Google has made its SynthID AI content detector available to everyone. The tool now lives on a new dedicated website, and it can read invisible watermarks from all of Google's partners, not just its own models. Until now, checking for these watermarks was a convoluted process.
What Is SynthID and How Does Google's AI Detector Work?
SynthID is not a visible logo or a label stamped on a file. It is an invisible signal encoded directly into the content itself. For images and videos, the signal sits in the pixels. For audio, it sits in the waveform.
This makes the watermark very difficult to remove or fake. Every piece of AI content from Google's Gemini models carries a hidden SynthID label, which makes it hard to pass that content off as authentic.
The scale is huge. Google says Gemini alone has integrated SynthID with over 180 billion images and videos, plus 240,000 years of audio.
Why the Global Rollout of the SynthID Detector Matters
The watermark only helps if people can actually check it. That was the problem. SynthID had buy-in from other major AI players, but verifying those invisible watermarks was a complicated process.
Google's announcement changes two things at once:
- The detector now supports watermarks from all of Google's partners, not just Google's own models
- It is available for anyone to use on a new dedicated website
In simple terms, a reader, journalist, or platform can now check whether a piece of content was made by AI — and the check covers more than one company's tools.
Our Take: A Small Website With a Big Job
To put it plainly, this is one of the more useful things to come out of the AI watermark push. Watermarks are only as good as the ability to detect them. A hidden label that nobody can read is close to useless in the fight against fake content.
In our view, the partner support is the more important half of this news. If every AI company watermarks its own content but nobody can check across companies, the system stays broken. Google opening the detector to partner watermarks is a step toward one shared checking layer.
There is still a limit. This detector only finds SynthID watermarks. Content made by tools that do not use SynthID will not show up. So a "no watermark found" result does not mean the content is real. Readers should treat this as one signal, not a final verdict.
For everyday users, the practical takeaway is simple: there is now a free, public place to check whether an image, video, or audio clip carries Google's AI label. That is a real improvement over the convoluted process that existed before.
