Google SynthID Detector Opens to Everyone to Spot AI Media

SynthID is not a Google-only standard. OpenAI, Nvidia and Kakao also support it, and OpenAI runs its own site for checking content. Apple is reported to be adding support soon, according to TechCrunch.

Rival efforts exist: Microsoft and Meta maintain their own watermarking and verification standards, so the field is still split between competing systems rather than one common rule.

These tools are not infallible, and they often fail to identify content created by their own makers' models. That is a key caveat: a negative result from the SynthID Detector does not prove a file is authentic, only that no SynthID watermark was found.

The limitation leaves room for AI-generated media that carries no watermark at all, whether because it came from a rival model or because the mark was removed or degraded.

For readers in Tunisia, the practical change is simple: the SynthID Detector is a free website, so no local launch or subscription is needed to try it on a suspicious image, clip or voice note circulating on social media.

Treat it as one signal, not a verdict. If a file is flagged as AI-generated, be cautious; if it is not flagged, stay just as cautious, since the watermark only covers content made by supporting models. There is no word yet on any Tunisia-specific rollout or partnership, and the tool is available directly through Google.