The legal war between the generative AI industry and the creative establishment has reached a high-stakes turning point. Midjourney, one of the most prominent players in the text-to-image AI space, has launched an aggressive legal maneuver in its ongoing copyright battle against Disney, Universal, and Warner Bros. The startup is now demanding that these entertainment conglomerates pull back the curtain on their own internal artificial intelligence practices, arguing that the studios may be engaging in the very behaviors for which they are currently seeking damages.
This latest development in the discovery phase of the litigation marks a shift from a defensive stance to a tactical offensive. By seeking to compel the studios to disclose their own AI development workflows, Midjourney is attempting to prove that the use of copyrighted material for model training is an industry-standard practice—even among those who hold the most valuable intellectual property (IP) on the planet.
Chronology: The Road to the Courtroom
The conflict began to boil over last year when entertainment giants Disney and Universal filed a joint lawsuit against Midjourney. The plaintiffs alleged systematic copyright infringement, asserting that the startup’s image-generation models were capable of churning out high-fidelity likenesses of iconic characters such as Darth Vader, Bart Simpson, and various Disney princesses.
The studios argued that this capacity for "character mimicry" was not a byproduct of creative AI synthesis, but evidence that the model had been trained on their proprietary catalogs without license or permission. The legal pressure intensified months later when Warner Bros. joined the fray, filing its own lawsuit against the company, citing the unauthorized generation of characters like Superman and Batman.
For its part, Midjourney has maintained a consistent defense: that its training process constitutes "fair use." The company contends that its models learn abstract concepts and visual styles rather than merely cataloging images, and that the resulting outputs are transformative works protected under United States copyright law.
As the litigation progressed into the discovery phase, a judge initially sided with the studios, placing a limitation on the documentation required to be produced. The court ruled that the studios only had to disclose information regarding their generative AI usage if those tools resulted in "consumer-facing" assets—videos, images, or products meant for the public. Midjourney is now seeking to overturn this limitation, claiming it creates an uneven playing field.
The Core Dispute: "Consumer-Facing" vs. Internal Practice
At the heart of the current motion is Midjourney’s contention that the court’s "consumer-facing" distinction is a loophole that allows studios to curate a narrative that suits their claims of market harm. In its latest filing, Midjourney argues that the studios are "cherry-picking" evidence.
"The documents [the studios] are withholding are precisely those that would reveal whether, behind closed doors, they are doing exactly what they are suing Midjourney for doing," the filing states.
Midjourney’s legal team is pushing for a broader discovery scope, specifically requesting information on how the studios utilize AI for internal processes. They hypothesize that studios are likely training their own proprietary models—or utilizing third-party tools—to assist in storyboarding, concept art, and preliminary ideation.
If Midjourney can prove that Disney, Universal, and Warner Bros. use unlicensed copyrighted material to train internal AI models for the purpose of film and television development, they believe they can establish that such behavior is an "industry custom." The legal logic follows that if the studios themselves engage in this practice for internal creative workflows, they cannot reasonably claim that Midjourney’s similar, albeit public-facing, training process is an unlawful or egregious violation of copyright.
Furthermore, the startup is pushing for access to all prompts used by the studios when interacting with Midjourney, rather than just the specific prompts that yielded allegedly infringing results. They argue that this data is essential to understanding the full scope of how the studios utilize the tool and to rebut claims that the platform is designed primarily to facilitate copyright infringement.
Official Responses and the "Fishing Expedition" Defense
The studios have vehemently pushed back against these demands. David Singer, the lead attorney representing the studios, has dismissed Midjourney’s motion as nothing more than a "fishing expedition."
According to Singer, the studios are not attempting to stifle the evolution of AI technology or dismantle Midjourney as a business entity. Instead, he characterizes the studios’ position as a focused defense of intellectual property rights. The studios’ primary demand, as articulated by Singer, is that Midjourney cease the unauthorized copying of their films and television shows and stop the generation and distribution of derivative works that rely on their iconic characters.
For the studios, the discovery process is a matter of protecting their specific assets, not a trial on the general legality of AI training. They argue that internal, experimental AI usage—if it even exists in the form Midjourney suggests—is fundamentally different from the public distribution of generative models that commoditize their IP. By focusing the discovery on "consumer-facing" products, the studios aim to keep the trial centered on the commercial impact of Midjourney’s outputs rather than a broad, distracting debate on industry-wide AI adoption.
Implications: The Future of AI and Creative IP
The resolution of this discovery dispute could have massive ramifications for the future of the media and technology sectors.
1. The Precedent of "Industry Custom"
If Midjourney succeeds in compelling the disclosure of the studios’ internal AI usage, and if those documents reveal widespread use of copyrighted training data, the legal landscape of generative AI could shift dramatically. A finding that "everyone does it" could weaken the studios’ claims of market harm and bolster the fair use argument for all AI startups, not just Midjourney.
2. The Scope of Discovery
The judge’s decision on whether to expand discovery will set a precedent for future copyright lawsuits involving AI. If courts allow discovery into internal, non-public workflows, it could force companies across every industry—from tech to finance to pharmaceuticals—to reveal their proprietary AI training methods whenever they engage in litigation. This could result in a chilling effect on internal innovation, as companies may fear that experimental internal AI tools could become evidence in unrelated legal disputes.
3. The Definition of "Derivative Work"
This case remains one of the most prominent vehicles for defining what constitutes a "derivative work" in the age of generative models. The studios argue that the output of an AI model, when based on their characters, is a direct violation of their rights. Midjourney argues that the model is merely a tool, and that the responsibility for the output lies with the user. By demanding data on the studios’ own prompts, Midjourney is attempting to demonstrate that even professional, studio-level users can generate similar results, thereby suggesting that the "infringement" is a function of user behavior rather than an inherent, malicious design of the software.
Conclusion: A High-Stakes Stalemate
As the case continues, the tension between the protection of creative legacy and the advancement of generative technology remains palpable. The studios are fighting to maintain the exclusivity of their creative assets, while Midjourney is fighting for the legal recognition of its model-training practices as a standard, transformative process.
Whether this discovery motion is granted or denied, the path forward is likely to be long and contentious. Both parties are digging in, clearly aware that the outcome of this litigation will establish the "rules of the road" for the next decade of AI development. For now, the public waits to see if the internal "AI labs" of Hollywood’s biggest players will be opened to the light of the courtroom, or if the court will uphold the current boundaries of discovery, keeping the focus squarely on the characters that have defined generations of entertainment.
