How to Remove “AI Info” Metadata From a Photo

What “AI info” means

Large social platforms look for industry-standard signals that an image was created or edited with generative AI. Two of the most common signals live inside the file itself as metadata: the IPTC DigitalSourceType field (values such as trainedAlgorithmicMedia) stored in XMP, and C2PA Content Credentials, a signed record of how the image was made. When these are present, a platform may show a label such as “AI info”.

Which apps write these fields?

• AI image generators that follow the IPTC or C2PA standards.
• Photo editors when you use generative features such as generative fill, expand or AI erase.
• Local tools like Stable Diffusion WebUI or ComfyUI, which store prompts in PNG text chunks.

That is why a mostly real photo can still carry AI metadata after a small generative edit.

Step 1: Check what your image contains

Open the image metadata checker and add your photo. It lists AI markers, C2PA manifests, XMP fields and GPS location so you know exactly what is inside before you change anything.

Step 2: Remove the AI metadata

Use the AI metadata remover in Complete mode, or Selective mode with XMP, C2PA and Software ticked. The image pixels are not modified. The cleaned file is scanned again so you can confirm the fields are gone.

Step 3: Upload the cleaned copy

Upload the downloaded file rather than the original. Keep your original somewhere safe in case you need its metadata later.

What metadata removal cannot do

• It does not remove invisible watermarks embedded in pixels (such as SynthID).
• It cannot stop a platform from using visual classifiers or other signals.
• It does not change a platform’s disclosure rules. If your content is realistic and AI-generated, the platform may require you to label it.

See our disclaimer for details. Questions? Email us.

FAQ