Does AI disclosure change ecommerce ad trust?
A practical 3-minute test design for measuring how AI disclosure changes trust, perceived quality, and purchase intent in matched ecommerce video ads.
Lamina Team
Product Team @ Lamina
Will disclosing an AI-generated product video ad cut brand trust and purchase intent?
AI disclosure can drag down trust, perceived quality, and purchase intent in some ad studies. It is not a universal penalty. A label may make people question the effort, authenticity, credibility, or content quality behind an ad; for people who see AI as capable, it can instead read as novel and draw a more favorable response.
That split is why you should test matched ecommerce creative yourself, rather than write a disclosure label off as either harmless or fatal. The supplied research includes no results from a Lamina 3-minute consumer test, so this report gives you a reproducible protocol, not a claim that Lamina creative has already passed one. Keep the product, offer, edit, music, and brand treatment fixed. Change the disclosure alone.
| Metric | Value | Source |
|---|---|---|
| Valid responses in an online short-form-video experiment that found a positive direct engagement path but an indirect quality-related penalty from disclosure | 479 | dl.acm.org |
| Participants in a study finding that AI disclosures increased AI and persuasion knowledge, with offsetting pathways to trust in the ad and organization | 304 | dare.uva.nlas of 2025 |
| Participants in an AI-generated video-ad experiment reporting no direct disclosure effect on purchase intention | 246 | run.unl.ptas of 2025-10-27 |
| Australian respondents who said they would trust a brand less if its ads were created mainly with AI; this was a hypothetical stated reaction, not a matched-creative experiment | 45% | yougov.comas of 2026-05-20 |
| Australian respondents who said they would be less likely to consider a brand whose ads were created mainly with AI; this was a hypothetical stated reaction, not a causal disclosure-label test | 44% | yougov.comas of 2026-05-20 |
What should a matched ecommerce AI-disclosure test measure?
Measure brand trust, perceived product quality, perceived production quality, and purchase intent separately. Those readings can pull apart. Published work finds that quality, credibility, authenticity, novelty, and attitudes toward AI can each account for part of a disclosure effect.
Do not mash those responses into one approval score. A shopper may still think the product looks well made while reading the ad as low effort; someone else may like the novelty of the disclosure and trust the brand less. You make different creative and disclosure calls from those two diagnoses.
How do you run a 3-minute matched-creative AI disclosure test?
Create one approved product-video master
Generate an on-brand ecommerce video in Lamina, then lock the SKU, product claims, price or offer, voiceover, edit, soundtrack, duration, and call to action. Check the item hard before testing: material, color, logo, packaging, and feature accuracy all need to hold. Bad product truth contaminates every response.
Build a disclosure-only variant
Duplicate the master exactly. Add a plain disclosure, such as “This ad was created using AI,” in the placement and format your team is considering; the control gets no AI-origin label. Do not combine the disclosure with a new hook, a lower-resolution export, or changed copy.
Randomly show one version to each respondent
Give every participant either the control video or the disclosed version. Never both in the same short test. That stops people from comparing labels instead of reacting as a shopper would after seeing one ad in a feed.
Ask outcome questions right after viewing
Ask one plain question apiece about trust in the brand, perceived product quality, perceived production quality, and likelihood of considering or buying the product. Use identical scales and wording in every condition. Add a simple recall check to confirm that respondents shown a label actually noticed it.
Test timing and wording separately
After you have the basic control-versus-label read, run disclosure timing and wording as their own variants. Video-ad research identifies timing as one possible way to alleviate negative effects. Other studies show the outcome can shift with perceived AI capability and attitudes toward AI.
Read the mechanism, not just the winner
Compare each outcome across conditions, then look for a drop that gathers around quality, authenticity, credibility, or trust. A label that leaves purchase intent flat while lowering production-quality perceptions calls for a different response than one that shifts brand trust. Publish only creative that clears your brand-critical review.
Why can one AI disclosure help an ad and hurt another?
Consumers read more into an AI label than the production method. In published advertising studies, lower perceived effort and authenticity explain negative reactions in some settings. Elsewhere, novelty and favorable attitudes toward AI create a positive path.
Product context changes the read. A 2026 image-ad study found a stronger positive novelty route and a weaker authenticity penalty for utilitarian products, while luxury-ad experiments found that disclosure could trigger lower-effort and lower-authenticity inferences unless the imagery was highly creative. Treat category, price point, and creative quality as test conditions. They are not background details.
How should you read a neutral purchase-intent result?
A neutral purchase-intent result says the disclosure did not move stated buying likelihood in that test. It does not prove trust or perceived quality held steady. One 246-participant AI-video-ad experiment found no direct effect on purchase intention, while other supplied research reports negative effects on advertising value and purchase intentions through credibility-related mechanisms.
Keep the interpretation tight. If intent stays flat and trust drops, the label may still matter for repeat purchase, premium positioning, or a brand with little room for credibility loss. If trust holds and quality rises or remains flat, the creative has earned a broader validation round, not a blanket policy.
What does the industry evidence say about disclosure labels?
MediaScience CEO Dr. Duane Varan’s view matters because it puts creative quality at the center of the decision rather than treating the label as the whole story. That fits the mixed academic record. Judge the actual ad, the audience, and the disclosure execution together.
There has been a lot of anxiety in the industry about what happens when you tell people an ad was made with AI. The data gives us a clear answer: if the creative is good, disclosure does not hurt it. Advertisers do not need to be afraid of the label.
What is the practical call for ecommerce teams?
Use AI generation for product-video work, then validate disclosure with a matched test before making it a brand-wide rule. Lamina can produce the on-brand variants. A human still needs to approve product fidelity, claims, and the final disclosure treatment, especially in a brand-critical hero placement.
Start with your highest-volume, lower-risk product category, and keep the asset exactly the same in every condition. If the label creates a quality or trust penalty, test placement, timing, and wording before you change the creative concept. If it does not, record the result alongside the audience and product context. That beats a generic promise that consumers either love or reject AI-made ads.
Continue reading
AI product reel benchmark: attention vs product truth
A reproducible Lamina test design for measuring whether AI product reels can win early attention while passing a strict SKU-level product-truth gate.
Lamina Team
Product Team @ Lamina
Do AI ecommerce ads feel less authentic?
A practical blind-test framework for measuring whether AI ecommerce ads affect authenticity, trust, emotional connection, and purchase intent.
Lamina Team
Product Team @ Lamina
Best AI for product video ads: a hands-on benchmark of on-brand ecommerce ad creation with Lamina
Lamina is strongest for brand-governed, multi-asset ecommerce production; the best underlying video model still depends on whether product fidelity or cinematic motion matters most.
Lamina Team
Product Team @ Lamina