SynthID Detector and Checker
Choose the provider that may have created the image. Neither checker is a universal AI detector, and “not detected” does not prove human authorship.
Check with OpenAI
Check an original image for supported OpenAI C2PA and SynthID provenance signals. This is not a general AI detector.
Check with Google
Google's public SynthID verification runs inside the signed-in Gemini app. Your file goes directly to Google, never through SynthID Remover.
Gemini currently recognizes SynthID from Google AI tools. It is not a universal detector for every AI provider.
How to use the SynthID checkers
The checker cards above deliberately remain separate. Google and OpenAI verify different provider-specific signals, use different upload destinations, and can return different answers for the same image.
Check an image with Google Gemini
- Open the official Gemini application using the Google checker button.
- Sign in with an eligible Google account and upload one original image.
- Ask: “Was this image created or edited by Google AI?”
- Read the response as evidence about Google AI tools only.
Google explains that Gemini currently recognizes SynthID in content created or edited by Google AI tools. Google does not document a public verification API for third-party websites, so SynthID Remover does not scrape Gemini or relay your file to it.
Primary source: Google Gemini Apps Help: Verify AI-generated media.
Check an image with OpenAI
- Choose an original JPEG, PNG, or WebP in the OpenAI checker card.
- Review and accept the separate OpenAI data-processing notice.
- Complete human verification and request one provenance check.
- Review C2PA and SynthID entries independently.
The OpenAI API checks supported OpenAI provenance signals. It is not a detector for every AI company, and its checks are not eligible for OpenAI Zero Data Retention. If native API access is unavailable, the card links directly to OpenAI’s official browser verifier.
Primary source: OpenAI Content Provenance guide.
What the results mean
| Result | Meaning | What it does not prove |
|---|---|---|
| Detected | The provider recognized a supported watermark or provenance signal. | It does not reveal the prompt, account, or full edit history. |
| Not detected | The provider did not find the supported signal it checked. | It does not prove the file is human-created or untouched by AI. |
| Inconclusive or unavailable | The provider could not return a reliable determination. | It should not be treated as a negative result. |
SynthID versus C2PA and ordinary metadata
SynthID is an embedded media signal intended to survive some transformations. C2PA Content Credentials are signed provenance records. EXIF, IPTC, and XMP are ordinary metadata containers. Removing one kind of information does not establish that the others are absent.
Privacy comparison
| Option | Upload destination | Scope |
|---|---|---|
| OpenAI checker above | OpenAI, after explicit consent | Supported OpenAI C2PA and SynthID signals |
| Google checker above | Google Gemini, uploaded directly by you | SynthID associated with Google AI tools |
| SynthID Remover | Private Cloudflare R2, then a temporary remote GPU | Fixed best-effort image processing; not detection |
Frequently asked questions
Is there one universal SynthID detector?
No. Public verification is provider-specific, and a negative result from one provider does not cover another provider’s models or every form of AI generation.
Can this page check a Google image without opening Gemini?
No. Google currently provides its public verification inside the signed-in Gemini application and does not document a public detector API that this site can call.
Does “not detected” mean an image is human-made?
No. Signals can be absent, unsupported, weakened, stripped, or created by a different provider.
Does the remover run these checkers automatically?
No. The processing workflow is not connected to a detector and does not tune itself against checker responses.