Google Gemini and Imagen image metadata
Google’s image stack uses two different transparency layers. Content Credentials (C2PA metadata) can ride in the file as a signed manifest. SynthID is an invisible watermark in the pixels. This page strips the first when the inspect report lists it. It does not remove SynthID.
Cleans metadata in your browser. SynthID and TrustMark stay in the pixels. A cleaned file does not force Instagram to drop a label.
1 image free · Pro $9.99 one-time (unlimited + ZIP)
Inspect the Google AI download
- 1 Select
- 2 Scrub
- 3 Verify & download
Free clean used
1 image free in this browser · Pro $9.99 one-time unlocks unlimited cleans + batch ZIP · How to use
Cleaning…
Clean every image with Pro
Unlimited images and folder ZIPs. Pay once.
Card, Apple Pay or Google Pay · No subscription
Do it in the tool, in this order
- Drop the JPEG, PNG, or WebP you will actually publish. Try sample only shows what a report looks like.
- Set the mode to AI only.
- Press Inspect only if you want to look. Press Clean this file to strip. The same press re-scans the result, and the After column is that new scan of the cleaned bytes.
- Press Download cleaned image. Upload that file. The name ends in
-cleaned.
What Google documents for file metadata
Google Cloud’s Content Credentials guide states that supported Google models automatically add signed C2PA metadata that can indicate AI generation, and points to Gemini image models such as Nano Banana and Nano Banana Pro ( Google Cloud documentation). Google’s blog on Gemini image verification adds that images from Nano Banana Pro (Gemini 3 Pro Image) in the Gemini app, Vertex AI, and Google Ads began receiving C2PA metadata embedded in the file ( Google blog). That is the class of AI metadata this tool removes when you see a C2PA chip on Inspect.
Do not assume every Gemini or Imagen export you already have includes that manifest. Rollouts are model- and surface-specific. Inspect the bytes you will publish. An empty C2PA line means this file has no embedded manifest for AI provenance mode to delete — not that Google never watermarked the picture.
SynthID is not file metadata here
DeepMind describes SynthID as a watermark embedded at generation time, designed to survive common edits ( SynthID overview). Gemini Apps Help separates SynthID from Content Credentials and notes Gemini can verify SynthID on uploads ( Gemini Apps Help). Cropping and recompression may weaken detection; they are not a removal method, and Remove AI Meta does not offer one. A clean After report can coexist with a positive SynthID result.
Workflow on this page
Download the image from Gemini (or export from Vertex / Ads). Drop the JPEG, PNG, or WebP into the tool above. Leave the mode on AI only (AI provenance). If Inspect lists C2PA or AI XMP, press Clean this file and download when the chip clears. If Inspect is empty for C2PA, you may still have SynthID or platform labeling — stripping again will not help.
The vocabulary for C2PA in general is what is AI metadata. OpenAI’s parallel case (metadata + SynthID) is ChatGPT image AI metadata. Adobe Firefly’s automatic credentials are Adobe Firefly metadata.
Remote or tampered credentials
Gemini Apps Help notes some third-party apps store Content Credentials remotely, separate from the file, and that Gemini Apps do not support remote metadata. This cleaner only rewrites bytes in the file you drop. It does not fetch cloud manifests.
Questions
Does every Google AI image have C2PA?
No. C2PA in the file is rolling out on supported models and surfaces. Inspect yours.
Will cleaning remove SynthID?
No. SynthID is a pixel watermark, not a metadata chip this tool strips. The standalone answer is can you remove SynthID from an image?
What if C2PA is missing but Gemini says “Google AI”?
SynthID verification does not require C2PA. Metadata stripping and watermark detection are different checks.
Will social platforms drop labels after a clean file?
No guarantee. They may use watermarks or classifiers beyond file metadata.
Clean the file, then publish that download
Use the tool above. Download only after the After report matches the tags you meant to remove.
Clean this downloadRelated guides
- Can you remove SynthID from an image?
- Remove Adobe Firefly Content Credentials from a download
- How to remove the “AI info” label from an Instagram post
- Open the Remove AI metadata tool on the homepage if you want the general cleaner.
- Short Instagram upload workflow if you only need the four-step version.
Questions people ask
Does every Gemini image include C2PA metadata?
No. Google’s blog states C2PA metadata rollout for images from Nano Banana Pro (Gemini 3 Pro Image) in Gemini, Vertex AI, and Google Ads. Other Google AI images may carry SynthID without a C2PA manifest in the file you saved.
Will this tool remove SynthID from a Google AI image?
No. Google documents SynthID as an invisible watermark embedded in AI-generated content. This site removes file metadata such as C2PA JUMBF. SynthID detection in Gemini or the SynthID Detector is separate.
Can I verify Google AI metadata elsewhere?
Yes. Google Cloud documents Content Credentials for supported models. Gemini Apps Help describes verifying SynthID and Content Credentials separately. C2PA validators read embedded manifests when present.
If After is clean, will Instagram drop an AI label?
Not necessarily. Platforms can label from watermarks and classifiers. A clean metadata report does not force a badge off.
Page last updated: