Google has opened SynthID Detector to the public, so anyone can now upload an image, video, or audio clip and see whether it carries a SynthID watermark, a hidden signal embedded in the media when it is generated. The tool is available worldwide, in English, starting 7 October.
Google puts the scale at more than 180 billion images and videos watermarked since SynthID launched in 2023, plus 240,000 years of audio, according to a post by Pushmeet Kohli on Google’s blog. Google describes the watermarks as imperceptible. That describes the design; the post does not say how well they survive editing.
Google says its built-in checks in Chrome, Search, and the Gemini app already process over a million requests a day.
The limit that matters is coverage. The Detector finds marks from Google and from the partners the post names: OpenAI, NVIDIA, and Kakao, with Apple listed as coming. A clean result therefore says only that none of those watermarks turned up. It does not show a file is human-made, and it cannot flag AI content from any other tool.
Google’s blog, by Pushmeet Kohli, published 7 October 2026.
