Consumer Research Highlights a Trust Gap as 29% Would Give an AI Advertiser Little or No Trust

Consumers may accept an AI-generated advertisement when the brand explains it upfront, while discovering that fact after a purchase can weaken trust and future buying intent.

A ContentEngine survey of 446 U.S. adults found a clear relationship between the timing of AI disclosure and consumer reaction. Respondents who learned about AI involvement after buying a product were less likely to trust the advertiser, less likely to buy from it again and more likely to feel misled than respondents who received the disclosure before purchasing.

One figure in the headline deserves a careful note. The survey materials supplied for this analysis report that 65.0% of the later-discovery group would trust the advertiser at least a fair amount. That means 35.0% selected not too much or not at all. The reported table does not show 29% for that measure. The headline’s 29% framing may come from a different calculation or source version, so the article relies on the detailed table and reports the 35% lower-trust share where precision matters.

The survey presented two samples of 223 adults with nearly identical purchase scenarios. Every respondent imagined seeing an online advertisement for a product they had not planned to buy. The ad convinced them to make the purchase, and every respondent was told they were satisfied with the product afterward.

The timing of the AI information changed between the samples:

The trust results favored the upfront disclosure group. Among those respondents, 76.7% said they would trust the advertiser a great deal or a fair amount. The corresponding figure for the later-discovery group was 65.0%, a difference of 11.7 percentage points.

Repeat purchase intent moved in the same direction. About 79.8% of respondents who knew about the AI before buying said they would be very or somewhat likely to buy from the advertiser again. That fell to 67.3% among respondents who learned about the AI afterward, a 12.6-point difference.

Feeling misled showed the reverse pattern. Fifty-seven percent of the upfront group felt misled a lot or some, compared with 70.0% of the later-discovery group. That 13-point gap was the largest difference measured in the survey.

The results suggest that a positive product experience does not fully repair the damage caused by a disclosure that arrives after the purchase decision. A consumer can like the product and still revise an opinion of the advertiser.

A survey of 446 U.S. adults found that consumers reacted more favorably when an advertisement disclosed its complete use of AI before purchase rather than revealing it afterward. Upfront disclosure was associated with higher advertiser trust and repeat purchase intent, while later discovery was associated with stronger feelings of being misled. The findings also show that disclosure alone is not enough: most respondents wanted a person to review and approve AI-generated advertising, and many remained uneasy even when they received information about AI in advance. Because the study used hypothetical scenarios, separate nonprobability samples and complete AI-generated creative, it offers a practical warning rather than a universal measure of consumer behavior.

Disclosure before a purchase gives consumers information while they can still decide whether the advertising approach fits their preferences. Disclosure afterward provides context, but it arrives after the main decision has already been made.

That timing distinction matters especially for an unplanned purchase. The scenario described advertising that persuaded respondents to buy something they had not intended to purchase. A later revelation may lead consumers to reconsider the influence that brought them to the checkout point.

The survey does not prove that timing alone caused every difference. The later-discovery scenario explicitly said that the ad had failed to disclose its AI use. That wording combines two ideas: learning about AI later and learning that the advertiser withheld the information. The second idea may have directly increased feelings of deception or manipulation.

Even with that limitation, the pattern gives advertisers a useful warning. A disclosure placed in a post-purchase notice, terms page or later brand communication cannot perform the same consumer-information role as a clear notice attached to the advertisement before the decision.

Transparency improved the measured outcomes, yet it did not remove discomfort. In the upfront group, 57.0% still said they felt misled a lot or some. Another 62.3% felt manipulated at least somewhat.

Several features of the scenario could explain those reactions. The purchase was unplanned. The ad was described as persuasive. The entire advertisement, including its words and images, was created using AI. Each detail may have influenced how respondents interpreted the experience.

The survey did not ask respondents to explain their answers, so it cannot identify the main source of the discomfort. Some respondents may have reacted to AI production. Others may have reacted to the persuasive force of an advertisement that changed an initial buying decision. Some may have viewed both factors together.

That uncertainty matters for how brands interpret the results. An AI label cannot compensate for questionable claims, aggressive targeting or creative that makes consumers feel pressured. Disclosure gives consumers information about production. It does not automatically make the underlying advertising feel fair or appropriate.

The most useful lesson is not simply to add an AI label; it is to treat disclosure as part of the consumer decision rather than as a post-purchase footnote. Advertisers should explain the scope of AI involvement before the relevant choice, assign a human owner to review the claims and creative, and measure trust and repeat behavior alongside conversion. The survey cannot isolate the effect of timing from the effect of being told that the advertiser failed to disclose AI, but that combination still reflects a real marketplace risk: consumers may interpret late transparency as concealment.

Support for transparency was broad across both samples. Of the 446 respondents, 402, or 90.1%, said it was very or somewhat important for brands to tell consumers when an advertisement had been created entirely using AI.

Human oversight received nearly the same level of support. A total of 394 respondents, or 88.3%, said it was very or somewhat important to know that a person had reviewed and approved an AI-generated advertisement before publication.

The human-review result was especially consistent between the two groups. It reached 87.9% in the upfront-disclosure group and 88.8% in the later-discovery group. The difference was less than one percentage point.

These responses point to two separate consumer expectations:

A label answers the first question. A documented review process addresses the second. Treating a disclosure as proof of human accountability leaves an important gap.

The distinction also fits the direction of current policy and industry guidance. The European Commission’s material on Article 50 of the EU AI Act discusses transparency obligations for AI systems, including the role of the legal person operating the system. The IAB’s AI Transparency and Disclosure Framework V2 takes a risk-based approach to deciding when and how AI involvement should be disclosed in advertising and marketing content.

Those frameworks do not turn every use of an AI tool into the same disclosure problem. Generating an entire advertisement raises different questions from using AI to resize an image or produce minor copy variations. The ContentEngine survey tested the stronger case, in which AI created all of the words and images.

The scenario describes an ad that worked. It convinced a consumer to buy, and the consumer was satisfied with the product. A campaign report focused only on conversion could therefore classify the creative as successful.

The survey adds a longer-term measure. Once AI involvement was discovered after the purchase, trust and repeat purchase intent fell. That creates a measurement issue for advertisers evaluating AI-generated creative. Immediate conversion may capture the first response while missing later effects on brand relationship.

Useful follow-up measures could include:

The survey found a smaller difference in recommendation intent: 73.1% of the upfront group said they would recommend the advertiser, compared with 67.7% of the later-discovery group. That six-point gap did not remain statistically significant after adjustment for the study’s eight comparisons. It still provides a possible direction for future research, especially with larger and more representative samples.

The survey offers a useful signal, but it does not represent all U.S. adults. The researchers used separate nonprobability online samples through SurveyMonkey Audience. Participants opted into online panels, the results were not weighted and respondents were not randomly selected from the full U.S. population.

The groups also were not randomly assigned to the two scenarios. Measured demographic characteristics did not differ significantly between them, but similar age, gender, income, device and regional profiles cannot rule out differences in unmeasured attitudes or experiences.

The wording creates another limitation. The later-discovery group was told that the ad had not disclosed its AI use. That condition may have produced stronger negative reactions than a scenario that simply revealed the production method later without describing the omission as a failure to disclose.

The survey was hypothetical. Respondents did not see a real advertisement, spend money or use a real product. Stated trust and purchase intentions may differ from behavior in an actual marketplace, where price, product quality, brand history and personal need also influence decisions.

The study also tested complete AI generation. Its findings may not apply to every use of AI in advertising. A human copywriter using AI for early ideas, a designer using AI to create image variations or a media team using AI for audience analysis may produce different consumer reactions.

The statistical analysis adjusted for eight comparisons using the Holm method. After that adjustment, trust, repeat purchase intent and feeling misled remained statistically significant. Differences in regret, recommendation, feeling manipulated and the perceived importance of disclosure were less conclusive. The survey did not establish a universal response to every form of AI advertising.

The clearest operational lesson is simple: disclosure should appear before the consumer makes the relevant decision. The notice should also describe the scope of AI involvement in plain language. A brand that uses AI for the full text and visual package faces a different transparency question from one that uses a tool for background editing.

Human review should have a real owner and a defined checklist. Reviewers can examine factual claims, visual accuracy, legal risk, audience targeting, emotional pressure and whether the disclosure itself is visible enough to matter. Approval should mean that a person accepts responsibility for the final creative rather than merely clicking through an automated workflow.

Brands can also test disclosure language before launching a campaign. A short notice may be easier to understand than technical language about models or production pipelines, but it still needs to be prominent and specific. Consumer research can compare reactions to different placements and levels of detail.

The ContentEngine findings place trust after the purchase alongside performance during the purchase. That broader view gives advertisers a better way to judge AI-generated creative. Consumers may accept AI involvement when they receive the information early, understand what it means and can see that a person remains accountable for the result.