Midjourney Asks Federal Court To Compel Disney, Universal, Warner Bros. To Disclose AI Training

AI image generator Midjourney has asked a federal court to compel Disney, Universal and Warner Bros. to disclose detailed internal generative AI training practices, including prompts and outputs on Midjourney, as part of a dispute over alleged copyright infringement.

“Last year, we reported that AI company Midjourney was facing the combined might of three of the biggest studios in Hollywood, as Disney,Warner Bros”

A.V. Club

The filings say the studios should not be allowed to sue Midjourney while potentially engaging in similar unlicensed training, and the dispute centers on discovery scope, copyright and fair use.

A.V. Club

Midjourney argues that training on publicly available images, including those featuring copyrighted characters, falls under the fair use doctrine, and it is now seeking to overturn a prior limitation that restricted disclosure to “consumer-facing” videos and images.

In mid-June, a judge limited what the studios must hand over by allowing them to retain most information about their AI use and transmit only information about applications “consumer-facing,” while Midjourney seeks broader access to internal practices.

Midjourney’s motion also seeks “all prompts” the studios used, along with resulting outputs, not just prompts tied to allegedly infringing images, according to the accounts of the filing.

Midjourney’s latest argument is that the withheld documents are the ones that would show whether the studios are “doing exactly what they are suing Midjourney for doing.”

In response, the studios’ lead attorney David Singer described Midjourney’s request as a “fishing expedition,” and emphasized that the studios “do not seek to stop AI technology or even shut down Midjourney’s business.”

CryptoRank

Singer said the studios “simply want Midjourney to stop copying their movies and TV shows and to stop distributing, publicly displaying, publicly performing, and creating derivative works that include copies of [their] famous characters without authorization.”

Midjourney’s position ties the discovery request directly to its fair use defense and its “unclean hands” theory, arguing that internal development for storyboarding or ideating would demonstrate an industry custom.

One account of the dispute also frames the court’s decision as potentially setting a precedent for AI training data transparency and corporate accountability across the generative AI industry.

Beyond the Midjourney case, Hollywood’s labor negotiations are also putting generative AI in the spotlight, with the Directors Guild of America reaching a four-year framework agreement with the Alliance of Motion Picture and Television Producers after negotiations began in May.

“Midjourney wants the Hollywood studios that sued it to show the court how they use AI The image generator argued that the companies are also training their AI on copyrighted data”

Engadget

The DGA said the agreement still must be approved by its national council and ratified by guild members, and it noted a 35% drop in the employment rate of television directors in 2024 as TV production is down in Hollywood.

The DGA also sought greater say over the use of generative artificial intelligence, and the agreement is described as longer than the three-year deals usually struck in the sector.

In parallel, the Midjourney dispute is being treated as a test of what information can be admitted in court, with one report saying the federal judge’s decision could have an effect on future lawsuits and set a precedent on what kind of information should and can be produced.

The stakes described in the coverage extend to how studios and AI companies approach the legal boundaries of AI training and fair use, as Midjourney presses for broader discovery while the studios argue for limits tied to consumer-facing outputs.