Do you have to disclose AI generated images in ads?

Is it a legal requirement to label AI ads in the UK?

At the time of writing, there is no UK law or CAP Code rule that says every ad made with AI must be labelled. The Advertising Standards Authority (ASA) has said that its rules apply to ads however they were made. So the real test is not "did you use AI?" but "is the ad misleading?"

That matters because AI can make an image look real when it isn't. If a reasonable viewer would take an AI image as a true record of what your product does, or of a real person using it, and that impression is false, the ad is likely to break the rules on misleading advertising. In that case a label might help, but it won't fix an image that misrepresents the product.

What does the ASA expect from beauty ads that use AI?

Beauty ads have a history here. The ASA has already ruled against cosmetics ads where filters or heavy retouching exaggerated what a product could do. The same thinking applies to AI. If an image is meant to show the effect of the product, it must not be enhanced beyond what the product really achieves.

In practice, that means:

• Don't use AI to create or improve before and after results.
• Don't make skin, lashes, hair or coverage look better than the product delivers in an image that's showing efficacy.
• Don't present an AI person as a real customer, reviewer or expert.
• Keep product details accurate: shade, size, finish and pack.
• Make sure any claim in the copy can be backed up, whatever the image shows.

Does Meta require AI labels on Facebook and Instagram ads?

Meta shows an "AI info" label on some content made or significantly edited with generative AI. For ads, the label can be added automatically when you use Meta's own generative AI ad tools, and Meta also uses industry metadata standards to detect content made with some third party tools. Meta has been expanding these labels for ads during 2026.

Separately, Meta requires advertisers running social issue, electoral or political ads to disclose when they use digitally created or altered realistic images, video or audio. Most beauty ads won't fall into that category, but it's worth knowing.

Because Meta's approach is changing, check the current advertising standards and transparency settings in Meta's business help centre before launch.

Does TikTok require AI labels on ads?

TikTok requires creators to label AI generated content that shows realistic scenes or people, and it has tools to add an "AI generated" label. It can also apply labels automatically when it detects AI provenance metadata on uploaded content.

TikTok's advertising policies have their own rules on AI generated content, including restrictions on misleading synthetic media and on using someone's likeness without consent. At the time of writing, if your ad shows realistic AI people or scenes, assume a disclosure is expected and check TikTok's current ad policies to confirm.

Does Google require AI disclosure in ads?

Google requires election advertisers to clearly disclose when their ads contain synthetic content that realistically depicts real people or events. It has also introduced information about how ads were made, including AI involvement, in its ad transparency tools, and it watermarks images made with its own generative tools.

For standard commercial ads, such as a beauty product on Search, Shopping or YouTube, the main rules remain accuracy and honesty. Shopping images must show the real product, and misrepresentation policies apply whether or not AI was involved. Check Google Ads policy pages for the latest position.

How do the main platforms compare?

A quick summary, at the time of writing. Treat this as a starting point, not a substitute for each platform's current policy.

If you advertise in the EU, the EU AI Act includes transparency duties for AI generated or manipulated content that could be mistaken for real, often called deepfakes. These duties are being phased in, so check the current position if the EU is a market for you.

Should you label AI ads even when you don't have to?

It depends on the image. A product sitting in a stylised, obviously imagined scene rarely misleads anyone. A realistic person with flawless skin holding your serum might. The more an image could be read as a true record of something, the stronger the case for disclosing.

Labels are also becoming more automatic. Platforms increasingly read provenance data from files, so it's worth assuming an AI label could appear whether you add one or not. Building ads that are honest from the start means a label never becomes a problem.

Think about your customers too. Beauty buyers often look closely at skin, texture and colour. An ad that feels too perfect can raise doubts even if it breaks no rules.

A good AI studio should help with this. At Niko x Studios, for example, real models are only featured with the brand's written permission, and we don't fabricate before and after results. See our AI ad creative service, and read can you use AI models in beauty ads and do AI images hurt customer trust for more.

What should you check before an AI ad goes live?

A short checklist before launch catches most problems. Run every AI made ad through it, alongside your usual claims review.

This guide is general information, not legal advice.

• Does the product look exactly like what the customer will receive, including shade, label and size?
• Is any part of the image showing a result? If so, is that result real and typical?
• Could a person in the image be mistaken for a real customer, expert or celebrity?
• If a real person appears, do you hold their written permission for this use?
• Have you checked the current AI rules for each platform you're running on?
• Would you be comfortable if the platform added an AI label to this ad?

Questions people also ask

Do I need to write "AI generated" on my product images?

+

Not as a blanket rule in the UK. You must not mislead, and some platforms will add their own labels. Disclose where an image could give a false impression.

Can Meta add an AI label to my ad without me choosing it?

+

Yes. At the time of writing, Meta can apply its AI info label automatically when you use its generative AI tools or when it detects AI metadata in your file.

Is it misleading to use an AI background behind a real product photo?

+

Usually not, if the product is shown accurately and the setting isn't presented as a real claim. The risk grows when images imply results or real customers.

Can I remove AI metadata to avoid a label?

+

We wouldn't. It doesn't change whether the ad is misleading, and platforms are moving towards stronger detection. Build ads that are fine to label instead.

Does the ASA have a specific AI rule?

+

Not at the time of writing. It applies the existing CAP Code to AI ads, with misleading advertising rules the most relevant.