Some of the largest movie studios have entered the fray of GenAI copyright litigation. Disney, Marvel, Lucasfilm, Twentieth Century Fox, Universal Pictures, and DreamWorks sued the image-GenAI company Midjourney, Inc. in the Central District of California, accusing the AI company of large-scale, willful copyright infringement. Disney et al. v. Midjourney Inc., No. 2:25-05275 (C.D. Cal. June 11, 2025). In a 110-page, star-studded complaint, Plaintiffs did not mince words, referring to Midjourney as “the quintessential copyright free-rider and a bottomless pit of plagiarism.”
Background
Plaintiffs’ complaint alleges that Midjourney trains its generative-image model on “innumerable unauthorized copies” of plaintiffs’ films, television shows and character artwork, then monetized that data by selling a tiered subscription service that distributes high resolution images with Plaintiffs’ copyrighted characters.
In use, a subscriber of Midjourney’s Image Service can enter a bare-bones prompt—e.g., “Darth Vader walking around the Death Star with a red lightsaber”—and receive a high-resolution, downloadable image featuring those iconic characters in seconds. Notably, a user’s prompt need not specifically identify the copyrighted character for Midjourney’s Image Service to output a copyrighted character. For example, according to the complaint (¶ 107), the following output was generated with the generic prompt “Superhero fight scene.”
Image Source: Disney et al. v. Midjourney Inc., No. 2:25-05275 (C.D. Cal. June 11, 2025))
Further, Midjourney automatically places many of these outputs on its public “Explore” page, using copyrighted characters to entice new customers and upgrade existing ones. Plaintiffs’ complaint is replete with side-by-side exhibits showing one-to-one recreations of Yoda, Iron Man, Elsa, Bart Simpson, and other well-known characters. The complaint further alleges that Midjourney was trained to output infringing content.
Legal Claims
The suit pleads direct and secondary infringement under the Copyright Act.
Direct infringement. Plaintiffs contend that Midjourney itself makes the infringing reproductions by (i) copying plaintiffs’ works into its training set and (ii) generating, displaying, and distributing derivative images in response to user prompts, all without license. The infringement is willful because Plaintiffs sent Midjourney letters, which it ignored, detailing the accused infringement, and thus, Midjourney was on notice.
Secondary infringement. If the factfinder deems users the “volitional actors,” the complaint alleges Midjourney is secondarily liable because it knowingly provides the tools and inducement that make the violations possible. Midjourney intentionally designed its model to recreate famous characters and ignored written takedown demands, thereby “materially contributing to and inducing” infringement.
- Right and Ability to Supervise or Control. The complaint stresses that Midjourney could have prevented—or at least curbed—copying. Midjourney chooses which copyrighted works to ingest and can exclude plaintiffs’ content from its training data. Specifically, the platform already deploys automated filters that block prompts that violate its “Community Standards” (e.g., prompts containing nudity, political propaganda, hate speech), returning a standard “AI Moderator” rejection message. Midjourney’s terms likewise state that it may terminate repeat infringers and that it can “block some text inputs automatically,” demonstrating full technical control. Plaintiffs allege that comparable services employ copyright filters, but Midjourney refused to adopt similar measures.
- Direct Financial Benefit from Infringement. Midjourney’s subscriptions are its sole revenue stream; the more images a user generates—especially of popular characters—the higher the tier Midjourney can sell. The complaint cites the Explore page and marketing materials as proof that access to unlicensed Disney and Universal imagery is a “draw” that attracts and retains paying customers. The availability of infringing outputs is alleged to be the key draw that helped Midjourney surpass $200 million in revenue in 2023 and hit $300 million in 2024.
Because Midjourney both profits from and can police the infringing conduct, Plaintiffs argue its liability is “textbook” infringement. They seek statutory damages of up to $150,000 per work or full actual damages and profits, plus injunctive relief halting further use of their content.
Key Takeaways
- Moderation Backfiring. Midjourney’s efforts to ensure improper content does not get generated and to enforce compliance with its Community Standards backfired. Plaintiffs are leveraging those efforts to show Midjourney had the ability to prevent infringing uses and output. If this allegation carries the day, GenAI companies may find themselves in a Catch-22 when it comes to content moderation.
- Fair Use. The complaint does not include any explicit discussion of fair use. The absence of any acknowledgment of fair use likely reflects the Plaintiffs’ view that Midjourney’s conduct falls squarely outside the bounds of this defense. But Plaintiffs nonetheless lay the groundwork for a rebuttal of the fourth Warhol v. Goldsmith fair use factor by focusing on the economic impact that the alleged infringement has on the Plaintiffs’ marketplace. For example, Plaintiffs argue that Midjourney illegally and unfairly competes with companies that license Plaintiffs’ copyrighted works for the purpose of creating authorized derivatives, and undercuts those existing and potential licensing markets.
- Infringement by the Unwary. GenAI users creating public-facing content (e.g., marketing materials) should be aware that GenAI can output copyrighted materials even when users’ prompts do not specifically call for it. Users could then unintentionally subject themselves or their company to liability for unauthorized use of that copyrighted material. Notably, Midjourney’s terms of service state that users “are responsible for all Content that you provide or generate, including ensuring that it does not violate any applicable laws or this Agreement, and that you have all necessary rights and permissions to provide the Content.”
