Should brands work with AI influencers?

For product visualisation, concept testing and openly stylised brand worlds, they can be useful. For testimonials, recommendations and expertise, real people remain more credible, because trust in influencer marketing rests on lived experience.

What are the biggest risks of virtual influencers?

Loss of trust when a synthetic person makes claims that sound like personal experience, backlash in categories such as beauty or health, and a generic look that makes the brand less distinctive.

Does advertising law apply to virtual influencers?

Yes. Paid content must be recognisable as advertising and claims must be accurate, whether a real or a virtual person presents them.

AI Influencers and Virtual Models for Brands

Q: Do AI generated models have to be labelled?

Yes in practice. Meta and TikTok require realistic AI generated content to be labelled, and the EU AI Act introduces transparency obligations for such content from August 2026. Check current details with legal counsel before publishing.

Brands can use AI influencers and virtual models, but only where the audience does not need a real human to believe the message. They work well for product visualisation, concept testing, campaign variations and stylised brand worlds. They work poorly wherever trust depends on lived experience: testimonials, recommendations, expertise and anything that sounds like a personal opinion. And wherever they are used, they need to be clearly labelled.

Virtual influencers are not new. Lil Miquela has been posting as a computer generated character since 2016 and has worked with major fashion labels. What has changed is the cost and speed of production. Generative tools now let almost any brand create a consistent synthetic person in days, which is exactly why the question deserves a clear answer instead of an experiment by default.

Where virtual models genuinely help

Product visualisation at scale. Showing a garment, a piece of furniture or packaging in many settings without a shoot for every variant.
Concept and pre production. Testing looks, moods and campaign ideas before investing in a real production.
Stylised brand worlds. When the brand is openly artificial, futuristic or illustrative, a virtual figure can be part of the concept rather than a substitute for a person.
Full control. No scheduling conflicts, no usage right renewals for a person's likeness, no off brand behaviour outside the campaign.

Where they cost you trust

Influencer marketing works because people trust people. A recommendation carries weight because someone with a real life, real skin and a real reputation stands behind it. A synthetic person saying that a serum changed their skin is not a testimonial, it is an advertisement dressed as one, and audiences increasingly notice. The backlash risk is highest in categories where the product promise is physical: beauty, health, fitness, food.

There is also a quieter cost. A virtual model trained to look like an ideal makes your brand look like every other brand using the same tools. Distinctiveness, the thing premium brands pay for, comes from specific people, specific places and specific light. We discuss this balance in more depth in AI images or photography.

Labelling: what applies

Platform rules

Meta and TikTok both require realistic AI generated or significantly altered content to be labelled, and both apply their own labels when they detect it. Unlabelled synthetic content that looks real can be restricted or removed. Use the built in disclosure tools rather than hoping nobody notices.

The EU AI Act

The EU AI Act introduces transparency obligations for AI generated and manipulated content in Article 50, applying from August 2026. In short, content that realistically depicts people who do not exist or things that did not happen must be disclosed as artificially generated. Details and possible transition rules are still being clarified, so check the current status with legal counsel before a campaign goes live.

Advertising law still applies

A virtual influencer does not change the basics. Paid content must be recognisable as advertising, and product claims must be true. If a synthetic model presents results the product cannot deliver for real people, that is misleading regardless of who or what delivers the line.

A practical rule for brands

Use virtual models where you would otherwise use a mannequin, a render or a mood board. Use real people where you would otherwise need a witness. Label generously, keep a human face for your expertise and testimonials, and for founder led brands remember that the most credible face is often your own, as we describe in personal branding for managing directors.

FADETOBLACK combines editorial photography and video with generative AI production, and helps brands decide where each one belongs. If you are considering AI models or a virtual brand figure, Book a free initial consultation.