The Mirror Move
Here's a legal tactic that feels almost too clever: when you're being sued for doing something, dig into your accuser's files and see if they've been doing the exact same thing behind closed doors. That's essentially what Midjourney is asking a federal judge to let it do in its ongoing copyright battle with Disney, Universal, and Warner Bros.
The AI image-generation startup filed a new motion this week seeking to compel the studios to disclose not just how they use generative AI in consumer-facing content, but every internal prompt and output they've ever generated using Midjourney's own tool. The goal? To prove that training AI models on copyrighted imagery is so widespread across Hollywood that it qualifies as an "industry custom" — a key element in Midjourney's fair use defense.
It's aggressive. It's clever. And the studios' lead attorney, David Singer, has already called it a "fishing expedition." Which, honestly, is exactly the kind of reaction you'd expect when someone's asking you to open your filing cabinets.
How We Got Here
The lawsuit chain is straightforward enough. Disney and Universal sued Midjourney first, last year, alleging that the startup's models could generate images of characters like Bart Simpson and Darth Vader — IP the studios own outright. Warner Bros. joined the fray a few months later with its own complaint.
Midjourney's defense has always rested on fair use. The argument: training AI models on publicly available images, including copyrighted ones, is transformative and protected speech. The studios say that's a convenient fiction — you can't just copy Homer Simpson pixel by pixel and call it innovation.
A judge already ruled that the studios would have to produce discovery documents related to their generative AI usage. But there was a catch: only when that usage resulted in "consumer-facing" videos or images. Internal storyboarding tools, concept art pipelines, ideation workflows — all of that stayed shielded.
Midjourney is now asking the court to remove that limitation entirely.
What Midjourney Is Actually Asking For
The motion is specific, and it's broader than most parties want them to be. Midjourney wants the studios to hand over:
- Every prompt they've ever typed into Midjourney's system
- All resulting image outputs, regardless of whether those images allegedly infringe on studio IP
- Documentation of any internal AI models the studios have developed themselves, particularly for storyboarding or content ideation
- Evidence of whether those internal tools were trained on unlicensed copyrighted material
The legal logic is transparent: if Disney and Warner Bros. are secretly using AI to generate images from copyrighted source material for internal production pipelines, then the practice can't be as harmful to their business interests as they claim in court. It becomes an industry norm, not a unique injury.
"The documents [the studios] are withholding are precisely those that would reveal whether, behind closed doors, they are doing exactly what they are suing Midjourney for doing," the filing states.
That's a serious accusation wrapped in legal language. It implies hypocrisy at best, and it suggests the studios' claimed market harm is performative — filed to establish a legal precedent rather than to address actual business damage.
The Studios Push Back
David Singer, representing the studios, didn't mince words. He characterized Midjourney's discovery request as a "fishing expedition" — a legal term of art for requests that go beyond what's relevant to the specific claims at issue.
Singer's framing is deliberate. He wants the court and the public to see this as a dispute about copyright infringement, not about AI industry practices. "We do not seek to stop AI technology or even shut down Midjourney's business," he said. "We simply want Midjourney to stop copying their movies and TV shows and to stop distributing, publicly displaying, publicly performing, and creating derivative works that include copies of [their] famous characters without authorization."
There's a meaningful distinction here that the legal battle will likely turn on: the studios aren't arguing against AI technology itself. They're arguing against unauthorized use of their specific characters and IP within those systems. That's a narrower claim, but it's also one that Midjourney's fair use defense has to directly confront.
The studios' position, as articulated by Singer, is essentially: we're not mad that you built a tool. We're mad that your tool reproduces our characters.
Why This Discovery Battle Matters
The specific question of what documents the studios must produce might seem like procedural minutiae. But it's actually where this case gets interesting.
If the judge allows Midjourney's broader discovery request, it opens the door to evidence that could fundamentally reshape how courts view AI training data. The studios' internal practices — if they mirror what Midjourney does publicly — would demonstrate that the entire industry operates on a shared assumption: copyrighted imagery is fair game for AI training, whether you license it or not.
That's a powerful fair use argument. Industry custom matters in copyright law, particularly when assessing the "market harm" element of the fair use test. If every major studio is doing it internally, how can Midjourney be causing unique market damage by doing it at scale?
But there's a counterargument, and it's the one Singer is pushing. Even if studios use AI internally for storyboarding or concept art, that's a fundamentally different use case than generating and distributing images of copyrighted characters to the public. Internal production tools don't compete with the studios' actual products — movies and TV shows. A Midjourney output that looks like Bart Simpson does.
The judge will have to weigh whether "industry custom" for internal workflows should carry the same weight as public-facing distribution. That's not a question with an easy answer, which is probably why both sides are fighting so hard over the discovery scope.
What Happens Next
This is discovery-stage maneuvering, which means we're still in the early innings of what could become a landmark case. The judge hasn't ruled on Midjourney's latest motion yet, and the broader fair use question remains unresolved.
What's clear is that Midjourney has chosen an aggressive posture. Rather than narrowly defending against the specific infringement claims, it's trying to flip the script — making the studios' own practices the central evidence in its case. It's a high-risk, high-reward strategy. If it works, the discovery process itself becomes a weapon. If it fails, the court could view Midjourney as overreaching and issue sanctions.
Either way, the studios are now in an awkward position. They want to win on copyright grounds without revealing that they might be doing the same thing internally. And Midjourney wants to prove that everyone in Hollywood is complicit, which means the discovery process will get messy regardless of how the judge rules on this motion.
The next ruling on discovery scope will tell us a lot about where this case is headed.