Google DeepMind has made its SynthID Detector available worldwide, according to infoq.com. The tool lets people check whether images, audio or video were produced with AI, and it now covers media from SynthID partners including OpenAI, NVIDIA and Kakao, with Apple expected to join soon.
The launch marks a shift in how Google's watermarking system is accessed. Previously, checking for a SynthID mark meant asking Gemini, and that route only worked for content made by Google's own AI. Anything from another company required that company's separate detection tool. The new site at SynthID.com puts those checks in one place.
Why it matters
The timing has drawn attention beyond the developer community. Businessday NG reported that Google opened the watermark detector to the public ahead of Nigeria's 2027 elections, a context in which verifying the origin of images and video carries obvious weight. Separately, Google's own blog noted that TIME named SynthID one of the Best Inventions of 2026, recognizing its approach to detecting AI content.
How the detector works, and what it cannot do
Google is explicit that the tool is not a general-purpose AI detector. It can only flag media from companies that have adopted SynthID, and the company says it is working to widen those partnerships. Meta and Microsoft are not covered because their AI systems use a different watermarking method, according to infoq.com.
There is a second limitation. Heavy editing can weaken the embedded signal, producing false negatives. In those cases, Google's guidance is to combine the detector with other tools rather than rely on a single check.
Skepticism and a competing approach
Reaction online has been mixed. Infoq.com quoted a Reddit user, Chicano_Ducky, who argued the approach has not worked in practice and pointed to what the user described as unacceptably high false positives on videos made by the companies' own models. Another commenter, Bulky-Employer, said the mark can be stripped with a low-denoising pass, describing SynthID as a speed bump against people seeking to defraud with AI images, while conceding that having it at all is useful.
A different camp favors provenance data over after-the-fact detection. C2PA metadata records where a file came from and how it was edited, which can help establish authenticity. Harry G, writing on LinkedIn, called that the only approach that scales, arguing that trying to spot synthetic content after the fact becomes an arms race. The weakness of metadata is that it can be stripped when a file is uploaded and does not survive a screenshot, infoq.com notes.
Engadget has published a guide on using the SynthID detection website, and Android Police rounded up detection tools for AI text, photos and videos, suggesting demand for practical guidance is already there.
What to watch: whether Apple's support arrives as expected, and whether Google extends SynthID partnerships to providers currently outside the system.
