Midjourney seeks broader discovery of Hollywood studios’ AI usage in copyright case

Midjourney has asked a court to broaden the scope of document production in its copyright infringement case against Disney, Universal, and Warner Bros., seeking details of the studios’ AI usage beyond consumer-facing outputs. The startup argues that limiting discovery to publicly visible AI-generated content deprives it of evidence that could demonstrate industry norms around unlicensed training data. In a July 2026 filing, Midjourney contends that documents withheld by the studios may reveal internal use of AI models trained on copyrighted content, which it says would support its fair-use defense.

The dispute centers on whether the studios must produce documentation of AI usage for internal processes such as storyboarding or ideation, not just final consumer-facing materials. Midjourney claims this broader disclosure is necessary to counter allegations that its image-generation models infringe on copyrighted characters like Bart Simpson and Darth Vader. The startup also seeks access to all prompts used with its tools and the resulting outputs, not just those tied to allegedly infringing images.

Studios’ lead attorney David Singer has characterized Midjourney’s request as a 'fishing expedition,' arguing that the company is attempting to uncover unrelated business practices rather than address the core issue of unauthorized copying. Singer stated the studios’ objective is to stop Midjourney from using their copyrighted characters without authorization, not to halt AI technology broadly.

A prior court ruling required the studios to disclose AI usage only when tied to consumer-facing outputs, but Midjourney is challenging that limitation. The startup asserts the current scope allows the studios to withhold documents that could equally demonstrate industry-wide practices of training AI on unlicensed content, thereby undermining its fair-use argument.