case-law

Disney and Warner Bros. v. Midjourney: The Character Generation Case

By Nora Whitfield · August 8, 2026

Ask Midjourney for an image of a lightsaber-wielding animated ogre and you'll probably get something that looks unmistakably like Darth Vader standing next to Shrek. That's the entire lawsuit, more or less, and it's why Disney, NBCUniversal, and Warner Bros. Discovery are now suing the AI image generator together in one of the highest-profile fights in this entire area of law.

The lawsuits

Disney and NBCUniversal filed the first major studio lawsuit against an AI company in June 2025, in the Central District of California. Warner Bros. Discovery followed in September 2025 with its own suit making similar claims. Both cases are now proceeding on parallel tracks against Midjourney.

The complaints are specific and visual in a way a lot of AI copyright filings aren't. Disney and Universal's 143-page filing names characters including Buzz Lightyear, Boss Baby, Shrek, Homer Simpson, Iron Man, Elsa, and Darth Vader, and describes Midjourney as functioning like "a virtual vending machine, generating endless unauthorized copies" of the studios' copyrighted characters. Warner Bros. Discovery's suit makes parallel allegations centered on characters like Batman, Superman, and Scooby-Doo.

The studios are seeking statutory damages of up to $150,000 per infringed work, the maximum available for willful infringement under US copyright law, and an injunction stopping Midjourney from continuing to generate these characters. They point to Midjourney's reported $300 million in annual revenue as evidence the company is commercially competing with, not merely inspired by, the IP it was trained on.

Why this case is different from the training-data fights

Most AI copyright litigation right now centers on the training side: was it lawful to use copyrighted material to build the model in the first place. The Disney and Warner Bros. suits lean harder on the output side. Their argument isn't primarily about what Midjourney was trained on, though that's part of the complaint too. It's about what the model produces on request, specific, recognizable, unmistakably identifiable characters that are core, heavily monetized IP for these studios.

That's a meaningfully easier case to visualize and argue than an abstract fair-use fight over training methodology. A jury doesn't need to understand transformer architecture to look at a Midjourney output next to Darth Vader and draw a conclusion. The studios are betting that specificity works in their favor, and it's part of why this case is being watched as a potential bellwether for character-IP claims against generative AI specifically, as distinct from claims about photorealistic style or general training-data sourcing.

The discovery fight over AI use

As the case moved into 2026, an unusual and genuinely interesting discovery dispute emerged. Midjourney sought to compel the studios to disclose their own internal AI use: training datasets, model weights, board-level presentations about AI strategy, and every prompt studio employees have entered into Midjourney's own platform. Midjourney's theory is that the studios' own AI practices are directly relevant to its fair-use defense and to an "unclean hands" argument, essentially that the plaintiffs are doing some version of the same thing they're suing over.

A magistrate judge ruled on June 16, 2026, limiting that discovery to AI use tied to the studios' own consumer-facing content, rejecting Midjourney's broader request for internal strategy documents and model weights. Midjourney has appealed, arguing the limitation lets the studios present evidence of harm from AI while blocking Midjourney from developing evidence that could support its own defense. That appeal remains pending, and it's a genuinely consequential procedural fight, because how much of the studios' own AI practices become discoverable could shape how the fair-use defense actually gets argued at trial.

What's still unresolved

As of mid-2026, there's no ruling on the merits, no trial date locked in for either the Disney/Universal or Warner Bros. tracks, and the case remains in active discovery. What's already clear, though, is that this litigation is going to produce one of the more direct tests of a question that matters enormously for anyone doing character-adjacent creative work with AI tools: how close can an AI-generated image get to a specific, recognizable, commercially significant fictional character before it crosses from inspiration into infringement.

What this means for AI authors

If any part of your creative work touches recognizable characters, whether you're a fan artist, a game designer, an illustrator, or a writer generating cover concepts, this case is the one to watch, because its outcome will shape how much legal room exists for AI-assisted work that evokes specific, owned IP.

In the meantime, the practical lesson is about specificity and intent. There's a real difference between generating original characters that happen to share genre conventions with existing IP, and deliberately prompting for a recognizable, named character owned by someone else. The former is closer to the kind of transformative, independently authored work that copyright protects. The latter is exactly the fact pattern this lawsuit is built around, and it's worth being honest with yourself about which side of that line your own AI-assisted work sits on, especially if you're publishing or selling it commercially.

If you do work in and around existing franchises, whether through licensed fan projects, parody, or commentary, keeping a clear record of your own creative additions, what you changed, what you added, how your work differs from the underlying IP, becomes more valuable the more this area of law gets litigated in public. A record showing genuine transformation and original human authorship is a meaningfully different position to be in than simply generating an image and hoping nobody notices the resemblance.

Copyrightable is not a law firm and doesn't provide legal advice. This is active, unresolved litigation. Confirm current status at official court dockets before relying on this summary for any commercial decision.