Midjourney has asked a federal court to overturn a magistrate judge's discovery ruling that shielded Warner Bros. Discovery, Disney, and Universal Studios from disclosing most of their internal AI operations — a move that could reshape the evidentiary landscape for AI copyright litigation broadly.

The image-generation company filed its Motion to Compel on March 27, 2026, targeting the studios that have sued it for copyright infringement. Disney and Universal filed the first lawsuit on June 11, 2025, alleging Midjourney generated images of their copyrighted characters; Warner Bros. Discovery followed with a separate action on approximately September 4, 2025, naming Superman and Scooby-Doo among the infringed works. Midjourney now wants the studios to produce AI business plans, research reports, training datasets, model weights, and board-level presentations on AI strategy.

In mid-June 2026, Magistrate Judge Joel Richlin ruled against Midjourney's discovery bid, holding that the studios need only hand over materials related to consumer-facing AI products. Midjourney is now asking the district court to reverse that order.

The "Unclean Hands" Gambit

The legal theory driving the request is straightforward: if the studios are themselves training AI systems on copyrighted material, that conduct could undercut both the moral and legal force of their infringement claims. Midjourney attorney Bobby Ghajar put it plainly in the filing — Engadget reported the quote directly: "If Plaintiffs are doing the very thing they seek to punish, that evidence goes to the heart of Midjourney's fair use and unclean hands defenses."

The "unclean hands" doctrine is an equitable defense: a party that has engaged in misconduct related to the subject of the dispute can be barred from obtaining equitable relief. Pairing it with a fair use argument creates a two-track strategy — either the studios' own AI training practices demonstrate that training on publicly available works is an accepted industry norm (supporting fair use), or those practices constitute the very infringement they are accusing Midjourney of (supporting unclean hands). Neither track requires Midjourney to win outright on the merits of fair use; together they complicate the studios' standing to pursue relief.

Midjourney's core fair use position — that training AI on publicly available images is transformative and therefore lawful — remains one of the central unresolved questions in U.S. AI copyright law. The company's discovery request is, in part, an attempt to build a factual record around how the content industry itself has answered that question internally.

What the Ruling Could Establish

The scope of permissible discovery in AI copyright cases is genuinely unsettled. Magistrate Judge Richlin's June 2026 order drew a line at consumer-facing applications, treating internal R&D, model weights, and strategic documents as out of scope. If the district court overturns that ruling, it would signal that defendants in AI infringement cases have considerably broader latitude to probe plaintiffs' own AI practices — regardless of whether those practices are public knowledge.

The studios are not without precedent in arguing that their internal AI work is irrelevant. Standard discovery doctrine requires relevance to the claims or defenses at issue, and a court could reasonably hold that how Disney uses AI internally has no bearing on whether Midjourney reproduced protectable expression. Midjourney's argument is that the studios' internal conduct is directly relevant — not incidentally, but centrally — because it speaks to the normative and legal status of the very training practices in dispute.

The broader litigation context adds weight. Disney, Universal, and Warner Bros. Discovery filed a parallel copyright action against Chinese AI company MiniMax in September 2025; that company lost its bid to dismiss the suit on May 26, 2026. The studios are clearly prosecuting a coordinated IP strategy across multiple AI defendants. Whether they can simultaneously do so while shielding their own AI development from scrutiny is now a live judicial question.

Worth flagging: the outcome of the district court's ruling on Midjourney's motion will likely influence how discovery is scoped in future AI copyright cases — particularly those where a media or technology plaintiff can plausibly be accused of parallel AI training activity. That covers a large fraction of the industry. Courts in subsequent cases will look to whatever reasoning the district court supplies here, making the written opinion, if there is one, as consequential as the outcome itself.