Midjourney Lawsuit: Plaintiffs, Defenses, and Fair Use

The Midjourney lawsuit is a consolidated copyright case in which Disney, NBCUniversal, and Warner Bros. Discovery accuse the AI image generator Midjourney of training its models on their copyrighted characters and producing near-identical reproductions on demand. Filed in the U.S. District Court for the Central District of California in June and September 2025, the case is now in pretrial discovery before Judge John Arnold Kronstadt, with mediation required by August 2026 and no trial date set.

Who Sued Midjourney and When

On June 11, 2025, seven Disney and NBCUniversal entities filed a 110-page complaint against Midjourney, Inc. The plaintiffs were Disney Enterprises, Marvel Characters, MVL Film Finance, Lucasfilm, Twentieth Century Fox Film Corporation, Universal City Studios Productions, and DreamWorks Animation.¹ The case was docketed as No. 2:25-cv-05275.²

On September 4, 2025, a second group filed a parallel complaint: Warner Bros. Entertainment, DC Comics, Turner Entertainment, Hanna-Barbera Productions, and The Cartoon Network. The Warner Bros. filing added an allegation that Midjourney had initially blocked video generation for its characters, then removed those content-moderation measures in August 2025, which the plaintiffs cited as evidence of knowing infringement. It also alleged Midjourney used copyrighted characters to market its service, including through a feature called “Midjourney TV.”³

Judge Kronstadt consolidated the two cases for all purposes, including trial, on November 4, 2025. The Disney complaint is the lead case.²

What the Studios Say Midjourney Did

Both complaints assert direct and secondary copyright infringement. The Disney filing described Midjourney’s conduct as “systematic, ongoing, and willful,” calling the company a “copyright free-rider” engaged in “mass piracy.”¹

The theory has two parts. First, that Midjourney scraped the studios’ copyrighted works to train its AI models without authorization. Second, that the resulting service generates recognizable reproductions of those works when subscribers type simple text prompts. Exhibits to the Disney complaint include Midjourney-generated images of characters like Darth Vader and the Minions produced from straightforward descriptions.¹

The catalog of characters cited across the two complaints is enormous. Disney and NBCUniversal identified figures from Star Wars (Darth Vader, Yoda, Stormtroopers, R2-D2, Chewbacca), Marvel (Iron Man, Spider-Man, Deadpool, Wolverine, Captain America, Groot, the Hulk), The Simpsons, Toy Story, Cars, Monsters Inc., Frozen, The Little Mermaid, Aladdin, Mulan, Lilo & Stitch, Snow White, Despicable Me, Shrek, How to Train Your Dragon, Kung Fu Panda, and Boss Baby. The exhibits cataloged more than 150 allegedly infringed works.⁴ Warner Bros. added Superman, Batman, The Joker, Wonder Woman, the Flash, Teen Titans, Bugs Bunny, Daffy Duck, Sylvester, Tweety, Tom and Jerry, Scooby-Doo, the Powerpuff Girls, and Rick and Morty.³

The Disney complaint also flagged Midjourney’s then-forthcoming video generator, alleging the company was training a video model that was “very likely already infringing” the studios’ copyrights. One week after the filing, Midjourney launched its V1 image-to-video model, capable of producing clips up to 21 seconds long.⁵

What the Studios Want

The complaints seek actual damages plus Midjourney’s profits, or alternatively statutory damages of up to $150,000 per infringed work. With more than 150 works listed in the Disney exhibits alone, statutory damages could exceed $20 million on that filing.⁴

The Disney complaint alleged Midjourney earned $300 million in the prior year with 21 million subscribers paying between $10 and $120 per month across four tiers.⁶ Outside estimates put Midjourney’s 2025 revenue at $500 million and its valuation at $10 billion. The company is self-funded, has never taken venture capital, and was founded by David Holz in 2021.⁷

How Midjourney Is Defending Itself

Midjourney answered the Disney complaint on August 6, 2025, and demanded a jury trial.² Its primary defense is fair use. The company argued that AI training is a transformative use of copyrighted material and that “the limited monopoly granted by copyright must give way to fair use, which safeguards countervailing public interests in the free flow of ideas and information.” It contended that copyright law does not grant studios “absolute control” over their works and that many outputs have “legitimate, noninfringing grounds,” including non-commercial fan art, experimentation, and social commentary.⁸

Midjourney also pointed at the plaintiffs. According to its filing, visual effects vendors working with Disney and Universal use the Midjourney platform, and “many dozens” of subscribers have email addresses tied to the studios. The company argued its terms of service, which require users to agree not to infringe intellectual property, shield it from secondary liability. In the Warner Bros. matter, Midjourney added an “unclean hands” affirmative defense and argued it does not control what its users generate.⁹

What Is Known About Midjourney’s Training Data

Midjourney has not disclosed the full contents of its training data, and the studios have signaled that the sourcing question will be a discovery battleground. CEO David Holz told Forbes in 2022 that “there isn’t really a way to get a hundred million images and know where they’re coming from.”¹⁰ Midjourney and similar image generators have been reported to draw on the LAION dataset, a collection of 5.6 billion images scraped from the open web that includes material from DeviantArt, ArtStation, Getty Images, and Pinterest.¹¹

Filings in a separate artists’ class action surfaced a Google Docs spreadsheet titled “Midjourney Style List” that Holz shared on the company’s Discord in February 2022. It cataloged more than 4,700 artist names with style labels, plus a tab listing 15,800 more names, and Holz encouraged users to type those names into prompts to emulate specific styles.¹² Internal Discord messages from a Midjourney staff member, also produced in that litigation, discussed scraping datasets and then “conveniently forget[ting] what you used to train the model” to avoid legal problems.¹³

Researchers Gary Marcus and Reid Southen ran experiments on Midjourney’s V6 model and published their results through the IEEE. Using indirect prompts that did not name specific characters or franchises, they got the system to generate recognizable images of characters from Star Wars, Toy Story, The Simpsons, Minions, Sonic the Hedgehog, and Super Mario Bros. The prompt “popular movie screencap” alone returned an image of Iron Man in a familiar pose. Marcus and Southen argued this was “powerful evidence” that Midjourney had trained on copyrighted material. After Southen published his findings, Midjourney banned his account.¹⁰

Where the Case Stands Now

The consolidated case remains in pretrial proceedings. There have been no settlements, no dismissals, and no dispositive rulings. Judge Kronstadt issued a scheduling order on November 14, 2025, and referred the parties to private mediation, which must take place no later than August 19, 2026. The parties’ joint discovery plan, filed in October 2025, estimated a trial length of 14 days.²

Key deadlines run through the rest of 2026. Initial expert disclosures are due in October 2026, rebuttal disclosures later that month, expert discovery through November, and deadlines for dispositive motions extend into late 2026.¹⁴ No trial date has been set.

How the Fair Use Question Could Decide It

Whether AI training on copyrighted works qualifies as fair use is the unresolved legal question at the heart of the case, and recent rulings in other AI copyright suits point in competing directions.

In Bartz v. Anthropic, Judge William Alsup ruled in June 2025 that using lawfully purchased books to train AI models was transformative fair use, but that using pirated copies of the same books was not.¹⁵ The Bartz case later produced a $1.5 billion settlement, preliminarily approved in September 2025, that set a benchmark of roughly $3,000 per pirated work and required Anthropic to destroy datasets sourced from pirate book libraries.¹⁶ In Thomson Reuters v. Ross Intelligence, a court rejected a fair use defense where the AI tool competed directly in the same market as the copyrighted work it trained on.¹⁷

The studios’ case touches both pressure points. If Midjourney’s training data was scraped without authorization, the Bartz framework cuts against fair use. If Midjourney’s outputs compete with or substitute for the studios’ own licensed merchandise and imagery, the market-harm factor, traditionally the most important element of a fair use analysis, could weigh heavily in the plaintiffs’ favor. Midjourney’s counter-argument, that its technology is fundamentally transformative and serves expressive purposes distinct from the originals, will be tested against those precedents as the case moves through expert discovery.

An analysis from Georgetown Law’s Institute for Technology Law and Policy noted one feature that sets this case apart from text-based AI disputes: visual evidence. Showing a judge or jury generated images that are nearly indistinguishable from official artwork makes the infringement more visceral than a text comparison, and Midjourney reportedly produces recognizable characters from simple direct prompts without elaborate workarounds. Disney’s general counsel said publicly that the Midjourney complaint was “our first case, but it likely won’t be the last.”¹⁸

1
Variety. Disney-NBCU v. Midjourney Complaint
2
CourtListener. Disney Enterprises Inc v Midjourney Inc
3
Copyright Alliance. Warner Bros Entertainment v Midjourney Complaint
4
CNN. Disney Universal Midjourney AI Copyright Lawsuit
5
TechCrunch. Midjourney Launches Its First AI Video Generation Model V1
6
RPJ Law. Hollywood Strikes Back: Disney and Universal Sue AI Platform Midjourney for Copyright Infringement
7
Contrary Research. Midjourney
8
Variety. Midjourney Disney AI Training Lawsuit
9
McKool Smith. AI Litigation Update
10
IEEE Spectrum. Midjourney Copyright
11
University of Toronto Libraries. AI Image Generators and Copyright
12
File 770. Court Exhibit Names Artists Midjourney Scraped to Train Its AI
13
The Register. Spreadsheet of Ripped Off Artists Lands in Midjourney Case
14
ForensisGroup. Disney and Universal v Midjourney: U.S. Generative AI Copyright Litigation
15
Norton Rose Fulbright. AI in Litigation Series: An Update on AI Copyright Cases in 2026
16
Susman Godfrey. Susman Godfrey Secures 1.5 Billion Settlement in Landmark AI Piracy Case
17
George Mason Law Review. Generative AI: When Fair Use Becomes Unfair Competition
18
Georgetown Law. Disney, NBC Universal, and DreamWorks File Major IP Lawsuit Against AI Image Generator Midjourney