Animoto’s 2026 State of Video report just confirmed what most marketers already suspected: consumers can spot AI-generated video, and many of them quietly trust your brand less when they see it. The fix is not avoiding AI. It is using it differently.
Animoto surveyed US consumers for its 2026 State of Video report and found that 83 percent believe they can spot AI-generated videos, with the biggest tells being robotic gestures (67%), unnatural voices (55%), and lack of emotional tone (51%). 36 percent say AI video lowers their trust in a brand. But 84 percent of marketers are using AI to create video anyway. The brands winning are the ones treating AI as an editing and production tool, not a casting decision. Real people, AI-assisted production. Here is the playbook.
Let me set the scene. Animoto, the video creation platform, surveyed 1,000 US consumers in early 2026 for its annual State of Video report.[1] The findings landed in late January and the marketing trade press picked it up immediately. The headline numbers:
The mismatch is the story. Marketers are racing to AI video. Consumers are quietly noting it and discounting accordingly. The gap is not loud. People are not commenting “this is AI” in the comments. They are just trusting you a little less and scrolling on. That is the worst kind of brand damage because you cannot see it in your engagement metrics.
The smart take: the data is not telling you to stop using AI in video. 84 percent of marketers using a tool is not a fad. The data is telling you that fully synthetic video (AI avatars, AI voices, no real human at any point) is the part that is hurting you, and AI-assisted video (real people, AI in the production pipeline) is not carrying the same penalty.
The Animoto report broke down the specific cues that signal “AI” to consumers. The top three:
Robotic gestures (67%). AI-generated avatars still struggle with the small, asymmetric movements real humans make: a half-shrug, a pause to think, a hand drifting up to scratch a temple. The avatars do “presenting” gestures well. Anything else looks programmed.
Unnatural voices (55%). AI voices have improved enormously since 2023, but they still over-articulate, hit beats too evenly, and rarely use the casual filler (“you know,” “honestly,” brief sighs) that marks unscripted speech. People notice.
Lack of emotional tone (51%). This is the real problem. Even the best AI presenters maintain a flat affective register. They are not actually delighted, frustrated, or curious about anything. The audience picks that up subconsciously and reads it as “this person does not actually care.”
If you have rolled out AI presenters in any of your content (training videos, product explainers, FAQ videos), this is what people are clocking. Watch your own clips with the sound off and look at the gesture vocabulary. If it looks like a presenter shifting their weight in the same range every 3 seconds, your audience is noticing.
The “uncanny valley” of AI video is not in the visuals anymore. The pixels are fine. It is in the timing, the movement vocabulary, and the affective range. That is harder to fix because it is invisible until you are watching for it.
Animoto’s own framing of the data is the right one: AI-assisted video is fine. Fully AI video is a brand liability.[1] The question is what “AI-assisted” actually means in your production pipeline.
Here are the four places AI is already earning its keep without spooking your audience:
1. Editing and post-production. Auto-trimming silences. Cleaning audio. Generating multi-language subtitles. Rough-cutting long-form interviews into short-form clips. Tools like Descript, Riverside Magic Editor, and Adobe’s AI features in Premiere are doing this work invisibly. Your audience will never know.
2. B-roll and stock-style visuals. AI-generated footage of generic scenes (an office, a city skyline, abstract animations) is a perfect use case. No one is being represented. No human is being replaced. You are just generating visual variety faster and cheaper than buying stock footage.
3. Voiceover for utility content. Internal training, accessibility narration on existing videos, language localisation of content originally produced in English. People accept synthetic voice for utility. They reject it for sincerity.
4. Personalised video at scale. Tools like Synthesia and HeyGen let you produce a base video with a real human, then automate name changes, tier swaps, or translation. The trick is that the base performer is real. The personalisation layer is AI.
The throughline: AI is in the production pipeline, not in the casting decision. The face, voice, and emotional register the audience sees and hears should still be human in the parts that ask the audience to trust you.
If you are running any of these plays, this is your sign to revisit them.
Fully AI-generated thought leadership videos. Some marketing teams are using AI avatars to create LinkedIn videos for executives who are too busy to record themselves. The audience reads it as “your CEO did not care enough to actually appear.” It hits trust harder than the time saved is worth.
AI-generated customer testimonials. This is now an active brand-safety issue. The FTC and the EU AI Act both consider undisclosed synthetic testimonials a deceptive practice. If you are tempted to “speed up” testimonial production with AI faces or voices, do not. Use real customers, even if it means fewer of them.
AI-narrated brand-story videos. The video on your homepage. The “this is who we are” video. The company anniversary reel. These are sincerity-heavy moments. Synthetic voice and AI imagery in this slot reads as cynical, even if it was actually faster and the script was great.
AI-generated influencer content. Running paid spend behind a fully AI-created influencer is a category that is growing fast and that is going to face significant regulatory pressure in 2026. If you are testing this, ringfence the spend, document the consent and disclosure path, and have your legal team look at it before it scales.
This one comes up in every marketing meeting on AI video and the answer is more nuanced than the loud voices on either side make it sound.
You should disclose when:
You generally do not need to disclose when:
The pragmatic recommendation: when in doubt, disclose. The reputational cost of being caught out is much higher than the reputational cost of an early “Made with AI tools” tag in your description. Recent research from a separate consumer study found that roughly 78 percent of consumers say they trust videos with real people more, but the trust gap closes meaningfully when AI involvement is explicitly disclosed.[4] Honesty is cheaper than discovery.
For more on the broader content authenticity shift, our analysis of the AI disclosure regulatory landscape covers the policy side.
Here is the working framework I am running with marketing clients right now.
1. Real people lead the trust-heavy moments. Brand story, founder thought leadership, customer testimonials, anything aimed at building genuine connection. Your CEO has 15 minutes and a Loom recording is fine. It is real. That is what matters.
2. AI handles the production pipeline. Auto-edit. Auto-caption. Generate the B-roll. Translate into 5 languages. Produce the cut-down clips for short-form. This is where AI saves you actual hours.
3. Personalisation is layered on top of real foundations. A real person records the base video. AI swaps in customer names, regions, or product tiers. You get scale without sacrificing the trust signal.
4. Disclose where it matters. A simple “Translated and adapted with AI” or “AI-edited from a longer interview” line in the video description costs you nothing and protects you from being called out.
5. Test trust signals as a metric, not just engagement. Most marketing dashboards track views, watch time, and CTR. Add a trust signal. Compare comments and sentiment on AI-heavy content versus human-led content. You will see the difference.
For brands using AI heavily in social and ad video specifically, our AI video ad ROI guide covers the paid side of this.
Three concrete actions that take less than an hour total.
Audit your last 5 published videos. Watch them with fresh eyes. Are any fully AI-generated (avatar, synthetic voice, no real person)? If yes, those are your highest-risk pieces. Plan the re-record with a real human in the chair, even if it is a 5-minute self-recorded Loom from your subject matter expert.
Update your AI disclosure standard. One paragraph in your brand guidelines: when AI is in the production pipeline (editing, B-roll, translation), you do not need to disclose. When AI replaces the human face or voice the audience expects to see or hear, you do. Get sign-off from legal. Get it in writing.
Run an A/B test. Take one piece of content you would have produced fully AI. Produce two versions: one fully AI, one with a real human in the central role and AI handling the rest. Run them on the same audience. Compare watch time, comments, and any conversion metric you have. Use the data internally to settle the “but AI is faster” debate.
Yes, more often than marketers expect. Animoto’s 2026 State of Video report found that 83 percent of US consumers say they can spot AI-generated videos, with robotic gestures, unnatural voices, and lack of emotional tone as the most common giveaways.
Animoto’s 2026 data found that 36 percent of US consumers say seeing AI-generated video lowers their perception of a brand. The hit varies by context: fully AI-generated content carries the most trust risk, while AI-assisted content (real human presenter, AI in editing) does not show the same penalty.
No. The data supports using AI in editing, post-production, B-roll, translation, and personalisation layered on real human content. The trust risk is concentrated in fully synthetic video where AI replaces the human face, voice, and emotional presence the audience expects.
Disclose when the AI replaces a human role the audience would assume is real (a presenter, an endorser, a customer testimonial). You generally do not need to disclose AI used purely for editing, captioning, B-roll generation, or accessibility narration. When in doubt, disclose. The EU AI Act and several US state laws have specific labelling requirements.
Use AI in the production pipeline (editing, captioning, language translation, B-roll generation, personalisation at scale) while keeping real humans in the lead role of any video that asks the audience for trust, attention, or belief. This keeps the trust signal strong while still getting most of the speed and cost benefits AI offers.
This guide is part of Future Factors AI’s ongoing effort to make AI useful for non-technical professionals. Written by Hina Mian, Co-Founder of Future Factors AI, an AI training company that has helped 2,000+ learners build practical AI skills through bootcamps, corporate workshops, and keynote sessions. Visit our AI Courses page to learn more.
Weekly tips. Real examples. Practical help for busy professionals.
We care about your data, check out our privacy policy.
