A high-stakes legal battle between Hollywood and the artificial-intelligence industry has taken a combative turn, with the AI image generator Midjourney demanding that the studios suing it — Disney and Universal — hand over evidence of how they use AI themselves, Variety reported. The move sharpens one of the most closely watched copyright disputes of the AI era.
How the case began
Disney and Universal sued Midjourney in June 2025, in what was described at the time as the first major lawsuit by Hollywood studios against an AI company, NPR reported. The studios allege that Midjourney trained its image-generating models on their copyrighted works without permission, and that the service can reproduce famous characters — the likes of Darth Vader, Shrek and the Minions — on demand, letting users conjure protected creations by simply typing a name. Warner Bros. Discovery later filed a similar suit of its own.
Midjourney denies the core allegations and rests its defense heavily on the doctrine of "fair use," the legal principle that allows limited use of copyrighted material without permission in certain circumstances. The company argues that training an AI model on existing images is "transformative" — more akin to how a human artist learns by studying the work of others than to outright copying.
The discovery gambit
The latest flashpoint is over discovery, the pre-trial process in which each side can demand documents and evidence from the other. Midjourney has pushed to obtain information about the studios' own use of generative AI, arguing that it is directly relevant to the case.
The logic is pointed: if the entertainment companies suing Midjourney are themselves using AI tools trained in similar ways, Midjourney contends, that bears on both its fair-use defense and its argument that the studios cannot condemn a practice they also rely on. The studios have resisted, maintaining that their internal AI practices are not relevant to whether Midjourney infringed their copyrights, and courts have so far been cautious about how far such demands can reach.
Why it matters beyond Hollywood
The dispute is about far more than one company or one set of characters. Courts across the United States are wrestling with a foundational question the AI boom has forced into the open: when an AI system is trained on vast quantities of copyrighted material gathered from the internet, is that fair use — a legitimate, transformative step — or is it infringement on an industrial scale?
The answer will shape the economics of an entire industry. If training on copyrighted works without a license is broadly permitted, AI companies gain enormous latitude; if it is not, they may face a future of costly licensing deals — or liability for what their models have already absorbed. The Midjourney case, pitting two of the world's most powerful copyright holders against a prominent AI firm, is one of the venues where that question is being tested.
An unsettled fight
For now, the outcome is genuinely uncertain, and reasonable observers disagree about where the law should land. The studios frame the suit as a defense of creators' rights against wholesale copying; Midjourney frames its defense as protecting a legitimate, transformative technology from owners trying to extend copyright into new territory. Both positions have found sympathy among legal scholars.
What the discovery fight underscores is how tangled the lines have become. The same studios challenging AI in court are, like much of the entertainment business, exploring how to use it in their own productions — a duality Midjourney is now trying to put on the record. However the judge rules on that request, the broader case is likely to be a landmark, helping to set the terms on which AI and the creative industries coexist.
