The AI Ad Perception Gap: Advertisers Think 82% of Young Consumers Like AI Ads. 45% Do.

The AI ad perception gap is now 37 points

Advertisers have the wrong number in their heads. In the IAB and Sonata Insights study, 82 percent of ad executives assume young consumers feel positive about AI-generated ads, and only 45 percent actually do. That AI ad perception gap has widened from 32 points in 2024 to 37 points in 2026, and it is now a measurable planning error that video teams can audit before the next campaign ships.

The mismatch is not a vibe, it is a measured number. IAB and Sonata Insights surveyed 505 US Gen Z and Millennial consumers who engage with ads, plus 104 ad industry executives at companies spending at least $1 million a year on media, between October 2025 and January 2026. Eighty-two percent of the executives said they believe young consumers feel very or somewhat positive about AI-generated ads. Only 45 percent of the consumers agreed. The 37-point spread is the single most actionable number in the study, because it quantifies how far the industry's working assumption sits from the audience's own report.

This is the second consecutive year the gap has grown. A comparison survey of 300 consumers and 75 executives ran between August and October 2024 with the same design, and the spread then was 32 points. Five points of drift in two years is not noise; it means the assumption is getting worse at the exact moment AI creative is shipping at higher volume. Teams that last checked audience sentiment in 2024 are planning against a number that was already wrong.

The cohort split matters as much as the headline. Gen Z consumers are nearly twice as likely as Millennials to feel negative toward AI ads, at 39 percent versus 20 percent, and the majority of Gen Z respondents describe AI-using brands as inauthentic, disconnected, or unethical. A single blended comfort score hides that. Anyone targeting 16-to-27-year-olds with generated creative is operating in the most skeptical band of the audience.

Advertisers credit innovation; consumers say manipulative

The two groups are not even describing the same thing when they evaluate AI in advertising. Advertisers associate AI use with innovation at 46 percent and uniqueness at 44 percent. Consumers reach for different words: 20 percent describe brands using AI as manipulative, double the 10 percent of ad executives who pick the same descriptor, and 16 percent call such brands unethical versus 7 percent of executives. These are brand-attribute words, not campaign feedback.

That distinction changes the stakes. A weak click-through rate on one campaign is a media problem you can fix next quarter. When your audience files your brand under manipulative or unethical, the cost lands on brand equity, pricing power, and the trust every future campaign borrows against. The study's authors frame it the same way: disclosure can play a decisive role in determining whether AI use in advertising becomes a long-term value driver or a short-term liability.

The gap is also asymmetric in a way that should worry media planners. Advertisers are not slightly off; they are off by a factor of two on the negative attributes while being roughly aligned on the positive ones. An audience model that overweights innovation and underweights manipulation risk will keep approving creative the audience reads as a warning label.

The gap widened while deployment scaled

The uncomfortable part of the study is the direction of travel. The perception gap grew from 32 to 37 points across a period in which the industry leaned harder into generative creative, not softer. The buying side scaled with it, and teams routed more volume through {{link}} while their read on the audience drifted the other way. Confidence and deployment rose together; calibration did not.

Stated sentiment is only half of the evidence, because {{link}} shows the behavioral side moving the same direction: younger viewers skip ads at high rates and increasingly name AI-made creative as a reason to disengage. When what people say and what people do both point away from the assumption, the assumption is not conservative, it is simply wrong.

There is a structural reason this drift happens. Audience comfort is nobody's KPI. Creative teams are measured on output volume and performance, procurement on cost per asset, and media on delivery. Nobody owns the sentence about the audience believing AI use makes a brand less trustworthy, so the number never enters a dashboard, and the assumption quietly hardens into a fact.

The buying side scaled with it, and teams routed more volume through agentic AI video media buying while their read on the audience drifted the other way.

Stated sentiment is only half of the evidence, because collapsing ad tolerance shows the behavioral side moving the same direction: younger viewers skip ads at high rates and increasingly name AI-made creative as a reason to disengage.

Measure audience comfort instead of assuming it

The fix is unglamorous and cheap: put the question in research you already run. Add two items to brand tracking, one on how positive or negative people feel about ads made with AI, one on whether a brand's use of AI changes how trustworthy it seems, and split every result by cohort. The study shows why the split is non-negotiable: 39 percent negative among Gen Z against 20 percent among Millennials is the difference between a passing grade and a warning.

Then benchmark against the published baseline instead of intuition. Forty-five percent positive is the 2026 reference point for ad-engaging young consumers; your audience may sit above or below it, but you cannot know that from the boardroom. A team that ran this question once in 2024 and got a friendly answer is five points of drift behind, planning the next quarter against data that expired.

Treat the output as a routing input, not a report. High-comfort segments can take generated volume without ceremony. Low-comfort segments get human-led creative, heavier disclosure, or a different format entirely. That is what a measured audience model buys you: the ability to direct AI video where it earns attention and hold it back where it taxes trust.

Disclosure is a net positive, not a tax

The strongest objection to measuring comfort is the fear that honesty costs sales. The study runs the other way. Seventy-three percent of Gen Z and Millennial respondents say clear disclosure of AI use would either increase or have no impact on their likelihood to purchase. More than half want brands to disclose when an ad is fully AI-generated or uses AI imagery or video. The audience is asking for the label the industry is afraid to print.

The industry now has a concrete shape for what to disclose. IAB's AI Transparency and Disclosure Framework, released alongside the research, is risk-based rather than universal: consumer-facing labels attach to material cases such as synthetic humans, digital twins, generated images and videos depicting real-world events, and AI chatbots simulating human interaction, while routine production assistance stays unlabeled. The machine-readable layer travels with the file as provenance metadata under the C2PA protocols, the Content Credentials standard maintained by the Coalition for Content Provenance and Authenticity.

For a commercial video team the practical reading is simple: disclosure is not where trust is lost, silence is. Labeling the material cases satisfies the majority who want to know, and the 73 percent figure says the label will not suppress purchase intent. What suppresses intent is being caught unlabeled.

What commercial video teams should change this quarter

First, instrument the assumption. One tracking question with cohort splits, benchmarked against the 45 percent baseline, costs a fraction of one generated asset batch and replaces the industry's worst-kept guess with your own number. Re-run it quarterly; the drift from 32 to 37 points proves the number moves.

Second, set a disclosure default before seasonal volume ships. Material cases get the label every time: synthetic performers, digital twins, chatbot interactions, generated depictions of real events. Routine assistance stays unlabeled, which keeps the label meaningful instead of triggering the disclosure fatigue a universal mandate would create.

Third, remember that transparency converts only when the work lands, which is why {{link}} remains the variable that decides whether any of this matters; IAB's own research lead pairs transparency with creative quality for exactly that reason. And none of it survives contact with production reality without an owner, because {{link}} is still why most teams cannot say where AI touched their last campaign. Measure the audience, label the material cases, ship the quality. Every part of that is checkable, and the 37-point gap says nobody is checking yet.

Third, remember that transparency converts only when the work lands, which is why the emotional premise remains the variable that decides whether any of this matters; IAB's own research lead pairs transparency with creative quality for exactly that reason.

And none of it survives contact with production reality without an owner, because the AI oversight gap is still why most teams cannot say where AI touched their last campaign.