ammunition for its fair use and unclean hands defenses, and it could pressure other studios to think twice about how their own AI experimentation might look under a legal microscope.
If the studios prevail instead, the case stays narrowly focused on what Midjourney’s own tool produces — and the question of how Hollywood uses AI behind closed doors stays exactly that: behind closed doors.
Either way, the ruling due after the August hearing will do more than settle a discovery dispute. It will signal how far courts are willing to go in scrutinizing not just what AI companies do with copyrighted material, but what the copyright holders themselves are doing with the same technology.
