Home » AI Automates Your Life » Google Opens SynthID Detector to the Public

Google Opens SynthID Detector to the Public

Google’s SynthID Detector now lets anyone check images, video, and audio for its AI watermark. It is a useful first check, not a full fake detector.

Woman checking her phone screen, illustrating how the Google SynthID Detector verifies AI-made photos and videos

Google opened its SynthID Detector to the public worldwide in the week of October 7, 2026. Search Engine Journal reported that Pushmeet Kohli of Google DeepMind announced the expansion in a blog post.

What does the detector actually do?

Anyone can upload an image, video, or audio file at synthid.com and check whether it carries SynthID, Google’s invisible watermark for AI-made media. A positive result shows “SynthID was detected” and says the media “was made or edited with Google AI,” and may note it “may have been edited further since then.” Google’s portal first launched in May 2025 for early testers, with a waitlist for journalists, media professionals, and researchers.

The detector can now also read watermarks from partners. Google says content from OpenAI, NVIDIA, and Kakao can be checked, and Apple is expected to join later. Google also says it has watermarked more than 180 billion images and videos and 240,000 years of audio since 2023, and that its checks in Search, the Gemini app, and Chrome handle over one million requests a day. Those are Google’s figures, as reported by Free Press Journal and Arise.

Is it free, and what can I upload?

The reports describe it as open to everyone and mention no charge, though you should check synthid.com for current terms. It takes images, video, and audio. The reports don’t list file formats or size limits. The portal supports English only, according to Search Engine Journal.

Why It Matters to You

You now get a quick extra check on a suspicious clip, which is handy for families and older adults. It’s a first check only. It finds media from companies that use the watermark, a clean result doesn’t prove a file is real, and it can’t screen a live call.

The scam numbers show why people want such a check. The FBI’s 2025 Internet Crime Complaint Center report, as summarized by HousingWire, counted 201,266 complaints from people 60 and older and $7.75 billion in losses, about $38,500 per victim. Experian’s fraud forecast, reported by Fortune in January 2026, expects emotionally convincing “family-in-need” scams to grow with AI. Those figures describe the problem. They don’t show that the detector prevents these losses.

What if it finds nothing?

That doesn’t mean the file is real. Google says SynthID is “not a general AI detector,” and the media may come from a tool that doesn’t use SynthID. Google’s FAQ advises gathering more than one signal before deciding.

What can’t it tell me?

A positive result doesn’t say whether the file is fully AI-made or only lightly edited, per Search Engine Journal. The reports give no accuracy or error rates. The detector checks files you upload, so it can’t screen a live phone call or video call as it happens. One more caveat on our side: we didn’t open Google’s October 2026 blog post, so the 2026 details come from news reports on it.

Sources

Related Reading