Midjourney's New AI Video Tool Generates Disney Characters Amidst Landmark Copyright Lawsuit

Midjourney's AI Video Ambition Collides with Hollywood's Copyright Fury

The world of artificial intelligence is evolving at a breakneck pace, pushing the boundaries of creativity and technology. Generative AI models, capable of producing stunning images, compelling text, and now, increasingly sophisticated video, are reshaping digital landscapes. Yet, this rapid advancement is not without its friction, particularly when it intersects with established legal frameworks designed to protect intellectual property. At the heart of a recent storm is Midjourney, a prominent generative AI startup known for its powerful image creation tool. The company finds itself locked in a high-stakes legal battle with two of the world's largest entertainment conglomerates, Disney and Universal, over allegations of copyright infringement. Adding a layer of complexity and controversy to this already tense situation, Midjourney has just rolled out its new AI-generated video tool, V1, a technology that, according to recent findings, appears capable of generating animated clips featuring the very copyrighted characters central to the ongoing lawsuit.

This juxtaposition – a major product launch coinciding with a landmark legal challenge directly related to the product's core functionality – underscores the profound and unresolved tensions between generative AI development and the rights of content creators and copyright holders. The ability of AI models to learn from and potentially replicate vast amounts of existing data, including copyrighted works, raises fundamental questions about fair use, derivative works, and the future of creative industries.

The Blockbuster Lawsuit: Disney, Universal, and the Allegations Against Midjourney

Just weeks before the debut of its V1 video tool, Midjourney was hit with a blockbuster lawsuit filed by Disney and Universal. The complaint, lodged by two titans of the entertainment industry, alleges that Midjourney's generative AI platform infringes upon their extensive portfolio of copyrighted works. The core of the studios' argument centers on the claim that Midjourney's AI model was trained on vast datasets that included copyrighted images and artworks without proper authorization. Furthermore, they contend that the output generated by Midjourney often amounts to infringing copies, reproductions, and derivatives of their intellectual property, including recognizable characters, styles, and settings.

Disney's legal head, Horacio Gutierrez, has publicly characterized Midjourney's output as amounting to “piracy,” a strong accusation that highlights the studios' perspective on the matter. They view the unauthorized use of their characters and creative assets by an AI model as a direct threat to their business model, which is built upon the creation, ownership, and licensing of valuable intellectual property. The lawsuit includes numerous examples of alleged infringing images generated by Midjourney, featuring characters and elements clearly identifiable as belonging to Disney and Universal franchises.

These examples were reportedly compiled, in part, from a report on Midjourney’s so-called “visual plagiarism problem” authored by AI critic and cognitive scientist Gary Marcus and visual artist Reid Southen. Marcus and Southen have been vocal about the issues surrounding generative AI and copyright for some time, pointing out the tendency of these models to reproduce or heavily borrow from existing styles and characters. Marcus commented on the situation, stating, “Reid and I pointed out this problem 18 months ago, and there's been very little progress and very little change... We still have the same situation of unlicensed materials being used, and guardrails that work a little bit but not very well. For all the talk about exponential progress in AI, what we're getting is better graphics, not a fundamental-principle solution to this problem.” His remarks suggest a frustration with the pace of ethical and legal development in the AI space compared to the rapid technical advancements.

The lawsuit is not the only legal challenge Midjourney faces. The company is also navigating additional legal challenges from visual artists who have filed their own lawsuits alleging copyright infringement. These cases, brought by individual creators and groups, often focus on the use of their specific artworks in the AI's training data and the generation of outputs that mimic their unique styles. Together, these legal actions represent a significant challenge to the current operational models of many generative AI companies, forcing a confrontation over the foundational data used to train these powerful systems and the nature of their output.

V1 Video Tool Launches: A New Dimension of Generative Content

Against the backdrop of this escalating legal conflict, Midjourney proceeded with the release of its sophisticated new video tool, V1. This tool represents a significant step for Midjourney, moving beyond static image generation into the dynamic realm of animated content. The current iteration of V1 requires users to start with an existing image, either one they generate using Midjourney's image tools or one they upload. Text-only prompts for video generation are not yet supported, meaning the animation is based on an initial visual seed.

The introduction of V1 was met with keen interest from the AI community, eager to experiment with Midjourney's distinctive aesthetic applied to motion. However, the timing of the release, so close to the filing of the Disney and Universal lawsuit, immediately raised questions about how the new video tool would handle copyrighted content, particularly the characters at the center of the legal dispute. The studios' complaint explicitly anticipated this, alleging that video capabilities would “only enhance Midjourney ability to distribute infringing copies, reproductions, and derivatives of Plaintiffs’ Copyrighted Works.”

It appears Midjourney may have attempted to implement some form of content moderation or guardrails specifically for the V1 video tool, perhaps in anticipation of or reaction to the legal pressure. During testing, certain prompts based on well-known Disney and Universal characters, such as Elsa from Frozen, Boss Baby, Goofy, and Mickey Mouse, were reportedly blocked from generating videos, even though Midjourney's image tool might still generate static images of these characters. Attempts to animate images of Elsa, for instance, triggered an “AI moderator” message stating, “Al Moderation is cautious with realistic videos, especially of people.” This suggests an awareness of the potential risks associated with generating animated content, particularly involving human-like characters, and an attempt to mitigate them.

Guardrails Tested: Circumvention and Copyrighted Characters in Motion

Despite these apparent efforts to implement guardrails, WIRED testing shows that V1 will generate animated clips of a wide variety of Universal and Disney characters. This indicates that the limitations, while present for some specific characters or prompt types, are far from comprehensive. Characters like Homer Simpson, Shrek, Minions, Deadpool, and Star Wars figures such as C-3PO and Darth Vader were successfully animated using the V1 tool.

The process often involved generating an initial image of the character using Midjourney's standard image tool and then using the new “Animate” button to bring it to life. For example, a prompt for an image of Minions eating a banana resulted in recognizable images of the characters. Clicking “Animate” on one of these outputs then produced a video clip of the Minions consuming the banana, peel and all. This demonstrates the tool's ability to not only reproduce the characters visually but also animate them performing actions.

Furthermore, testing revealed that it was sometimes possible to circumvent the potential guardrails by using variations in spelling or repeating prompts. This suggests that the moderation system might rely on specific keyword detection rather than more sophisticated visual recognition or contextual understanding, making it vulnerable to creative prompting techniques. More concerningly, Midjourney's V1 allows users to provide a text prompt to inform the animation process, even when starting from an image. Using this feature, testers were able to generate clips depicting copyrighted characters engaging in adult or inappropriate behaviors. Examples included the beloved Pixar character Wall-E brandishing a firearm and the iconic Star Wars character Yoda smoking a joint.

Photograph: Anadolu/Getty Images

These findings are particularly problematic for Midjourney in the context of the ongoing lawsuit. Generating images of copyrighted characters is one thing; generating animated clips of those characters, especially in scenarios that are wildly out of character or potentially harmful (like a child-friendly robot holding a gun), could be seen as a more egregious form of infringement and creation of unauthorized derivative works. It also raises ethical questions about the responsible use of generative AI and the potential for misuse when guardrails are ineffective or easily bypassed.

The Legal Battle Heats Up: Implications for Generative AI

The lawsuit filed by Disney and Universal against Midjourney is poised to be a significant legal test case for the generative AI industry. The outcome could have far-reaching implications for how AI models are trained, how their outputs are regulated, and the extent to which AI companies are liable for copyright infringement. The studios are likely to argue that Midjourney's training process constitutes unauthorized copying of their works and that the generated images and videos are infringing derivative works. They may seek damages, as well as injunctions to prevent Midjourney from generating content based on their intellectual property.

Midjourney's defense will likely involve arguments related to fair use, claiming that training an AI model on publicly available data, even if copyrighted, is a transformative use and therefore permissible. They might also argue that their output is not a direct copy but a new creation, albeit one influenced by the training data. However, the clear visual resemblance of the generated characters to their copyrighted counterparts, and the ability to prompt specific characters, weakens the transformative use argument in the context of the output.

The fact that the V1 video tool can generate these characters in motion, and in potentially inappropriate contexts, could further complicate Midjourney's legal position. Animation adds another layer to the concept of a derivative work, as it involves bringing static images to life and depicting characters performing actions. The studios could argue that these animated clips are even more direct and potentially damaging infringements than static images.

The legal challenges faced by Midjourney and other generative AI companies highlight a fundamental disconnect between the rapid technological advancements in AI and the slower pace of legal and regulatory frameworks. Existing copyright law was not designed with generative AI in mind, leading to complex questions about its application in this new context. Key issues include:

**Training Data:** Is it legal to train AI models on copyrighted data scraped from the internet without explicit permission?
**AI Output:** When does AI-generated content that resembles existing works constitute an infringing derivative work?
**Liability:** Who is liable for infringing content generated by an AI model – the company that developed the model, the user who provided the prompt, or both?
**Fair Use:** How does the doctrine of fair use apply to the training and output of generative AI models?

The outcomes of these lawsuits could set precedents that determine the future trajectory of generative AI development. A ruling in favor of the copyright holders could force AI companies to fundamentally change their training practices, potentially requiring licensing agreements for the data they use. This could significantly increase the cost of developing and operating generative AI models and might limit the scope of content they can generate. Conversely, a ruling that favors AI companies could weaken copyright protections in the digital age, making it more difficult for creators and companies to protect their intellectual property from being used to train and generate new content.

The Technical and Ethical Tightrope of AI Guardrails

The testing of Midjourney's V1 video tool also sheds light on the significant technical challenges involved in implementing effective guardrails for generative AI. While Midjourney appears to have made some attempt to block certain prompts related to specific characters for video generation, these efforts were clearly insufficient and easily bypassed. This is a common challenge across generative AI platforms.

Implementing robust guardrails requires sophisticated AI models capable of understanding not just keywords but also visual likenesses, contextual nuances, and potential for misuse. Blocking specific character names is relatively simple, but preventing the generation of a character's likeness through descriptive prompts or image uploads is much harder. Furthermore, preventing the generation of inappropriate or harmful content involving characters requires a deep understanding of context and potential interpretations, which current AI models often lack.

The ease with which testers could generate clips like “Wall-E with a gun” or “Yoda smoking a joint” highlights the ethical tightrope AI developers walk. While the technology enables creative possibilities, it also opens the door to the creation of content that can be disturbing, offensive, or damaging to the brands and characters involved. For companies like Disney, whose brand is heavily associated with family-friendly content, the unauthorized depiction of their characters in violent or adult scenarios is particularly harmful.

The development of effective and ethical guardrails is crucial for the responsible deployment of generative AI. This involves not only technical solutions but also clear policies, user education, and mechanisms for reporting and addressing problematic content. The current situation with Midjourney suggests that the industry still has a long way to go in this regard.

Looking Ahead: The Future of Generative AI and IP

The convergence of Midjourney's V1 video tool launch and the Disney/Universal lawsuit marks a critical moment in the evolution of generative AI. It forces a public confrontation between the ambitions of AI developers and the established rights of intellectual property holders. The outcome of this legal battle, and others like it, will likely shape the future landscape of generative AI.

Several potential scenarios could unfold:

**Licensing Models:** AI companies might be compelled to enter into licensing agreements with major copyright holders, similar to how music streaming services license music. This could create a new revenue stream for content owners but might also concentrate power in the hands of large corporations.
**Stricter Regulation:** Governments might step in to enact new laws specifically addressing AI and copyright, providing clearer guidelines on training data, output, and liability.
**Technological Solutions:** AI developers might invest heavily in developing more sophisticated guardrails and content filtering mechanisms to prevent the generation of infringing or inappropriate content.
**Shift in AI Development:** The focus of generative AI development might shift towards creating models trained on licensed or public domain data, or towards generating content that is more abstract and less likely to mimic existing styles or characters.
**Continued Legal Battles:** The current lawsuits might be just the beginning, leading to years of litigation as courts grapple with these complex issues.

The situation also highlights the need for ongoing dialogue between AI developers, artists, copyright holders, and policymakers. Finding a balance that fosters innovation while protecting the rights of creators is essential for the healthy development of the digital economy.

For now, Midjourney's V1 video tool stands as a powerful demonstration of the rapid progress in generative AI, but also as a stark reminder of the unresolved legal and ethical challenges that accompany this technology. The ability to generate animated clips of beloved, copyrighted characters, especially in questionable scenarios, while simultaneously defending against a major copyright lawsuit, encapsulates the complex and often contradictory state of generative AI in 2025. The world is watching to see how this unfolds and what it will mean for the future of creativity, technology, and intellectual property.