A few years ago, the idea that a feature could be generated — written, shot and “acted” — largely by artificial intelligence felt like science fiction. Now, OpenAI’s Sora 2 can spit out hyper-real video from a prompt, “AI talent” like Tilly Norwood is courting representation, and the entire industry feels like one update away from apocalypse.
“I got a full treatment which was: ‘We’re going to make a 90-minute movie all with AI, will you finance it for a couple of million bucks?” says Guy Danella, president of film at indie genre stalwart XYZ Films (The Raid, Skylines). “What if we said yes? What are the implications?” he asks, “That was really the turning point for me. I call it Skynet cinema, because it literally feels like we’re coming for the humans.”
Bryn Mooser, head of nonfiction studio XTR, agrees the indie industry may be facing its judgment day, but he doesn’t think it needs to be a Terminator-style event. Mooser, through XTR, has spun up AI animation group Asteria with a slate that includes projects from Natasha Lyonne and Toy Story 4 writer Will McCormack.
“There is a confluence of two technologies happening right now that are enabling this massive transformation. One is that NVIDIA chips are making rendering really fast, even real time. That’s not even an AI thing, it’s just a transformational moment in what it costs to render,” says Mooser. “The second real revolution is in training custom AI models so they can become an extension of the creative team’s hand, compressing the time it takes to do storyboards or previs, animatics, backgrounds, whatever.”
For the studios, Mooser argues, the combination of ultra-fast rendering and custom models will make big projects cheaper and faster. His pitch with Asteria is that the same combination, in the indie space, will make projects possible that would have been out of reach before.
“This shouldn’t be about how you can make Anora cheaper, which I don’t think AI could do anyway, but about how can you help indie filmmakers to make their projects bigger and get them done on a budget,” he says. “What we have with the potential of AI is the democratization of studio-level films.”
An indie-financed $80 million animated feature is a non-starter. The same project, made with AI tools for under $10 million, says Mooser, could be a viable proposition.
“Is there a way to use [AI] as a tool to make x amount of savings so we can get another day of shooting into our budget?” asks XYZ Film’s Danella, “That’s the sort of conversation that makes more sense.”
Asteria’s approach — developing bespoke AI models in collaboration with creators and training them on licensed or original material — stands in sharp contrast to the ChatGPT/Sora 2 opt-out model of scraping existing IP, a practice studios and agencies have warned could breach long-standing intellectual property protections. This week, a court in Munich ruled that ChatGPT violated German copyright laws by reproducing lyrics from nine popular songs, a landmark win for rights society GEMA that could set the tone for European litigation over AI training data. OpenAI is appealing the ruling.
Contractually, buyers are catching up. At AFM, international buyers and sales agents say disclosure language is creeping into reps-and-warranties: Was AI used, where, and on what datasets? “It’s just insurance against possible future legislation,” notes one sales rep.
That growing paper trail reflects a deeper anxiety running through the business: That the real battleground over AI isn’t creative but legal. After years of data scraping done in the shadows, IP protection has become the industry’s new red line.
“Our whole business is predicated on IP. If you do not protect the IP, no more business,” says Darren Frankel, who oversees AI initiatives at Adobe. The argument shouldn’t be “for” or “against” AI, but “ethical versus unregulated,” he says. But if the entertainment industry waits for the government legislation, “[you’ll] be waiting a while,” says Frankel. His advice is to sue early and sue often.
“I spoke at the DGA, and what I opened with was that four companies: Amazon, Meta, Microsoft and Google are projected to spend $364 billion this year alone, most of that in AI infrastructure. Now look at the size of our industry. Where do you think we fit in that arena? Basically, that’s my passive-aggressive way of saying: You’re not going to hold that tide back. So how do you fight the good fight? How do you pick what to target?”
The AI-led disruption of the film industry, Mooser argues, is inevitable. “Technicolor shut down before AI even had its effect,” he notes, “we actually haven’t seen the AI effect on the industry inside of VFX or animation. [But] the question is: Do you fight for an old way of doing something, or do you fight really hard to build opportunities for the next generation of this industry?”
Danella, for one, isn’t abandoning his flesh-and-blood sets and human-focused filmmaking.
“Fundamentally, I believe in the human flaws, in the good and bad that comes with human filmmaking,” he says, “The hope is that there’s a way to have synergy, to make better movies creatively with more humans working more days by embracing a component of AI. “
Says Frankel: “If you don’t have that humanity in it, it’ll be all frosting and no cake.”
