Google opens SynthID Detector to everyone: Here’s what it actually checks

Google has made SynthID Detector available worldwide in English. The tool lets people check images, videos and audio files for hidden AI watermarks associated with supported tools from Google, OpenAI, Nvidia and Kakao. Support for media created with Apple’s AI tools is planned but is not available yet.

How the detector works

SynthID embeds an imperceptible watermark into supported media when it is created. Users can upload a file to SynthID Detector, which looks for that signal and reports whether it finds one. For video and audio, it can also indicate where in the file a watermark appears.

Google introduced SynthID in 2023 and previously offered an early version of the standalone detector to selected media professionals. It says it has since watermarked more than 180 billion images and videos, as well as audio amounting to 240,000 years of playback.

What a result means

Finding a SynthID watermark is a useful indication that a file was made or edited with a supported AI tool. Finding no watermark does not prove a file is genuine. It may have been generated by a tool that does not use SynthID, or the signal may be harder to detect after substantial changes to the file.

That makes SynthID Detector a verification aid, not a universal AI-content test. If the origin of an image, recording or video matters, its result should be considered alongside the source of the file and any other available evidence.

Availability

The standalone detector is available globally, but its interface is currently in English. Its usefulness depends on whether the tool that created the media applied a compatible watermark—not simply on whether the file is an image, video or audio recording.