AI in Film Divides Hollywood as Studios Shift from Clip Generation to Full Production Workflows

AI video is moving beyond clip slop

This is Lowpass by Janko Roettgers, a newsletter on the ever-evolving intersection of tech and entertainment, syndicated just for The Verge subscribers once a week. Hollywood is cooked -- or so a growing number of people on social media would like you to believe. Their purported proof: AI-generated clips of Daniel Craig riding a Vespa through an Italian city, Godzilla fighting King Kong, or The Avengers zooming through Manhattan. In reality, cheap slop like this won't replace Hollywood blockbusters any time soon. However, a new generation of AI video solutions could upend how studios work. That's because, until recently, AI companies basically tried to sell Hollywood on the same idea as those Twitter guys, with a slightly more palpable spin. The pitch, in a nutshell: AI video will allow everyone to make movies faster, cheaper, better -- one prompt at a time. "The premise was: Substitute your camera for our video model," says Luma AI CEO Amit Jain, whose company used to make that very same pitch to studios. But when it began partnering with the entertainment industry, it received a crash course in the way Hollywood actually works. "It's not sufficient to just produce a clip," Jain says now. "Because then what?" Clips generated by video models are typically 10 to 16 seconds. "That's not a shot. That's not a sequence. That's not a scene," Jain says. "Churning out short videos is not enough." Now, AI companies like Luma believe they have found a better way to sell Hollywood on AI. The gist? Don't just use AI for video -- use it for everything. Luma has been working on AI agents that can help with the entire production process. Jain compares this transition to the way software development with AI has evolved, with companies like Anthropic moving from simple vibe coding to agentic workflows. "It's not sufficient to just generate a little bit of code," Jain says. "We need these systems to do long-horizon, end-to-end work. That's what solves problems for people." AI agents can do the same for Hollywood, he believes. Luma isn't alone with that approach. Just this week, Google unveiled a new version of its AI media authoring platform Flow that also emphasizes agentic end-to-end work over simple clip generation. "There's this huge evolution that's happening in generative tools," says Google Labs VP Elias Roman. "Moving forward, they're going to become much more like agents." In the new version of Flow, an agent guides the user through multiple steps, from starting with a concept to fleshing out plotlines to developing characters to setting the desired look and feel. And when it's ultimately time to generate video, the agent uses the things it learned along the way to achieve a specific outcome without having to be prompted about every single detail. One issue this is aiming to solve is consistency. Generative AI has long struggled with keeping characters looking the same from clip to clip. In the new version of Flow, users can add a character they developed for a project to a prompt simply by tagging it, just like you'd add a colleague to a Slack conversation. A new generation of video models is also better at understanding physics, the look of a certain era, and cinematic languages. Google's Flow is powered by the company's new Gemini Omni world model, while Luma has developed Uni-1 as a unified model that doesn't need extremely complex prompts anymore to make sense of an envisioned world. Luma recently teamed up with Amazon to produce The Old Stories: Moses, a companion special for MGM's House of David show. While shooting Moses, actors would perform in front of LED walls showing backgrounds generated with Luma's video models, while their costumes were rendered with AI as well. If a shot turned out to look not quite right, all it took was one new prompt to generate a new asset. "This level of production would take about six weeks to eight weeks per hour of television," Jain says. "Now, it's taking them a week." Some studios are increasingly embracing that change. Netflix acquired Ben Affleck's AI company InterPositive in March, and launched its own AI animation studio the same month. Two major Hollywood studios already use Luma's AI agents, Jain claims. He declined to name names, but the company has been publicly celebrating some smaller wins: Luma recently announced the launch of a joint venture with indie studio Wonder Project, which made Moses. These developments undoubtedly will lead to job losses, even though the scale of the impact is still unknown. If studios can make a TV show in a month instead of 10, it won't be sending out checks for those other nine months. The counterpoint AI boosters like to raise is that this will lead to more productions. This could be a silver lining for Los Angeles in particular, which has seen production days plummet in recent years. Now, we'll just have to see if Hollywood uses this tech for something people actually want to watch.

Most mainstream films already use AI. The new Oscars rules won't stop that

The Academy of Motion Picture Arts and Sciences has adjusted the eligibility criteria for films vying for Oscars from 2027 onward. Films featuring actors generated by artificial intelligence (AI) are now ineligible, as are scripts that aren't demonstrably human-authored. Crucially, the rules do not ban AI - generative or otherwise - altogether. The Academy explicitly acknowledged the widespread adoption of generative AI, and has left it to voters to determine whether a film's creative direction is substantively driven by humans. Academy president Lynette Howell Taylor framed it simply: "humans have to be at the centre of the creative process". The rules were imposed following specific controversies: the 2025 awards season surfaced AI voice modification in The Brutalist, AI voice cloning in Emilia Pérez, and varying degrees of AI use in A Complete Unknown and Dune: Part Two. The resulting public debate has focused almost entirely on what audiences can see: generated faces, synthetic voices and digital resurrections. But this focus ignores the main areas of film production in which AI actually plays a key role. Cinema's automation history Automation tools have been embedded in cinema for longer than most people realise. When non-linear editing software Avid Media Composer launched in 1989, it replaced the physical cut-and-splice flatbed editing process that had defined post-production for decades. This all but eliminated traditional linear editing suites, and the skilled labour that went with them, within a few years. Computer-generated imagery (CGI) entered mainstream production with Tron (1982) and Jurassic Park (1993). And by the mid-2010s, deepfake and neural rendering technology (distinct from generative AI) made it possible to manipulate human likenesses directly - such as de-aging Jeff Bridges in the Tron sequel, Tron: Legacy (2010), and Michael Douglas and Kurt Russell in the Marvel Cinematic Universe. Generative AI aids and improves these existing processes. Director Robert Zemeckis used AI company Metaphysic to de-age Tom Hanks and Robin Wright in real time, on camera while shooting Here (2024). And the forthcoming film As Deep as the Grave will feature an AI-generated likeness of Val Kilmer, who died in 2025. This is the most prominent case yet of AI being used to "generate" a performer who couldn't physically be present. These examples sit on an ethical spectrum, from de-aging a living actor to synthesising a performance based on a deceased star. The Academy's new rules attempt to draw a line somewhere on that spectrum. Where exactly that line should go is an ongoing question. Where AI is usually used in film The most pervasive uses of AI in contemporary mainstream productions are usually invisible to audiences. For most studio-produced feature films, AI-assisted platforms handle script breakdown in pre-production. They help with extracting production requirements based on the screenplay, scheduling shoots and modelling budgets. These tasks once took hours, or even days, to complete. On set, AI may be used to generate the background environments for virtual production stages. Originally used for TV shows such as The Mandalorian, these sets are increasingly used as a cost-effective option in studio productions. In post-production, AI tools handle first-pass editing, audio clean-up, VFX, and voice modification. For instance, Dune: Part Two (2024) used machine learning to achieve the blue-eye effect - a minor enhancement audiences don't immediately register. Arguably the most widespread (yet invisible) use of AI is for localisation, as streaming platforms pursue global audiences at scale. AI-assisted dubbing and subtitle generation are now standard processes. AI can also influence what films get made. Companies use algorithmic analysis to inform greenlight decisions, predict trailer performance, and optimise release windows. As Vue Cinemas CEO Tim Richards put it, "[AI] determines what we show at what cinema, on what screen and at what time". Automation and AI-driven processes have a long history in film. Yet it is only now, amidst the deluge of AI-generated content, that these modes of filmmaking have attracted the public's attention. Not a new problem, but a newly visible one The Academy's changes do not reflect a resistance to innovation - but a centring of human authorship. The most publicly legible creative contributions to a film should remain demonstrably human in origin. That's a reasonable position, with broad support across audiences and industry professionals alike. But it sits alongside the harder question of consent: not just whether AI was used, but whether the people whose likenesses, voices, and labour are implicated actually agreed to it. The Oscars' rules are about disclosure as much as eligibility - and they aren't alone. Many publishers and creative bodies are now formulating AI principles that emphasise the need for transparency. The question isn't whether AI is being used in film and TV production; it is. The question is whether audiences have the right to know -- and whether public discourse has caught up enough to know what it's even asking about. Generative AI is the latest inflection in a long history of automation -- more powerful and fast-moving than its predecessors, but no different in the questions it raises. The pressing task now is making sure the frameworks for consent, disclosure and creative attribution keep pace with AI technology.

A filmmaking tool or an existential threat: Cannes Film Festival weighs the rise of AI

CANNES, France (AP) -- The Cannes Film Festival can function like a global water cooler for movies, with prevailing issues and anxieties tending to come to the surface at the event. This year, the topic du jour is artificial intelligence. The 79th Cannes may go down as the time the world's grandest film festival for the first time wrestled with the onset of AI -- its arrival has been felt like a tsunami on the French Riviera. Its potential to remake the movie industry, for good or bad, has been an ongoing debate since the festival opened. And in many quarters, the tone is softening. "The buzz in Cannes and the buzz in the industry, it does feel like it's definitely a turning point," said Scott Mann, co-chief executive of Flawless, a company that specializes in assistive AI programs for post-production. On screen and off, AI is much more present. For the first time, Cannes has partnered with Meta in a new multiyear deal. The company has set up camp at the Majestic Hotel. And its AI tools were used to help produce a festival entry: Steven Soderbergh's "John Lennon: The Last Interview." The documentary is about a lengthy and insightful interview Lennon and Yoko Ono gave on the day Lennon was shot and killed in 1980. To add imagery to match Lennon's conversation, Soderbergh used Meta's AI programs to create surreal graphics. The choice brought scorn from most critics in Cannes, but Soderbergh, a highly skilled innovator who has shot movies on iPhones, believes its time for such experimentation. "We haven't seen yet someone with a certain amount of creative credibility go full-metal AI on something, and see how people react. I think it's necessary," Soderbergh said in an interview. "How do you know where the line is until somebody crosses it? I don't think what I'm doing crosses it. Some people may disagree. I don't know where my line is yet. I'm waiting to see." Filmmakers, actors and others at Cannes have been drawing their own lines, or at least making pronouncements about AI. On opening day, Demi Moore, a juror, said fighting AI "is a battle we will lose." The next day, honorary Palme d'Or recipient Peter Jackson, said: "I don't dislike it at all. To me, it's just a special effect. It's no different from other special effects." Filmmaker James Gray, whose starry family drama "Paper Tiger" was one of the standouts over the weekend, said he's not worried. "In some cases, it can be a very helpful tool," said Gray in an interview. "I don't think in our lifetime, or even our children's lifetimes, it will come close to mirroring the only true infinite we know, which is the soul." "The answer I think is that most young people should be studying the humanities," added Gray. "People should be reading Tolstoy in their spare time to understand the human soul." Cannes is unfolding in the wake of some significant new developments for AI in Hollywood. Earlier this month, the Academy of Motion Picture Arts and Science announced new guidelines, ruling that only performances "demonstrably performed by humans with their consent" will be considered for acting nominations. At the same time, the Oscar group also said AI tools "neither help nor harm the chance of a nomination." The Screen Actors Guild-American Federation of Television and Radio Artists recently reached a tentative agreement with studios detailing and clarifying AI guardrails for things like the use of digital replicas and synthetic performers. Some advancements have sent shudders through Hollywood. The unveiling of Tilly Norwood, an entirely AI-created so-called "actress," sparked outrage through the industry. Earlier this year, the first look at a posthumous AI resurrection of Val Kilmer, for a film made with his family's consent, spawned another round of debate. But while more extreme uses of generative AI continue to prompt worry, other iterations continue to make inroads. "It is going to be a part of our business," Kent Sanderson, Bleecker Street chief executive, said in a panel discussion. "It is going to lower production costs, and yes, you probably will be able to make something that looks like a Marvel movie in your basement in a couple of years." While Cannes has stringent rules for what can and can't be worn on its red carpet, it's issuing no decrees banning AI from film selections -- for now. The day before the festival began, Cannes' artistic director Thierry Frémaux wryly responded to a question about AI, noting that he had also heard James Cameron had used special effects for "Avatar." "What I can say with certainty in relation to artificial intelligence is that we are on the side of the artists, the screenwriters, actors and voice actors," said Frémaux. "We stand with everyone whose job could be negatively impacted by artificial intelligence. It requires legislation. We need to control this." Mann, the Flawless executive, was sitting on the Cannes beach outside a party his company was throwing in one of the seaside clubs that regularly host movie after-parties. Since 2019, Flawless has set out to demonstrate that AI can be used thoughtfully. Unlicensed generative AI is bad, he states unequivocally. "But what we've found is that the way people don't understand is part of the problem. AI as a term is seen as a catchall, but it's not that simple," says Scott. "The truth is, our industry needs saving. It needs a technological evolution, and this is offering it."

'We're expanding the cinematic toolbox': AI fault lines on show at Cannes

Darren Aronofsky among proponents of using technology, while Guillermo del Toro says he would 'rather die' Under a white marquee on Cannes' Croisette beach, with the Mediterranean glistening behind him and superyachts drifting across the horizon, the director Darren Aronofsky addressed an audience of executives and tech evangelists gathered for an "AI for Talent" summit. "There's so much pushback against AI," said Aronofsky, who has faced criticism over his embrace of generative AI projects though his new studio, Primordial Soup, at a time when artificial intelligence has become one of the film industry's most divisive fault lines. "AI is a terrible word, because it's a catchphrase for so many different things," continued the director of Requiem for a Dream, The Wrestler, and Black Swan. "The thing we deal with when we're talking to Chat GPT about the weather, or how to spend three days in Cannes, is very different to the AI we're using to generate images. It's not impersonating a person, it's actually a tool." If Cannes is a barometer for the film industry's anxieties and obsessions, this year the subject of AI dominated more than any other. From beachside summits and yacht parties to press conferences, leading figures debated whether AI was cinema's next creative revolution or an existential threat to film-makers. Aronofsky's Primordial Soup has partnered with Google DeepMind on projects including Dustin Yellin's short Goodnight Lamby, which premiered in Cannes. He argued that the technology could solve practical and ethical production problems, citing one project in which AI tools allowed film-makers to avoid using a real newborn baby on set by digitally transforming what an actor was holding into "a live baby". "None of these movies would exist without this technology," he said. "They're not replacing anything, they're purely additive." Elsewhere in Cannes, AI startups and studios competed to position themselves at the forefront of Hollywood's next transformation. During an event hosted on a yacht by the generative video platform Higgsfield, the film-maker Chuck Russell unveiled two AI-driven sci-fi features by his company Neumorphic AI. "AI technologies are expanding the cinematic toolbox to a scale we've never had before," he said. AI also became a talking point because of Oscar-winner Steven Soderbergh's new documentary, John Lennon: The Last Interview. Created in partnership with Meta, the film reconstructs Lennon and Yoko Ono's final radio conversation on 8 December 1980, using AI for about 10% of its imagery. Soderbergh described the stylised sequences - including crying infants in 1960s clothing and cavemen acting out Lennon's reflections on masculinity - as "thematic surrealism", insisting they were intended as metaphor rather than deception. "It's essentially in the way that you would use VFX or CGI or any sort of non-photographic technology," he said recently. "My moral obligation is honesty in how we achieved certain things." But the divisions over AI now run across the industry. Guillermo del Toro recently said he would "rather die" than use AI in his films, while others such as Reese Witherspoon have invested heavily in AI storytelling tools. The late Val Kilmer recently appeared posthumously in a trailer using an AI-generated recreation of his performance. Studio executives have said hybrid AI productions could allow several mid-budget films to be made for the price of a single blockbuster. Last year, the unveiling of Tilly Norwood, an AI-generated "actress" marketed as a potential Hollywood star, prompted backlash from actors and unions. While films primarily generated by AI have been banned from Cannes' main competition, leading figures urged the industry to adapt. Aronofsky described fears that computers would soon replace human storytellers as "science fiction", and framed the technology as part of cinema's long technological evolution, comparing it to the arrival of sound, portable cameras and visual effects, and said it could create countless jobs. "A Guillermo del Toro, a Leonardo DiCaprio movie at Imax will always exist, and I will continue to make movies like that," he said. "Storytelling is not going away. These tools are hopefully going to make it easier for many new storytellers to tell stories and connect." Serving on this year's competition jury, Demi Moore said: "AI is here, to fight it is a battle that we will lose." Moore acknowledged concerns about protecting artists, but insisted technology could never replace the "human soul and spirit" at the centre of film-making. Peter Jackson, who received an honorary Palme d'Or, defended more limited uses of AI, comparing it to the stop-motion techniques pioneered in early cinema. "AI used in the right way, it's just a tool like any other tool," he said. "But like anything, it's going to come down to the imagination and originality of the person feeding the instructions into the AI program. Is it actually interesting? Is it funny? Is it imaginative?" Meanwhile Seth Rogen dismissed the idea of AI-assisted screenwriting while promoting his animated film Tangles. "If your instinct is to use AI, you shouldn't be a writer," he said, mocking the flood of AI-generated clips online as "the most stupid dog shit I've ever seen in my life". Dominic Lees, leader of the Synthetic Media Research Network, said Cannes had "embraced" the AI controversy for several years. In 2024, the competition film Emilia Pérez used AI voice modification technology to extend the vocal range of Karla Sofía Gascón. "And the American Academy of Motion Picture Arts and Sciences is trying to steady the boat after major rows over the use of AI to improve Adrien Brody's Hungarian accent in The Brutalist, but it is making a mess of it," Lees added. "New rules say that acting must be 'demonstrably performed by humans' - but no one knows if the tweaking of Brody's accent would have lost him the Oscar that he won for the film."

Cannes Film Festival Says the Wall Street Journal Is Wrong, and It's Not Debuting an AI-Generated Feature Film This Week

Can't-miss innovations from the bleeding edge of science and tech This week, the Wall Street Journal ran a provocative story claiming that a fully AI-generated movie called "Hell Grind" was being screened at the iconic Cannes Film Festival, dropping a nuclear bomb in the middle of the already-heated debate on the tech's intrusion into the art and business of cinema. "Four street thieves are on the road to hell, literally, in an action-adventure movie debuting at the Cannes Film Festival Thursday," the newspaper wrote. "But what's compelling about 'Hell Grind' isn't the campy plot: It's that every character, setting and prop in the 95-minute movie was generated by AI." But we couldn't find the AI movie on the official schedule of the prestigious event, which is held every year on the French Riviera. So we reached out to the organizers of the festival, who denied that they're showing it at all, saying instead that the film was presented in a third-party screening at a local theater in the town of Cannes. "We can confirm that 'Hell Grind' was not screened as part of the official Festival de Cannes program," a Cannes spokesperson told us. "As publicly reported by Screen Daily and other media outlets, the project was presented during an industry event organized by third parties in Cannes." The company behind the film, Higgsfield AI, doesn't seem interested in clarifying the confusion. When the company's founder Alex Mashrabov posted on LinkedIn this week that "we just premiered at Cannes our first 95-minute feature film" -- and boasted that "for decades, Cannes has been the room where new cinema gets legitimized" -- certain reactions were sharp. "This isn't screening at the Festival de Cannes, which is what you're implying," director John Washburn shot back in the replies. "The [Cinéma Olympia] is a movie theatre that happens to be in the town of Cannes but it isn't a venue for the festival. The suggestion that paying for a screening at some random theatre in the same town and at the same time as a major festival is somehow the same thing as being selected by that festival -- the actual 'room' where new cinema gets 'legitimized' -- is misleading at best. Spurious bullshittery, really." Neither the WSJ nor Higgsfield replied to requests for comment, but we'll update if they do. This kind of chicanery is not an uncommon tactic for filmmakers trying to hijack some of the buzz of one of the most paid-attention-to events in cinema every year. But it's also emblematic of the misleading hype that AI companies feed off, making grand claims and awing the public with exhibitions that aren't quite what they seem. That a major newspaper bought the hype and reported it was showing at Cannes when it wasn't is a testament to where things are at. The tentacles of AI have ensnared themselves in all manner of industries, but the tech has an especially strong hold on film. AI video generators promise to upend traditional modes of filmmaking as we know it, a sentiment glibly expressed by the common AI bro refrain that "Hollywood is cooked." No longer will you need expensive actors, cameras, or sets to make that story idea you've had rolling around in your head since you were a teenager a reality. You just need to know what to whisper to an AI model. And since film is a popular medium, it's easy for AI boosters to show something that on a surface level appears to transform it. Earlier this year, a purportedly AI-generated video of Tom Cruise and Brad Pitt fighting on a rooftop went viral for its blockbuster-like quality that had even Marvel screenwriters prophesying the death of the industry. It turned out, however, that the video was an AI-reskin of existing footage of two human performers fighting in front of a green screen. In other words, it was pure theater. The creators of "Hell Grind" are also making striking claims. The film supposedly took just two weeks and $500,000 to make, per the WSJ. A full 80 percent of that went to compute costs; its creators describe it as a proof-of-concept for how AI can empower creatives who wouldn't otherwise have the means to make a movie, a common pro-AI spiel. Of course, given the specious marketing around it being shown in Cannes, all of these should be taken with a grain of salt. "The main aim as a filmmaker is I just wanted to tell stories," Adilet Abish, an in-house director and creative producer at Higgsfield who worked on the film, told the WSJ. "This is the case where AI can give you the tool to show the world your story." Right off the bat, it's clear that "Hell Grind" isn't the type of flick designed to garner awards at prestigious festivals like Cannes. It's a 95-minute sci-fi action movie heavy on epic slow-mo shots and irreverent dialog -- or that's what the three minute trailer Higgsfield released this week suggests, at least. It follows a perennially blood-soaked dude named "Roco" and three other street thieves we care about whose "heist goes catastrophically wrong when an ancient artifact tears one of them into the underworld." Battles against demons with overwhelming odds ensue. "Hell Grind," in other words, is exactly the kind of cheesy spectacle you'd expect AI bros to make. While the visuals are impressively shiny at times, they can't cover up the aesthetic predilections of generative AI or the people who use it. Roco's love interest looks like every other "photorealistic" but bordering-on-anime AI-generated waifu, less a character and more an amalgamation of attractive features that an algorithm averaged together. Ditto for the generic demon antagonists. Its creators at Higgsfield insist that a lot of work and skill goes into making an AI film. The company makes a tool that harnesses other video generating models like Google's Veo 3 to fine tune their outputs and ensure their images remain consistent across the thousands of short, 15-second clips they churn out that the tool later stitches together. "You have to understand camera composition, which shots are changed. Like you can't have two close-ups back to back, you have to start with an establishing shot," Adil Alimzhanov, a content lead at Higgsfield, told the WSJ. "You still need those filmmaking skills." It's entering a polarizing atmosphere on AI in the industry broadly, and at Cannes specifically. Some big names at the festival have come out in support. "The Substance" star Demi Moore declared that "AI was here," calling on filmmakers to "find ways that we can work with it." Nicolas Winding Refn, the director of "Drive" (2011) and this year's "Her Private Hell," compared AI to a painter's brush, saying it came down to the artist to figure "what you're going to do with it." Others were fiercely opposed. "AI doesn't have a chance," actress Tilda Swinton said, adding: "Humans make cinema, right?" Guillermo Del Toro, the director of "Pan's Labyrinth," favored brevity.

The Filmmakers at Cannes Who Are Learning to Love AI

Jafar Panahi to Again Face Trial in Iran on "Propaganda" Charges Nikola Todorovic has heard all the panic. He's scrolled through the Instagram reels of AI-generated shots, read the think pieces predicting cinema's imminent extinction, and sat through more than a few conference panels at the Cannes Film Market where the word "disruption" was wielded like a cudgel. He gets it. He also thinks most of the conversation is completely missing the point. The Bosnia-born co-founder, of Wonder Dynamics, now an Autodesk company, has spent a decade trying to build something more precise than the blunt instrument the industry argues about: AI tools designed to reduce the technical drudgery of visual effects production without pulling the human artist out of the equation. His pitch at Cannes this week is the same one he's been making since 2016. "We were really focused on how do we make AI that coexist with the existing pipeline," he says. It is, in its way, the defining argument of the 79th Cannes Film Festival -- one that has run as a persistent undercurrent beneath the red-carpet premieres and the Palme d'Or competition. The question is no longer whether AI will reshape the film industry. That debate, to many here, already feels settled. The more urgent and thornier question is which version of AI Hollywood is actually talking about -- and whether the industry has the sophistication, or the will, to tell the difference. At the opening jury press conference, The Substance star Demi Moore, serving this year on the nine-member feature film jury led by Park Chan-wook, offered her own pragmatic read on the moment. "I think the reality is that to resist -- I always feel that against-ness breeds against-ness," she told reporters. "AI is here. And so to fight it is to fight something that is a battle that we will lose. So to find ways in which we can work with it I think is a more valuable path to take." She acknowledged that the industry was "probably not" doing enough to protect itself, but added that whatever ground AI might gain, "what it can never replace is what true art comes from, which is not the physical. It comes from the soul. It comes from the spirit of each and every one of us sitting here." Moore's nuance was appreciated by at least one camp on the Croisette -- the faction of filmmakers and technologists who have spent years trying to carve out a middle ground between wholesale resistance and uncritical embrace. Todorovic knows that ground intimately. In 2016, he and actor Tye Sheridan -- fresh off the set of Steven Spielberg's motion-capture-heavy Ready Player One -- founded Wonder Dynamics to automate the most technically and economically burdensome parts of visual effects production without removing the filmmaker from the creative act. Motion capture without a suit. Lighting and compositing driven by machine learning. The company attracted seed funding from Peter Thiel's Founders Fund, board members including Joe Russo and early involvement from Spielberg himself. The Russo brothers used the technology on The Electric State, their Netflix production starring Millie Bobby Brown and Chris Pratt. In 2024, Autodesk acquired the company. The practical results, Todorovic says, have been considerable. "We've had animation studios tell us they basically increased their productivity from doing 30 seconds a day of animation to three and a half minutes a day," he says. "And we've had production companies of five to seven people that said they could now do things they could never do before." But what Wonder Dynamics is emphatically not building, Todorovic insists, is a replacement for the filmmaking act itself. "We're not in the business of 'let's prompt a movie,'" he says. "I'm not a believer that you can prompt a movie, prompt a performance, prompt a camera move. Tye is an actor, and I think the beauty of filmmaking is having an actor give a performance, having a cinematographer set up the frame, having a director direct." The economic logic for AI-assisted production is hard to argue with. One French director, Xavier Gens, told Reuters here that AI tools could have cut the visual effects budget of his Netflix hit Under Paris in half and shaved eight months off the schedule. Morgan Stanley analysts have estimated generative AI could reduce film and TV production costs by as much as 30 percent. The broader absurdity of current industry economics drives Todorovic. "We think it's normal that a film should cost $90 million, $100 million, $200 million -- which to me is absolutely absurd," he says. "A hospital will cost you $20 million to build, and it's there for two hundred years. And one movie, maybe based on a terrible idea, can cost $200 million plus marketing. And it could flop." But Todorovic is equally clear-eyed about the disruption that is coming regardless of how the debate is framed. "I think a lot of you'll hear people saying, 'Oh, we're accelerating, we're not replacing' -- I think there's just being politically correct. There's a shift, 100 percent, and I think visual effects get hit first, because it's very tech dependent." The jobs that were created because CGI was slow and difficult -- character creation artists, rigging artists, texture artists -- face real risk. "Those are the jobs that are in danger," he says. "It would be a lie to say otherwise." And yet he remains, by his own account, optimistic about the longer arc. His vision is less about Hollywood's incumbents and more about the filmmaker who currently has no path into the industry. "Me coming from a small country, it took me a long time to break in," he says. "I'm very optimistic long term, and it is going to open up for more people to tell stories, more voices to be discovered." There is, however, a precondition. "I think we need to step in and build AI how we want it to be used," Todorovic says, "otherwise we're going to have the tech industry leading it for the wrong reasons -- to create videos for TikTok in a day versus creating an actual film." That friction -- between Silicon Valley's incentives and cinema's -- was visible throughout the Marché du Film this week, where startups pitched everything from automated VFX pipelines to AI-generated simulated focus groups. Representatives from Alphabet, Disney Accelerator, NVIDIA, and OpenAI all made appearances in the market's new Innovation Village, overlooking the yacht-filled harbor. But for every figure willing to embrace AI as a production tool, others on the Croisette were drawing harder lines. Seth Rogen, here to present Tangles -- the hand-drawn animated film he co-produced with his wife Lauren Miller Rogen about a family navigating Alzheimer's disease -- was direct when asked about AI in filmmaking. "Every time I see a video on Instagram that's like, 'Hollywood is cooked,' what follows is the most stupid dog s -- I've ever seen in my life," he told Brut. "And if your instinct is to use AI and not go through that process, you shouldn't be a writer. Because you're not writing. Go do something else." On Tangles specifically, Rogen was unequivocal: "Not at all. It's hand-drawn animation. Every frame has a human touch to it." The film received a seven-minute ovation at its Cannes premiere. It is, in its own way, the same argument Guillermo del Toro has been making for years -- and never more pointedly than this week. Del Toro returned to Cannes to present a 4K restoration of Pan's Labyrinth, marking the 20th anniversary of its world premiere and the still-unbroken record of a 23-minute standing ovation, having personally supervised every stage of the restoration from the original 35mm negative. Speaking to The Hollywood Reporter, del Toro offered what may be the most precise diagnosis of the festival's central argument. "The thing is, to have the discussion, we have to start defining AI, because they're trying to pass five types of things under one single name," he said. "So what are you talking about? Are you talking about a tracking program? Are you talking about rotoscope? Or are you talking about generative AI, where you remove the artist from the equation? The nomenclature has to change before we can have a real discussion. Otherwise it's just a headline." Taking the stage before the restoration screening, del Toro took a more impassioned tone. "We are, unfortunately, in times that make this movie more pertinent than ever," he said, "because they tell us everything is useless to resist, that art can be done with a f -- ing app." Todorovic, for his part, would not disagree with the sentiment, even if his instrument is different. "We like art because it's difficult," he says. "We like film because it's difficult. We admire work where only that one person can do this, because that's their voice, and it's so hard to do that. We're not going to go pay to watch something someone spent a few days prompting." What he is betting on -- and what Wonder Dynamics was built around -- is a version of cinema's future that looks less like a prompt box and more like a wider door. "Technology has always driven the film industry," he says, "and there's always been resistance, but we've always depended on each other." The first Hollywood studios, he notes, were founded by producers fleeing Thomas Edison's technology monopoly on the motion picture camera. "The first studio was created by producers who were saying storytelling should be for everyone, and there shouldn't be one company that holds the technology and holds us hostage." In the Innovation Village, on the Croisette, and in the Debussy Theatre where a Cannes audience gave Pan's Labyrinth a hero's welcome all over again, the argument at this festival is clarifying. The question isn't really about AI. It's about what makes a film worth making in the first place -- and whether the tools Hollywood reaches for help answer that question, or quietly make it disappear.

"The future is one person making a whole film," says director of "first AI movie at Cannes"

Even just a few months ago the idea of making an entire feature-length movie using generative AI was fairly risible. If Coca-Cola couldn't achieve consistent visuals over a 60-second Christmas ad, how would anyone produce a 90-minute film? But months are years in AI time, and this week we suddenly had headlines claiming that a 100% AI-generated movie had "premiered at the Cannes Film Festival": a 95-minute action-fantasy film called Hell Grind. Hell Grind (trailer above) was produced entirely on the Higgsfield AI video platform by a 15-person team in Kazakhstan, purportedly in just 14 days. To clarify, the AI movie didn't premiere at the Cannes Film Festival, despite even the Wall Street Journal suggesting it did. Rather it was presented at a side event held in the geographical vicinity of the festival. That's a bit like a band hiring the back room of a pub in Glastonbury and then claiming to have played "at Glastonbury". It's cheeky, a bit sad and it makes it hard to know how much to believe anything else the band says. Higgsfield says Hell Grind cost "about $500,000" to make, of which $400,000 was AI compute costs. It estimates that traditional movie production for a comparable film would cost about $50 million. Judging from the trailer above, the movie is camper than a field of AI-generated tents, but the objective was to prove it could be done and that generative AI is ready for production. Hell Grind isn't intended to be a blockbuster but to promote the AI pipeline to Hollywood (in tandem, The Mask director Chuck Russell has announced two AI features, also to be made with Higgsfield). Such a pipeline completely changes what it means to work on a movie. Instead of directing actors and other artists, the various directors of Hell Grind (a sizeable portion of that 15-strong team has director billing) prompted AI models. Instead of retakes, prompts were tweaked to get a closer result to what they wanted. In the behind-the-scenes video below, Hell Grind's directors are very clear about what they see as the benefits of generative AI for making movies: no equipment, no physical exhaustion on set, no arguments over budget... and no people? "The future is one person making a whole film," Mikhail Kumarov says. "Like a writer, like a manga artist. Their stories, their fans: the only thing that matters is that a person has light inside; a story they can tell the world." The big challenge for making movies with AI video has been achieving consistency in characters' appearances and in world physics over anything longer than a short clip. Higgsfield is an orchestration platform hosting AI video models from various developers, including Google's Veo 3, but it adds tools intended to help make output more controlled and consistent across multiple shots. In its pipeline for the Hell Grind, each prompt generated about 15 seconds of footage. The first 25-minutes of what was originally intended to be a series required 16,181 video generations to produce 253 final shots To get usable footage, prompts had to very specific in terms of detail and were often around 3,000 words long. Prefixes were used to define a shot's style, the type of camera and lens and the lighting, the later reportedly being the key to avoiding the AI slop look. Prompts also had to specify that the output should respect natural laws like gravity and inertia - even to the point of using phrases like "no floating props", according to the Wall Street Journal. Another director, Adilet Abish, talks of being surprised to find himself moved to tears even though Hell Grind's characters are AI-generated. Aitore Zholdaskali, another of the directors, previously made movies the traditional way, including a 2025 movie called Sicko. In a press release, he's quoted as saying: "It took me ten years to get my first traditional feature made. Most people who started with me never made it to directing movies. I started out shooting weddings and music videos, slowly building trust until someone gave me a TV series. "Making a movie today feels like making an album twenty years ago: you need investors and big studios. But then the laptop changed music forever and gave us Billie Eilish. That's what Higgsfield is doing for filmmakers. The next generation won't have to wait ten years." Let me know what you think in the comments below. This vision of solo filmmakers making entire movies on their own may sound alarming to many who have fears about the impact of generative AI on jobs. Just ahead of Cannes, the artist Alma Haser launched Empty Red Carpet, a series of photos intended to highlight what she sees as the "creative theft" posed by generative AI. The work was created for Human Made Mark, an initiative that proposes a certification programme for movies that have real people both in front of and behind the camera. In the US, the Academy of Motion Picture Arts and Sciences recently updated the rules for the Oscars to clarify that generative AI cannot win acting or script writing awards.