Sledru

News explained

AI personas and the new label: what changed on 31 August

Instagram renamed its label for AI-generated profiles and began pulling unlabelled ones out of recommendations. Here is who the rule actually covers, what happens to reach, and how to recognise an AI account without relying on a label.

Alexey Rusakov6 minutesupdated September 3, 2026

In short

Since 31 August, profiles built around invented people must carry the “AI-generated profile” label or lose recommendations in Reels and Explore. Ordinary use of AI tools is untouched. The label is a useful signal but not a complete one: its absence proves nothing, and telling artificial accounts apart still comes down to audience signals.

What changed

On 31 August 2026 Instagram renamed the “AI creator” label to “AI-generated profile”. The wording is now direct: the old one read as “this author works with AI tools”, the new one says “the person on this profile was generated by AI”.

At the same time the platform started limiting reach for profiles that skip the label. Their posts stop being recommended to non-followers: they disappear from Reels and Explore.

The reason was stated plainly. People complained that they mistook AI personas for real humans — followed them, read them, trusted them, and found out by accident.

The label is not a penalty. A labelled profile stays in recommendations like any other — the platform restricts silence about AI, not AI itself.

Who the rule covers

The main confusion of the first days: many assumed everyone who has ever touched an AI tool now needs the label. That is not the case.

What you doLabel needed
Retouch your own photos with AIno
Generate graphics and coversno
Write captions with AI helpno
Edit video in an app with AI featuresno
Run a profile as an invented personyes
The face on the avatar is generatedyes

The rule is about identity, not tooling. If the audience sees a person who does not exist, the label applies. If the person is real and AI only helped with editing, it does not.

What happens to reach

The mechanism is familiar to anyone who has followed recommendation limits before. The account is not banned, posts are not deleted, followers keep seeing everything. Only the inflow of new people stops.

For an AI persona that is a hit to the foundation: such projects live almost entirely on cold traffic from Reels and Explore rather than on a returning audience.

  • Detection and owner notifications roll out gradually over the coming weeks.
  • The current status of a profile is visible in Account Status.
  • There are two ways back into recommendations: add the label or appeal.
  • Posts that already carry an AI-created label do not need a second one.

If your reach dropped in early September and your profile is an ordinary human one, this rule is almost certainly not the cause. A different August change explains it more often: recommendations shifted towards original photos and carousels, and aggregator accounts reposting other people’s work were removed from Explore.

Why virtual influencers multiplied

The rule did not appear out of nowhere. Over the past year the barrier to running such a project fell close to zero: a generated face, staged interiors and a voice come together in one evening, and the profile can be run from anywhere with no camera and no person on set.

The economics are simple. Virtual influencers do not get ill, miss deadlines or argue about revisions, so advertisers try them willingly. Audiences, meanwhile, kept assuming they were watching a real person — and that mismatch is what the platform decided to close.

Hence the shape of the rule. Instagram does not ban AI projects and does not demote them for existing. The restriction fires in exactly one case: when the artificiality is hidden.

How to apply the label

  1. Open your profile and tap “Edit profile”.
  2. Find the “AI-generated profile” toggle and turn it on.
  3. Check the profile: the label appears under the name and next to posts.

The same toggle removes it. But if the profile really is AI-generated, removing the label brings the reach limit back as soon as automatic detection catches up.

How to spot an AI account yourself

The label will not appear everywhere, and not at once: detection is automatic, which means incomplete. Until then the work falls back on indirect signals — the same ones that reveal ordinary fake audiences.

What the posts show

  • The face is identical across every frame to an unnatural degree: same geometry, same lighting.
  • Hands, jewellery and background text — the places where generation still slips.
  • No everyday debris: no random passers-by, no reflections, no unflattering angles.
  • The author’s replies are formulaic and uniformly warm to everyone.

What the audience shows

Signals on the follower side are more reliable than staring at fingers in photos. AI projects almost always inflate their audience, and that leaves traces in the lists.

  • Many followers with no avatar, no posts and names made of letters and digits.
  • A sharp jump in the follower count while reactions stay flat.
  • Followers barely overlap with the profile’s topic.
  • The same accounts leave reactions under every post.

Checking these signals by hand across several thousand followers is not possible. Sled collects the public follower list and shows how many of them carry suspicious signals — without logging in, and without inventing a “bot percentage” that does not exist.

Check a profile’s audience

What it means for an ordinary user

The label is the first time the platform has offered a public marker of an artificial profile. Until now this could only be guessed at.

Relying on it as a guarantee is premature, though. The label is voluntary and detection is automatic, so some profiles will stay unlabelled — by oversight or on purpose, until a check catches them.

A missing label does not prove the person is real. A present label proves the opposite — that the person is invented.

Does the label apply to the account or to individual posts?

To the whole profile. It appears under the name and next to posts. Individual posts may carry their own AI-created label, in which case the profile label is not repeated beside them.

What happens if I skip the label?

Posts stop being recommended to non-followers: they disappear from Reels and Explore. The account itself is not banned and followers keep seeing the posts.

I retouch my own photos with AI. Do I need the label?

No. The rule covers profiles where the person themselves is generated. Editing real photographs, making graphics and getting help with text do not require it.

How do I remove the AI label?

With the same toggle in profile editing. But if the profile really is AI-generated, automatic detection will restore the reach limit.

Can the restriction be appealed?

Yes. The status is shown in Account Status and the appeal is filed there. The other route is simply to add the label.

Sources

Check a creator before a paid post