AI image detector by generator
Gemini and Imagen image detector: what Google puts in the file
Google marks Gemini and Imagen images with an invisible SynthID watermark and, for newer models such as Gemini 3 Pro Image, with C2PA Content Credentials signed by Google, and isGenAI validates that signed record when it is present.
Illustration · PNG file
- Signature + IHDR
- iTXt XML:com.adobe.xmprecord
- tEXt parameters
- tEXt prompt / workflow
- tEXt invokeai_metadata
- eXIf
- caBX C2PA manifest
- IDAT pixels
- IEND
AI source type declared
The file labels itself as AI-made.
IPTC DigitalSourceType such as trainedAlgorithmicMedia is read from XMP and shown with its exact value.
Check an image →What Google documents
Google states that images generated or edited with Gemini carry a SynthID watermark, and that images from Nano Banana Pro (Gemini 3 Pro Image) in the Gemini app, Vertex AI and Google Ads include C2PA metadata. Google Cloud’s documentation adds that Content Credentials are automatically added and signed by Google LLC for supported models on Vertex AI. The Gemini app can also be asked whether an uploaded picture carries a SynthID mark.
A visible watermark on Gemini images can be switched off by the user, while SynthID and the C2PA record remain. Which Gemini or Imagen release writes which signal has changed several times, so inspect the actual file.
What isGenAI shows for a Gemini image
When the download carries a Google-signed manifest, the report lists a signed creation record with validation state, trust state, the claim generator and the recorded AI action. When only the SynthID watermark is present, isGenAI cannot read it: the checker reads file records, not pixel patterns, and the report says that no file evidence was found.
An IPTC DigitalSourceType declaration in XMP, if present, is listed as a declared source type of moderate strength.
Screenshots and reposts
A screenshot removes the C2PA record. Google designed SynthID to survive common edits, and the Gemini app is the place to check for it. For the file record, isGenAI needs the original export from the Gemini app, AI Studio or Vertex AI.
Use the matching-copy check to tie a stripped copy back to an original that still carried the credential.
How to check an image from this generator
- Get the original download, not a screenshot or a copy saved from a chat or social feed.
- Upload it to the free AI image detector. Files up to 20 MiB in PNG, JPEG, WebP, AVIF or TIFF are read on the server and are private by default.
- Open “How we checked” and read each finding with its source field: signed record, declared source type, saved prompt, AI statement or matching copy.
- Treat a clean result as “no evidence found”, never as proof that a person made the picture.
Frequently asked questions
Does Gemini add metadata to generated images?
Google documents a SynthID watermark in Gemini images and C2PA Content Credentials for Gemini 3 Pro Image outputs in the Gemini app, Vertex AI and Google Ads, with Vertex AI credentials signed by Google LLC, as of October 2026.
Can isGenAI detect SynthID?
No. SynthID is a pixel-level watermark that Google’s own tools check. isGenAI reads the file’s C2PA record, IPTC source type and saved prompts, and reports when none are present.
Is a Gemini image detectable after a screenshot?
Not from the file record. The C2PA manifest is gone. Google states that SynthID is built to survive edits, so the Gemini app check may still work; isGenAI will report no file evidence.
What does a valid Google credential prove?
That Google signed a record saying an AI tool generated or edited the picture and that the file has not changed since. It does not say anything about who requested the image.
Sources
Vendor behaviour changes. These pages describe what the vendor documents and what isGenAI reads from a file; they do not test every release.