AI in film industry news has spent three years ping-ponging between panic and hype. This September it managed both at once, in the same city, on the same weekend.
At the 2026 Venice Film Festival, George Clooney warned that AI would gut crew jobs at scale. Maggie Gyllenhaal revealed she’d thrown away an AI-generated Marilyn Monroe because the result felt dead on screen. A short boat ride down the canal, a rival festival premiered ten AI-made finalists picked from more than 3,000 submissions, with Twilight director Catherine Hardwicke on its jury. Both scenes are real. The tension between them is the actual story.
So what does AI in film mean, precisely? It covers any machine learning tool touching the production pipeline: generative video models, voice and face manipulation, previsualization software, and likeness systems trained on licensed footage. Almost none of it produces a finished movie today. All of it is quietly changing who gets hired, what productions cost, and which stories get funded.
Key Takeaways
| Point | Details |
|---|---|
| Two festivals, one canal | The main Venice festival warned about AI while the Reply AI Film Festival celebrated it, screening 10 finalists from 3,000+ entries |
| The tech failed its biggest test | Maggie Gyllenhaal discarded a licensed AI Marilyn Monroe for Flesh Impact because Dakota Johnson’s human take was far more alive |
| Hollywood already uses AI quietly | Steven Soderbergh and Darren Aronofsky lean in publicly, and one AI filmmaker says a studio asked him to fill a budget gap in secret |
| Economics drive the threat | Clooney predicts phone-made films will rival $200 million studio productions, replacing crew and VFX supervisors along the way |
| Disclosure is the real battleground | The Brutalist backlash and today’s NDAs show the fight is about honesty, not just jobs |
In This Article
- What AI Is Actually Doing in Film Right Now
- How AI Gets Used Across a Film Production
- Why the Venice Split Matters
- Common Misconceptions About AI in Film
- What This Means for Jobs, Budgets, and Your Watchlist
- The Hypocrisy Problem Nobody at Venice Solved
- Frequently Asked Questions
What AI Is Actually Doing in Film Right Now
Strip away the dread and the evangelism, and 2026’s AI filmmaking looks less like replacement and more like augmentation. Directors use generative tools for backdrops and previsualization. VFX houses lean on models for crowd work, set extensions, and cleanup that once consumed junior artists’ weekends. The premieres still star humans.
The most revealing case came from Gyllenhaal herself. Her short Flesh Impact, starring Dakota Johnson as Monroe and Ellen Burstyn as an imagined older version of the star, was commissioned with AI baked in. The producers used what she calls licensable AI, trained on Monroe films the team had properly licensed, with the blessing of the Monroe estate holders. She tried it in good faith. She threw the result away.
“It fundamentally did not work. All the reasons we decided to make this project to begin with were killed off in the AI version.”
That failure is instructive. The tooling was legal, paid for, and estate approved. It still lost to an actor doing her job. Speaking at the Future of Creativity summit in Venice, Gyllenhaal pointed to collaboration as the difference, citing how Jessie Buckley’s performances come from bouncing off the people around her. No model fakes that yet. Worth noting too, Genesis presented the short. An automotive brand underwriting prestige cinema fits a broader pattern, since AI in automotive industry news shows those companies pushing AI into design, marketing, and now content.
The Two Venices, Explained
Symbolically, the city hosted a schism. The main festival’s symposium, organized by Finch & Partners and CAA, gathered Clooney, composer Alexandre Desplat, and Signal president Meredith Whittaker to warn about job losses running from art departments through visual effects. Gyllenhaal told that panel the AI version of her film lacked everything that made the project worth making.
Down the canal, the Reply AI Film Festival ran its own program, premiering 10 finalists including UK filmmaker Mike Bennion’s Kev. Hardwicke, judging there, said every director she talks to, including Academy-nominated blockbuster names, is curious about AI, and many already use it for backdrops and previsualization. Both camps describe the same industry. They just sit at different points on the adoption curve.
How AI Gets Used Across a Film Production
Here’s where the tools plug in, from first pitch to final render.
- Development and pitching. Producers generate concept art, mood reels, and synthetic backdrops to sell projects before a single frame is shot.
- Previsualization. Directors sketch sequences with generative video instead of paying specialist vendors for weeks of animatics.
- Performance augmentation. AI refines accents as on The Brutalist, blends likenesses as in Gyllenhaal’s Monroe test, and de-ages actors.
- VFX and environments. Models handle crowds, sky replacements, and architectural elements, like the Brutalist’s buildings drawn in its protagonist’s style.
- Budget-gap filling. Bennion told Semafor that a cash-strapped Hollywood production asked him to use AI to “fill that gap at a cheaper price.”
| Production stage | Traditional approach | AI-assisted approach in 2026 |
|---|---|---|
| Previsualization | Specialist vendors build animatics over weeks | Generative video drafts sequences in hours |
| VFX cleanup | Junior artists fix frames by hand | Models handle crowds, extensions, rotoscoping |
| Dialogue and voice | Dialect coaches and ADR sessions | Accent refinement and synthetic touch-ups |
| Likeness work | Casting a physical lookalike | Licensed models trained on estate-approved footage |
Pro Tip: Scan a film’s end credits for unfamiliar vendor names. AI work usually hides behind technical credits like digital facial replacements or machine learning services, and those lines tell you more than any press release.
Why the Venice Split Matters
Follow the money and the warnings make sense. Clooney predicted audiences will soon watch films made on a phone “that costs the studio a couple of 100 million dollars,” with actors, crew, and VFX supervisors replaced in the process. Gyllenhaal framed the same threat from the funding side. If you can deliver a decent movie for $40,000 or $4 million, nobody hands you $10 million to do it the old way.
Her sharpest line was about efficiency. “You can efficiently make an ok movie, probably at some point,” she said. That’s the real fear. Not terrible AI films, but adequately average ones produced at a fraction of the cost, crowding out everything that needs an actual budget.
The pattern isn’t unique to film. Animation industry AI news keeps circling the same squeeze on entry-level roles, and AI in music industry news has spent two years fighting voice cloning without a clear resolution. Even AI and the news industry, this publication’s own trade, is restructuring around similar tools. Games got there first, where AAA studios now bake AI directly into development pipelines using the exact efficiency arguments Venice heard last week.
Common Misconceptions About AI in Film
“Hollywood has rejected AI”
The backlash moments get the coverage. Brady Corbet’s The Brutalist was demonized when its AI use surfaced, and he had to issue a public defense. But Semafor’s reporting from Venice found Steven Soderbergh planning heavy AI use on a Spanish-American war film, Darren Aronofsky launching an AI-focused studio called Primordial Soup, and Lionsgate letting Runway train models on its catalogue. Bennion, who has worked inside one of these productions under NDA, put it bluntly. He finds the industry “full of hypocrites who use the tech but are reluctant to say how and why.”
“AI movies are already good enough”
Gyllenhaal’s discarded Monroe is the cleanest counter-evidence available. A funded, licensed, estate-approved attempt at an AI performance lost to a human take, decisively, in the director’s own judgment. Generative models produce impressive clips. Sustained, emotionally coherent performances across a feature remain out of reach, which is why AI film festivals showcase shorts rather than epics.
“Licensing solves the ethics”
Licensable AI, trained on properly paid-for footage with estate sign-off, sounds like the tidy answer. It isn’t. The Monroe experiment was fully licensed and still raised the question of who profits when a dead star’s face performs. Licensing also covers the likeness but not the livelihoods of the VFX artists and crew whose workload the tool compresses. Consent and compensation are different problems, and the industry keeps conflating them.
Pro Tip: When coverage says “licensable AI,” read it as legal clearance, not a human-free production. It confirms the training data was paid for. It says nothing about who worked on the film or who got cut.
What This Means for Jobs, Budgets, and Your Watchlist
The practical fallout lands in three places.
First, jobs. Clooney’s warning that “a lot of people are going to get replaced” targets crew categories first: VFX supervisors, art departments, previz artists. The industry’s entry ladder, where juniors learned by doing grunt work, is exactly what these tools automate. Animation already shows what that pipeline squeeze looks like.
Second, budgets. Genre films and commercials adopt AI fastest because their margins are thinnest. Expect the mid-budget film, the $10 million to $40 million range Gyllenhaal described, to become Hollywood’s most endangered species. Studios will fund either cheap AI-assisted bets or tentpoles, with little in between.
Third, discovery. A flood of cheap content is coming, and finding the good stuff gets harder. Independent creators gain tools the gatekeepers once hoarded, which we’d count as a net positive given how indie games matter to the industry’s health for similar reasons. Meanwhile, AI-curated feeds and search bars rebuilt around AI will increasingly decide what audiences ever see at all.
Pro Tip: Check production notes and festival program listings for disclosure statements. The gap between films that advertise AI use and films that hide it has become a useful quality signal.
The Hypocrisy Problem Nobody at Venice Solved
I’ll be honest about where I land. Neither camp convinced me. The doomsayers undersell how quickly working directors are already experimenting, and the evangelists undersell how far the technology sits from a performance worth watching. Gyllenhaal, by actually running the experiment and publishing her failure, contributed more to this debate than any keynote in either venue.
But Bennion’s “sworn to secrecy” story is the detail I can’t shake. If using AI on a Hollywood production is legitimate, why does it require an NDA? Secrecy is rational right now because The Brutalist showed what happens to films caught using AI after release. That incentive structure guarantees the next wave of AI in film industry news will be dominated by exposures rather than honest adoption, and exposures breed the exact backlash that keeps everyone hiding.
Disclosure is the fight worth having, in my view. Let directors say they previsualized with generative tools without career risk. Let VFX houses list model-assisted shots in the credits. Once the silence breaks, the labor conversation can happen on real terms, with data about which jobs actually vanished instead of which ones people feared would. Until then, the quiet adoption problem gets louder, one leaked production at a time.