Midjourney has escalated its legal battle with Hollywood studios by demanding Disney, Universal, and Warner Bros. Discovery disclose how they train their own generative AI models. The AI image generation service isn’t just defending itself against copyright infringement claims-it’s turning the tables, arguing that if these studios use similar AI practices behind closed doors, their lawsuits lose credibility.

According to TechCrunch, Midjourney is pressing the court to compel the studios to produce internal documents related to their AI efforts. This includes business plans, training datasets, model weights, board presentations, and records of how studio employees have used Midjourney itself. The rationale is straightforward: if these media giants have trained AI systems on copyrighted content for storyboarding, scene visualization, or other pre-production tasks without proper licensing, their case against Midjourney becomes much weaker.

The suit against Midjourney was filed in June 2023, when Disney and Universal accused the service of enabling the creation of images featuring iconic characters like Bart Simpson and Darth Vader. Warner Bros. Discovery joined the suit in September with franchises such as Batman and Superman. If found liable, the studios could seek damages up to $150,000 per infringed work under US copyright law-a hefty penalty, even by American standards.

Midjourney’s defense rests on the fair use doctrine, asserting that training AI on publicly available images can be lawful. But the company has added a twist by invoking the ”unclean hands” doctrine-claiming that studios can’t sue over practices they themselves engage in secretly. Midjourney’s lawyer, Bobby Gajjar, said this strikes at the heart of the plaintiffs’ case if they’re doing exactly what they accuse Midjourney of doing.

Midjourney’s clash with Hollywood studios over AI training data

This isn’t the first legal fight over AI training data, but it’s one of the most high-profile given the players involved. In 2023, Getty Images sued Stability AI in both the US and UK for allegedly using millions of licensed photos without permission. These battles all focus on a fundamental question: can generative models be trained on copyrighted material without explicit consent?

What sets this case apart is that Hollywood studios are caught in a dual role-as plaintiffs claiming infringement and as potential AI users themselves. Major media companies have been experimenting with generative AI across pre-production, marketing, and archiving for years. For example, in 2024, Lionsgate announced a partnership with Runway to develop AI models trained on its own catalog, ensuring control over the rights of the training materials.

This puts Midjourney’s challenge in sharp relief. If studios are found to have trained their internal models not only on their own content but also on external copyrighted works without clear licenses, the legal dispute transcends ”tech startup versus rights holders.” Courts may be forced to scrutinize the studios’ AI usage as much as Midjourney’s output. For Hollywood, the stakes are both legal and reputational.

Meanwhile, the commercial market for ”safe” generative AI tools aimed at enterprises is booming. Adobe’s Firefly, for instance, is touted as trained exclusively on licensed content and suitable for commercial use, with contractual guarantees for corporate clients. For studios, such services offer a cleaner alternative to public AI models trained on the chaotic, unvetted content of the internet, which carries a long trail of legal risks.

Midjourney’s legal strategy may extend the litigation timeline, but it raises the stakes for the entire media industry. Should the court order wide disclosure of AI training documentation, the case will effectively audit Hollywood’s own AI practices. If not, the ruling will signal how US law treats AI training on public data-either affirming or limiting fair use in this context.

For studios, the outcome is also a business concern. The global market for generative AI in media and entertainment is already valued in the billions and growing rapidly thanks to automation in visual content production. This lawsuit is far more than a dispute with a startup-it’s a test of the rules for the next generation of content creation.

Looking ahead, this case could redefine how AI is integrated into Hollywood’s creative workflows. Will studios face tighter scrutiny over their own AI training data sources? Or will courts carve out clear exceptions for fair use in AI development? The decision could set a precedent impacting not only entertainment but all industries relying on generative AI technology.