Instagram AI Info Label on a Real Photo

Last updated · 6 min read

Quick answer

The label usually comes from hidden marks in the file. Meta says it adds "AI info" when it detects industry-standard AI indicators, which means C2PA or IPTC metadata. Editing apps with generative tools write these marks, even into real photos. Check your file with the AI metadata checker. If no AI was used, a copy without the marks usually avoids the metadata-based label, but Meta may use other signals.

What the label means

"AI info" is the label that Meta shows on Facebook, Instagram and Threads for content that it believes AI made or edited. Meta's newsroom post on labeling says it adds the label in two cases: when it detects industry-standard AI indicators in the file, or when the person who posts discloses that the content is AI-generated.

DateWhat Meta changed
February 2024Announced that it would label images when it detects C2PA or IPTC "AI generated" information from companies such as Google, OpenAI, Microsoft, Adobe, Midjourney and Shutterstock
May 2024Started labeling organic content as "Made with AI"
1 July 2024Renamed it "AI info", because minor changes such as retouching had received the old label
12 September 2024Moved the label into the post's menu for content that AI only modified or edited. Content that AI generated keeps a visible label.

The wording may be different in your app today. One 2026 guide reports that Instagram's help pages now say "AI content". We could not confirm that on Instagram's own site.

Why a real photo gets the label

The label does not judge how the photo looks. It reacts to marks in the file. Two kinds matter:

  • C2PA (Content Credentials). A signed record of how a file was made or changed. Adobe tools and some AI generators write it.
  • IPTC DigitalSourceType. A field in the XMP data. The value "trainedAlgorithmicMedia" says that a generative model made the image.

Meta told TechCrunch in July 2024 that it did not change the technology that detects these marks. It said that if photographers use a tool like Adobe's Generative Fill to remove objects, their photos might still get the label. PetaPixel tested this in May 2024. Removing a speck of dust with Generative Fill triggered the label. Doing the same with the Spot Healing Brush, Content-Aware Fill or the Clone Stamp tool did not.

Other apps can write these marks too. A March 2026 article by Roboin says that images from ChatGPT and Gemini, and images edited with Magic Eraser in Google Photos, contain C2PA data. We have not tested each app. Check your own file.

How to see the marks in your file

  1. Take the exact file that you plan to upload, not a screenshot of it.
  2. Open the AI metadata checker and add the file.
  3. Look for a C2PA record, the name of the app that wrote it, and a DigitalSourceType value.

The checker runs in your browser. It does not upload your photo. If it shows nothing, the file has no marks for Meta to read. Meta can still label a post that you made with Meta AI or that you marked yourself.

What you can do

  • Check the file first. Know which app added the marks.
  • If you did not use AI, export the photo again from the original without the generative step. Or remove the AI metadata from a copy before you upload it. A copy without the marks does not carry them to Instagram.
  • If you used a generative tool, the label tells viewers the truth about the edit. Meta made "AI info" to cover images that AI modified, not only images that AI generated.
  • If you added the label yourself, a 2026 guide by Narqo says you can edit the post and change the AI label setting. The exact menu may differ in your app version.
  • If Meta added it and you think it is wrong, we found no Meta page that describes an appeal. You can send feedback with Report a problem in the Instagram help menu. Meta does not promise a review.

What a clean file does not guarantee

Meta said in February 2024 that people can strip invisible markers, and that it is working on classifiers that detect AI-generated content without them. It also uses invisible watermarks in images from its own AI tools. A file without metadata can still get a label.

Cleaning a file does not remove pixel watermarks such as Google SynthID. WipePic removes file metadata and does not change the picture. See Does removing metadata remove SynthID? for the detail. WipePic removes C2PA records, AI labels in XMP and prompt text in PNG files, and it does not upload your photos.

Sources

Checked on 2026-11-08. Meta's own posts are marked.

Questions

Why does Instagram say "AI info" on a photo that I took myself?

Meta says it adds the label when it detects industry-standard indicators that AI made or edited the content. An editing app can write these marks into the file when you use a generative tool, for example to remove an object. In a 2024 test by PetaPixel, removing one speck of dust with Photoshop Generative Fill triggered the label.

Will removing the metadata stop the label?

If the label came from marks in the file, a copy without those marks does not carry them. That is usually enough for the metadata-based label. But Meta also uses self-disclosure and its own AI tools, and it has said that it is working on classifiers that do not need metadata. So no tool can promise that the label will not appear.

Can I remove the AI info label from my post?

If you turned on the AI label yourself, a 2026 guide by Narqo says you can edit the post and change that setting. We found no Meta page that describes a way to remove a label that Meta added from the file. Instagram has a Report a problem option in its help menu, but we cannot say that Meta will review the label.

Does the label mean that my photo is fake?

No. Meta says that "AI info" covers content that AI made or modified. It renamed the label from "Made with AI" in July 2024 because minor edits, such as retouching, were getting the old label.

Remove the metadata from your photos

Free. In your browser. Your photos do not go to a server.

Clean a Photo