SDSignal Desk

Google’s new SynthID website can identify AI-generated media

Oct 7, 2026, 7:00 AM · TechCrunch

Image: TechCrunch

Google just handed everyone a free AI-media checker, which is a real public good, as long as nobody treats a blank result as proof something is human-made.

Why it matters

Google has opened a SynthID verification site to the general public. Anyone can upload an image, video or audio clip and check whether it carries the SynthID watermark. Until now, access was limited to select journalists, media professionals and researchers who had been testing it since Google I/O last year.

The tool covers a broad list of everyday formats, including JPG, PNG, WEBP, HEIC and GIF images, MP4, MOV and WEBM video, and MP3, WAV, FLAC and M4A audio. Google already builds SynthID checks into the Gemini app and Chrome, and says people make about 1 million verification requests a day. A standalone public site makes that capability available to anyone with a suspicious file and a browser.

From the desk

We think this is a good move, and we want to say so plainly. When a fake clip is spreading, the people who most need a quick check are usually not the ones with research access: a local reporter, a school administrator, a family member looking at a viral video. Putting the checker on the open web lowers that barrier, and the format coverage suggests Google is thinking about real-world files, not just pristine test images.

The ecosystem is also wider than one company. Google's own Nano Banana, Veo and Lyria models, along with tools like Gemini, Flow, ProducerAI and Vids, watermark what they generate with SynthID. OpenAI, Nvidia and Kakao support it too, and Apple is reportedly adding support. That is the closest thing we have to a shared signal across major generators.

Now the caution. A SynthID checker can only find SynthID. Microsoft and Meta run their own watermarking standards, OpenAI keeps its own checking site, and plenty of generators mark nothing at all. These tools are also not infallible, often missing content made by their own makers' models. So the most dangerous misreading of this site is the negative one: "no watermark found" does not mean "real." If the public learns to treat a clean result as a certificate of authenticity, bad actors will simply use unmarked tools and enjoy a false stamp of credibility.

Our read: the provenance fight is shifting from whether to watermark to how many checkers a person has to visit. The more useful future is one verification step that reads several standards at once. Until then, this is a solid, free tool that answers one narrow question well, and that is worth having.

Context

Google introduced SynthID in 2023 and has already built verification into the Gemini app and Chrome. This launch adds a public website alongside those built-in checks.

Who feels it

Journalists and fact-checkers
A free, public checker across images, video and audio speeds up first-pass triage, though a negative result still needs other reporting.
Everyday users
Anyone can now check a suspicious file for Google's watermark, but should understand it only detects SynthID-marked content.
Platforms and generator makers
Pressure grows on companies outside the SynthID camp to make their own marks easy to check, or to interoperate.

What to watch

  1. Whether Apple's reported SynthID support actually ships
  2. Any move toward a single checker that reads SynthID alongside Microsoft, Meta and OpenAI signals
  3. Independent tests of how well SynthID survives cropping, compression and re-recording
  4. Whether daily verification volume grows meaningfully now that the site is public

Read the original

Continue at the source.

TechCrunch

Companies: Google