Instagram's “AI info” label explained: why it appears and how to handle it honestly

2026-10-16 · 5 min read

More and more creators post AI-made pictures on Instagram and then notice a small tag on the post, shown as AI info (earlier it was called “Made with AI”). People ask where it comes from and what they can do about it. This is an honest explanation. It is based on what Meta says publicly, and platform rules change, so check Instagram’s current help pages too.

What the label is

The label is Meta’s way of telling viewers that a post was made or changed with AI. It is a transparency feature, not a penalty. Plenty of creators who make AI art use it openly and keep their audience.

How a post gets the label

Meta says it uses several signals together:

Signals inside the file. Some AI tools write notes into the picture file, such as Content Credentials (C2PA) or an IPTC “digital source type” entry that says the image was made by AI.
What you tell Instagram. When you upload, you can say that a post contains AI-generated content. Meta asks people to do this for realistic AI content.
Other signals. Platforms do not publish all of them, so you cannot know from the outside what else is used.

What is inside an AI picture file

This is the kind of information an AI picture can carry. The list below is a real report from Image Cleaner on a test file that we prepared with typical notes (GPS is only reported as “found”).

Before (the file we tested) 8

Camera maker Example Phone Co
Camera model Example Model 1
Software Example AI tool 1.0
Date and time 2026:10:14 18:20:11
GPS location found
Colour profile 128 bytes
Comment made with an example AI tool
Content Credentials (C2PA) a signed record that says how the image was made

After cleaning 0

Nothing left. The cleaned file was checked and carries no hidden data.

What cleaning the file does

Image Cleaner rebuilds the picture from its pixels, so these notes are not in the new file: camera and software names, dates, GPS location, prompt text, colour profile and the content-credential marker. That is good for privacy: you do not post your location or your prompt by accident. It also removes the small sparkle logo from the corner of Gemini images.

The picture is a demo picture we made for this website (no real people, no client work). The sparkle was drawn by us in the same shape as the Gemini one, and the list comes from a test file we prepared. The cleaning and the report are the real Image Cleaner.

Example portrait picture with a white sparkle in the bottom-right corner — Before: sparkle in the corner

The same example picture after cleaning, the corner is clean — After: clean corner

What cleaning does not do

It does not make an AI image a real photo. It is still AI-generated, and it still carries invisible watermarks such as SynthID if the tool added one. See what SynthID is.
It does not replace your own disclosure. Instagram asks you to label realistic AI content, and it can add a label from other signals. We do not offer, and will not build, a way to hide the AI origin of a picture.
It cannot promise what a platform will show. We do not know, and nobody outside the platform knows, exactly how every post is labelled.

The honest way to post AI images

Use Instagram’s own AI label when you upload realistic AI content.
Say it in the caption too, for example “Made with AI (Gemini)”. A short line is enough.
Clean the file for privacy and a tidy look, not to look “real”.
If you work with brands, tell them. See the disclosure checklist.

If a real photo of yours gets an AI label

Some editing tools use AI for small steps (for example generative fill or object removal) and write an AI note into the file, so a mostly real photo can be labelled. If that happens to you: check which editing step added the note, keep your original and raw files as proof of how the photo was made, and use Instagram’s help and support options to ask for a review. Be accurate about what you used.