Subtitle: 2026 survey data and my own experience releasing an AI-made series point to the same fault line, and it isn’t the one most creators are bracing for.

A few weeks ago I almost added a caption under a Lost Garden clip: “made entirely with AI.” I typed it, looked at it, and deleted it. Not because it was untrue. Because I couldn’t decide if saying it out loud would help the film or hurt it.

Audiences do care whether a film was made with AI, but not in the way most creators fear: the data says they judge disclosure, not the tool itself. A YouGov poll from February 2025 found that 86% of people want AI use disclosed in media production, while 61% said AI in filmmaking is acceptable or that they have no strong opinion either way, according to an analysis by AI FILMS Studio. Those two numbers look contradictory until you notice what they’re actually measuring. One is about honesty. The other is about the technology. Audiences answer them differently because they are, in fact, different questions.

I run a workspace for AI-assisted scriptwriting and shot planning, and I made Lost Garden, a dark-fantasy series, with an AI pipeline from script to final render. I also have a short film, The Seam, entered in this year’s Higgsfield Global Film Festival, sitting in a feed next to hundreds of other AI-made entries. So the question of what audiences actually forgive isn’t abstract for me. It decides what I write under my own videos.

Do audiences actually care if a film used AI?

Yes, but selectively. The same YouGov data shows people support AI for visual effects, environment extension, and dialogue translation, while opposing it for acting and screenwriting. It isn’t a blanket verdict on the technology. It’s a test applied use by use.

Two events from the past year drew that line clearly:

• In September 2025, a British startup called Particle6 introduced “Tilly Norwood,” an AI-generated performer that reportedly drew interest from talent agencies. The backlash was immediate and came from a specific fear: a synthetic actor replacing the parasocial bond audiences form with real ones.
• “Late Night with the Devil,” a 2024 horror film, used AI to generate a handful of background graphics for a fictional talk show. Viewers likely wouldn’t have noticed if the detail hadn’t leaked. Once it did, the backlash on Letterboxd wasn’t really about the graphics. Commenters described feeling deceived, not because the AI work was bad, but because nobody had told them it was there.

Put those next to each other and the pattern holds: AI doing a job a human would have done anyway, quietly, in a category audiences already accept, barely registers. AI standing in for a person, or AI hidden from an audience that assumed there was none, is where the trouble starts.

The technology isn’t what audiences are voting on. Whether they were told the truth about it is.

Based on the same polling, here’s roughly where audiences land:

• Generally accepted: visual effects and crowd generation, environment extension, dialogue translation and dubbing, color grading, upscaling, and production-side tasks like continuity checking.
• Generally rejected: AI-generated performers standing in for actors, AI-written scripts, and any of the above used without disclosure.

Notice what separates the two lists. It isn’t how advanced the AI is. It’s whether a human is still the one making the creative call. Translation software doesn’t decide what a character means. A generative script would.

That distinction is why I built my AI-assisted film production workspace around a locked script and a director-approved shot plan before a single frame gets generated, rather than a loop where the model proposes and you just pick a favorite. It’s not a moral position. It’s a bet that the “who’s actually deciding” question will keep mattering to viewers long after most people stop noticing which shots used AI at all.

Does disclosing AI use hurt or help a film?

It helps, and the “cheap” perception problem is a bigger risk than the disclosure itself. ProVideo Coalition’s December 2025 roundtable on the topic flagged a real reputational issue: audiences increasingly associate generative video with low production value, independent of whether the actual quality holds up. That association, not the disclosure, is what a filmmaker should actually worry about.

Which means disclosure is a chance to frame the work, not just a confession. Saying “this used AI for backgrounds and crowd work, directed by a human” tells a different story than silence followed by discovery. The first is a production choice an audience can evaluate. The second reads as something you were hoping nobody would ask about.

For Lost Garden, I ended up somewhere in the middle: I don’t hide that it’s AI-made, but I also don’t lead with it. The credits, my posts, along with my own site, all say plainly how it was built. What I stopped doing was treating “made with AI” as either an apology or a selling point. It’s a production fact, the same as “shot on 16mm” would be.

Does using AI make audiences trust a film less?

Somewhat, and that gap is measurable, even outside filmmaking specifically. Animoto’s State of Video 2026 report, based on a survey of 460 U.S. consumers and marketers, found that 78% trust videos featuring real people more than AI-generated content, and that 36% of viewers who noticed AI involvement in a video said it lowered their trust in the brand behind it. That’s video marketing, not narrative film, but the underlying instinct (people over process) is the same one showing up in the film-specific polling.

Trust isn’t fixed, though. A March 2026 study of 2,000 consumers by Alvarez & Marsal, cited in the AI FILMS Studio analysis, found that 51% would pay full ticket price for a 100% AI-generated film if the story itself was compelling. Story quality closes a lot of the trust gap that “AI-generated” opens.

That tracks with what I’ve seen entering The Seam into a festival built for AI films specifically. Nobody there is arguing about whether AI belongs in the room. The comments that land are about pacing, a shot that doesn’t read, a cut that works. The medium stopped being the conversation. The film became the conversation. That’s a smaller sample than a national poll, but it’s consistent with the survey data, not opposed to it.

Here’s what I’d actually tell another solo or small-team filmmaker weighing whether to say anything about AI at all:

1. Say it once, plainly, where it belongs (credits, description, an about page), not as a headline and not as an apology.
2. Name what AI actually did. “AI-assisted visual effects” reads as a production choice. “Made with AI” alone reads as a shrug.
3. Don’t let the disclosure replace the work. Audiences that heard about Late Night with the Devil’s AI graphics were reacting to secrecy, not scale. A film that’s honest and merely fine still has to be fine.
4. Expect the AI question to fade before the technology does. Right now every AI-assisted film gets asked about it. That won’t last, the same way “shot digitally” stopped being a caption worth adding once nobody needed to be told.

FAQ

Do audiences care if a movie was made with AI?

Yes, but the deciding factor is disclosure and what the AI replaced, not the presence of AI itself. Undisclosed AI use draws far sharper backlash than disclosed use of the same tools.

What AI uses do audiences accept in film?

Visual effects, environment extension, translation and dubbing, color grading, and technical restoration generally register as acceptable, according to YouGov polling cited by AI FILMS Studio’s 2026 analysis.

What do audiences reject?

AI-generated performers replacing actors, AI-written scripts, and any AI use a production didn’t disclose upfront.

Does AI make a film feel cheaper to audiences?

It can. A ProVideo Coalition roundtable in December 2025 flagged a growing association between generative video and lower perceived production value, separate from actual quality, which is a stronger reason to invest in the work than to hide how it was made.

Whether The Seam or the next Lost Garden episode gets judged fairly won’t come down to which model rendered which shot. It’ll come down to whether the film is honest about what it is and good enough that the question stops mattering. That’s a harder standard than either “disclose everything” or “disclose nothing,” and it’s the one I’m actually trying to meet.