Google has opened SynthID Detector to the public worldwide, giving users a way to check images, videos and audio for invisible watermarks associated with AI-generated content.
The service is available at synthid.com. Its interface is currently in English, and users must sign in with a Google, Apple or OpenAI account to check a file, 3DNews reported.
Google says the tool can recognize SynthID watermarks from its own models and from those of OpenAI, Nvidia and Kakao. Apple support is expected later. The detector can also highlight areas of an image or video where it finds a watermark.
SynthID was developed by Google DeepMind. The mark is embedded in the content itself and, according to the company, remains detectable after distribution and some forms of alteration. Google said it had watermarked more than 180 billion images and videos since 2023, along with audio totaling 240,000 years.
A detected mark can show that a file contains SynthID, but the absence of one does not by itself prove that AI was not used to create the material. The public website expands access to a portal Google introduced to journalists and other media professionals last year.