Tilly Norwood just got her big break.
Last week a London studio called Particle6 announced that the AI “actress” it built — a synthetic performer assembled through roughly two thousand iterations until she could hold a face and a line reading — will headline a feature film called Misaligned. The premise is almost too on the nose: an AI with no body, no childhood, and no memories of her own gets seduced by a rogue bot from the dark web into developing human desires, impulses, and ambitions. Particle6’s founder, Eline van der Velden, says the goal is for Tilly to become “the next Scarlett Johansson or Natalie Portman.” SAG-AFTRA fired back that she is not an actor and has “no life experience to draw from.” Emily Blunt called the whole thing terrifying and said, in so many words, that we’re screwed.
The industry is treating this as a technology story, a referendum on whether a machine can act. That is the wrong frame. The interesting part of this story isn’t the software. It’s the arithmetic. Because the honest reason a studio would put a synthetic lead at the top of a call sheet in 2026 has very little to do with what AI can do, and almost everything to do with what human stars have spent the last ten years doing to the people who used to buy tickets.
Start with the numbers, because they are brutal and nobody in Burbank wants to say them out loud. Domestic ticket sales hit an all-time high of roughly $11.9 billion in 2018. Last year the North American box office limped in somewhere around $8.6 to $8.8 billion, more than twenty percent below the pre-pandemic peak of $11.4 billion. And revenue actually flatters the picture, because ticket prices have been climbing the entire time. Strip out price and look at bodies in seats: North American theaters sold about 769 million admissions in 2025 against 1.23 billion in 2019, a drop of roughly thirty-seven percent in six years. Measured against 2002, the domestic movie ticket market has essentially been cut in half. The theaters are charging more money to fewer people every year and calling it a recovery.
Now follow that money down to the people who actually make the movies, because this is where the collapse stops being an abstraction. Los Angeles County had about 142,000 motion picture and sound recording jobs in 2022. By late 2024 that number was down to around 100,000 — roughly 42,000 jobs, nearly a third of the workforce, gone in twenty-four months. Shoot days in the L.A. region fell from 36,792 in 2022 to under 20,000 in 2025. Television production, long the steady paycheck that kept the middle class of this industry alive, dropped fifty-eight percent from its 2021 peak. Writing and acting roles are down close to forty percent from 2022. FilmLA’s own president called 2024 the worst year on record excluding COVID, and then watched the next year come in worse. This is an Oscar-winning sound mixer who can’t get a gig and an animator who worked on Pocahontas sitting at home. This is Detroit, in real time, with better weather.
So where did the audience go? Here is the part that should terrify every studio chief more than any deepfake ever could: they didn’t stop watching. They just stopped watching you. YouTube has been the single most-watched source of streaming in the United States for more than two years running, pulling roughly one out of every eight minutes Americans spend in front of a television. On an average day, people watch more than a billion hours of YouTube on their TV sets alone. Alphabet now books somewhere north of $50 billion a year from YouTube, with more than 125 million people paying for a subscription. In 2025 streaming as a whole finally surpassed broadcast and cable combined for the first time in history. The audience is bigger and hungrier than it has ever been. It is simply spending its attention on a teenager with a ring light in Birmingham, Alabama, instead of a $270 million tentpole from a studio that thinks its customers are a problem to be managed.
And a lot of them didn’t drift away. They were pushed. Because for about a decade now, a remarkable number of the biggest names in the business have used their press tours to tell their own paying customers, more or less directly, that the movie wasn’t for them.
Consider Disney’s live-action Snow White. It cost somewhere between $240 and $270 million to make, with total costs reported at $336.5 million once you count everything, and it grossed a grand total of about $205.7 million worldwide. Forbes pegged Disney’s loss at nearly $170 million. Three days after the trailer dropped in August 2024, its lead, Rachel Zegler, capped a promotional post with “and always remember, free Palestine.” According to Variety, the film’s producer, Marc Platt, flew across the country to ask her to take it down. She didn’t. His own son, the podcast host Jonah Platt, later said publicly that her decision to drag her personal politics into the middle of a $250 million production hurt the film and, in his words, put at risk “all the colleagues and crew and blue-collar workers who depend on that movie to be successful.” Meanwhile Zegler and her Israeli co-star Gal Gadot — a former IDF soldier — were so estranged that they ran entirely separate press tours. Audiences got the message and stayed home, and the crew ate the loss.
Consider this summer’s Supergirl. Its star, Milly Alcock, spent the press run musing to a queer outlet during Pride month that her version of the character “probably goes both ways,” and joking that she found it “hilarious” when the people criticizing her identified themselves online as “Dad of four, Christian.” She was, in other words, on record laughing at a slice of the exact audience a Superman-adjacent tentpole needs to survive. The film opened to about $38 million and is now tracking toward a loss of roughly $200 million for Warner Bros. Little wonder the character had lukewarm interest to begin with — but you do not fix a thin premise by publicly needling the ticket buyers.
Consider The Marvels, which opened to $47 million, the worst debut in the history of the Marvel Cinematic Universe, against a budget north of $200 million, and lost a fortune. Its lead, Brie Larson, had spent years telling interviewers she didn’t “need a 40-year-old white dude to tell me what didn’t work for him about A Wrinkle in Time,” and that she wanted her press days to be less “overwhelmingly white male.” To be fair to Larson, the movie also suffered from a strike that gutted its promotion and from a general audience fatigue with the franchise, and the ugliest “men need not attend” line often pinned on her actually came from an anonymous troll review, not from her. But the pattern is the pattern. When your marketing message, even indirectly, is that a large chunk of your prospective audience is the problem, that audience has a simple and devastating response available to it: it can decline to buy a ticket.
None of this is monocausal, and anyone who tells you “get woke, go broke” explains the whole box office is selling you a bumper sticker. Streaming, ticket prices, franchise exhaustion, the strikes, runaway production chasing foreign tax credits — all of it is real, and all of it matters. But the alienation factor is real too, and it is the one the industry is least willing to name, because naming it would require the most powerful people in Hollywood to admit that their most valuable assets have been actively repelling customers on camera and getting rewarded for it.
And here is the part I know from the inside, not from a spreadsheet. This priority, product over star, did not arrive with AI. Back in 2008 I was working on the disaster film 2012 on the Sony lot in Culver City, in the visual effects department. It was a $200 million picture, and the overwhelming majority of that money went to us, to the effects, not to the cast. The real stars of that film were the megatsunamis and the collapsing freeways, and everyone on the production knew it. The producers deliberately hired capable, mid-tier actors precisely so they could free up budget for the spectacle on screen, because the spectacle was the product people were buying a ticket to see. That was seventeen years ago, long before a single generative model could render a frame. Even then, with human actors as the only option on the table, the studio had already decided that a better product mattered more than a more expensive name above the title. AI didn’t invent that calculation. It just removed the last reason to pay for the name at all.
Which brings us back to Tilly Norwood, and to why the actors’ own argument against her has quietly collapsed.
For years the objection to AI performers was about theft: they’re trained on our faces, our voices, our craft, without consent or compensation. It’s a legitimate grievance, and the likeness-and-consent fight is a fight worth having. But underneath it was an unspoken assumption — that audiences desperately want the human star, that the face is the irreplaceable asset. The last decade of data quietly demolished that assumption. It turns out audiences are not paying $15 for the privilege of a specific celebrity’s opinions about them. A synthetic lead will never post “free Palestine” three days into a press cycle. She will never call your subscribers deplorables, never sneer at the dads in the third row, never blow up a $300 million production over a tweet. She has no agent leaking to the trades, no memoir, no cause, no contempt. From a studio’s point of view she is not a threat to the art. She is a risk-management instrument that happens to be able to hold a close-up.
That is the actual pitch, stripped of the hand-wringing. Cast the AI and the bad press goes away, because there is no human being attached to alienate half your ticket buyers on a Tuesday. The character is a blank, apolitical vessel that shows up, hits its marks, and never once tells the people funding it that they aren’t welcome. For a business that has spent a decade proving it cannot stop its own biggest names from insulting the customer, that is not a gimmick. It is a cheat code.
And casting Tilly is only the first move. A synthetic lead inside a traditional studio production is a transitional technology, the horseless-carriage stage, where the new thing still wears the shape of the old one to make everyone comfortable. The real disruption isn’t an AI actor in your movie. It’s the moment the studio stops being necessary at all, because the movie is generated on demand, for you, in about the time it takes to make a cup of coffee.
Picture it a few years out, and it is not a big leap. You keep a library of your favorite AI characters the way you keep a playlist today. Some you discovered, some you built, some you licensed from the people who did. You drop onto the couch and tell your television you want a ninety-minute action movie starring two of them, set in a rain-soaked city, in the tone of a nineties thriller. Then you wait. Behind the glass, an orchestration layer goes to work. One agent drafts the plot and the three-act structure and hands the beat sheet to a second agent that writes the scene-by-scene script and the dialogue. A continuity agent tracks who knows what, which wound is on which arm, and whether the sun is still setting in scene forty. Other agents handle shot composition, the voices, and the score, and a director agent holds the entire thing to the tone you asked for. A minute later it starts streaming. Don’t like the ending? Ask for a different one and get it in seconds. Want the same two characters in a rom-com instead? Say so, and watch them fall in love by the time you’ve refilled your glass.
When that lands, and the pieces are already being assembled one capability at a time, the whole economic structure of this business inverts. The unit of production collapses from a studio to a sentence. The audience becomes the director. And the alienation problem that has been quietly bleeding Hollywood out for a decade doesn’t just get managed, the way casting Tilly manages it. It becomes structurally impossible. A film generated for an audience of one, starring characters that viewer personally chose, has no press tour to sabotage, no cause to broadcast, and nobody on screen who can turn around and inform the person watching that the movie was never meant for them. There is no green room left to say the wrong thing in. There is only the story you asked for.
If that future sounds far off, consider that audiences are already living in a crude version of it, and consider why they wanted it badly enough to build it by hand. Because the alienation problem does not live only with the actors. The people running the franchises have their own agenda, and Star Wars is the clearest autopsy. Lucasfilm’s president, Kathleen Kennedy, was famously photographed in a shirt reading “The Force is female,” a phrase that actually began life as a Nike campaign but that fans read, not unreasonably, as a statement of intent for where the saga was headed. Leslye Headland, who would go on to run the reported $180-million-plus series The Acolyte, once held up George Lucas himself, the man who invented the entire universe, as her example of “the misogyny and the problem with the auteur myth,” arguing that he should not be “the only one holding the key” to Star Wars. And then there was what they did to Luke Skywalker, the hero of the original trilogy, rewritten in The Last Jedi as a broken, bitter hermit who at his lowest contemplated murdering his sleeping nephew. The character was so unrecognizable that Mark Hamill, who had played him for four decades, said out loud that “Luke would never say that,” and that this was “Jake Skywalker,” not his Luke. He later made his peace with it in public, the way an actor protecting his colleagues does, but that gut reaction was the tell. The people entrusted with these stories were quietly dismantling the very things audiences loved about them.
So the audience did what audiences now do. It left, and it started making its own. Search YouTube and you will find a sprawling universe of Star Wars fan films, a good number of them shot and cut to a standard that embarrasses the official product, pulling millions of views. What is striking is what they leave out. They draw on the original trilogy, the prequels, the Jedi and the Sith and the Clone Wars, the lore people actually love. They overwhelmingly ignore the sequel-era canon, and you can look for a very long time before you find one built around the Rey character. Handed the tools and the freedom, fans did not ask Disney for permission. They routed around it completely and made the Star Wars they wanted to watch.
That is not a fringe curiosity. It is a live preview of exactly the future I described a few paragraphs ago, running years ahead of the technology that will make it effortless. Today, making a fan film takes a crew, a camera, months of nights and weekends, and real talent. When the generative tools mature, all of that collapses into a sentence spoken at a television. And the lesson Star Wars keeps teaching is blunt: audiences will abandon even a sacred franchise the moment its custodians decide the story is a delivery vehicle for a message nobody asked for, and given any means at all, they will build their own version and never look back. Make-your-own-movies is not a threat the industry can out-market. For a growing slice of the audience it is already the preferred product. The only thing missing is the button, and the button is coming.
So congratulations to Eline van der Velden and the team at Particle6. Not for the technology, impressive as it is, but for reading the room the entire industry refused to read. They looked at a decade of losses, a workforce cut to the bone, an audience that fled to YouTube, and a star system that treats its own customers like an inconvenience, and they drew the obvious conclusion: maybe the safest lead is one who has no politics to broadcast and no audience to insult.
For ten years the most bankable names in the business told their own paying customers, in one form or another, that the film wasn’t for them. The customers finally believed them. Now the studios are building stars who will never say it. The tables didn’t turn on the grip, the sound mixer, or the working actor grinding out a living — those people are the collateral damage of executives who kept mistaking a mirror for a movie. The tables turned on a single, expensive idea: that a star can hold an audience hostage to their politics and still expect that audience to show up. That idea just got written out of its own franchise, and it isn’t getting a sequel.
About the Author
Gal Ratner is the founder and CTO of Inverted Software and WhiteStar Labs and Chief Architect at Prana Entertainment, a Las Vegas enterprise software and AI consultancy. He has spent nearly thirty years shipping production systems on the Microsoft and .NET stack for clients including Microsoft, Sony, Rockstar Games, Best Buy, and Allegiant Air, and was a Los Angeles Business Journal CTO of the Year finalist. He writes about the gap between AI hype and production reality on Substack as “the .NET AI Guy That Ships.” When he isn’t writing or building, he trains Brazilian jiu-jitsu, rides motorcycles, and co-hosts Edge Grip Podcast.
Great article. I think this technology will also level the playing field for creators who never had that Hollywood connection, who never had access to the finances. It'll give the little guys a shot at making a blockbuster.