Google opens SynthID Detector globally in English

Public access to a watermark check

Google has opened SynthID Detector to the public globally in English. Its October 7 announcement says users can check images, video and audio for watermarks associated with AI media from Google and participating partners.

The service extends access beyond the early version introduced for media professionals. Google names OpenAI, NVIDIA and Kakao among its partners, with Apple described as forthcoming. Those names indicate the announced scope; they do not turn the tool into a universal test for every generator or every file found online.

Read a result in its stated scope

SynthID identifies supported watermarks embedded in generated media. A check is most useful when the person interpreting it understands what the tool can identify. Someone reviewing a disputed image should keep the original file and record the detector result alongside the source of the image.

An absence of a supported watermark should not, by itself, settle whether a file is authentic. The wider verification task still involves identifying who supplied the material and whether its context matches the claim attached to it. That is an editorial interpretation of the detector’s stated scope, rather than a performance claim from Google.

A practical place in a verification workflow

Google says the detector supports image, video and audio checks. A newsroom or community moderator can use that extra signal when examining material from an unfamiliar source, while retaining the evidence behind any decision to publish or remove it.

For personal use, start with a file whose origin you already know. Read how the interface describes a positive or inconclusive result before applying it to something contentious. Save the result and the circumstances of the check. Keeping those details makes a later discussion about provenance more useful than sharing a bare screenshot labeled as proof.