The face recommending a product in your feed may not belong to a person. Brands are deploying AI-generated influencers across social media, and presenting the results as genuine customer experiences.
A Guardian investigation found paid posts and short videos in the user-generated style that give no obvious sign the people featured are synthetic. The deception is the design, not a side effect.
Where it has surfaced
The investigation flagged specific cases. At Once, a photo app, analysis by Reality Defenders pointed to likely use of AI-generated talent. Once did not respond to requests for comment.
Maket said it had experimented with AI influencers to test creative ideas at small scale. The Dubai fashion label Ashle pulled images after the paper asked about them, saying early launch shots had used AI while insisting the garments are real.
Why brands are making the switch
The motive is cost. Creators and agencies in the investigation described a move to AI because it is cheaper and easier to scale than a traditional shoot.
A real shoot needs models, a crew, a location and a budget. A synthetic influencer needs none of that and can be produced at volume. Non-disclosure agreements keep the arrangement quiet, which is why the practice spread before most viewers noticed it.
The regulator's gap
This is where the story turns from clever marketing to something closer to a loophole. The Advertising Standards Authority said its rules do not explicitly require AI content to be labelled.
Complaints are judged on whether an ad misleads, not on whether AI was used. That distinction sounds reasonable until you apply it. A synthetic person posing as a satisfied customer is built to mislead, yet the framework treats the AI itself as neutral.
What viewers can actually tell
The case for disclosure rests on a simple fact. People cannot reliably spot the difference.
Which? Tech cited a consumer test in which 70% of people failed to correctly identify all the real and fake videos shown to them. If most viewers cannot separate genuine from generated, the choice not to label is a choice to let them assume it is real.
Which? has called for clearer disclosure when promotional content features AI-generated influencers.
Who benefits from the silence
The current arrangement favours one side. Brands get cheaper, scalable content and face no requirement to admit how it was made. Viewers carry the burden of detection they keep failing.
The technology will improve, which means the synthetic faces will get harder to catch, not easier. The rules were written for a world where an advert featuring a person featured a person.
That world is ending, and the regulator has not yet decided whether anyone needs to be told.
