Prompt. Generate. Infringe? Midjourney and the War for Creative Control

A micro start-up of Sans Frangois, with less than a dozen staff members and reportedly $300 million in revenue in 2024, is now facing a lawsuit from the biggest giants in Hollywood. This could become the most defining copyright battle in the history of IPR. In June 2025, several leading Hollywood powerhouses, including Disney, Marvel, Lucasfilm, and Universal, launched a high-profile lawsuit against a San Francisco-based AI firm named Midjourney for creating images from textual inputs. The studios assert that this start-up model wrongfully imitates the intellectual property by reproducing copyrighted characters. Midjourney trains its AI model on extensive datasets extracted from the web. This case comes up as the first of its kind, as this case has emerged as a pivotal point in the dispute over copyright’s scope intersecting with AI-generated creations, potentially redefining the boundaries of innovation, authorship, and intellectual property.

AI’s Memory Trap: When Innovation Echoes Original Work Too Closely

Over the last several years, generative AI models like Midjourney, DALL·E, and ChatGPT have revolutionized how creative industries operate. With simple text prompts, these systems can instantly produce high-quality images, music, writing, and even animations, initiating a transformative shift in modern content development. As these systems are trained on massive datasets retrieved from the internet, which typically contain copyrighted material used without proper clearance, prompting concerns over legal rights and content ownership. As per the U.S. Copyright Office’s 2025 report, “Widespread use of the content may amount to copyright infringement unless it qualifies as transformative or is used under a valid license.” It is also noted that when AI output closely resembles the original characters or style, it may not be counted as fair use. Moreover, another technical study revealed that as the AI model advances, it increases the danger of direct copyright violation as it “memorizes” training data and reproduces it verbatim. With AI models attaining greater sophistication and precision, the case serves as a key test of the adaptability of copyright law in the context of algorithmic creativity.

Plagiarism by Prompt: Midjourney in the Crosshairs of Copyright Law

The lawsuit pulls no punches, calling Midjourney “the ultimate copyright-free-rider” and “bottomless pit of plagiarism.” According to the plaintiffs, it doesn’t matter if the infringing content is generated by AI; piracy is still piracy. Filed by Disney and other major studios, the case brings the unresolved question of AI and copyright law into sharp focus, challenging courts to draw legal boundaries that have yet to be clearly defined. This specific case does not rest simply with image generation of characters that are copyrighted because it goes deeper into the realm of copyright ownership, who and what can infringe such ownership. The core issue of this case revolves around Midjourney’s way of training its AI model. The complaint references a wide array of iconic characters allegedly misused, ranging from intergalactic heroes and animated royalty to beloved superheroes and family-friendly figures. While the filing refrains from listing every name, it includes some of the most recognizable figures in modern entertainment, signaling the lawsuit’s broad scope and high cultural stakes. The plaintiffs have provided many examples of output claiming forgery of output from Midjourney’s platform. The plaintiff also alleges that this model of Midjourney is an act of “calculated and willful” copyright infringement.

Virtual Vending or Copyright Offending? The Midjourney Dilemma in Court

Midjourney works as a virtual vending machine where users can input a prompt and, as output, can get a polished image, frequently echoing well-known artistic elements, personas, or motifs. But what supplies the system’s creativity?  Like most Generative AI systems, it has to depend on enormous datasets scraped from the internet, which most likely contain copyrighted images. The plaintiff contends that using their characters, even indirectly through training, amounts to creating unauthorized derivatives. While this training has been labelled “fair use” by tech companies, the claim is still murky. The U.S. Copyright Office, in its April 2023 policy statement, said, “not taken a position on whether training AI models on copyrighted works without a license constitutes infringement,” but maintained it is a matter that demands attention. Another legal issue is whether outputs generated by AI are considered “derivative works.” If the system produces an image that strongly resembles a copyrighted character or style, even if not directly replicated, it may still be viewed as infringing according to U.S. law. Ultimately, the case brings up the question of platform liability, whether Midjourney itself is accountable for the infringement, rather than just its users. Due to the design of the tool and the absence of content filters, studios claim it facilitates and benefits from ongoing infractions. This lawsuit could push courts to clarify how copyright applies when machines create, copy, or mimic protected content instead of humans.

Protect the Past or Power the Future? The Copyright Question in the Age of AI

The decision in Disney v. Midjourney is more than a difference in opinion over electronic art only; it could be the tone for how copyright still exists after the new creative machine was invented. If the court sides with the studios, AI developers would have to get a license for data, do an audit of the sources of their training, or install stronger content filters before they launch generative tools for the public. It means that the court will put a legal limit on the current “scrape now, ask later” practice in AI training. Conversely, if Midjourney wins, it would give a green light to the whole bunch of AI platforms to carry on using public content as a creative training resource, thus radically changing the notion of authorship. It might spur innovation, but with a drawback: artists, studios, and copyright holders would lose the right of control. In a way, Napster revolutionized music sharing, and Google Books challenged access to literature, and this case could become the point when copyright law has to choose to either protect the past or adapt to the future. In the end, this case is not only about the law but defining the boundaries between innovation and theft. While the technology can produce works similar to human creativity, who owns that work? The outcome of Disney v. Midjourney won’t just affect one AI company. It will shape how every artist, platform, and developer moves forward in a world where machines don’t just assist creativity, they compete with it. This is the moment where copyright law either evolves or gets left behind.

Authored By: Ms. Mitna Gupta. She is a second year B.COM L.L.B student studying at Institute of Law Nirma University.

Editor – Aeshita Marwah

Be the first to comment

Leave a Reply

Your email address will not be published.


*