Disney and Universal Sue Midjourney Over AI-Generated Character Copies: What's at Stake

Disney and Universal Studios have filed a federal lawsuit against Midjourney, claiming the AI image generator illegally reproduces their copyrighted characters without permission. The complaint, filed on June 11, 2025, includes over 30 side-by-side visual comparisons showing Midjourney outputs that allegedly match beloved characters from Marvel, Star Wars, and other franchises nearly unchanged .

What Exactly Are the Studios Accusing Midjourney Of?

The lawsuit centers on two legal concepts: direct infringement and secondary infringement. Disney and NBCUniversal argue that when users prompt Midjourney to generate images of characters like Elsa, Iron Man, or Shrek, the tool reproduces those characters in ways that violate copyright law. The studios claim Midjourney profits from subscriptions while artists and character creators receive nothing .

The complaint is substantial. It spans 110 pages and includes comparative exhibits showing source material alongside AI-generated outputs. The studios argue that the similarity extends beyond surface details; they point to matching composition, lighting, and costume elements as evidence of infringement .

"Piracy is piracy," stated Horacio Gutierrez, Disney's General Counsel.

Horacio Gutierrez, General Counsel at Disney

NBCUniversal's Kim Harris framed the dispute differently, emphasizing that creators deserve protection regardless of technological change. Both studios presented the litigation as an existential stand for creative labor in the age of generative AI .

How Is Midjourney Defending Itself?

Midjourney filed its answer on August 6, 2025, denying nearly every allegation. The company's defense rests primarily on the concept of "fair use," a legal doctrine that permits limited use of copyrighted material for transformative purposes .

Midjourney's legal team at Cooley LLP argues that training AI models on diverse reference images is similar to how artists study thousands of reference photos to develop their craft. They contend that the outputs are transformative mashups, not direct replacements of copyrighted works. The company also argues that minimal market harm occurs because the generated images serve different purposes than official licensed merchandise .

However, the studios counter that consumers already download finished posters and artwork directly from Midjourney without licensing, creating immediate market harm. This disagreement over whether the technology is transformative or infringing sits at the heart of the case .

Why Does This Case Matter Beyond Midjourney?

Tech companies across Silicon Valley are watching this lawsuit closely because the outcome could reshape how all AI image generators operate. A victory for the studios could require AI companies to license datasets and obtain permission before training models on copyrighted content. Such requirements would significantly increase development costs and timelines for startups .

Conversely, a defense win for Midjourney could embolden other AI companies to release open-weight models without licensing agreements. However, creators and industry groups warn that such an outcome might normalize large-scale copyright infringement across entertainment and media .

The Motion Picture Association supports aggressive enforcement, while policy groups like the Chamber of Progress view the lawsuit as an overreaction. This split reflects genuine uncertainty about how copyright law should apply to generative AI .

How Will Courts Evaluate Fair Use in This Case?

U.S. copyright law allows judges to consider four factors when determining whether a use qualifies as fair use. The first examines the purpose and character of the use, including whether it is transformative. Here, Midjourney's commercial subscription model could weigh against the company's fair use defense .

The second factor reviews how creative the original work is. Marvel and Star Wars characters are highly creative content, which typically tilts the scale toward copyright holders. The third factor compares how much of the original work was used. The studios' exhibits show entire characters, not small snippets, which undermines claims that the use is transformative .

Finally, courts assess market harm. The studios argue that AI image generation already siphons licensing revenue for posters, t-shirts, and streaming thumbnails. Midjourney may struggle to prove that the substitution effect is negligible, though legal precedent on generative AI remains unsettled .

Steps to Reduce Legal Risk in AI Image Generation

Companies developing or deploying AI image tools can take concrete steps now to reduce their exposure to copyright claims:

  • Dataset Audits: Review training data for protected content and maintain detailed provenance logs showing where each image came from and whether it was licensed.
  • Opt-Out Mechanisms: Offer users and rights holders the ability to exclude their content from training datasets and honor takedown requests quickly.
  • Licensing Agreements: Negotiate blanket licenses with major studios and content creators when feasible to avoid accusations of infringement.
  • Internal Red-Teaming: Conduct cross-functional reviews involving engineers, legal counsel, and marketing teams to spot potential copyright conflicts early.
  • Governance Frameworks: Reference emerging best practices from the U.S. Copyright Office and document diligent oversight efforts that may lower insurance premiums.

What's the Timeline for Resolution?

Initial discovery conferences are scheduled to begin in early 2026 under Judge John A. Kronstadt's oversight. The studios are expected to move for a preliminary injunction once document production starts, meaning the court could rule on interim relief before a full trial .

A mediation session with retired Judge Suzanne Segal is slated for late summer 2025. The case could resolve through settlement, a partial injunction limiting specific franchise prompts, or a full trial addressing fair use and infringement .

The coming year may yield the first major judicial opinion on whether generative AI training constitutes fair use. That decision will ripple through every AI image generation roadmap and influence how investors evaluate copyright risk in AI startups .