AI Imagery Statistics 2026: What 502 US Shoppers Think of AI Product Images
Prateek Gupte
Last updated 12 November 2026. Source: Caimera Consumer Report 2026, a survey of 502 US consumers fielded April to May 2026.
85% of US shoppers could not reliably tell AI-generated product images from real ones, and 79% trust the brand that labels its AI imagery over the one that doesn't. Those are the two headline AI imagery statistics from Caimera's 2026 survey of 502 US consumers, matched to the US census on age, gender and ethnicity.
This page publishes every figure from the study in one place. You can quote any of it; the citation line is at the bottom.
Key statistics
70% of US consumers are open to AI imagery in commerce: 14% see it as just another tool and 54% say it is fine in some contexts.
10% say AI imagery is harmful, and about 5% have stopped buying from a brand because of it.
85% (428 of 502) could not reliably distinguish AI-generated images from real ones in a six-image quiz.
2 of 502 people got a perfect score on that quiz.
79% trust the brand that labels its AI imagery when two brands both use it.
74.7% say brands should always disclose AI imagery.
47% prefer a product page that mixes real photos with clearly labelled AI images. 37% want all real photos.
75% are neutral to positive about AI imagery on social media, the most accepting channel.
35% are neutral to positive about AI imagery for beauty and skincare, the least accepting category.
How do consumers feel about AI imagery overall?
Most shoppers are conditionally open. Outright rejection is a minority view.
| |
| Attitude toward AI imagery in commerce | Share of respondents |
|---|---|
Just another tool |
14% |
Fine in some contexts |
54% |
Uncomfortable in commerce |
20% |
Harmful |
10% |
No opinion |
2% |
The report combines the first two groups into an "openness score" of 70%.
The 10% who call AI imagery harmful are not all acting on it. About half of them, roughly 5% of all respondents, say they have stopped buying from a brand over it.
Where is AI imagery accepted?
Respondents rated AI imagery on a 1 to 5 scale for each channel. The figures below are the share who gave a 3, 4 or 5, meaning neutral to positive.
| |
| Channel | Neutral to positive (3–5 of 5) |
|---|---|
Social media (Instagram, TikTok) |
75% |
Billboards and outdoor |
68% |
Paid online ads |
67% |
Brand emails |
63% |
Influencer collaborations |
54% |
Editorial and news |
38% |
Acceptance drops where people expect the image to be a record of something real. Editorial and news sits 37 points below social media.
Which product categories do shoppers accept AI imagery for?
Same scale, by product category.
| |
| Product category | Neutral to positive (3–5 of 5) |
|---|---|
Home and furniture |
65% |
Electronics and gadgets |
60% |
Apparel, flat lay |
57% |
Jewellery and watches |
52% |
Apparel, on model |
51% |
Food and beverage |
49% |
Supplements and health |
41% |
Beauty and skincare |
35% |
For fashion brands, two numbers matter. Apparel shown as a flat lay scores 57% and apparel on a model scores 51%, so a majority is neutral to positive in both cases. The six-point gap suggests shoppers are a little more careful once a person is in the frame.
The pattern across the table: acceptance is higher where a shopper can check the product against the picture, and lower where the picture is the evidence of a result, as with skincare or supplements.
Can people tell AI images from real ones?
Mostly, no. Every respondent took a six-image quiz mixing AI-generated and real images.
| |
| Quiz result | Respondents |
|---|---|
Could not reliably distinguish AI from real |
428 of 502 (85%) |
Could reliably distinguish AI from real |
74 of 502 (about 15%) |
Perfect score |
2 of 502 (0.4%) |
Do consumers want AI imagery disclosed?
Yes, and by a wide margin, even though most can't spot it.
| |
| Should brands disclose AI imagery? | Share |
|---|---|
Yes, always |
74.7% |
Only in certain cases |
18.4% |
No, it doesn't matter |
4.4% |
Unsure |
2.0% |
Among people who see AI as "just another tool", 56% still want disclosure.
Respondents were also shown two brands that both use AI imagery, one that labels it and one that doesn't.
| |
| Which brand do you trust? | Share |
|---|---|
The brand that labels |
79% |
Avoid both |
9.4% |
See no difference |
7.4% |
The brand that doesn't label |
4.0% |
The report notes this preference holds across age, gender and attitude group, and calls it the most universal preference in the dataset.
If you are working out what the law requires, as opposed to what shoppers prefer, see our legal and ethical guide to AI fashion imagery and the New York AI disclosure law explainer.
What do shoppers want on a product page?
| |
| Preferred product page format | Share |
|---|---|
Real photos plus clearly labelled AI images |
47% |
All real photos |
37% |
No preference |
17% |
A labelled mix beats an all-real page by 10 percentage points.
What is driving the concern?
The survey collected 496 open-text responses and grouped them by theme.
| |
| Theme in open responses | Share |
|---|---|
Authenticity |
34% |
Deception |
21% |
Misrepresentation |
17% |
Transparency |
14% |
The complaint is rarely about AI as such. It is about whether the product will match the picture. In respondents' words:
"AI doesn't bother me as long as it's not being used to mislead people."
"What we see should be what we get, and I don't always experience that with AI generated photos."
"As long as the product is good and as described, I think AI is fine."
What the numbers mean for a brand
The report reduces the findings to three working rules.
Disclose by default. Label AI imagery on product pages and in ads. On Instagram and TikTok the platform label can do the job.
Use AI for the setting, not the substance. Backgrounds, lighting, room scenes and seasonal mood are accepted. Changes to skin, colour accuracy, scale or claimed results break trust.
Build hybrid product pages. Keep a real photo on the page and add labelled AI images alongside it.
Caimera adds disclosure at download: a text label, the European Commission's AI labels, and image metadata.
Methodology
| |
Sample |
502 US consumers |
Representativeness |
Matched to the US census on age, gender and ethnicity |
Panel |
Prolific, with screeners |
Fieldwork |
April to May 2026, in three waves |
Cleaning |
528 responses collected; 26 removed for failed attention checks, screeners, or completing in under 120 seconds |
Quiz |
Six images, AI-generated and real, shown to all 502 respondents |
Open text |
496 responses coded by theme |
The survey measures stated attitudes and a short recognition quiz. It does not measure purchase behaviour, and it covers the US only.
How to cite
Caimera Consumer Report 2026: Insights from 502 US Consumers for AI Imagery in Commerce. Caimera, 2026. https://www.caimera.ai/blogs/ai-imagery-consumer-statistics
You are free to quote the figures and reuse the charts with a link to this page. The full report, with the channel-by-channel permission map, is available at caimera.ai/consumer-report-us-2026.
FAQ
What percentage of people can tell AI images from real ones?
About 15%. In Caimera's 2026 survey, 74 of 502 US consumers could reliably tell AI-generated images from real ones in a six-image quiz, and 428 (85%) could not.
Do consumers trust brands that use AI images?
Most are open to it. 70% of US consumers are open to AI imagery in commerce, and 10% consider it harmful. Trust depends on disclosure: 79% trust the brand that labels its AI imagery over one that doesn't.
Do shoppers want AI-generated images labelled?
Yes. 74.7% say brands should always disclose AI imagery, and a further 18.4% say they should in certain cases.
Is AI imagery accepted for clothing?
By a majority. 57% are neutral to positive about AI imagery for apparel flat lays and 51% for apparel on a model.
Where is AI imagery least accepted?
Beauty and skincare (35% neutral to positive), supplements and health (41%), and editorial or news content (38%).
About Caimera. Caimera is an AI visual production platform for fashion teams, from designer sketch to live campaign. Try it free.
