State of AI Filmmaking 2026: What 40 AI Film Leaders Told Us May Surprise You!
- 9 minutes ago
- 11 min read
After conversations with filmmakers, technologists, studio leaders, educators, festival organizers, and creators around the world, you might be surprised what they say.
At the beginning of 2026, my brother Andrew and I started Brothers’ Saga, a weekly video podcast about AI filmmaking and the future of storytelling.
We expected to spend a lot of time talking about models.
Which image generator is best? Which video model has the best character consistency? How long until we can generate a feature film with one prompt? Will AI replace VFX jobs or will studios use video models? What happens when the next model launches?
We certainly talked about all of those things at times.
But after conversations with more than 40 people working at the edge of this medium, a different theme kept surfacing.
The technology is getting better extraordinarily quickly. Making something people actually want to watch is not getting easier.
In some ways, it is getting harder. I wasn’t convinced. How could that be —and will it stay like this in the near and long term?
Our guests on the show have included filmmakers making work for major studios, award-winning AI directors, traditional Hollywood veterans, film professors, AI film festival organizers, creative technologists, entrepreneurs, and independent creators experimenting far outside the studio system.
They do not agree on everything. Far from it.
But here are the ideas we kept hearing again and again.
1. AI is making filmmaking easier. Great filmmaking is getting harder. It’s “harder to impress.”
Jiajian Min, Chair of the MIT Global AI Film Hack, described one of the central paradoxes of 2026 better than almost anyone we spoke with.
AI is making execution easier. But making something exceptional is becoming harder.
The reason is simple: everyone else received the same superpower. We love removing gatekeepers and reducing barriers, but there is another side to that coin.
An auteur filmmaker who once needed a crew, equipment, locations, actors, and a significant budget can now create imagery that would have been astonishing only a few years ago. But millions of other people can do it too. Billions! And they are, with backlash building towards “AI slop” and a wave of unintentionally generated mass content at scale.
Jiajian pointed out that a piece of AI-generated work that might have attracted 100,000 views several years ago could struggle to get 1,000 today. Audiences on YouTube, Reddit, and X have seen too much, too quickly. Their expectations are rising along with the technology.
This may be one of the most important changes happening in creative AI.
Technical scarcity is being replaced by creative scarcity.
When almost anyone can generate a beautiful cinematic shot, a beautiful cinematic shot stops being enough.
The question shifts from:
Can you make this?
to:
Why should anyone care?
2. Story is becoming more important, not less
If there was one theme we heard more than any other, it was this.
AI filmmakers have spent the last few years obsessing over visual quality. Some Creative Partner Programs (CPP) generated thousands of hours of content with free credits just to hype the latest sub-version of a different video model every week.
Character consistency. Lip sync. Physics. Camera movement. Prompt adherence. Resolution. Those problems matter.
But the closer the models get to solving them, and the more the price drops in orders of magnitude, the more obvious the next bottleneck becomes.
Story.
Aashay Singh, a director at GRAiL who spent roughly 15 years making films before the generative-AI wave of the 2020's, told us that he can often recognize a weak AI film almost immediately.
“It’s the writing that gives it away. In the first thirty seconds you already know.”
His criticism was not that AI creators lack access to good enough models. It was almost the opposite.
Generation has become so fun and immediate that creators are tempted to skip the difficult work that normally comes first: character development, structure, conflict, motivation, pacing, and rewriting.
Aashay described that temptation as jumping straight into video execution because that is now the enjoyable part. He praised structured writing workflows precisely because they force creators to confront questions they might otherwise avoid.
Jimmy Woolley, founder of 404 Films and an AI filmmaker with a background in traditional British television at the BBC, arrived at essentially the same conclusion from another direction:
“We don’t make films for machines, we make films for people.”
His point was that every piece of content still needs a purpose and an audience. A human driving the story forward with an intention and target audience. The technology may change radically, but the person watching on the other side of the screen has not disappeared.
That sounds obvious.
In 2026, it is surprisingly easy to forget.
3. The age of the “AI demo” is ending
For the first phase of generative video, novelty itself was a form of entertainment.
A giraffe skateboarding through Times Square could be interesting simply because a machine generated it.
A cinematic tracking shot created without a camera crew could go viral because people wanted to understand how it was possible.
That window is closing.
Tiffani Lee Joseph was one of the first to work with AI media professionally and has judged multiple AI film competitions over the years. Yet one of the provocations from our conversation was that even people immersed in this field often cannot sit through many AI films.
The industry, she argued, also created a messaging problem by implying that AI production could simultaneously be fast, cheap, and good.
That promise encouraged people to confuse reduced production friction with reduced creative effort.
They are not the same thing.
AI can make a shot faster.
It cannot automatically make the choice of that shot meaningful.
It can make ten alternatives cheaper.
It cannot decide which one belongs in the film.
It can shorten parts of production.
It does not remove the work of human taste, even if or when it ever can.
The era when audiences rewarded a film because it was “made with AI” is giving way to an era where most audiences will not care.
They will ask the same question they have always asked:
Is it any good? After all, we don’t watch Pixar movies to see if they used CGI. It’s all about story, Ed Catmull knew it. Steve Jobs knew it. You know it. Even AI knows it. So get to writing your story (and AI can help).
4. Traditional filmmaking knowledge may become more valuable
This was another confirmation.
The conventional fear is that generative AI devalues traditional skills. That it will replace the jobs of top VFX artists at ILM and other top Computer Animation & Special Effects shops. Those creative artists have mastered many skills on the computer that make them top users for apps like ComfyUI, Saga, and Adobe Firefly.
Many of the strongest AI creators we interviewed suggested something closer to the reverse.
When technology makes execution abundant, knowing what to execute becomes more valuable.
Composition matters.
Editing matters.
Performance matters.
Sound design matters.
Film history matters.
Writing matters.
Knowing why a close-up should follow a wide shot matters.
Knowing why a scene feels emotionally flat matters.
Knowing when not to generate another shot matters.
Jiajian came into filmmaking with a background in architecture and visual thinking. He described the challenge of learning how to turn world-building ability into character and narrative structure. The technology could help him visualize a world, but storytelling remained a discipline he had to learn.
That pattern appears repeatedly across AI filmmaking.
The best results often come from people who combine new technical fluency with an older craft.
My brother Andrew and I see the same dynamic from opposite sides of our own backgrounds. I spent my career building technology and AI products. Andrew went through film school and has spent years working professionally on film and television sets, serving as Saga’s Chief Story Officer, and my Brothers’ Saga Co-Host.
AI is increasingly capable of producing pixels.
But a filmmaker still has to know what those pixels are for. Cinematic art. audiences will pay (in time and/or tickets) to watch.
5. The winning workflow is becoming hybrid
The conversation around AI film is often framed as a binary:
Traditional filmmaking or AI filmmaking?
The people actually making this work tend to be far less ideological.
They use whatever works.
Live action plus generative VFX.
Traditional editing plus AI-generated inserts.
Human actors plus synthetic environments.
Photoshop plus image generation.
Motion capture plus video-to-video transformation.
A screenplay written and rewritten by a person, with AI helping organize, challenge, visualize, or accelerate parts of the process.
The future looks less like a giant prompt box that replaces a production and more like AI becoming another layer of production.
Percy Leon gave us one particularly practical example.
He had a documentary project sitting unfinished for roughly a decade. Using a modern AI-assisted workflow, he was able to organize the material and assemble multiple episodes in a fraction of the time the project had previously demanded.
That is not “press a button, make a movie.”
It is arguably more interesting.
It is a filmmaker finally finishing something that otherwise might never have existed.
6. The model is becoming less important than the workflow
The AI filmmaking stack is fragmenting and consolidating at the same time.
Creators now move between large language models, image generators, video models, voice systems, music generators, upscalers, editing tools, compositing tools, and increasingly specialized applications.
And the “best” model can change almost weekly “I thought it was Seedance 2.5? Did you try the latest Luma? I heard the new one from Runway is going to be better next week.” Versions and subversions getting one step better every week, available in aggregators like Saga and Magnific.
That makes betting your entire creative identity on one model increasingly difficult.
Our conversation with Machine Cinema’s Fred Grinstein and Minh Do touched on how quickly the creative AI landscape is expanding, from AI VFX and filmmaking to 3D and world models like Google DeepMind Genie and World Labs Marble. Their market view is increasingly about categories and workflows, not one universally dominant tool.
I think this is where the next generation of creative applications becomes important.
The question is no longer simply:
Which model should I use?
It is:
How do I get from an idea to something finished?
That requires continuity between steps.
An idea becomes characters.
Characters become scenes.
Scenes become a screenplay.
The screenplay informs storyboards.
Storyboards become shots.
Shots need voices, music, sound design, editing, revision, and ultimately distribution.
This is also the thesis behind what we have been building with Saga since 2021: not another foundation model, but an opinionated workflow around storytelling and filmmaking.
The individual models will keep changing.
The creative process still has to connect.
7. AI changes the economics of what stories can exist
Jan-Willem Blom made a point on the podcast that has stayed with me: AI does not just make individual shots cheaper. It can change the math of storytelling itself.
If production becomes dramatically less expensive, creators can think beyond a single short film.
A character can continue.
A world can expand.
A film can become a series.
A series can become an entire story universe.
His advice was essentially to stop thinking only in terms of one isolated film and start thinking about larger narrative structures.
We are already seeing this emerge in short-form episodic content.
Percy Leon, for example, uses structured tools to organize series names, loglines, and 2-to-3-minute episodic stories as part of a much broader AI stack.
Microdramas and vertical series may be particularly important here.
The economics of traditional television demand enormous audiences because production and distribution are expensive.
The economics of AI-native production may allow increasingly specific audiences to support increasingly specific stories.
That could create more content.
It could create a lot more bad content.
But it could also create stories that were previously economically impossible to make.
8. Democratization does not guarantee quality, but it does expand who gets to try
This is where the conversation becomes bigger than production efficiency.
Luke Minaker, also known as Magenta Rúne, talked with us about independent animation, marginalized voices, finding paid work as an AI filmmaker, and whether AI is actually “killing creativity” or simply removing repetitive production tasks.
One possibility is that the technology gives artists more time for designing, writing, experimenting, and directing rather than spending all of their available resources executing the mechanics of production.
Machine Cinema’s Fred and Minh similarly described an optimistic case for AI built around more voices and more stories emerging from more places around the world.
That does not mean every person with access to a video model becomes a filmmaker.
A smartphone did not make everyone a photographer.
YouTube did not make everyone a great director.
Word processors did not make everyone a novelist.
But each technology dramatically expanded the number of people who could participate.
AI filmmaking may do the same thing at a much larger scale.
9. AI will not eliminate creative work. It will move the work.
One of the misleading questions in this industry is whether AI makes filmmaking effortless.
It doesn’t.
It moves effort around.
You may spend less time physically setting up a camera.
You may spend more time iterating.
You may spend less time building one expensive visual-effects shot.
You may spend more time choosing between 50 generated possibilities.
You may spend less time executing an idea.
You may therefore need to spend more time deciding whether the idea was worth executing in the first place.
Jimmy Woolley described an interesting version of this. He uses LLMs not merely to generate ideas, but to test his own ideas. He mixes human concepts with machine-generated alternatives and uses another model to rank them blindly, effectively turning AI into a creative quality-control mechanism.
That may be a better metaphor for the future than “AI makes the movie.”
AI becomes:
a collaborator,
a critic,
an intern,
a storyboard artist,
a VFX tool,
a research assistant,
an editor,
a generator,
and sometimes an adversary you have to wrestle into giving you the shot you actually wanted.
The filmmaker still directs.
So what is the state of AI filmmaking in 2026?
We are no longer waiting for AI video to become “good enough.”
In many circumstances, it already is.
The harder questions now sit one layer higher.
Can you sustain a character?
Can you sustain a story?
Can you make someone feel something?
Can you create 20 minutes that people actually want to watch rather than a 20-second clip that makes them stop scrolling?
Can you build a repeatable workflow instead of reinventing your stack for every shot?
Can creators make money?
Can the industry develop fair models for actors, writers, artists, and rights holders?
Can filmmakers use these tools without allowing the tools themselves to dictate the aesthetic?
And perhaps most importantly:
When everyone can create, what makes your creation worth someone’s time?
Those are no longer primarily model questions.
They are filmmaking questions.
What I think happens next
Based on our conversations so far, I expect the next phase of AI filmmaking to be defined by several shifts.
From shots to sequences.
The impressive unit will stop being a single generated clip and become a coherent scene.
From sequences to stories.
Consistency and narrative structure will matter more as creators move toward longer work.
From individual tools to integrated workflows.
Creators will increasingly expect writing, visualization, generation, audio, and editing to work together.
From AI-native spectacle to invisible AI.
The most successful work may not advertise itself as AI-generated at all.
From short films to recurring formats.
Microdramas, episodic vertical video, animation, series, and persistent story worlds are particularly well matched to falling production costs.
From technical advantage to taste advantage.
Access to the best model will become temporary. Knowing what to do with it will remain scarce.
And finally:
From “Can AI make a movie?” to “Who has something worth making?”
That may be the biggest transition of all.
For several years, the AI filmmaking community has been trying to prove that the technology can create.
In 2026, that argument is increasingly settled.
Now we have to prove that we can create something worth watching.
Russell Palmer is the co-founder and CEO of Saga (by CyberFilm®), an AI filmmaking platform founded in 2021, and co-host of the Brothers’ Saga podcast with filmmaker and screenwriter Andrew Palmer. Brothers’ Saga features weekly conversations with filmmakers, creators, technologists, and industry leaders shaping the future of AI entertainment.
© 2026 CyberFilm®
“Saga” and “Cyberfilm” are each a trademark and/or registered trademark of Cyberfilm AI Corporation or its affiliates in the United States and/or various other jurisdictions.
Comments