What does the “AI content” label mean?
“AI content” (formerly “AI info”) is the label Meta adds to posts on Instagram, Facebook and Threads when it has a sign that AI was used to make or change the image or video. It tells viewers that AI played a part somewhere. It is a note on that one post, not a mark on your whole account. To see what your own file says, drop it into the tool above.
A two-word label can’t show the difference between a fully generated picture and a real photo where AI removed one small object. That is why it so often feels unfair to photographers and small businesses. Other apps word it differently: TikTok shows “AI-generated”, Pinterest shows “AI modified”, and LinkedIn shows a Content Credentials “CR” icon.
From “Made with AI” to “AI info” to “AI content”
In 2024 Meta began labeling posts on its apps with the words “Made with AI”. Photographers soon complained that real photos with small edits were getting it. In July 2024 Meta changed the wording to “AI info”, explaining that a tagged image is not always created with AI and may only have been edited with AI tools.
More recently the label changed again, to “AI content”. Older help pages, videos and forum answers still use the earlier names, so advice written for “AI info” usually applies to the new label too. The main trigger is still the same: AI metadata inside the file.
What triggers the AI content label?
Meta can add the label for one of four reasons. The first two are tags inside the file, which Meta reads when you upload, and on real photos they are the usual cause:
- Content Credentials (C2PA) in the file that record an AI step, such as a generative fill or an image created by an AI tool
- An IPTC Digital Source Type value of “trainedAlgorithmicMedia” (created with AI) or “compositeWithTrainedAlgorithmicMedia” (edited with AI)
- Your own choice: the AI label setting you can turn on when you share a post
- Meta’s own judgment, which can label a post even when the file carries no tags
Why does my Instagram post say “AI content” when the photo is real?
Because many everyday editing tools now use generative AI, and they record that in the file. Photoshop and Lightroom Generative Fill, Generative Remove and AI Denoise, Adobe Firefly and Express, Canva Magic tools, Google Photos Magic Editor and Pixel edits, Samsung Galaxy AI and Apple Clean Up can all add AI tags. Removing one stray person from the background is enough.
The IPTC standard has a separate value for exactly this case, “compositeWithTrainedAlgorithmicMedia”: a real capture that was edited with AI. But a short label on a post can’t show that nuance, so viewers may read it as “fake”. Remember: the label describes the file, not your camera.
How to check what your file says
You don’t have to guess. Drop the original photo or video into the tool above. You get a plain list of every AI tag in the file, which app wrote it, and how likely an Instagram or Facebook label is. For example, a photo edited with Generative Fill can show a Content Credentials record written by Photoshop and an IPTC value that means edited with AI.
If the report finds tags, one tap removes them. If it shows no AI tags but your post was still labeled, the cause is outside the file: your own AI label setting, or Meta’s own judgment of the picture.
How to remove the label, and when to keep it
Before you post, clean the file here and post the clean copy. If you added the label yourself, edit the post and switch it off. If Meta added it to a live post, there is no switch: request a review where the app offers one, or delete the post, clean the original and post it again. Reposting costs you the old post’s likes and comments.
And if the picture was fully generated by AI, keep the label. It is accurate, and the people who see your post deserve to know. This site exists for real photos with false labels.
Does “AI content” mean the same thing on Facebook and Threads?
Yes. Meta uses the same label across its apps, so the answer to “what does AI content mean on Facebook” is the same as on Instagram, and a file with AI tags can get the label on all three.
Instagram also has a separate account-level label, “AI-generated profile” (earlier “AI creator”), for accounts whose main identity is an AI-generated person. That one is set on the account, not read from a file, so no file tool can change it.
SynthID, the invisible pixel watermark
Some generators, including Google’s AI tools, also put an invisible watermark called SynthID into the pixels. Cleaning removes it: one tap takes out both the AI tags and the SynthID mark. Meta says it reads industry-standard indicators such as C2PA and IPTC, so for Instagram and Facebook the file’s metadata is the first thing to check.
SynthID matters elsewhere too, because Google’s own tools can read it. The tool above clears it along with the tags, so the clean copy carries no AI marks for any checker to find.