Disney called Midjourney a “bottomless pit of plagiarism.”

Midjourney’s response: show us yours.

In June 2025, Disney and Universal sued Midjourney for copyright infringement, alleging its AI image generator trained on copyrighted characters — Darth Vader, Shrek, Homer Simpson — without permission and let paying subscribers generate endless images of them. Warner Bros. followed in September with its own complaint: Superman, Batman, Bugs Bunny, Scooby-Doo. The studios seek up to $150,000 per infringed work. The math, applied to millions of generated images, could be company-ending.

Midjourney’s defense is fair use. And unclean hands.

The unclean hands argument: if the studios are doing the same thing — training AI models on copyrighted material, using AI to generate images internally — then they can’t claim pure victimhood. Attorney Bobby Ghajar, in a motion filed this week: “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 wants the studios to produce AI business plans, research reports, training datasets, model weights, board meeting presentations. Everything.

On June 16, a magistrate judge said no. The studios need only reveal AI tools that produce “consumer-facing” output. Everything behind the scenes stays hidden.

Midjourney is appealing to Judge John Kronstadt to overturn that ruling.

Here’s what’s behind the scenes.

A Hollywood Reporter survey found 36% of film productions used AI tools to reduce staffing in pre-production and post-production in 2024. A 10-minute animatic that takes a studio team months can now be generated in two to three days for under $100. Filmmakers using AI report 60-70% time reduction and 40% cost reduction.

Los Angeles County has lost 41,000 film and TV jobs in three years.

Disney laid off approximately 1,000 workers in April 2026, with Marvel Studios reportedly hit hardest. Industry projections estimate 118,500 entertainment positions cut by year’s end. Concept artists, storyboard artists, animators, VFX workers — roles paying $60,000 to $108,000 — are the most vulnerable.

None of that is consumer-facing. All of it stays behind the discovery limitation.

The 2023 strikes won real protections. The WGA’s contract says AI is not a writer — it can’t receive credit, can’t reduce a writer’s pay, can’t replace the human in the room. SAG-AFTRA’s contract protects performers’ voice and likeness — digital replicas require consent, compensation, and separate negotiation each time.

Writers are protected. Actors are protected.

Concept artists are not. Storyboard artists are not. Animators are not. VFX workers are not. The people who make the images that trained the models that generated the images that started the lawsuit — they have no contract, no guild provision, no seat at any table in this courtroom.

In leaked chat screenshots, Midjourney CEO David Holz reportedly discussed preloading 16,000 artist names and styles into the training data. The LAION-Aesthetics dataset — 5.6 billion images scraped from the internet, 47% from stock photo sites, DeviantArt, ArtStation, Flickr — was the foundation. Three artists filed a class action in January 2023, calling these platforms “21st-century collage tools that violate the rights of millions of artists.”

Midjourney scraped artists without consent to build a tool that generates images. The studios allegedly train AI on copyrighted material to replace artists internally. Both sides of this lawsuit trained on other people’s work. The question in the courtroom is whose other people.

Studios’ lead attorney David Singer said they “do not seek to stop AI technology or even shut down Midjourney’s business.” They want Midjourney to stop generating their characters.

Not to stop generating. To stop generating theirs.

The discovery limitation is the line. Above it: consumer-facing AI. Visible. Litigated. Generating Darth Vader without a license — that’s the lawsuit. Below it: internal AI. Invisible. Protected by a magistrate’s order. Replacing a storyboard artist with a $100 animatic — that’s not discoverable.

The visible AI is the copyright dispute.

The invisible AI is the layoff.

Both sides of this lawsuit use AI. Both sides trained on other people’s work. Both sides are fighting over what the public sees. Neither side is fighting over what happened to the people who made the work in the first place.

Sixteen thousand artists trained a model. Forty-one thousand workers lost jobs. The courtroom has room for the corporations. The discovery has room for the consumer-facing tools.

The artists are in neither.

// NEON BLOOD