Why Facebook and Instagram show “AI info” on your photo
Meta labels photos whose files say AI was used, even for a small retouch. Why the label appears, and what you can and can’t do about it.
Facebook, Instagram and Threads add an “AI info” label to some photos; until July 2024 it read “Made with AI”. Meta adds it when the image file says AI was used, and when it detects AI in other ways. Photographers soon found it on real photos too.
Why a real photo gets the label
Meta reads two industry standards in the file: Content Credentials (C2PA) and the IPTC “digital source type” field. Editing apps write them when you use an AI tool, even for a small change: taking out a stray object with Photoshop’s Generative Fill, or with an AI eraser in your phone’s photo app, can be enough. For photos only edited with AI, Meta shows the label in the post’s menu rather than on the photo; for images it considers made with AI, it shows it more prominently.
Check what your file says
Drop the photo into PhotoCleanr. The free report says whether the file says it was made or edited with AI, and which apps its Content Credentials name.
If the label is wrong for your photo
If you took the photo and only retouched it, you can post a copy without these records: PhotoCleanr’s clean copy has no Content Credentials and no AI field in its metadata, and the picture isn’t re-encoded. The label can still appear:
- if the picture carries an invisible watermark, which is part of the pixels and which PhotoCleanr doesn’t change; Meta’s own AI images have one, for example;
- if Meta recognises AI in some other way;
- if PhotoCleanr leaves the record in: it leaves an image unchanged when its Content Credentials contain author or rights information, and keeps the data an HDR photo needs to display. The report tells you when that happens.
Cleaning a file doesn’t change a post that’s already up.
When the label should stay
If the image was generated with AI, or AI changed what it shows in a way that could mislead, such as a person, a place or an event that looks real, don’t hide it. Platform rules can require a label for such content, and in the EU, since 2 August 2026, the AI Act requires anyone who uses AI to make a deepfake, other than for purely personal use, to disclose it. Removing metadata changes none of that, and our terms don’t allow using PhotoCleanr to deceive people about where an image comes from.
More on the record Meta reads: what are Content Credentials.