Picture a profile. A smiling young woman, wellness tips every morning, flawless skin, hundreds of thousands of followers. She recommends a food supplement, with a link to buy it.
She does not exist. Her face, her voice and her videos were made by AI.
Since 31 August, Instagram has required this kind of account to say so. A month on, we can look at what this rule really changes, and above all what it lets through.
What has changed
Instagram already had an optional label, "AI Creator". It has been renamed "AI-generated profile", a much clearer name. And above all, it becomes mandatory for accounts whose main character is a person invented by AI.
The label appears on the profile, in search results, next to posts and videos, and even near comments and messages.
Accounts that do not play along risk a specific penalty: they are no longer recommended to people who do not follow them. No more promotion in Reels, in the Explore tab or in account suggestions. Instagram acknowledges that its systems can get it wrong, and provides for an appeals process.
The network explains its decision simply: its users do not like coming across a profile that seems human, only to discover later that the person does not exist.
The real danger: fake experts
AI-generated profiles have promoted apps, brands or wellness advice, sometimes presenting themselves as healthcare professionals. In August, an investigation by the New York Times catalogued hundreds of fake AI-generated doctors selling food supplements.
This is where the question becomes serious. A fake fashion influencer is a piece of theatre. A fake doctor advising you on your health is a danger.
What the label overlooks
It relies on good faith. The system depends first and foremost on creators declaring themselves. For the others, Instagram says it detects the accounts concerned, without explaining how, or how well it manages to.
It penalises visibility, not lying. A fake doctor who displays their label can carry on selling to their followers. The label tells you the person does not exist. It does not tell you that their advice is dangerous.
It does not cover the most common case. Real people who post images and videos made by AI do not need the label. Yet that is the bulk of the artificial content in circulation: not invented characters, but invented images posted by very real people.
It remains vague on the penalty. Instagram has not said by how much an undeclared account's visibility drops, or for how long.
What to take away
This label is genuine progress. It meets a legitimate demand, and it aligns with the transparency rules that Europe now imposes on artificial content.
But treat it as a clue, never as a guarantee. The absence of a label does not prove that a person exists. And for anything touching your health or your money, the real question remains the same, with or without AI: who is advising me, and what are they trying to sell me? As for the images themselves, our five checks still hold.