By Stuart Kerr, Technology Correspondent, LiveAIWire

Open Midjourney’s own public Explore page and, according to a 2025 lawsuit, you will find Shrek, Deadpool, Spider-Man, and Bart Simpson staring back at you, generated by users who typed little more than a character’s name into the prompt box. That gallery of user-made images is now Exhibit A in one of the largest copyright cases the AI industry has faced, a Midjourney copyright lawsuit brought jointly by Disney, NBCUniversal, and Warner Bros. Discovery that could reshape how every AI image generator handles the characters studios have spent decades building.

The case has moved well beyond the artists-versus-scraped-data framing that defined the early AI art copyright debate. It now sits alongside a landmark UK court ruling, a separate artist class action heading toward trial, and a UK government decision to abandon its own proposed fix, together forming the clearest picture yet of where AI image generation and copyright law actually stand in mid-2026.

What makes the Midjourney copyright lawsuit different from the artist-led cases that preceded it is scale and resources. Disney, NBCUniversal, and Warner Bros. Discovery are not asking a court to recognise a novel harm to individual creators. They are asserting ownership over some of the most commercially valuable fictional characters in existence, backed by legal budgets that dwarf anything an independent illustrator could bring to bear, and that asymmetry is already shaping how the case is being litigated.

The Midjourney Copyright Lawsuit at the Heart of Hollywood’s AI Fight

Disney, NBCUniversal, and DreamWorks filed their complaint against Midjourney in June 2025 in federal court in Los Angeles, a 110-page filing alleging what the studios called mass piracy of their characters. According to Georgetown Law’s Institute for Technology Law and Policy, the complaint documented that Midjourney would generate recognisable images of Star Wars characters, Bart Simpson, Shrek, Ariel from The Little Mermaid, Wall-E, and the Minions from Despicable Me from prompts as simple as a character’s name, with no apparent guardrails preventing it. Testing cited in the case found the platform could also generate Homer Simpson, C-3PO, and Deadpool, both as still images and as short animated clips.

Warner Bros. Discovery followed in September 2025 with its own suit, adding Batman, Superman, and other DC properties to the list of characters the studio said Midjourney reproduced on demand, describing the platform’s conduct in its complaint as systematic and wilful. An exhibit attached to the Disney and Universal complaint lists more than 150 allegedly infringed works, and statutory damages for wilful infringement can reach 150,000 dollars per work, putting potential exposure for Midjourney into the tens of millions of dollars before any additional claims are counted.

The complaints also allege that Disney and Universal asked Midjourney directly to stop the infringing behaviour, or to build technical safeguards preventing users from generating named copyrighted characters, before either studio filed suit. According to the studios’ filings, Midjourney did neither, and the platform’s Explore page continued to surface user-generated images of the same characters named in the complaint for months after the initial warnings were sent.

What This Means for You

If you use Midjourney or a similar AI image generator for personal projects, generating an image of a named copyrighted character is exactly the activity now at the centre of a major lawsuit, and platforms are under growing legal pressure to block it rather than tolerate it. If you work in a creative or marketing role, the safest assumption is that AI-generated images resembling any studio-owned character carry real legal risk regardless of how the image was produced, since liability arguments in these cases are increasingly aimed at the platform rather than the individual user, but that protection is not guaranteed to last as the litigation develops.

Midjourney’s Defence and the Discovery Fight Over Hollywood’s Own AI Use

Midjourney has argued that generating images of characters from publicly available training data falls under fair use, and it has gone on the offensive in discovery, demanding that Disney, Universal, and Warner Bros. disclose how their own internal teams use generative AI. Legal commentary on the case describes the strategy as an “unclean hands” defence: if the studios use unlicensed AI tools internally for storyboarding or early concept work, Midjourney’s lawyers argue that undercuts any claim that AI training on copyrighted characters is categorically improper. The studios have resisted the request, arguing their internal production tools have nothing to do with whether Midjourney itself infringed their copyrights.

A magistrate judge partly ruled on that discovery fight in mid-June 2026, limiting Midjourney to information about the studios’ consumer-facing AI tools rather than their internal production use, and finding Midjourney’s broader demands largely irrelevant to whether the platform itself infringed. A hearing set for 17 August 2026 is expected to shape how much further that discovery fight goes, and the outcome will influence disclosure obligations in future AI copyright cases well beyond this one. Whatever the magistrate decides, it will not resolve the underlying Midjourney copyright lawsuit itself, which is still in its early stages and unlikely to reach trial before 2027 given how long the comparable Andersen case has already taken.

The Artists’ Case Moving in Parallel

Before the studios sued, individual artists had already spent more than three years fighting Midjourney, Stability AI, DeviantArt, and Runway over the same underlying question: whether training an image generator on copyrighted work without a licence is infringement. Andersen v. Stability AI, filed in January 2023 by cartoonist Sarah Andersen alongside Kelly McKernan and Karla Ortiz, has survived three rounds of motions to dismiss.

Court filings tracked through the case’s docket show Judge William Orrick allowed direct copyright infringement, induced infringement, and Lanham Act trade dress claims to proceed in an August 2024 order, while dismissing the artists’ DMCA claims. The plaintiffs filed a third amended complaint in February 2026 adding new allegations drawn from training-data discovery, including details about Midjourney’s use of a dataset called Datacomp xLarge. Trial, originally set for September 2026, has since slipped to April 2027 as discovery has expanded, and the case’s outcome is expected to directly shape the Disney, Universal, and Warner Bros. litigation, since both rely on the same underlying theory that an AI model can be found to embody the copyrighted works it was trained on.

Getty’s UK Ruling Adds a Data Point, Not an Answer

The most closely watched ruling to land since Midjourney’s Hollywood lawsuit began came not from a US court but from London. In November 2025, the UK High Court ruled on Getty Images’ long-running case against Stability AI, and the result was a split decision that clarified less than either side hoped. Getty’s own statement on the ruling confirmed the court found that Stable Diffusion’s outputs had infringed Getty’s trademarks by reproducing its watermark, and rejected Stability AI’s attempt to shift responsibility onto users rather than the model provider.

On the core copyright question, Getty lost. The court held that an AI model’s internal weights are not a “copy” of the training images in the sense UK copyright law requires, and Getty was unable to show that the actual copying occurred on UK soil, since Stable Diffusion’s training took place abroad. Getty’s statement afterward argued the case still established that “intangible articles such as AI models are subject to copyright infringement claims in the same way as tangible articles,” a genuinely new legal foothold, even though it fell short of the broad precedent Getty had sought.

Getty said it would carry the UK ruling’s factual findings, including the finding that its copyrighted works were used in training regardless of where that training legally occurred, into its still-active US case against Stability AI. That transatlantic strategy is one the Disney, Universal, and Warner Bros. legal teams are watching closely, since any factual finding that a model can be shown to have used specific copyrighted material in training, rather than simply having learned generalised patterns from it, strengthens the same theory their own complaints depend on.

Where the Law Stands, and Doesn’t

Governments on both sides of the Atlantic have spent the past year trying, and largely failing, to settle these questions through policy rather than litigation. The UK government published its long-awaited report on copyright and AI in March 2026, and its central conclusion was retreat: the opt-out mechanism it had previously floated, letting rightsholders exempt their work from AI training while allowing a broad exception otherwise, was abandoned after creative industry opposition, with no replacement policy put forward. The government said it would gather further evidence rather than legislate.

In the United States, the Copyright Office has held its ground on a single, consistent principle: its published guidance states that works generated entirely by AI, without meaningful human creative input, are not eligible for copyright protection, even when a detailed human prompt shaped the output. The US Supreme Court declined to hear a challenge to that position in March 2026, leaving the human-authorship requirement intact for now. Neither development resolves the separate, harder question both the Andersen and Disney cases are actually litigating: not whether AI output can be copyrighted, but whether training on copyrighted work in the first place was ever lawful.

Where This Leaves Artists, Studios, and Everyone Using Midjourney

LiveAIWire’s earlier coverage of artists’ resistance to AI training found that individual creators have far less leverage than large rights holders in forcing AI companies to change their practices, a gap the Disney, Universal, and Warner Bros. lawsuit throws into sharp relief. The Hollywood studios waited two years after the artist-led cases began before filing their own, and did so only once Midjourney’s outputs were demonstrably reproducing their most valuable characters at scale. A similar pattern played out in music, where Universal Music Group, Sony, and Warner Music waited for AI vocal cloning tools to mature before suing Suno and Udio in 2024, using resources and legal teams individual musicians could never match.

LiveAIWire’s reporting on Bryan Cranston’s deepfake consent win with OpenAI showed one route beyond litigation: organised professional pressure that changes platform behaviour faster than a court case can. LiveAIWire’s coverage of the parallel Reddit and New York Times training-data lawsuits found the same pattern of large, well-resourced plaintiffs setting the terms of AI copyright law that individual creators then have to live with, whatever those terms turn out to be.

None of the current litigation guarantees Shrek and Deadpool will stop appearing in Midjourney-generated images by the time the case reaches trial. What it guarantees is that the answer will be decided by the studios with the deepest pockets and the clearest evidence, not by the artists whose original copyright complaints started this fight three years ago.

About the Author

Stuart Kerr is Technology Correspondent at LiveAIWire, covering artificial intelligence, emerging technology, and their impact on business, society, and everyday life. LiveAIWire publishes original AI journalism every weekday at liveaiwire.com.