Hollywood Takes on AI: Major Film Studios Sue Midjourney Over AI Copyright Infringement

This blog was written with contributions from Leo Lichtman, Partner at ESCA Legal, and Lea Leisure, Intern.

In the latest fight between rights holders and AI companies, a number of major film studios, including Disney, Universal, and its subsidiaries, have filed a copyright infringement lawsuit against Midjourney, the popular AI image-generation platform founded in 2021. At the heart of the case is a key question: Can generative AI legally train on and reproduce well-known characters like Spider-Man, the Simpsons, Elsa, or Shrek without a license? It’s the first time major Hollywood studios have taken a generative AI company to court, and the outcome could reshape how copyright law applies to AI-generated visuals. 

What’s the Case About?

            The lawsuit alleges that Midjourney illegally trained its model on Disney and Universal’s vast libraries of copyrighted content and now allows users to generate unauthorized replicas with just a few words. The studios argue that Midjourney functions as a “virtual vending machine” distributing endless, unlicensed version of their protected characters. For example, users can easily prompt Midjourney to produce images of Darth Vader in new cinematic scenes, Po from Kung Fu Panda reimagined in different settings, or Despicable Me’s The Minions, rendered with striking similarity to the originals.

            The complaint emphasizes that Midjourney, which reportedly earned over $300 million in revenue last year, has crossed the line between inspiration and mass commercial reproduction. The studios are asking the court for a preliminary injunction to block Midjourney from reproducing the protected content and from offering its services without safeguards against infringement. They are also seeking damages.

Why This Case Stands Out:

While a number of recent lawsuits have challenged how AI platforms use copyrighted material, this case stands out in a few key ways. First, it’s the first such suit brought by major Hollywood studios. Second, it largely focuses on the outputs. While the studios allege that Midjourney trained on their content, the complaint includes extensive side-by-side comparisons of Midjourney-generated images and original studio-owned character art, aiming to demonstrate near-identical copying, and alleging that this is a feature and not a bug. Accordingly, this case is significant because it’s about more than just how AI is trained; it’s about what it produces. It’s the first time a court will weigh whether generative AI can legally replicate the distinctive visual styles of beloved franchises without infringing IP rights. 

Midjourney’s Likely Legal Defense:

As other commentators have noted, the defenses here will likely depend on the theory of infringement. As to Midjourney’s use of the studios’ materials for training, Midjourney will likely argue that its conduct is protected under the “fair use” doctrine, which allows limited, transformative use of copyrighted content in some instances. Here, Midjourney could argue that its use of a wide swath of materials for training purposes is fundamentally different from the original purpose in which the materials were created, because unlike their original expressive purpose, the new purpose is purely to serve as training data. The studios, however, may counter that Midjourney’s use supplants the market for its copyrighted materials since Midjourney itself ultimately serves as a repository for those materials, making the degree of transformativeness far more limited. Indeed, this would be in line with the U.S. Copyright Office’s views, set forth in the recently published Part 3 of its Copyright and Artificial Intelligence report. Therein, the Copyright Office opined that while the use of copyrighted works for training AI can be transformative (particularly where the data is used to train the AI to handle non-substitutive tasks), training data to generate outputs that are substantially similar to the copyrighted works in the dataset renders the use far less transformative.

As to the creation of unauthorized derivatives in the output, questions will remain over (1) whether the outputs copy enough of the original to constitute infringement (i.e. are the outputs substantially similar to the copyrighted works); and (2) the extent of Midjourney’s liability for said infringement. On the first question, if the studios’ allegations are true, it will likely have no issues demonstrating substantial similarity. Indeed, if a picture is worth a thousand words, the studios’ image-laden complaint, much like the Star Wars universe they allege was infringed, presents an epic saga.

On the second question, Midjourney will likely take the position that the infringing outputs were generated by subscribers, and therefore it cannot be held directly liable. While the studios may present arguments for why Midjourney’s creation of the platform and refusal to enact reasonable technical measures to limit the infringement renders it liable as a direct infringer, much of the focus may well be whether Midjourney can be held liable as a secondary infringer for its subscribers’ conduct, under theories of contributory liability (which will require the studios to show that Midjourney knowingly contributed, enabled, or induced its subscribers’ infringement) and vicarious liability (which will require the studios to show that Midjourney had the right and ability to control the infringement, and derived a direct financial benefit from it).

The Broader Picture:

This lawsuit comes amidst a recent wave of AI-related copyright lawsuits. Check out the ongoing NYT v. OpenAI copyright suit here. Meanwhile across the pond, the first major copyright trial involving generative AI is underway in London, with Getty Images suing Stability AI over the use of its stock photo library to train image-generation tools. Industry voices across music, publishing, and visual art are demanding clearer rules around how AI systems are trained and deployed. The Midjourney suit reflects rising pressure from rights holders across creative industries to draw firmer boundaries around the use of their work in AI systems.

What Could Happen Next?

Whatever the outcome, this case is poised to set an important precedent. If the studios win, AI platforms may be required to implement filters that could block protected content, including recognizable characters. Licensing agreements could also become a standard prerequisite before training or generating content, and Midjourney itself may face damages, a court-ordered resign of its tool, or both.

If, on the other hand, Midjourney succeeds in defending its position, courts may affirm a broad interpretation of transformative fair use in the AI context. This could allow AI developers to retain greater flexibility in how their systems are trained and used, though they would still face significant ethical and legal scrutiny in light of growing industry and public concern. Either way, the decision is likely to shape future regulatory and commercial frameworks for generative AI for years to come.

What This Means for Creators, Developers, and Users:

            This case carries implications for everyone involved in the AI ecosystem. Developers may face increased oversight of how their models are trained and what kinds of outputs they generate. Rights holders could gain stronger tools to license and enforce protection of their work. Finally, users might see more restrictions on what AI tools can create, but also benefit from more transparency surrounding how the tools are trained and what they can (and can’t) do. Whether you build with AI, create the content it learns from, or simply use these tools in your creative workflow, this case marks a turning point. Courts are beginning to shape the legal boundaries of generative AI, and the ripple effects will likely be felt across tech, entertainment, and beyond. 

            As this case unfolds, one thing is clear: the future of AI won’t just be written in code. It’ll be negotiated in court.

This post isn’t legal advice—just general insight. Always consult a qualified lawyer about your specific situation. ESCA Legal specializes in intellectual property, global brand strategy, and fighting counterfeits; reach out to us at info@esca.legal if you’ve got questions.