Disney and Universal sue Midjourney over AI copyright violations
Midjourney is accused of training its AI on copyrighted film content. The lawsuit could redefine how copyright laws apply to generative AI.
Disney and Universal have filed lawsuits against Midjourney, a prominent generative AI company, accusing it of copyright infringement over its use of proprietary content in training its image-generation model.
The lawsuits were filed in a California federal court, alleging that the startup used copyrighted movie stills, character designs, and promotional visuals without proper authorisation.
According to the filings, the studios claim that Midjourney’s AI has been trained on datasets that include “tens of thousands” of copyrighted assets owned by both companies.
These assets allegedly include visual material from popular franchises such as Frozen, Jurassic Park, Marvel, and Despicable Me, enabling the model to produce images closely resembling protected characters and scenes.
Legal focus on training datasets
The lawsuits centre on the contentious issue of whether using copyrighted materials as part of a training dataset constitutes fair use.
While Midjourney maintains that its AI was trained on publicly available internet data—falling under fair use exemptions—Disney and Universal argue that the outputs generated by the model cross a legal boundary by replicating the distinctive features of their intellectual property.
“The model is capable of generating images that are substantially similar to copyrighted works owned by the plaintiffs,” the legal filing states. “This capability is not coincidental, but the result of ingesting copyrighted material at scale.
Wider industry context
The case comes amid growing backlash from artists, studios, and rights organisations regarding generative AI tools that can replicate or remix existing creative styles. Visual artists and illustrators have previously filed similar lawsuits against other companies, including Stability AI and DeviantArt, citing unauthorised use of their work.
Meanwhile, regulatory bodies in the EU and U.S. are exploring guidelines that may require AI firms to disclose the origins of their training data—a move that could standardise transparency across the sector.