Midjourney Seeks Court Order to Force Studios to Disclose Internal AI Use in Copyright Discovery
Midjourney seeks court orders compelling Disney, Universal and Warner Bros to produce internal AI usage records, prompts and outputs in ongoing copyright litigation.
Midjourney filed a motion asking a federal court to broaden discovery in its copyright battle with major studios, arguing the companies should disclose how they themselves use generative AI. The startup says withheld records could show whether studios train or operate image-generating models using unlicensed copyrighted material, a fact it asserts is relevant to its fair use defense.
Court previously limited studios’ disclosure
The judge overseeing the case earlier required studios to produce documents about generative AI only where that use resulted in consumer-facing videos or images.
Midjourney now asks the court to lift that restriction, calling the limitation unfair because it permits the studios to produce only materials that support their market-harm claims. The startup contends the narrower order prevents it from obtaining evidence that could back its position.
Allegations and parties involved
Disney and Universal sued Midjourney in mid-2025, alleging the company’s image models can reproduce copyrighted characters owned by the studios.
Warner Bros. later filed its own suit, adding to litigation that centers on whether image-generation systems infringe by producing likenesses of characters such as those owned by the studios.
Midjourney frames disclosure as central to fair use defense
In its latest filing, Midjourney argues that internal studio use of unlicensed copyrighted images for AI training or ideation would support industry practice as a mitigating factor.
The company says evidence that studios use image-generation tools internally—such as for storyboarding or creative ideation—would show that similar practices are common and therefore relevant to assessing fair use and market harm.
Specific materials Midjourney demands
Midjourney is pressing the studios to produce not only records of AI training but also the prompts provided to its systems along with the resulting outputs.
The startup says limiting production to prompts that allegedly produced infringing images would be incomplete and could obscure exculpatory evidence that demonstrates broader industry norms.
Studios call request a fishing expedition
Studios’ lead counsel, David Singer, has argued the discovery attempt amounts to a “fishing expedition” and that the studios are not trying to halt AI development.
Singer said the studios seek to stop the copying and public distribution of their films, television works and famous characters without authorization, rather than to shut down generative AI technology or Midjourney’s operations.
Legal stakes and industry implications
How the court resolves this discovery dispute could shape future litigation tactics and transparency norms around AI training.
If the court requires expansive production, studios and other rights holders may face pressure to disclose internal AI workflows, which could influence settlement positions and industry policies on training data.
Midjourney maintains that access to a fuller record is critical to mounting a comprehensive defense, while studios warn that overly broad discovery would reach sensitive internal materials and trade secrets.
The dispute underscores a larger question facing courts and companies: whether internal uses of generative AI by rights holders themselves affect claims that third-party AI systems infringe by training on copyrighted works.
As discovery unfolds, judges will weigh the relevance of internal studio practices against claims of confidentiality and competitive harm, a calculation that could set precedents for how AI training materials are treated in copyright cases.
Final negotiations and rulings on the scope of discovery will be closely watched by technology firms, entertainment companies and legal observers because they may determine whether similar requests for prompts, outputs and internal training records become standard in AI-related litigation.
