Should Brands Use AI Influencers.

A tool can now build an AI influencer from a single photo, so the question is no longer whether a brand could launch one, but whether it should.

Our answer is that an AI influencer works as a clearly labelled brand character and fails as a fake person. For anyone in Brands & campaigns work, that line matters more than the technology, because UK advertising guidance already points to it.

The Advertising Standards Authority, which regulates ads in the UK, counts a virtual persona as an influencer in its advice. It has also warned advertisers that an AI-generated celebrity endorsement that looks real enough to mislead people is likely to breach its rules. Platform rules on AI content are a separate layer, covered in what every platform now does about AI content.

What the tools can now do

Take Higgsfield, an AI image and video tool. Its AI Influencer product says it can create an influencer from one photo or from scratch.

With a photo, Higgsfield says the builder keeps that person’s identity while you change everything else, such as age, body, hair and wardrobe. Without one, you build a new person from settings, including age, body type and skin and face details.

The character can then perform a motion taken from a trending video, or from a clip of 3 to 30 seconds that you upload, and the finished video comes out at 1080p. Higgsfield charges in its own credits, and says creating or editing a character costs 1.12 of them.

So making a convincing synthetic presenter is no longer the hard part. Deciding what it is allowed to pretend to be is.

The ASA already counts a virtual persona as an influencer

The ASA’s advice on influencer marketing says it typically defines an influencer as “any human, animal or virtually produced persona” active on a social media platform. There is no follower threshold and no required job title.

The advice says plainly that it is not legal advice and does not bind the ASA. But it tells you how the regulator is likely to look at an AI character: as an influencer, judged by the same rules as a person.

Why a separate persona account is where the risk sits

The ASA’s advice draws a distinction worth understanding. It contrasts content on a brand’s own social media page with an influencer talking about a brand, where it is not self-evident whether that is an independent opinion or something the brand has incentivised.

Its rule for that second case is strict. When an influencer is personally or commercially connected to a brand, for example as its owner or an employee, content featuring the brand must be obviously identifiable as advertising. The advice says this is a requirement under consumer protection law, enforced by the Competition and Markets Authority and Trading Standards.

Our reading is simple. An AI character a brand built is about as connected to that brand as anything can be. If it runs its own account, presented as a personality in its own right, every post that features the brand’s products should be labelled as an ad.

What a proper ad label looks like

The ASA’s advice is specific about labels, and none of it changes because the influencer is synthetic:

• The ASA is likely to expect a prominent ad label upfront, which usually means at the beginning.
• People should know a post is an ad before they click, expand it or start watching.
• The ASA and the Competition and Markets Authority prefer labels that say it outright, such as Ad.
• The advice recommends against the word sponsored, because people read it in different ways.
• Tagging the brand, or saying in the bio that the account is an ambassador, has not been enough on its own.

Both the brand and the influencer are responsible for getting this right. With an AI influencer, the brand is both.

Deepfakes are where it fails

The ASA’s June 2026 note on AI and deepfakes makes three points that matter here.

First, the advertising code applies however the content was made. The ASA describes it as media-neutral.

Second, an AI-generated celebrity endorsement that looks real enough to mislead people is likely to breach the code. The ASA does not regulate personality rights or intellectual property, but it tells advertisers to consider the legal risk of using a public figure’s likeness without permission, and to seek legal advice.

Third, the advertiser stays responsible for any AI output it uses in its ads, and should check that output before it runs.

Put that next to a tool that keeps a person’s identity from one photo, and the danger is clear. The feature that lets a founder turn themselves into a presenter is the same one that could put a real person’s face on a brand’s product without their say.

Do you have to say it is AI?

Not always, on the ASA’s reading. Its May 2025 article on AI disclosure says there is no blanket legal requirement in the UK to disclose the use of AI in ads, and that the existing rules apply however content is made.

Instead, it suggests two questions. Is the audience likely to be misled if the use of AI is not disclosed? And if so, does the disclosure clarify the ad’s message or contradict it?

The article is clear that a label cannot rescue a misleading ad. Its example is an AI image showing the effect of a cosmetic product that does not reflect real results: saying AI was used does not make that acceptable. But it also says that making clear an influencer is AI-generated “could well help to negate an otherwise misleading impression”.

For an AI influencer, our answer to the first question is almost always yes. A persona that looks human and talks about products is exactly the case where an audience could be misled. So we would label it as AI, in plain words, wherever it appears.

A brand character or a fake person

This is the line the whole decision turns on.

A brand character is presented as the brand’s creation. It is labelled as AI, its posts about the brand’s products are marked as ads, and it never claims an experience it cannot have.

A fake person is presented as someone real. It shares routines, reactions and results as if it lived them. That is where the ASA’s warnings land: an endorsement that looks real enough to mislead, and results that do not reflect what the product actually does.

An AI presenter that says it uses a moisturiser every morning is a testimonial from nobody. We would not sign that off for any brand.

Where an AI character fits, and where it does not

It fits where the character is a format, not a witness. A labelled presenter who explains, demonstrates or fronts a recurring series can be useful, because the audience knows what it is watching.

It fits badly where personal experience is the claim. Skincare results, fitness journeys and before and after stories all depend on a real person having lived them.

Take extra care with young audiences. The ASA says influencer content aimed at children under 12 is likely to need more than an ad label upfront, because children under 12 lack the critical understanding to recognise marketing that is closely built into the content around it.

If the founder’s real question is how to grow the brand’s social presence, rather than whether to try this one idea, a social audit and strategy is the better place to start.

The questions to answer before saying yes

When a founder asks for an AI influencer, these are the questions we would put back:

1. Will it be labelled as AI everywhere it appears?
2. Will it post from the brand’s account, or from its own? If its own, every post featuring the brand needs an ad label upfront.
3. Whose face is it built on? If it is a real person, do you have their permission and legal advice?
4. Will it ever describe using the product? If so, rewrite the script.
5. Who checks each output before it goes out?
6. Will the audience include children under 12?

If any answer is unclear, the answer to the founder is not yet.

How NBK thinks about AI influencers

An AI character is not a shortcut around the work. It is a new content pipeline that produces output at speed, and the ASA is clear that the advertiser is responsible for any AI output it uses in its ads.

That makes it an operations question. Someone writes the scripts, someone checks the output, someone confirms the labels, and someone keeps a record of what ran. If those steps do not exist for your human content, they will not appear for a synthetic presenter. Our social media approval process template is a free way to put them in place.

Next step

If your team is being asked to say yes to new formats faster than your process can check them, NBK can help rebuild the workflow behind the content.