Google launched the public version of its AI-generated media verification tool, SynthID, on Tuesday, October 7, 2026. Initially available to a limited group of journalists, researchers, and media professionals for testing since last year’s Google I/O event, SynthID now allows anyone to verify whether digital content is AI-generated.
How SynthID Works
SynthID identifies AI-created content by detecting invisible watermarks embedded during the generation process. The tool currently supports a wide range of file types including common image formats like JPG, PNG, and GIF; video formats such as MP4 and WEBM; and audio formats like WAV and MP3.
Partnerships Expand Reach
Google’s Nano Banana, Veo, Lyria models, Gemini, Flow, ProducerAI, and Vids all utilize SynthID to watermark content. Beyond Google's own tools, OpenAI, Nvidia, and Kakao also support the system. Apple is expected to add support in the near future. The technology has been integrated into Google’s Gemini app and Chrome browser, processing over 1 million verification requests daily.
Limitations of Detection
While a valuable step towards transparency, SynthID isn't foolproof. The detector only identifies watermarks from participating AI models; a negative result does not guarantee content is human-made. Tools developed by Microsoft and Meta also exist for verifying content, but have similar limitations in identifying media created by other systems.
To date, Google has watermarked over 180 billion images and videos, along with the equivalent of 240,000 years of audio. The SynthID detector is currently available globally in English at synthid.com.
💬 Comments (0)