Meta now adds an “AI info” label (formerly “Made with AI”) to a lot of images on Instagram, Facebook and Threads. Sometimes it's right; sometimes an ordinary photo that was lightly retouched gets tagged too. Here's what's actually happening — and what you can and can't do about it.
How the label is decided
Meta reads the provenance metadata inside the file — primarily C2PA Content Credentials and IPTC fields — plus its own detection signals. If a file says “edited with an AI tool” in its manifest, that's a strong hint for the automatic label.
Why real photos get mislabeled
This is the common frustration: you open a normal photo in an editor that uses AI features — even just to remove a spot or expand the frame — and the exported file now carries a Content Credentials manifest. Instagram reads that and applies the label, even though the photo is essentially your own.
What you can do about it
Because the automatic check leans on the metadata embedded in your file, removing that provenance data from your own file before you upload means there's less for the automatic check to read. You can strip the C2PA manifest and EXIF in your browser first.
See what's in your file first
Not sure whether your photo even carries a manifest? Drop it into the C2PACleaner tool — it shows the Content Credentials and EXIF it finds before you remove anything, all locally in your browser.
Clean a file now
Strip C2PA, EXIF, GPS and AI text watermarks in your browser — nothing is uploaded.