AI Image Detector
Detect whether an image was generated by AI.
Your image
AI Image Detector — evidence first, opinion second
Most AI detectors hand you a percentage and no reasoning. This one separates the two things that actually matter: what the file provably says about itself, and what the picture looks like on inspection. The first is checked in your browser without uploading anything.
Why use this tool
Reads Content Credentials and generator metadata
C2PA manifests, the IPTC trainedAlgorithmicMedia marker, Stable Diffusion parameter blocks and ComfyUI workflows are all written by the tools that made the image. When one is present, the answer is not a guess.
Metadata is parsed locally
The file is read in your own browser. It is only transmitted if you choose to run the second, visual step.
Camera evidence too
EXIF make, model and capture software are surfaced alongside the AI signals, so you can see what points the other way.
Honest about what it cannot do
Where the evidence runs out, the result says inconclusive instead of inventing a number. That is more useful than false confidence in either direction.
Frequently asked questions
Two kinds of evidence
The metadata check reads what the file says about itself — signed Content Credentials, generator parameter blocks, camera EXIF. That is fact. The visual check is a judgement about how the picture looks, and it is shown separately because it can be wrong.
Good to know
- Screenshots and social media downloads have their metadata stripped.
- No visual detector is reliable. Treat the second half as a hint.
- Metadata is easy to remove and possible to forge.
- Nothing here can prove an image is real — only that evidence is missing.