Warner Bros. Discovery (the Hollywood studio behind Superman, Batman, and Bugs Bunny) has taken legal action against Midjourney, a popular AI image generator. In early September 2025, Warner Bros. Discovery filed a copyright infringement lawsuit claiming Midjourney “brazenly” uses the studio’s characters without permission. This lawsuit marks a new flashpoint in the growing clash between creative industries and generative AI technology. Why is a major studio suing an AI company? And what does it mean for artists, tech innovators, and fans? Let’s break down the story.
![]() |
| Warner Bros. Logo | Photo by Silas Lundquist on Unsplash |
What Is Midjourney AI and How Does It Work?
Midjourney is an AI-powered image generator – essentially a computer program that creates pictures based on text instructions. Users type in a prompt (for example, “a comic book superhero battle in space”), and Midjourney’s algorithms produce a custom image matching that description.Midjourney was launched in 2022 by founder David Holz. The service quickly gained popularity; by late 2024, it had about 21 million users and was generating an estimated $300 million in revenue. Many artists, designers, and hobbyists use Midjourney via a simple chat interface to create artwork, concept designs, or just for fun. The tool can even generate short video clips from prompts.
How does Midjourney create these images?
It uses a type of generative AI model (often called a diffusion model). In simple terms, the AI was trained on a huge collection of images from the internet, learning patterns that connect words to visual concepts. For instance, it has seen many images labeled “Batman” or “Scooby-Doo,” so it learned what those characters typically look like. Later, when you ask Midjourney for “Batman fighting the Joker in Gotham City”, it draws on what it learned to produce a new image in that vein. The Midjourney team describes their tool as “an independent research lab” expanding the imagination, and Holz has analogized it to a search engine for images that pulls in knowledge from across the web. In other words, Midjourney learns like an artist might – studying existing art to inspire new creations.
![]() |
| Generated AI image of “Batman fighting the Joker in Gotham City” | Image by Katakita on Pinterest |
However, this learning process is exactly where legal trouble arises. Midjourney’s training likely included countless copyrighted images (from comics, cartoons, movies, etc.) without explicit permission from the rights holders. This has led to concerns that the AI is regurgitating protected artwork.
Why Warner Bros. Discovery Is Suing Midjourney
Warner Bros. Discovery (the parent company of Warner Bros. Studios, DC Comics, Cartoon Network, and more) says Midjourney crossed the line from inspiration into infringement. On September 4, 2025, the company filed a lawsuit in a Los Angeles federal court accusing Midjourney of stealing its intellectual property. The complaint alleges that Midjourney “brazenly dispenses” Warner Bros.’ copyrighted characters as if they were its own. Simply put, Warner Bros. alleges that the AI is generating unauthorized imitations of iconic characters such as Batman, Superman, Wonder Woman, and Scooby-Doo, among others.According to the lawsuit, Midjourney isn’t just occasionally spitting out look-alikes; it has generated “countless” infringing images and videos of Warner Bros. characters. The studio even provided side-by-side comparisons of its original cartoon characters and the AI’s reproductions to highlight identical details, for example, the exact color of Scooby-Doo’s collar and fur appears in a Midjourney image. In some cases, users didn’t even have to mention a character by name to get a similar result. The suit notes that a generic prompt like “classic comic book superhero battle” yielded high-quality images unmistakably depicting Superman, Batman, and The Flash. This suggests Midjourney’s model has these famous figures deeply embedded in its knowledge base.
Warner Bros. Discovery frames Midjourney’s actions as deliberate and knowing. The complaint points out that Midjourney initially restricted users from generating certain video animations with famous characters – an early attempt to avoid obvious infringement – but then lifted those safeguards in August 2025, advertising the change as an “improvement”. In the studio’s view, this move proves that Midjourney knew it was enabling copyright violations, yet chose to remove protections in a “calculated and profit-driven decision”. The company even alleges that Midjourney updated its terms of service to forbid “red teaming” (a safety-testing practice), perhaps to prevent users or outsiders from probing its model for these IP violations. In short, Warner Bros. says Midjourney knows it’s doing something wrong but “thinks it is above the law”.
What Warner Bros. want from this lawsuit
Primarily, two things: money and control. The studio is seeking unspecified damages, which could theoretically be as high as $150,000 per infringed work if willful infringement is proven. Given the “breathtaking scope” of images involved, that sum could be enormous. Perhaps more importantly, Warner Bros. wants a court injunction – essentially an order to stop Midjourney from using its characters altogether going forward. The suit asks the court to block Midjourney from copying, displaying, or distributing any of Warner’s intellectual property via its AI tools. They even demand that Midjourney implement proper copyright protection measures in its software if it wants to continue operating.
This case isn’t an isolated attack, either. Warner Bros. joined forces with other studios that share similar grievances. Back in June 2025, Disney and Universal (another two Hollywood giants) filed a joint lawsuit against Midjourney on nearly identical grounds. Those companies that own characters like Darth Vader, The Simpsons, Shrek, and more, described Midjourney as “a bottomless pit of plagiarism” and “textbook copyright infringement”. Warner Bros. is now represented by the same law firm handling the Disney/Universal case, signaling a coordinated effort. In fact, Warner Bros. stepping in means the three studios behind DC Comics, Marvel/Star Wars (Disney), and major animation franchises (Universal) are all teaming up to challenge this AI tool. Together, they control a huge chunk of Hollywood’s most beloved characters – making this a high-profile showdown.
This case isn’t an isolated attack, either. Warner Bros. joined forces with other studios that share similar grievances. Back in June 2025, Disney and Universal (another two Hollywood giants) filed a joint lawsuit against Midjourney on nearly identical grounds. Those companies that own characters like Darth Vader, The Simpsons, Shrek, and more, described Midjourney as “a bottomless pit of plagiarism” and “textbook copyright infringement”. Warner Bros. is now represented by the same law firm handling the Disney/Universal case, signaling a coordinated effort. In fact, Warner Bros. stepping in means the three studios behind DC Comics, Marvel/Star Wars (Disney), and major animation franchises (Universal) are all teaming up to challenge this AI tool. Together, they control a huge chunk of Hollywood’s most beloved characters – making this a high-profile showdown.
Legal Analysis: Is AI-Generated Art Fair Use or Theft?
At the heart of this clash is a burning legal question: Can an AI legally “learn” from copyrighted works and produce new content, or is that simply digital piracy? Warner Bros. Discovery and its peers argue that what Midjourney does is clear-cut infringement. When Midjourney produces a picture of Superman or Scooby-Doo on demand, it is essentially creating an unauthorized derivative work – something copyright law forbids without permission. The studios say Midjourney built its entire business by exploiting decades of their creative labor, offering subscribers an endless supply of knock-off images of protected characters to which it has no rights. In Warner’s words, Midjourney is “purposefully exploiting [our] valuable intellectual property to attract subscribers”. The company even warns that Midjourney’s practices cause consumer confusion, misleading users into thinking these AI-generated images are somehow officially authorized or legal.Midjourney, on the other hand, is expected to mount a fair use defense. U.S. copyright law allows limited use of copyrighted material without permission under certain conditions, for example, for commentary, teaching, or transformative creations. Midjourney’s team contends that training an AI on billions of publicly available images to learn artistic concepts is a “transformative” use of those works, not an outright copy. In an August 2025 court filing (responding to the Disney/Universal suit), Midjourney argued that its AI “extracts statistical information” from the original images to learn patterns, a process it likened to how a human artist learns by studying others’ art. “Training a generative AI model to understand concepts by extracting... information embedded in copyrighted works is a quintessentially transformative fair use,” Midjourney’s filing stated, citing recent court rulings that supported this view. In simpler terms, Midjourney is saying: we’re not stealing art, we’re learning from it, and the outputs are new and different. David Holz (Midjourney’s CEO) once drew an analogy: if a person can look at someone else’s picture and then create a similar picture from what they learned, that’s generally legal – so if an AI “learns” like a person, it should be okay too. From the AI company’s perspective, its tool is more like an inspiration engine than a piracy machine.
So is generative AI art more like a student or a scammer?
Legal experts note that this is uncharted territory. Copyright law has never definitively answered whether ingesting millions of images to train an AI counts as fair use. Some early court decisions have leaned in favor of AI developers – for instance, in separate cases involving AI trained on books, courts recently suggested that using copyrighted text to teach an AI might fall under fair use. These rulings imply that an AI’s learning process could be treated similarly to a human researcher reading books in a library (which is legal). On the other hand, the outputs of AI raise another question: If the AI’s generated image is too close to the original artwork, it might be considered an infringing derivative. Warner Bros. argues that Midjourney’s outputs are virtually indistinguishable from the originals in many details, effectively replacing the need for authorized art. They also point out that Midjourney operates commercially; it’s not a nonprofit research project, but a paid service that arguably competes with licensed artwork and official merchandise.
Courts will likely examine factors like:
Courts will likely examine factors like:
- How transformative are Midjourney’s images (do they add new expression or just remix old art)?
- Is Midjourney’s use of the originals affecting the market for those originals (e.g., could someone use AI art instead of paying DC Comics for a poster or commissioning an artist)?
Industry Impact and What’s at Stake
This lawsuit is more than just Warner Bros. vs. Midjourney – it’s seen as a test case for the entire entertainment and media industry. The Motion Picture Association (which represents major Hollywood studios) quickly voiced support, warning that unchecked AI copyright infringement “threatens the entire American motion picture industry” and the millions of jobs it supports. In other words, Hollywood is fearful that if AI tools can freely use their characters and scenes, the value of those franchises could be undermined. Why would someone pay for an official comic or licensed artwork if an AI can spit out something very similar for free (or for a small subscription fee)? Warner Bros., in its statement, emphasized that protecting its content also means protecting the artists and creators behind those characters. The company said it brought this suit to “protect our content, our partners, and our investments”. This resonates with many human creators (from comic book artists to filmmakers) who worry that generative AI systems could appropriate their style and work without compensation.On the flip side, tech innovators and AI proponents are watching nervously. If the courts side strongly with Warner Bros. Discovery, AI companies might be forced to impose strict filters or licensing agreements whenever an output even resembles a copyrighted character. This could slow down the progress of open-ended AI creativity or require expensive deals between AI firms and content owners. Some in the tech community argue that being too restrictive could stifle innovation – comparing it to telling human artists they can never be inspired by existing art. The ideal outcome for many technologists would be clear guidelines that allow AI development to continue while respecting reasonable bounds of IP law.
This isn’t the only legal battle of its kind. In addition to Disney, Universal, and Warner Bros. teaming up against Midjourney, numerous authors, artists, and media companies have filed suits against AI firms in recent years. For example, groups of visual artists have sued makers of other AI image models for allegedly scraping their artwork. Book publishers and authors have sued AI developers for using novels to train text generators. Even music labels are on alert. All these cases boil down to a common question: Where do we draw the line between learning from existing works and unlawfully copying them? It’s a tricky balance. The U.S. legal system is just beginning to grapple with these questions, and so far, there’s a mix of opinions and preliminary rulings, but no definitive answer.
For now, Midjourney remains operational – users shouldn’t notice any immediate changes to the service while the legal process unfolds. These lawsuits can take months or years to resolve, especially if they set a precedent. The court in California will have to weigh novel arguments and perhaps even bring in expert witnesses (such as AI experts or copyright scholars) to understand how Midjourney works under the hood. It’s possible the case could even be settled out of court if the parties reach some agreement (for instance, an industry-wide licensing framework for AI training data, which some have proposed). However, given the high stakes and strong language (“textbook infringement” vs. “transformative fair use”), both sides seem prepared to battle it out.


Post a Comment