Skip to content
Pixvisor
Inspect a photo

FreeNo account, up to 25 MB

AI image detector: is this image AI generated?

Drop an image to read what its file says about how it was made: Content Credentials, a declared AI source, or the prompt a generator left behind. When there is no trace, the report says so instead of guessing.

  • Deleted right after analysis
  • Original never modified
  • No sign-up
Drop an image here
or click to browse (up to 25 MB)
FormatsJPEGPNGWebPHEICTIFFGIF

Your original file is never modified. Every action produces a new image.

How do I check if an image is AI generated?

  1. Drop the image into the checker above, choose a file, or paste its URL.
  2. Read the verdict. It names each signal it found and where it sits in the file.
  3. Open All metadata to see the raw fields for yourself.

The checker gives one of two kinds of answer. Either the file declares an AI origin, and you see the declaration, or it carries no such trace, and the report says there is no metadata evidence. It never turns an absence into a percentage.

What does the AI photo detector look for?

The checker reads four families of AI fingerprints from the metadata layer:

  • C2PA / Content Credentials: the provenance standard used by DALL·E, Adobe Firefly and a growing list of cameras and editors, embedded as a JUMBF block in the file.
  • Generation parameters: the prompt, negative prompt, model, seed and sampler that Stable Diffusion (AUTOMATIC1111) and ComfyUI write into PNG files. The checker decodes them for you.
  • IPTC digitalSourceType: the standard field that declares an image was made by generative AI.
  • Software tags: the declared creator tool, which often names the generator outright.

Can it detect Midjourney, DALL·E or SynthID images?

Only if they left a trace in the metadata. This checker works on the metadata layer only.

Invisible pixel watermarks such as Google’s SynthID stay out of reach. How that watermark works, and how to check for it with Google’s own tools, is covered in what is SynthID. So does any file whose metadata has been stripped: a screenshot of an AI image shows no evidence at all.

Where other tools hand back a confident-sounding percentage, we show you the signals we found, or say plainly that there are none.

Why does metadata beat pixel-based detection?

Pixel-based detectors estimate probabilities, and they can be wrong in both directions. Metadata fingerprints work differently.

When an image carries C2PA credentials or a Stable Diffusion parameter block, that is direct evidence of where it came from, and you can inspect it. You read the fields yourself in the report.

Found fingerprints in your own work and want them gone? Use the remove EXIF tool, or look through the whole file first with the EXIF viewer.

What does a clean result mean?

If this checker finds nothing, the file is clean. That does not mean a human made the picture.

Almost every image online carries no Content Credentials, no declared source type and no generator settings. Messaging apps, social platforms, format conversion and screenshots all strip metadata as a matter of routine, with nobody hiding anything.

So we show you the raw field and where we found it, rather than a percentage. A score invites you to read an absence as a verdict. A field you can read invites you to weigh the evidence yourself.

For the three kinds of evidence available and their error rates, see what AI image detectors can and cannot prove.

Which generators leave which trace?

Three families, and they behave differently.

  • Content Credentials are a signed manifest, written by some generators and some editing software. They sit in the APP11 segment of a JPEG or the caBX chunk of a PNG.
  • A declared source type is an IPTC and XMP field, digitalSourceType. The value trainedAlgorithmicMedia is a file stating outright that a model made it.
  • Generator settings are working data nobody removed. Stable Diffusion writes its whole block into a PNG text chunk called parameters; ComfyUI writes prompt and workflow.

The third family is often the most detailed thing in the file, and it was never meant as evidence at all.

Frequently asked questions

How do I check if an image is AI-generated?
Drop the image above. Pixvisor reads its metadata for AI-generation fingerprints (C2PA / Content Credentials, Stable Diffusion and ComfyUI parameters, IPTC digitalSourceType and declared software) and returns a verdict with the exact signals found.
How accurate is this AI checker?
When a fingerprint is present, the verdict is based on hard evidence, not statistics: the image declares its own AI origin through C2PA, generation parameters or IPTC fields. When no fingerprint is found, the answer is "no metadata evidence", which does not mean the image is necessarily human-made.
Can it detect ChatGPT, Gemini or Midjourney images?
It detects what the generator writes into the file. Original downloads from ChatGPT and Gemini carry a signed C2PA manifest, which the checker reads and names (tested on 27 September 2026). For other generators, and for any copy that lost its metadata, the checker reports exactly what it finds, or that it finds nothing.
What if the metadata was stripped?
Then there is nothing left to read: screenshots, social-media re-uploads and aggressive optimisers all remove metadata. The checker is transparent about this limit: no signal is reported as "no evidence found", never as "not AI".
Is this AI image detector free?
Yes. Checking an image is free and needs no account. The file is analysed on our server and deleted right after.
Can it tell if a photo is fake or edited?
It shows what the file declares: the editing software that saved it, its dates, and any Content Credentials that record edits. It cannot prove that the pixels were not altered, because metadata can be removed or rewritten.
Can it detect SynthID watermarks?
No. SynthID and similar invisible watermarks live in the pixels, not the metadata, and require the provider's own detector. Google offers its own check in the Gemini app for content made with Google AI.