Somewhere in the last year, you’ve probably watched a video, heard a voice, or scrolled past a face and felt a half-second of doubt: is that real? If you genuinely can’t answer that anymore, it changes something about how you watch everything else too, the movies, the television, the ads, the clips your friends send you. That low-grade uncertainty, the one you’ve probably learned to live with by now, just found its way into a courtroom.
In its recent motion, the AI company Midjourney asked a judge to force Disney, Universal, and Warner Bros. to reveal exactly how much AI they use internally, as part of the studios’ own copyright lawsuit against the company. On paper, it’s a legal fight over discovery, the pretrial process where each side has to hand over evidence to the other, not a story about AI actors at all. Underneath it, though, is a much older, more personal question. You’ve probably already asked it yourself, without putting it into words: how many AI actors are already on your screen, and who decides whether you get to know?
The Fight Nobody Asked to Watch
Disney and Universal sued Midjourney back in June 2025, later joined by Warner Bros. Discovery. Their argument: its AI image tools let anyone generate their copyrighted characters on demand, Homer Simpson and Darth Vader among the examples named in the case. In mid-June 2026, a magistrate judge partly sided with the studios. They only had to hand over information about their “consumer-facing” AI tools, meaning anything shipped as a product or feature the public can actually touch. The tools used quietly behind the scenes, to write, storyboard, or market a film, stayed off-limits.
Midjourney’s lawyers want more. Their recent motion, covered by TechCrunch, asks Judge John Kronstadt to overturn that limit. It wants the studios to hand over internal AI business plans, research reports, training datasets, model weights, and even their own board presentations on generative AI. The argument leans on a legal principle called unclean hands. It’s the idea that a party asking a court for help has to have clean hands itself.
Midjourney’s attorneys put it plainly: “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,” according to Gizmodo’s reporting on the filing. Studios’ counsel countered, in Variety’s account, that they aren’t trying to shut Midjourney’s business down, just to stop it from copying their work without permission. Both sides, in other words, agree that AI is already deeply embedded in how movies and shows get made. What they disagree on is how much of that either of them has to admit to the other, let alone to you.
Strip away the legal wrapping and this isn’t really a story about a courtroom filing. It’s the first time studios have been forced to answer, under oath, a question audiences have been asking informally for two years: how much of this is AI, and why won’t you just say so?
You’ve Already Met an AI Actor
The disclosure fight Midjourney just picked isn’t hypothetical. It already played out in public, as one of the most talked-about AI actors debates in recent Hollywood history, and you might have missed it. In September 2025, a fully AI-generated “actress” named Tilly Norwood was introduced to Hollywood, created by Eline Van der Velden, CEO of the European production company Particle6, through its AI division Xicoia. Van der Velden built Norwood a full public presence to match, including a Tilly Norwood Instagram account with a bio describing her as the “world’s first AI Actress.” Talent agencies were reportedly interested in representing her, a detail that alarmed the entertainment industry as much as Norwood’s existence did.
SAG-AFTRA, the actors’ union, wasn’t having it. The union said, as reported by Variety, that Norwood “has no life experience to draw from, no emotion,” and that she was built from the work of professional performers without their permission or pay. Real working actors reacted like the controversy was personal, because to them, it was.
When a Variety interviewer told Emily Blunt that agencies wanted Norwood to be “the next Scarlett Johansson,” Blunt shot back, “but we already have Scarlett Johansson,” before adding, according to Variety, “That’s an AI? Good Lord, we’re screwed. That is really, really scary.” Whoopi Goldberg took a more mixed view, calling Norwood “a little bit of an unfair advantage” since an AI performer can be built from thousands of real actors at once, then landing on “bring it on,” with a confidence that human performers can always be told apart from AI because human faces and bodies “move differently.”
Van der Velden defended her creation directly, writing on Tilly’s Instagram account. Norwood “is not a replacement for a human being, but a creative work, a piece of art,” she wrote, closer to “a new tool, a new paintbrush” than to human acting itself. AI characters, she added, “should be judged as part of their own genre, on their own merits, rather than compared directly with human actors.”
Tilly Norwood was the loud, visible version of this fight. A quieter one has been happening on real film sets for years. Background actors are already having their likenesses digitally scanned under studio contracts, a practice most audiences have never heard of. That’s exactly the kind of AI actor this piece’s title is really about.
Why Disclosure Is the Word Everyone Keeps Using
Disclosure sounds like a simple word. In practice, it’s already splitting into at least three separate fights, and none of them fully covers what you’d actually want to know as a viewer.
- A union contract, already in effect. SAG-AFTRA’s agreement with the major studios requires clear, separate consent before a background actor’s digital replica appears in scenes. The union must be notified 48 hours ahead of the scan, according to Rolling Stone’s breakdown of the contract terms. Those replicas can’t be used to avoid hiring background actors, and the performer still gets paid as though they worked the full day. The same contract lets studios build fully synthetic performers too, AI-generated characters resembling humans, as long as the union is notified. Actress Justine Bateman has criticized that provision for letting productions “use human-looking AI Objects instead of hiring a human actor.”
- A state law. Utah’s AI Policy Act was the first law of its kind in the US, requiring companies to disclose generative AI use in certain consumer interactions. It wasn’t written with Hollywood in mind, but it’s a working model for what a legal disclosure requirement looks like once lawmakers decide people are owed one.
- A courtroom. Midjourney’s discovery push is different again. It isn’t about telling an audience what they’re watching. It’s about forcing studios to admit, internally, how much AI they already use.
Here’s what’s still missing from all three: no rule yet requires a studio to tell you, the person on your couch, that the performance in front of you is AI-generated or a real actor’s digital replica. Every disclosure fight so far is happening in contracts and courtrooms, not on the screen itself.
That’s the same gap Google’s Gemini ran into with its personal intelligence features: disclosure only means something if you get to see it before you need it, not after the fact. YouTube has already tested a version of this for music, building an artist opt-in consent system before use happens, rather than asking forgiveness after.
What Not Knowing Does to How You Watch
Emily Blunt’s word for Tilly Norwood was “terrifying.” That reaction wasn’t really about the technology itself. It was about losing the ability to locate the line between a person and a program. Watching a performance has always depended on trusting there’s a person behind it, someone whose actual choices, mistakes, and lived experience shaped what you’re seeing.
SAG-AFTRA’s own framing gets at exactly this: a performance moves you partly because it comes from someone’s real life, and that’s the one thing an AI-generated performer, by definition, cannot bring to the screen. Even Whoopi Goldberg’s confidence that human performers can always be told apart from AI is really a bet, not a guarantee. That bet gets riskier every time the underlying models improve. It gets riskier, too, every time a background actor’s scan quietly stands in for a body that would once have been hired fresh for the day.
Not everyone will draw the same line, either. Some viewers may not care whether a performance is human, if it moves them anyway. Is that a real shift in what AI in Hollywood means for entertainment? Or does it only start to matter once you find out the truth? Here’s the tension in one sentence: an AI actor can borrow the shape of a thousand real performances, but it can’t borrow the life that made any one of them true, and that difference is exactly what disclosure is supposed to protect.
There’s a version of the same tension in debates over AI and creative jobs, where the debate isn’t whether AI can produce something competent, but whether competent was ever really the point of watching in the first place. Knowing who you’re watching was never just trivia. It’s part of how a performance earns your trust, and that’s the thing actually at stake in every one of these disclosure fights.
What Actually Changes From Here
Midjourney’s motion is still sitting in front of Judge Kronstadt, unresolved. However it’s decided, it will shape how much studios have to reveal about their own AI use going forward, which makes this a case worth watching rather than a settled matter. When a real disclosure requirement for AI performances eventually shows up for the audience, not just for the union or the courts, where will it start? Most likely, exactly where the background-actor contract language already lives: synthetic performers and digital replicas, not a broad catch-all label reading “this scene used AI.”
Expect more states to follow Utah’s lead on AI disclosure in general contexts too, which makes an entertainment-specific version more plausible over time, even without one on the books yet. Watch, too, for how streaming platforms and end credits handle this in the next year or two. A line crediting a “digital performance” or a “synthetic character” would be a small, testable signal that the industry is moving toward telling you something at the point of watching, rather than only in a courtroom or a union filing.
None of this is settled, and nobody’s required to tell you today, whether the AI actors in question are named stars like Tilly Norwood, working alongside human actors on a Hollywood set, or an unnamed background face that never got a callback. So here’s the honest question worth carrying with you the next time something on screen feels a little too perfect: would you actually want a disclosure label on an AI-generated performance, or might you not check it even if it existed? Whatever else this fight resolves, that answer is still yours to give, and Nollytech’s ongoing tech coverage will keep tracking where the rules land next.
