What is AI metadata in an image?
Many AI image tools and editors write a label into the file itself. It can be a C2PA Content Credentials manifest (used by OpenAI, Adobe Firefly, Google and others), the IPTC “digital source type” value trainedAlgorithmicMedia, the name of the generator in the XMP data, or the full prompt and sampler settings that Stable Diffusion front ends store in PNG text chunks. Platforms and search engines read these labels to tell AI images from photos.
How to check an image
- Drop one or more images on the box above, or press Choose images.
- Each image is read in a second. You see one of three results: AI labels found, Content Credentials found or No AI labels found.
- The list under each image shows exactly what was found, such as “Made with Midjourney” or “Generation prompt stored in the file”.
- If you want the labels gone, use the AI Image Metadata Remover.
What it can and cannot do
- It reads metadata only: C2PA Content Credentials, the IPTC / XMP “digital source type” label, generator names written into the file, and the prompt or settings blocks some Stable Diffusion tools save inside PNG files.
- It cannot see invisible watermarks that are woven into the pixels themselves (for example Google SynthID), so a file with no labels can still have been made with AI.
- Supported formats: JPEG, PNG, WebP and GIF, up to 8 MB per image, ten images at a time.
Your images
Unlike our other tools, the image is sent to our server so it can be read. It is processed in memory, never saved to disk or a database and never shared. Requests are rate limited to keep the service free for everyone.
Frequently asked questions
Does this tool prove an image was made by AI?
No. It only reports labels stored in the file. Many AI images have none (screenshots and re-saved copies lose them), and a few real photos carry Content Credentials because a camera or editor signed them.
What is C2PA / Content Credentials?
An open standard for signed provenance information inside a file. It records which tool created or edited an image, and whether a generative AI model was used.
Is my image stored?
No. It is read in memory on our server and discarded as soon as the result is sent back.
Can it detect SynthID or other invisible watermarks?
No. Those live in the pixels, not in the metadata, and they need the issuer's own detector.
Related tools
Remove C2PA, AI labels, prompts, EXIF and location data from images without re-compressing them.
Use tool