Disney and Universal Launch Pivotal Copyright Lawsuit Against Midjourney Over AI Image Training

On June 11, 2025, The Walt Disney Company and Universal Studios jointly initiated a landmark federal lawsuit against Midjourney, an artificial intelligence company specializing in advanced image-generation technology. The complaint was filed in the U.S. District Court for the Central District of California and alleges that Midjourney engaged in "massive and deliberate copyright infringement" by enabling users to generate highly realistic depictions of proprietary characters such as Elsa, Shrek, Darth Vader, and Homer Simpson without any license or authorization.

This litigation transcends a conventional copyright dispute, directly challenging the foundational business model of generative AI companies that train their algorithms on vast amounts of publicly scraped internet content. It represents a critical juncture in the ongoing conflict between protecting creative ownership and fostering AI-driven innovation.

Central to the plaintiffs’ allegations is the claim that Midjourney trained its AI models using tens of thousands of copyrighted images owned by Disney and Universal, all sourced without permission. The studios contend that the AI-generated outputs are not merely inspired by but "virtually indistinguishable" from the original copyrighted works, thereby violating both the letter and spirit of U.S. copyright law.

According to the complaint, cease-and-desist letters previously sent to Midjourney were disregarded, and the company continued to facilitate user prompts that produced images of copyrighted characters and scenes. The plaintiffs argue this conduct amounts to willful infringement, justifying enhanced statutory damages under 17 U.S.C. § 504(c).

The studios seek injunctive relief requiring Midjourney to halt generating or training on their copyrighted content. Additionally, they demand the implementation of proactive copyright filtering mechanisms to prevent future unauthorized use.

This lawsuit is the first major action by leading Hollywood studios targeting a generative AI firm but arrives amid a growing and evolving legal landscape concerning AI and intellectual property rights. Notable precedents include:

- Getty Images v. Stability AI (2023): Getty sued Stability AI in both the U.K. and U.S. for training AI models on millions of watermarked photographs without licenses. The U.K. court refused early dismissal, indicating a willingness to examine the case’s merits.

- The New York Times v. OpenAI & Microsoft (2023): The Times alleged that its journalism was ingested and reproduced nearly verbatim by large language models, raising critical questions about fair use and data licensing.

- Andersen v. Stability AI (2023): A class-action suit by artists claimed that their personal styles and images were copied and reconstructed by generative AI tools. While partially dismissed, the case introduced the novel concept of "style as property" into copyright debates.

These cases, alongside Disney v. Midjourney, underscore a pivotal unresolved legal question: Does training AI on copyrighted material, even without exact copying, constitute infringement?

Midjourney and similar AI companies may invoke the fair use doctrine, arguing that their use of copyrighted works for training is transformative and does not substitute the original market. However, courts have traditionally limited fair use where the AI-generated output diminishes the commercial value of the original, especially in the entertainment industry where licensing is a multi-billion-dollar market.

What distinguishes the Disney and Universal lawsuit is the clear identifiability of the copyrighted characters. The complaint includes direct visual comparisons demonstrating that AI-generated images closely replicate the originals with minimal variation, reducing the likelihood that courts will deem such use transformative or de minimis.

- Increased scrutiny of AI training data, potentially compelling companies to disclose sources and obtain licenses akin to music and media licensing in streaming platforms.

- Court-mandated technical safeguards, similar to YouTube’s Content ID system, to prevent real-time generation of infringing content.

- The acceleration of industry-wide licensing regimes for AI training and prompt-based usage, ensuring IP holders receive compensation.

Conversely, a ruling favoring Midjourney could embolden AI developers to train models on publicly available content with fewer restrictions, significantly altering the balance between rights holders and AI innovators.