Google Gemini now has the ability to determine whether a photo, video, or audio clip was created by artificial intelligence. Users can upload media directly to the chatbot, which will analyze it and reveal if it was generated by an AI model, edited by one, or is an authentic original. The tool extracts provenance data, identifies the camera or phone that captured the image, and lists every generative edit applied, including modifications made in Google Photos.
Expanded Verification Across Google Services
The same verification layer is being integrated into Google Search and Chrome, where AI-generated results will display a flag. Pixel phones are also expected to receive this feature in the near future. This move aims to increase transparency and trust in digital content.
Industry Adoption of SynthID
Behind Gemini's new capability is SynthID, Google's watermarking system. In a significant development, OpenAI, ElevenLabs, and Kakao have agreed to adopt SynthID for their own generative tools. This marks the first time direct competitors of Google in the AI space have committed to tagging content using a rival's system. Images from OpenAI, voices from ElevenLabs, and content from Kakao's models will now carry an invisible SynthID watermark, detectable through Google's verification systems. Earlier this year, NVIDIA became the first external partner, watermarking videos from its Cosmos models.
Combining SynthID with C2PA
Google is also leveraging C2PA, the Content Credentials standard backed by Adobe, Microsoft, and others. Pairing SynthID with C2PA addresses two distinct tasks: tracing the origin of a file and detecting whether AI has manipulated it afterwards. The reasoning is straightforward: AI-generated video and audio are becoming too convincing for the human eye alone. Google argues that the solution cannot come from a single company; it must be a system that the entire web can verify against.
