Artificial intelligence startup Midjourney is escalating its legal battle with some of Hollywood’s biggest studios by asking a U.S. court to compel Disney, Universal and Warner Bros. to disclose the full extent of their own use of generative AI, arguing that the information could undermine the studios’ copyright infringement claims.
The latest court filing marks a significant shift in one of the most closely watched copyright disputes involving generative artificial intelligence, with Midjourney attempting to demonstrate that the entertainment companies may themselves be relying on AI systems trained on copyrighted material while simultaneously suing AI developers over similar practices.
Disney and Universal filed suit against Midjourney last year, accusing the image-generation platform of widespread copyright infringement by enabling users to create images featuring iconic characters owned by the studios, including Bart Simpson, Darth Vader and numerous other copyrighted properties.
Register for Tekedia Mini-MBA edition 20 (June 8 – Sept 5, 2026).
Register for Tekedia AI in Business Masterclass.
Join Tekedia Capital Syndicate and co-invest in great global startups.
Several months later, Warner Bros. joined the litigation, further raising the stakes in what has become a landmark legal battle over whether AI companies can lawfully train models using copyrighted works without obtaining licenses from rights holders.
The dispute is centered on Midjourney’s argument that using copyrighted material to train artificial intelligence models constitutes fair use, a long-established legal doctrine under U.S. copyright law that permits limited use of protected works under certain circumstances without the copyright owner’s permission.
The latest disagreement focuses on the discovery phase of the lawsuit, during which both sides are required to exchange evidence. A federal judge previously ordered the studios to provide documents relating to their use of generative AI. However, the court limited that disclosure to AI-generated videos and images intended for consumer-facing products.
Midjourney is now asking the court to substantially broaden that order.
In its filing, the company argues that the restriction allows the studios to selectively disclose only information supporting their claims while withholding evidence that could strengthen Midjourney’s defense.
According to the startup, the documents currently being withheld could reveal whether the studios are internally using AI systems trained on copyrighted content in much the same way they accuse Midjourney of doing.
“The documents they are withholding are precisely those that would reveal whether, behind closed doors, they are doing exactly what they are suing Midjourney for doing,” the company argued in its court submission.
Midjourney specifically contends that if the studios are developing image-generation tools for internal creative purposes such as storyboarding, concept art, visual development or film and television production planning, such evidence would demonstrate that using copyrighted content to train AI models has become an accepted industry practice.
The company argues that this information would directly support its fair use defense by showing that even major copyright owners consider such practices necessary for developing modern AI systems.
Beyond internal AI development, Midjourney is also seeking disclosure of the studios’ own interactions with its platform.
The company wants Disney, Universal, and Warner Bros. to produce every prompt they entered into Midjourney, along with every image generated in response, rather than only those examples they claim infringe copyrighted works.
Midjourney argues that limiting production to allegedly infringing outputs creates an incomplete picture of how its technology performs and prevents the company from presenting evidence that many generated images do not violate copyright.
The studios have strongly opposed the broader discovery request.
Their lead attorney, David Singer, previously described Midjourney’s demands as a “fishing expedition,” arguing that the company is seeking information unrelated to the central copyright issues before the court.
Singer has also rejected suggestions that the lawsuit represents opposition to artificial intelligence itself.
According to him, the studios are not attempting to stop AI development or shut down Midjourney’s business. Instead, he argues that they want the company to stop copying copyrighted films, television programmes and famous fictional characters without authorization and distributing AI-generated derivative works based on those protected properties.
The case has become one of the defining legal battles shaping the future of generative AI. A ruling in favor of the studios could strengthen the ability of copyright holders to demand licensing fees from AI developers and potentially reshape how future models are trained.
Conversely, if Midjourney succeeds in establishing that AI training constitutes fair use, the decision could provide significant legal support for the broader AI industry, where leading companies including OpenAI, Anthropic, Meta and Google also face lawsuits over the use of copyrighted material in model development.
The discovery dispute also highlights growing tensions between traditional media companies and AI developers. Several Hollywood studios have publicly criticized generative AI firms for allegedly using copyrighted content without permission, while simultaneously investing heavily in AI tools to improve production workflows, visual effects, animation, script development and creative planning.