Skip to content
Try ZeroTell

Built for students, writers and researchers

AI Image Detector by ZeroTell · evidence before guesses

Drop in an image. We first look for proof the image carries itself: signed Content Credentials and generator metadata, checked in your browser without uploading anything. If that settles nothing, a pixel classifier can take a second look.

Your image

Step 1 runs in your browser. The image is not uploaded.

Step 1 · Credentials and metadata

No detector is proof on its own. Signed credentials are the strongest evidence; everything else is a signal to weigh, not a verdict about a person.

How the AI Image Detector works

  1. Add an image

    Drop, paste or choose a JPG, PNG or WebP file up to 10 MB. It stays on your device for the first step.

  2. We read what the file says about itself

    Signed Content Credentials (C2PA) are verified against the official trust list, and metadata written by tools such as Stable Diffusion WebUI, ComfyUI, InvokeAI or NovelAI is read.

  3. A clear answer when there is proof

    A verified credential that declares AI generation is the strongest evidence there is, and we tell you who signed it.

  4. An optional pixel check

    When the file carries no proof, a detection model can analyze the pixels and answer likely, can't tell or unlikely.

Why credentials come first

  • Signed proof beats guessing

    Generators such as ChatGPT, Gemini and Adobe Firefly sign their images. A valid signature is checkable; a model's guess is not.

  • Private by default

    The credential check runs entirely in your browser. Nothing is uploaded unless you choose the pixel check.

  • Honest answers

    The pixel check gives three plain answers instead of a fake precise percentage, because no image detector is that exact.

  • Evidence you can see

    Every finding lists where it came from and whether it was signed or can be edited, so you can judge it yourself.

Who uses it

  • Students

    Check whether an image you want to cite or reuse in an assignment was made with AI.

  • Teachers

    Get a starting point for a conversation about an image in submitted work, with the evidence behind it.

  • Journalists and fact-checkers

    See at a glance whether a picture carries a verified origin before digging deeper.

  • Social media readers

    Take a second look at a viral picture before you share it.

More ways to check work

Frequently asked questions

Only when the image carries a signed Content Credential that says so, verified against the C2PA trust list. Metadata can be edited, and a pixel check is a likelihood. Use those as signals, never as the only evidence against someone.

Most real photos have no credentials yet, and screenshots, edits and social media uploads remove them from AI images too. Missing credentials tell you nothing either way.

An open standard for signed information about how a file was made. Tools such as ChatGPT, Gemini, Adobe Firefly and some cameras attach them. We verify the signature and check that the signer is on the official C2PA trust list.

No. Google does not offer a public way to check SynthID from other websites. Google's own SynthID Detector can check images made with Google and partner tools.

Step 1 (credentials and metadata) is free and needs no account. The pixel check calls a paid detection model and uses 2 credits per image, on every plan.

Step 1 never uploads your image. For the pixel check, the image is sent to our detection partner for analysis. ZeroTell keeps only the result, never the image itself.

Look for proof first

Drop in an image and see what it says about itself.

Check an image