Overview
- Users can now upload an image in the Gemini app or on the web and ask if it was created or edited by Google AI, with detection powered by SynthID watermarks.
- Google says SynthID has marked more than 20 billion items and that its SynthID Detector was tested with journalists and media professionals before this public release.
- The detector reliably flags images carrying Google’s SynthID but does not consistently identify content from other models, as hands-on testing found with ChatGPT-generated images.
- Google has begun embedding C2PA metadata in images produced by its Nano Banana Pro model across Gemini, Vertex AI and Google Ads, with plans to add support for C2PA content credentials over time.
- Google plans to expand verification beyond images to video and audio and to bring these tools to more surfaces such as Search, though researchers have shown watermarks and metadata can be removed.