Midjourney Demands Studios Reveal AI Use in Copyright Fight

In a surprising reversal of the typical power dynamic between tech companies and Hollywood studios, Midjourney is turning the tables in its copyright lawsuit by demanding that Disney, Universal, and Warner Bros. disclose the full extent of their own generative AI usage. The move exposes a glaring contradiction at the heart of the entertainment industry's war on AI: the very studios filing aggressive lawsuits against AI companies are quietly deploying similar technologies in their own operations, raising uncomfortable questions about whose innovation deserves protection and whose deserves litigation.

The legal strategy marks a pivotal moment in the ongoing battle over artificial intelligence, copyright, and fair use. Rather than simply defending itself against accusations of training models on copyrighted content without permission, Midjourney has chosen to shine a spotlight on the studios' hypocrisy, forcing them to account for their own generative AI experiments. This discovery request could fundamentally reshape how the entertainment industry is perceived in the court of public opinion and potentially undermine the legal credibility of copyright claims being levied against AI companies.

The Discovery Demand: What Midjourney is Asking For

The discovery process in the Midjourney copyright case requires both parties to exchange relevant documents and information. Midjourney's legal team has filed requests demanding that Disney, Universal, and Warner Bros. provide comprehensive details about any generative AI tools they've developed, tested, or deployed internally. This includes information about what training data these studios used, how they obtained that data, and whether they sought explicit permission from copyright holders before training their models.

The specificity of these discovery requests is noteworthy. Midjourney isn't simply asking whether the studios use AI—they're demanding detailed documentation of the scope and scale of AI implementation. The requests likely cover everything from visual effects departments experimenting with AI-assisted animation tools to script analysis systems powered by machine learning algorithms. According to reports on the case, the discovery requests specifically target any AI systems that could be considered "generative," meaning systems capable of creating new content rather than simply analyzing existing content.

This legal maneuver serves a strategic purpose beyond gathering defensive information. By forcing the studios to disclose their AI usage, Midjourney's attorneys can potentially argue before the court that the studios are applying a double standard—suing AI companies for the exact same activities they themselves engage in quietly behind closed doors. The argument would suggest that the studios' legal position is motivated by competitive advantage rather than genuine copyright protection, a distinction that could carry significant weight in fair use determinations.

Hollywood's Hidden AI Arsenal

The irony of studios suing AI companies while simultaneously developing their own generative AI tools is hardly secret within the industry. Over the past 18 months, major studios have increasingly experimented with AI for visual effects, scriptwriting assistance, and post-production workflows. The Hollywood Reporter has documented several instances of studios testing AI systems, though executives have generally avoided making these initiatives public, likely anticipating the kind of backlash now unfolding in Midjourney's legal response.

Consider the practical applications: AI can assist in rotoscoping (the tedious frame-by-frame animation process), background generation for visual effects, and even preliminary storyboarding. Some studios have quietly experimented with machine learning models trained on their own archived content—essentially doing what they're accusing Midjourney of doing, but with tighter control over training data. The difference, studios would argue, is one of consent and ownership. When Disney trains an AI system on footage from its own movies, the company owns the copyright to that footage. But this distinction becomes murkier when we consider whether the underlying creative elements—cinematography techniques, composition styles, narrative structures—can be truly owned by a single entity.

The discovery process could expose specific instances of studios using third-party AI tools or licensing AI services that themselves might rely on broad training datasets. For example, if a studio licensed an AI-powered visual effects tool that was trained on publicly available internet images without explicit permission from each photographer or artist, that studio would be participating in the exact behavior they're condemning in AI companies like Midjourney.

This potential exposure explains why major studios are likely to resist these discovery requests vigorously. Their legal teams understand that public disclosure of their own AI usage could devastate the moral authority of their copyright claims. It transforms the narrative from "evil AI companies stealing from creators" to "established companies fighting to protect their monopoly on AI innovation while hoarding tools for themselves."

Broader Implications for AI Defense Strategy

Midjourney's approach is likely to become a template for other AI companies facing copyright litigation. The Verge has noted that similar copyright cases involving Stability AI and other generative AI companies could adopt comparable discovery strategies. The precedent being set here is that accused AI companies need not simply accept copyright maximalist arguments—they can interrogate whether their accusers are genuinely committed to the principles they're asserting.

The strategy also highlights how copyright litigation increasingly requires asymmetry analysis. Courts will need to evaluate not just whether Midjourney trained its models on copyrighted content, but whether the studios pursuing the lawsuit would face similar infringement claims if their own practices were scrutinized under the same legal framework. This comparative analysis could influence fair use determinations, particularly the fourth factor—whether the accused use harms the market for the original work.

If studios are simultaneously developing competing AI systems, one could argue that their lawsuits are motivated by market competition rather than copyright enforcement. This distinction matters enormously in fair use law. A court may be more sympathetic to fair use arguments when the plaintiff is engaged in parallel conduct, as opposed to a scenario where the plaintiff is a pure content creator with no involvement in AI development.

The case also raises questions about what the U.S. Copyright Office considers "fair use" in the context of generative AI. If major studios are training AI systems on their own archives for internal use, what distinguishes that from Midjourney training on publicly available images for commercial purposes? The answer likely hinges on questions of permission, attribution, and market impact—but discovery could reveal whether studios have satisfactory answers to these same questions about their own practices.

Ultimately, Midjourney's discovery demands represent a shift from defensive to offensive legal positioning. Rather than arguing that AI training should be protected as fair use, the company is arguing that its accusers lack standing to make copyright claims because they've engaged in parallel conduct. This strategy may prove more effective than appealing to abstract principles about innovation and fair use, particularly if discovery reveals documented instances of studios using AI tools in ways that would violate their own copyright standards.