Midjourney asks court to force studios to disclose internal AI use in copyright battle
Midjourney is asking a judge to compel Disney, Universal and Warner Bros. to produce internal documents about their own AI use, arguing such records are essential to its fair-use defense in ongoing copyright suits.
Midjourney seeks broader discovery in studio lawsuits
Midjourney filed a motion asking the court to expand the scope of documentary discovery in the copyright lawsuits brought by Disney, Universal and Warner Bros.
The startup argues the studios should not be allowed to limit production to materials tied only to “consumer-facing” images and videos.
The company says that restriction would let the studios withhold documents that could show whether they themselves use unlicensed copyrighted works to train internal AI systems.
Midjourney contends that such internal practices are directly relevant to its claim that training models on copyrighted characters can fall within fair use.
Studios sued over AI-generated images of copyrighted characters
Last year the major studios sued Midjourney, alleging the company’s image-generation tools produced images of copyrighted characters without authorization.
The complaints named examples of characters the studios say appear in outputs from Midjourney’s models, prompting litigation over whether model training and output creation infringe copyright.
Those cases have turned on both the legal contours of infringement and on what discovery the studios must produce to prove market harm and copying.
A judge earlier ordered production of studio materials only when they resulted in publicly available or consumer-facing content, a limitation Midjourney now seeks to overturn.
Midjourney frames internal studio use as probative of industry custom
In its filing, Midjourney argues that evidence of studios using generative AI internally — for storyboarding, ideation or other production tasks — would show industry practice and weaken claims of exclusive harm.
The startup says internal use of unlicensed copyrighted materials would be material to whether training models on such content is an accepted industry custom.
Midjourney also asserts that documents showing how studios train their own models could undercut claims that Midjourney’s techniques are uniquely harmful or unusual.
The company maintains this discovery is necessary to avoid a one-sided evidentiary record that favors the studios.
Midjourney demands prompts, outputs and related records
Beyond production of internal AI policies and training procedures, Midjourney seeks the studios’ prompts and generated outputs submitted to its platform.
The startup requests not only prompts tied to allegedly infringing images but all prompts and outputs that passed through Midjourney, arguing a full dataset is needed to test the studios’ infringement theories.
Midjourney says a broader production would allow forensic comparison of studio outputs and the models’ behavior, and would help rebut conclusions drawn from a narrow subset of materials.
The company argues that restricting production to select consumer-facing items gives the studios the ability to “cherry-pick” evidence that supports market-harm claims while hiding material that could support Midjourney’s defenses.
Studios call request a “fishing expedition” and stress limits
The studios’ lead counsel has described Midjourney’s bid for broad discovery as an improper search for evidence rather than a focused request tied to specific claims.
They have emphasized that their objective is not to halt AI development broadly but to prevent unauthorized copying and public distribution of characters and other copyrighted elements.
The studios argue that the court’s prior limitation to consumer-facing outputs protects genuinely private or internal materials from unnecessary disclosure.
They maintain that public-facing uses are the most relevant to claims of market substitution, distribution and public performance.
Legal and industry implications if discovery is expanded
If a judge requires studios to disclose internal AI use broadly, the decision could set a precedent that shapes discovery in future AI copyright cases.
Expanded disclosure might reveal common practices across entertainment companies and influence how courts assess fair use defenses in model-training disputes.
Such a ruling could also pressure studios and other rights holders to formalize internal AI policies and to document training practices more transparently.
Conversely, limiting discovery to public outputs would make it harder for defendants to demonstrate industry-wide practices and could narrow defenses centered on accepted norms.
Outcome will influence fair use and AI transparency debates
The dispute over discovery is likely to play out alongside the substantive copyright claims, with each side arguing the other’s practices matter to liability and damages.
A ruling either way will affect not only Midjourney and the studios involved but also the wider technology and entertainment sectors as they navigate AI development, rights management and litigation risk.
Final courtside decisions on these discovery issues will help define how much internal AI development information companies must disclose in future copyright suits and could influence settlement dynamics across the industry.
