Google has opened its SynthID Detector to the general public. The company said on 7 October 2026 that the tool, which checks whether an image, video or audio file carries Google's invisible AI watermark, is now available globally in English. Until now an early version, launched last year, was mainly offered to journalists and media professionals for testing.
The idea is simple. You upload a file and the detector tells you whether it was made with AI from Google or one of its partners. Google names OpenAI, NVIDIA and Kakao as partners whose watermarked content the tool can recognise, and says Apple is coming soon.
In practice, that means a picture made with a partner's tool can be checked in the same place as one made with Google's own Gemini or Imagen models, as long as the partner has added the SynthID mark. Google's announcement focuses on images, video and audio, and does not say that the public detector checks text.
How SynthID works
SynthID is a digital watermark that Google DeepMind adds to AI-generated content at the moment it is created. For images and video, Google says the mark does not change the visible quality and is designed to survive common edits such as cropping, filters, changed frame rates and lossy compression.
For audio, it covers music from Google's Lyria model and the podcast feature in NotebookLM. Google says the audio mark cannot be heard and holds up against added noise, MP3 compression and speed changes. Text from the Gemini app and its web version is also watermarked, by slightly adjusting the probability of the words the model picks, which Google says readers cannot notice.
The scale is large. Google says it has watermarked more than 180 billion images and videos, along with 240,000 years of audio, since SynthID launched in 2023.
Other places you can run the check
The detector sits alongside checks already built into Google's own products. In the Gemini app, Search and Chrome, you can upload an image, video or audio clip and ask whether it was created or altered by Google AI. Google says these built-in checks now handle more than 1 million requests a day.
Google also used the same announcement to introduce SynthID Bio, which applies its watermarking idea to synthetic biology. That is aimed at researchers and has no use for everyday readers yet.
What a result can and cannot tell you
This is the part worth remembering. SynthID can only find its own watermark. A positive result tells you the file came from Google's tools or a partner that uses SynthID. A negative result only tells you that this particular watermark was not found. It does not prove a picture or clip is genuine, because content from AI tools that do not use SynthID would not carry the mark at all.
So the detector is one check among several. It still makes sense to ask where a clip first appeared, look for the original upload and see whether an official source has said anything. That matters most with forwarded videos of public figures pushing money schemes, the kind of material regulators flagged during World Investor Week.
What it means for readers in India
The tool is in English for now, and Google has not said when other languages will follow. Because the check works on the file itself, a Hindi or Tamil caption on a forwarded image makes no difference to the result, but the interface and the explanation of the result are in English.
For families, the practical use is clear: before sharing a startling photo or voice note in a group, run it through the detector, and treat a 'not found' result as a reason to keep checking rather than as proof. Schools and newsrooms that already teach image checking now have a free, public tool to add to the routine.
Google has been adding other safety features across its products in recent weeks, including the way Chrome now blocks certificates after domain registry hijacks. Accessibility features in its AI tools, such as Gemini Live's guided vision mode, show the other side of the same technology.
