2 in 5 display and video ads are now made with AI: here's the data

Our estimate: around 2 in 5 display and video ads running today have been made at least partly with generative AI, and roughly 1 in 10 is made entirely by it.

That is not a guess plucked from the air. It is built from the best industry data available in late 2026, platform disclosures from Meta, and what we see across our own clients' accounts. Nobody publishes an exact figure, because no ad server labels creative by how it was made. So below we show our working, split by format, and explain what it means for brands planning 2027 budgets.

What the data says

The most reliable number we have is for video: a third of digital video ad assets use generative AI in 2026, up from a quarter in 2025. That comes from the IAB's annual survey of US video buyers, and it is the closest thing the industry has to a census.

Source	Published	Finding
IAB Digital Video Ad Spend & Strategy Full Report	July 2026	Nearly 2 in 3 video buyers use GenAI for creative; 1 in 3 video ad assets use it, projected to reach 43% by 2027
IAB / Sonata Insights, The AI Ad Gap Widens	January 2026	83% of ad executives have deployed AI in the creative process (60% in 2024); 73% use it for display ads, 85% for social
Advertiser Perceptions, AI and Creative	Undated	72% use AI for static images such as banners; 24% use AI for all their creative assets
Meta Q2 2026 results, via PPC Land	July 2026	9 million small businesses now use at least 1 of Meta's AI ad creative tools

Two cautions. First, these are mostly US surveys, but UK budgets run through the same platforms and tools, so we see no reason the UK picture differs much. Second, a survey of buyers measures what advertisers say they make, not what audiences actually see. Platforms that generate variants automatically push the share of impressions higher than the share of briefs.

Clicky's estimate

We estimate that display ads are further along than video: roughly half of display creative now has AI in it, against just over a third of video.

By "in part" we mean any ad where generative AI produced or altered something the viewer sees or hears: a generated background, an extended or resized image, rewritten headline variants, a synthetic voiceover. By "totally" we mean no camera and no designer's original artwork were involved. AI used only for targeting or bidding does not count.

In practice, the line between "AI ad" and "real ad" has already blurred. A typical campaign today might use filmed footage with an AI-extended background, a product shot cleaned up by AI, a synthetic voiceover and headlines written by a model. No single element gives it away, and the finished ad carries no label. Even trained eyes struggle: in a Columbia Business School study of display ads, viewers judged most AI-generated images to be human-made or could not tell. That is why we give ranges rather than a single figure, and why most ads are now neither purely human nor purely AI, but a mix of the 2.

Format	Partly AI-generated	Fully AI-generated	Main basis
Display (banners and static social)	45 to 55%	10 to 20%	73% of advertisers already use AI for display; static images are the easiest asset to generate, and platforms now create variants automatically
Video (online, social and CTV)	33 to 40%	5 to 10%	IAB's 33% asset share, nudged up for UK social video, where Meta and TikTok tools do the most automatic editing
All display and video combined	around 40%	around 10%	A blend of the 2 rows above

Why is fully AI-generated so much lower? Because most brands use AI to adapt, extend and version work that started with a real shoot or a designer's master file. A product still needs to look like the product. Fully synthetic work is growing fastest in lower-cost social formats and among smaller advertisers, where the alternative was often no video at all.

Watch: a real shoot, finished with AI. Original Source's 2025 TV ad, made by the agency Fold7, filmed a real actor on real sets for the opening and closing shots. About 70% of what you see in between was generated with AI, working from footage of his expressions. It was the brand's first TV ad in years, and few viewers would guess how it was made.

Why video is catching up fast

Video's AI share is rising by roughly 10 points a year: a quarter of assets in 2025, a third in 2026 and a projected 43% in 2027, according to the IAB. The share of buyers using it has gone from half to nearly 2 in 3 in a single year.

3 things are driving it:

• Platforms build it in. Meta, TikTok and YouTube now turn still images and product feeds into video inside the ad manager. Meta says 9 million small businesses use at least 1 of its AI creative tools.

• Variants beat hero films. Algorithms reward fresh creative, and AI makes 20 cut-downs as cheap as 1. In 2025 the IAB found the top use was building different versions of the same ad for different audiences.

• Demand outruns supply. Among smaller US buyers, 96% say they are not yet using GenAI as much as they want to. That is pent-up demand, not resistance.

On current trends we expect the gap between video and display to close by 2028, as video models catch up with filmed footage.

Why your Netflix and Prime Video ads look different

If the ads interrupting your film lately have felt a little synthetic, you are probably not imagining it. Both of the UK's biggest ad-supported streamers have publicly committed to generative AI in their ads.

Netflix announced at its May 2025 upfront that it would launch midroll and pause ads using generative AI to blend advertisers' ads with the world of its programmes, rolling out to every ad-supported market by 2026. Netflix later clarified that the AI personalises the experience around the ad, such as placement and overlays, rather than creating the ad itself.

Prime Video goes further. Amazon has announced AI-generated contextual pause ads, created on the fly to match the scene you are watching. It also hands advertisers the tools to make the ads: in February 2026 it launched Creative Agent for UK advertisers, which researches, storyboards and produces video and display ads, alongside its Video Generator and Image Generator. One industry report says over 60% of the video generator's users are first-time video advertisers, many of whom then appear on Prime Video.

You may also have started spotting small print on streaming ads saying they were made partly with AI. Those labels come from advertisers, not the platforms, and there are good reasons they are appearing. There is no blanket legal requirement in the UK to disclose AI in ads, but the ASA expects it where viewers could otherwise be misled. Across the Channel, the EU AI Act's deepfake transparency rule has applied since 2 August 2026, and it can catch UK ads that are accessible in the EU. Brands running one creative across Europe often label it everywhere. The fear that a label hurts results also looks overblown: a MediaScience study of video ads found no decline in any performance measure when AI labels were added.

Labelling is still patchy, so for most spots no viewer can be certain either way. But the direction is clear: streaming TV is where AI creative is moving from social feeds onto the biggest screen in the house.

Do audiences actually care?

They say they do, but their behaviour suggests most barely notice. Surveys paint a sceptical picture: a 2026 Bentley University and Gallup poll found 49% of Americans view AI-created ads negatively, against 19% who view them positively. Yet when researchers measure what people actually do, the effect largely disappears. A MediaScience study found adding AI labels to video ads caused no decline in any performance measure. And when Coca-Cola's AI Christmas ad drew a storm of YouTube criticism, a Truescope survey found 83% of people were simply neutral.

The IAB's January 2026 research found 82% of ad executives believe younger consumers feel positive about AI-generated ads. Only 45% actually do, and that gap has widened from 32 points to 37 in a year. Gen Z is the most sceptical group, with 39% feeling negative.

Watch: the backlash in action. Coca-Cola rebuilt its 1995 "Holidays Are Coming" ad with generative AI in 2024 and again in 2025. Both drew heavy criticism in YouTube comments, even though the 2024 version scored the maximum rating on System1's ad testing platform. This is the 2025 AI version.

There is a gap, though, between what people say about AI ads and how they actually respond to them. Many people say they are turned off by AI advertising, yet most have no reliable way of knowing which ads are AI-made. In the Columbia Business School study of more than 4,600 display ads, AI-generated images got click-through rates on a par with human-made ones, and converted those clicks no worse. What hurts is the look of AI, not its use. Ads that feel cheap, uncanny or generic get punished, whoever or whatever made them, while well-crafted AI work passes unnoticed. The Coca-Cola case makes the same point: loud criticism online, but a top score when the ad was tested properly.

And even where some viewers are put off, the commercial case still wins. AI lets a brand test dozens of versions of an ad for different audiences instead of 1 or 2, and make ideas that would otherwise be impossible or unaffordable to film. After Original Source's part-AI ad aired, the brand's value sales rose 4.5% and it reached its highest market penetration in 4 years (The Grocer). Coca-Cola's AI studio said its 2024 Christmas ad took 2 months instead of the year traditional CGI and filming would have needed (Tom's Guide). A small dip in sentiment among a vocal minority is easily outweighed by more variants, faster testing and creative that stops the scroll.

The honest answer: a vocal minority cares a lot, most people are indifferent, and almost nobody changes what they buy because an ad used AI. What they do notice is whether an ad feels cheap or deceptive.

What it means for brands

The risk is no longer whether to use AI in creative but whether audiences trust what you make with it. Advertisers are far more positive about AI ads than the people seeing them.

It is worth separating 2 very different kinds of AI ad. The first is deliberately fantastical: a lion playing basketball, a farmer floating in a pool of eggs. Nobody thinks it is real, and before AI the same idea would simply have been made with CGI, which took far longer but was no more "real". Whether a talking animal comes from an animation studio or a video model makes little difference to the viewer. The second kind is AI that looks like real life: synthetic people giving testimonials, or product shots that make an item look better than it is. That is where trust is genuinely at stake. The IAB's own guidance draws the same line, advising disclosure when AI creates a material risk that consumers will be misled.

Disclosure helps more than many brands assume. 73% of Gen Z and Millennial consumers said knowing an ad was made with AI would increase or make no difference to their likelihood to buy. Yet fewer than half of advertisers using GenAI always disclose it.

Our advice for 2027 planning:

1. Use AI to raise quality, not just cut cost. Audiences punish creative that looks cheap, not creative that is AI-assisted.

2. Keep a human sign-off on every asset. 40% of video buyers say they want humans in the loop, and brand risk sits with you, not the platform.

3. Have a disclosure policy before you need one. Decide when you label AI imagery, voices and avatars, and apply it consistently.

4. Test AI variants against your control. Measure them on sales and leads, not clicks.

The bottom line

AI is already in around 2 in 5 of the display and video ads you scroll past, and in 1 in 10 it did all the work. By the end of 2027 it will be close to half of video. The brands that win will not be the ones producing the most AI creative, but the ones using it to make better work, faster, with a human hand on the wheel.

At Clicky we have built an AI Video capability that pairs the latest video models with our creative and paid media teams, so you get the speed and scale of AI with the craft and sign-off of an agency. It can also create shots that would be impossible or uneconomic to film. If you want to see what it could do for your next campaign, get in touch.

Sources

• IAB, Business Outcomes Are Just the Beginning: 2026 Digital Video Ad Spend & Strategy Full Report (14 July 2026)
• IAB and Sonata Insights, The AI Ad Gap Widens (15 January 2026)
• IAB, 2025 Digital Video Ad Spend & Strategy Full Report (15 July 2025)
• Advertiser Perceptions, AI and Creative Report
• PPC Land, coverage of Meta's Q2 2026 results
• Netflix, Upfront 2025: The Center of Attention (14 May 2025), with MobileSyrup's clarification
• Fast Company via Owler, Prime Video contextual AI-created pause ads
• Advanced Television, Amazon Ads launches Creative Agent (12 February 2026)
• Nova Analytics, Prime Video ad tier and AI video generator (22 December 2025)
• The Grocer, Original Source's AI-generated TV ad boosts sales and sentiment (2 July 2025)