Overview
- Users can upload an image to the Gemini app and ask if it is AI-generated, with Gemini checking for a SynthID watermark and returning context-rich explanations.
- The initial capability verifies only content created or edited by Google AI that carries SynthID, requiring manual checks by users.
- Google is starting to embed C2PA metadata this week in images from its Nano Banana Pro (Gemini 3 Pro Image) model across the Gemini app, Vertex AI and Google Ads.
- The company says it will expand verification to video and audio and bring the feature to more surfaces such as Search, with longer-term plans to support C2PA credentials from tools beyond Google.
- Google says SynthID has marked over 20 billion pieces of AI content and has tested a detector with journalists, while outside reporting notes broader impact depends on platforms automatically surfacing provenance signals.