Why Your Real Photo Gets an “AI Info” Label (and How to Fix It)
· 4 min read
You post a real photo — shot on your camera, lightly retouched — and Instagram, Facebook, Threads or LinkedIn marks it as AI. It’s one of the most common complaints from photographers and designers, and it almost always comes down to metadata rather than the picture itself.
Why platforms add the label
Platforms don’t visually inspect your photo to decide. They read signals inside the file:
- C2PA Content Credentials that say a generative tool was used.
- IPTC “digital source type” fields such as trainedAlgorithmicMedia or compositeWithTrainedAlgorithmicMedia.
- Invisible watermarks from AI tools, where the platform can detect them.
The usual culprit: a small AI edit
Using generative fill to remove a stray object, AI noise reduction or an “expand canvas” feature can add an AI record to the whole file — even if 99% of the pixels came from your camera. Meta renamed its label from “Made with AI” to “AI info” in 2024 precisely because so many lightly edited real photos were being tagged. An AI label remover takes the tag out without touching the pixels.
How to fix it
- Check what the file carries. Run it through the free Unmarkr check to see which AI fields are present.
- Export without the AI record. In many editors you can turn off Content Credentials on export or choose not to include metadata.
- Or strip the metadata. The free clean removes C2PA, IPTC, EXIF and XMP and leaves the pixels identical.
When the label is accurate
If the image is substantially AI-generated, platforms generally expect you to disclose it — and Meta asks users to label photorealistic AI content themselves. Removing the metadata doesn’t change that. The fix above is for real photos that were mislabelled, and for creators who want control over their own files.
See why your photo is flagged — free
Drop the file into Unmarkr to see exactly which AI fields it carries before you post it.
Check an image — free