Warner Bros. v. Midjourney

The third studio group in five months

Disney and Universal sued Midjourney first. Warner Bros. Discovery, through Warner Bros. Entertainment and DC Comics, followed in early September 2025 — the third major studio group to bring the same core claim against the same image-and-video generator inside less than half a year.

The consistency across the three complaints is itself a signal. Studios that compete fiercely for box office and streaming subscribers are aligned on one point: Midjourney’s outputs reproduce their characters closely enough, and often enough, that they consider it worth the joint legal exposure of coordinated litigation.

What the complaint alleges

The Warner Bros. Discovery complaint alleges Midjourney willfully generates images and videos of DC characters — Superman, Batman, Wonder Woman — and Looney Tunes and Hanna-Barbera characters including Bugs Bunny, Daffy Duck, Tom and Jerry, and Scooby-Doo. It alleges that even a generic prompt like "classic comic book superhero battle" returns outputs recognizable as specific DC characters, not generic superheroes.

That specificity matters legally. A generator producing a generic caped figure raises different questions than one reliably producing Superman specifically from an unbranded prompt. The complaint is built around examples of the latter.

“"Classic comic book superhero battle."”— Prompt example cited in the complaint, Warner Bros. Entertainment v. Midjourney, C.D. Cal.

The guardrail allegation

The complaint’s sharpest addition to the studios’ shared playbook is the allegation that Midjourney removed guardrails that had previously blocked infringing video generation, timed to its video product launch. If accurate, that is a decision to loosen restrictions the company had already built and presumably knew were necessary — evidence aimed squarely at willfulness, which multiplies statutory damages under the Copyright Act.

Midjourney answered with the same fair-use defenses raised in the Disney case and reportedly added an unclean-hands defense of its own. The consolidation that followed on November 4, 2025 means all three studio groups’ claims — and Midjourney’s defenses to them — now proceed together, with the Disney case as lead.

Video outputs raise the stakes for training-data provenance

This case sharpens a point the earlier studio suits only gestured at: video generation is now squarely in scope, not just still images. Any operator of a video model, and any supplier of video training data, should assume character-level output audits — the kind that produced the "classic comic book superhero battle" example — will become standard litigation exhibits going forward.

The willfulness angle also raises the stakes for how model operators document their own safety decisions. A record showing guardrails were loosened rather than tightened is difficult to explain away after the fact. For data buyers and licensors, that argues for keeping a clean, timestamped record of what filtering and provenance controls were in place at every stage of a model’s development — not just what the training corpus contained.