AI images and video in campaigns: where they help and where they hurt the brand

Why campaigns are full of AI images and video

In two years, tools that make images and video with AI have gone from “interesting experiment” to everyday work. Google’s Nano Banana and Veo, OpenAI’s Sora and Adobe’s Firefly now produce results that, at first glance, are hard to tell apart from a photo shoot or a film set.

AI has also moved inside the ad platforms themselves. Meta swaps backgrounds, builds variations and, since October 2026, makes video from static images for every advertiser. Google does similar things in Google Ads and Merchant Center. Often you don’t even need another tool: the platform suggests “improved” versions of your ad on its own.

There’s a more practical reason too. On Meta, creative has become the new targeting: the system needs many, varied ads to find your audience. More ads mean more production, and AI is the fastest way to produce. The question is no longer whether you’ll use it, but where.

Where they genuinely help

Let’s start with the good part, because there’s a lot of it. AI is excellent when an idea and a style already exist and you want to take them further, faster:

• Variations and sizes: the same ad in square, vertical and horizontal formats, or in versions for different seasons and offers, without a new shoot each time.
• Backgrounds and settings: your product, unchanged, in a different space or setting, to show where and how it’s used.
• Motion from still images: a short video for Reels and Stories from photos you already have.
• Ideas before production: sketches, scripts and “rehearsals” of a campaign, so you can decide what to shoot before you pay for a shoot.
• Things you can’t photograph: ideas like security, speed or savings, which a conventional photo struggles to show.

In all of these cases, AI works on something that has already been decided: a product, a style, a message. That’s where it really saves time and money. The problems start when we let it decide the basics too.

Where they hurt the brand

The most common problem is no longer six fingers or strange lettering. Those are rare now, and when they happen, they show. It’s something subtler: ads that look like no one in particular. The same models, given similar instructions, produce similar results: the same glossy light, the same “perfect” faces, the same colours. If your ads look like your competitors’, customers won’t remember who was who.

The second problem is consistency. Ten images from ten different prompts look like ten different companies. A brand is built through repetition: the same colours, the same style, the same tone, again and again. When every ad has its own “personality”, recognition slowly slips away without you noticing.

The third is more serious: AI changes whatever it shows. A garment with a slightly different texture, packaging in another shade, a dish more “generous” than the real one. For an e-shop, that isn’t a matter of taste. It means returns, complaints and, at worst, misleading advertising.

When the audience can tell it’s AI

At Christmas 2024, and again in 2025, Coca-Cola released ads made with AI. Technically they were polished, but many viewers called them “soulless”, at exactly the time of year when the brand sells emotion. In December 2025, McDonald’s in the Netherlands released an AI-made Christmas ad and pulled it three days later after a strong backlash.

In both cases, the mistake wasn’t the tool. It was using it where the audience expects human care: warmth, celebration, family. When viewers spot that something was made with AI, they often conclude the brand cut corners where it shouldn’t have.

This matters even more in sectors built on trust. For a medical practice, a clinic or a beauty service, an AI face that looks like a real patient, or a “result” that never happened, isn’t just a poor choice. It can be misleading, and the consequences go well beyond the comments section.

Labelling and rules: what applies from 2026

Since 2 August 2026, the AI Act’s transparency rules apply. Put simply: images, video or audio created or altered with AI that resemble real people, places or events, to the point that someone could take them as real, must be disclosed as such. The changes to the law adopted in 2026 pushed back other deadlines, not this one.

The platforms add their own labels too. Meta labels ads created or significantly edited with its own AI tools, and tries to detect automatically those made with outside tools, from the “invisible” signals those tools leave inside the file. Google asks that AI-generated product images keep those signals when they’re uploaded to Merchant Center.

Then there’s ownership. An image made entirely by AI may not be protected the way a photo shoot or a design is, and every tool has its own terms for commercial use. For a banner, that changes little. For a mascot, a character or anything that will become part of your identity, it changes a lot.

The rules your brand needs

The answer isn’t to ban AI. It’s to decide once how you use it, so you don’t have to rethink it for every ad. That belongs in your brand book, just like your colours and typography:

• Where yes and where no: which channels, topics and campaigns use AI, and where you always want real photography.
• What never changes: the product, the faces of your team, the results you show.
• One style, not ten: a specific visual style that’s recognisable in every image, whoever makes it.
• Who signs off: every image and every video is checked by a person before it goes out, and that person knows what to look for.

The images on this blog, for example, are made with AI. But their style was set once and stays the same in every article, which is why they look like they belong together. How to make this concrete for a business, which style, which limits and how every partner sticks to them, depends on the sector, the audience and how mature its identity already is. That’s where the work we do comes in.

FAQ about AI images and video

Do I have to say an ad was made with AI?

If it shows people, places or events that look real and could be taken as real, yes: since August 2026 the AI Act requires it. For a clearly graphic or illustrative result things are simpler, but platforms may add a label on their own.

Do AI-made ads perform worse?

Not in themselves. Neither Meta nor Google “penalises” an ad because it was made with AI. What decides the result is, as always, whether the ad is good, distinctive and believable.

Can I make all my product photos with AI?

The background and setting, often yes. But the product itself has to look exactly as it is. Whatever AI changes in the product, the customer will see when it arrives.

Is it cheaper than a photo shoot?

In production, usually yes. But the real cost is measured differently: how many revisions it takes, how consistent the result is, and what it costs when an ad makes your brand look cheap.

AI is neither a threat nor a magic wand for your brand. It’s a very fast tool, and the faster it is, the more it matters to know where it’s heading. If you already make creative with AI, or are thinking of starting, at CNCTECH we look at your ads with you and tell you where AI helps, where it pulls you away from your brand, and which rules you need.

This article describes the tools, platform policies and rules as they stood in November 2026. AI tools change very quickly. For legal obligations, such as labelling AI content and copyright, consult a legal adviser.