Trained to Imagine
Rendering
Creative Intelligence
AI Production / Hollywood / Legal

Midjourney moves to expose studios' own AI use in copyright defence

Source: Variety ·

Defending against an infringement suit from Disney, Universal, and Warner Bros., Midjourney is seeking discovery into how those studios themselves use AI, arguing fair use and pointing to the plaintiffs' own practices.

TTI Takeaway

Whatever the outcome, discovery is the story. Studios are being asked to produce records of their own AI usage, which raises the internal cost of undocumented generative work across the industry. Expect documentation requirements to tighten in vendor contracts well before any judgment lands.

Analysis

Midjourney’s motion is a procedural filing, but it points at something the industry has largely avoided saying out loud: the studios suing AI companies are also users of AI, and in litigation, what a plaintiff does in private can become evidence. The legal theory is that if Disney, Universal and Warner Bros. rely on the same practices they are trying to punish, that supports Midjourney’s fair use and unclean hands defences. Whether or not a court accepts that, the attempt alone changes how studios and their vendors should think about records.

The sequence matters. The studios sued Midjourney last year, accusing it of enabling mass infringement of their characters. In a June 15 ruling, magistrate judge Joel Richlin limited discovery to the studios’ consumer-facing AI applications and found broader information irrelevant to whether Midjourney infringed. Midjourney has now asked Judge John Kronstadt to overturn that, seeking AI business plans, research reports, training datasets, model weights and board presentations about AI. Its attorney, Bobby Ghajar, argued that if the studios are building image models trained on unlicensed third-party data for internal storyboarding or ideation, that would show such training is an industry custom. The studios’ lead attorney, David Singer, has called the effort a fishing expedition, and has written that the plaintiffs do not seek to stop AI technology or shut Midjourney down, only to stop it copying and distributing their characters without authorisation.

Note the examples Midjourney chose: storyboarding and ideation. Those are exactly the early-stage uses where generative tools have spread fastest and been documented least, often inside art departments, previs teams and outside vendors rather than through any formal studio programme. If discovery widens, the question will not only be what tools a studio built, but what was used on its productions and by whom. Vendors working on studio projects should assume their own tool choices may one day need to be explained, and keep a plain record per project: which models, which versions, what inputs, and whether any training or fine-tuning drew on material the vendor did not have rights to.

This sits alongside a run of disclosures that make internal AI use harder to treat as invisible. Netflix said in July that roughly 300 titles used generative AI somewhere in production this year. Amazon MGM built Project Nara around Maya, Nuke and Unreal. And the Eli Roth episode in August showed how quickly an undocumented use becomes a credibility problem once someone spots a name in the credits. Discovery is a more formal version of the same pressure.

The limits should be stated plainly. A magistrate judge has already found the broader request irrelevant, and district judges often defer on discovery rulings; Midjourney may well lose this motion. Even if it wins, discovery is not a judgment on the merits, and a studio’s internal use of a model does not by itself establish that Midjourney’s use was lawful. What to watch is Kronstadt’s ruling and, more broadly, whether studio vendor agreements begin asking for AI usage logs as a standard deliverable. Our expectation is that the second happens regardless of how the first goes.