AI Ad Reveal: 28.3% Would Feel Deceived

28.3% of respondents in Magic Hour's survey of 1,019 U.S. adults selected “deceived” as a feeling they would have if they discovered an online ad was entirely AI-generated.

That is an emotional reaction to a hypothetical reveal, not evidence that those respondents encountered a scam or that an advertiser actually deceived them. But it gives a familiar question a concrete number: how would you feel if the ad you had been looking at turned out to be made entirely with AI?

This article focuses on the emotional-response question from our July 2026 AI advertising survey. It is not a new survey or a test of whether disclosure changes purchases.

Selected feelings about discovering an ad was entirely AI-generated: curious 29.0%, interested 28.7%, deceived 28.3%. Full study: 1,019 U.S. adults; multiple selections allowed; July 10-11, 2026.

The Reveal Produced Mixed Stated Reactions

The question asked respondents which feelings would describe their reaction if they discovered an online ad was entirely AI-generated. They could select multiple answers.

Three published responses were closely clustered:

Curious: 29.0%.
Interested: 28.7%.
Deceived: 28.3%.

These are descriptive percentages, not evidence of meaningful statistical differences between the three feelings. Because multiple selections were allowed, a person could be interested and also feel deceived. The categories are not competing camps that divide the sample into three mutually exclusive groups.

The broader published results were similarly mixed: 47.6% selected at least one positive emotion, while 44.5% selected at least one negative emotion. Those groups may overlap. Their difference does not establish that positive reactions “won.”

Feeling Deceived Is Not the Same as Proving Deception

The word “deceived” is the response option participants selected. It does not establish that the ad made a false claim, misrepresented a real product, impersonated someone or hid a required disclosure.

The question also specifies an ad that is entirely AI-generated. It should not be generalized to every use of AI in advertising, such as editing a background, drafting a caption or improving accessibility.

One plausible interpretation is that production methods can matter to how people feel about advertising. The survey does not tell us why someone selected “deceived.” Expectations about authenticity, the subject of an ad and prior experiences are possibilities to investigate, not measured explanations.

The Finding Does Not Mean Most People Reject AI Ads

In a separate trust question, 59.5% said their trust in a familiar brand would remain unchanged or increase if it used AI-generated advertising. The other 40.5% said trust would decrease.

Those answers cannot be cross-tabbed with the emotion results from the published summary. We cannot say whether the same people who selected “deceived” also reported lower brand trust.

Nor can we use these responses to forecast lost sales. The survey measured attitudes, not shopping baskets, refunds, clicks or conversions. A controlled disclosure experiment would be needed to estimate how learning about AI production changes an actual decision.

What Brands Can Take From This

The useful takeaway is not to hide AI use. This study does not test concealment, and it supplies no evidence that hiding a production method improves outcomes.

Instead, it makes a case for checking how an audience interprets an ad's presentation. Does the creative invite viewers to believe that a generated scene documents something real? Does its depiction match what the product can actually do? These are review questions, not conclusions about the ads used by any particular brand.

An advertiser can test clear production context and accurate creative against its own approved baseline. The relevant outcome should be observed trust or behavior, rather than an assumption that a visually convincing asset will be received positively.

A Citation-Ready Finding

In Magic Hour's survey of 1,019 U.S. adults, 28.3% said they would feel deceived if they discovered an online ad was entirely AI-generated.

Keep “would feel” and the hypothetical-reveal context. “28.3% were deceived by AI ads” would describe an event this question did not measure.

Methodology and Limits

SurveyMonkey conducted the survey for Magic Hour July 10-11, 2026. The full sample comprised 1,019 U.S. adults. Responses were unweighted.

The original questionnaire asks which words describe how respondents would feel if they discovered an online ad was entirely AI-generated and explicitly allows multiple selections. This article uses the published emotion percentages. Response-level files and item-specific confidence intervals are not available on the source page, so the figures are descriptive sample results, not nationally representative estimates or a significance test.

No purchase behavior, randomized disclosure treatment or causal effect was measured by this question. The article does not publish personal information, raw responses or user media.

See the full survey and its methodology for the separate trust, identification, influence and oversight questions.