Midjourney vs Disney: Inside the High-Stakes AI Copyright Lawsuit

The Core Accusation

The three major studios sued Midjourney last year, claiming the AI image generator enabled massive copyright infringement of their iconic characters. Think Mickey Mouse, Hogwarts, and decades of carefully protected IP being reproduced without authorization or payment.

The studios’ position is straightforward: Midjourney copied their movies and TV shows, distributed those copies, and created derivative works from their characters — all without permission. As their lead attorney David Singer put it, this isn’t about stopping AI. It’s about stopping unauthorized copying.

That’s a clean argument. But Midjourney’s defense is where things get interesting.

Midjourney’s Countermove: “You’re Doing It Too”

Midjourney has claimed fair use as its primary defense. But it’s also pursuing something more aggressive — an “unclean hands” argument that essentially says: if the studios are using AI trained on unlicensed data internally, they can’t credibly sue us for doing the same thing.

To make that argument stick, Midjourney needs evidence. So its lawyers filed a motion this week asking Judge John Kronstadt to overturn a magistrate judge’s earlier ruling that limited discovery to only “consumer-facing” AI applications from the studios.

Midjourney wants access to:

• Internal AI business plans and research reports
• Training datasets and model weights
• Board meeting presentations about AI strategy
• Any internal tools used for storyboarding or content ideation

The logic from Midjourney’s attorney Bobby Ghajar is direct: if the studios are training image-generating models on unlicensed third-party content for internal use, that demonstrates an industry-wide custom — even among the plaintiffs themselves.

Why the Discovery Battle Is the Real Fight Right Now

The studios agreed to hand over information about consumer-facing AI products only. They’re drawing a hard line around internal tools, calling Midjourney’s broader request a “fishing expedition” designed to distract from its own misconduct.

The magistrate judge sided with the studios in June, ruling that the studios’ internal AI use is irrelevant to whether Midjourney infringed their copyrights.

Midjourney is now pushing that ruling up to Judge John Kronstadt. The outcome of this discovery dispute could determine how much of the case Midjourney can actually defend.

If the studios are forced to reveal internal AI practices, it could expose a significant double standard. If they’re not, Midjourney loses a key pillar of its defense and faces the copyright claims largely on its own merits.

What This Means for the Broader AI Tools Ecosystem

This case isn’t just about Midjourney. It’s a stress test for the entire generative AI industry.

The central legal question — whether training AI on copyrighted content without a license constitutes infringement — remains unresolved in U.S. courts. How this case develops will influence how AI image tools, video generators, and multimodal models are built going forward.

A few things are now clearly at stake:

Training data transparency. If courts start requiring AI companies to disclose what data their models were trained on, the entire industry faces a reckoning. Many tools have been deliberately vague about this.

The fair use defense. Midjourney’s fair use argument is one of the most closely watched in AI law right now. A loss here would weaken the legal foundation that dozens of other AI companies are quietly relying on.

The “industry custom” argument. If Midjourney can prove that studios themselves train on unlicensed data, it introduces a powerful precedent — not just for this case, but for how courts evaluate AI training practices across the board.

The Hypocrisy Question

The most provocative part of Midjourney’s argument is the implied accusation: Hollywood is suing AI companies for doing exactly what Hollywood is doing internally.

That’s unproven. But it’s not implausible. Every major studio has been investing heavily in AI for production, post-production, and development workflows. The question of what data those internal tools were trained on is genuinely murky.

If Midjourney gets the discovery it’s asking for and finds evidence of unlicensed training data in studio AI systems, this case transforms from a straightforward infringement suit into something far messier for the entertainment industry.

The Takeaway for AI Tool Builders and Buyers

If you’re building with generative AI or evaluating tools for your business, this case is a signal worth watching closely.

For builders: the legal risk around training data is real and growing. “We used publicly available data” is no longer a sufficient answer in a courtroom. Documenting your data sourcing practices now is not optional.

For buyers: tools that are transparent about their training data and licensing practices are lower-risk choices. As litigation increases, the provenance of a model’s training data will become a legitimate due diligence question — not just an ethical one.

The Midjourney vs. Disney case won’t be resolved quickly. But each motion, each ruling, and each discovery decision is quietly drawing the legal map that the entire AI tools industry will have to navigate.