Google has expanded its SynthID AI content detector and opened it to the public through a new website, making it easier to check images, video, and audio for invisible AI watermarks from Google and its partners. The update also broadens support beyond Google’s own output, with the company saying the tool can now recognize SynthID marks from OpenAI, Nvidia, Kakao, and, soon, Apple.
Google’s SynthID detector gets a wider reach
SynthID is Google’s invisible watermarking system for AI-generated media. It is embedded in the pixels of images and videos and in the waveform of audio, allowing content to be flagged later without relying on visible labels or metadata.
According to Google, Gemini alone has already integrated SynthID into more than 180 billion images and videos, along with 240,000 years of audio. Until now, people outside Google’s trusted tester program had to use Gemini itself to look for the mark, which was awkward and often produced more context than a simple verification check needed.
A dedicated website replaces the old workaround
The new detector is available at SynthID.com, where users can upload content for a check. Google says the site will tell users whether SynthID was used to produce or edit the media, rather than breaking down the result in the more detailed way its internal tools reportedly can.
That earlier approach had a major limitation: it only checked for Google’s version of SynthID. In practice, that meant content created by other systems with compatible watermarks could be missed. Google says the updated tool now supports the different SynthID watermarks used by its partners, creating a single place to check a file against several major AI systems.
What the new SynthID tool can and cannot do
The expanded detector is meant to make AI watermark checks more practical, but it is not a universal detector for all synthetic media. Google’s own disclaimer acknowledges that plenty of AI-generated content will not be identified by the tool.
That matters because not every model provider uses SynthID or any similar watermarking method. Some use other schemes, some use none at all, and many open-weight models can be run locally to generate highly realistic content without even basic metadata labels.
- Supported through SynthID: Google, OpenAI, Nvidia, Kakao, and soon Apple
- Media types covered: Images, video, and audio
- Result shown: Whether SynthID was detected, not a detailed forensic breakdown
- Access requirement: Login is required
Login and usage limits still apply
Although the detector is publicly available, users must sign in to use it. Google says the site supports Google, OpenAI, and Apple accounts, and each user gets a daily quota of “approximately 10 image, video, and audio checks.”
The company’s earlier tools had similar limits. Google engineers have said access is restricted to make it harder for people to build methods for removing SynthID watermarks. The “approximately” language likely reflects the fact that the system may cut users off earlier if it detects repeated checks on similar files, which could suggest someone is testing a bypass.
Why Google is opening the detector now
Google’s move makes SynthID more useful as a practical screening tool, especially for journalists, creators, and platform moderators trying to understand whether content may have AI involvement. Having a single site that checks multiple major partners is a meaningful improvement over the previous setup, which required more manual steps and produced less direct answers.
Still, the limits are important. The detector only works within the ecosystem of supported watermarking systems, and there is no guarantee that all AI-generated content will carry a detectable signal. Even within that ecosystem, Google says its protections are designed to reduce misuse and make removal harder, not impossible.
The larger verification problem remains
Google’s broader point is that watermarking alone is not enough to solve content authenticity. The company suggests the stronger long-term approach may be the reverse of AI watermarking: cryptographically signing genuine images, video, and audio at the moment of creation.
That idea is already being explored through C2PA, a standard designed to preserve provenance data for media. Google says its Pixel phones are among the few devices using that method. For now, though, SynthID remains one of the most accessible tools for spotting AI fakes, even if it cannot catch everything.
What this means for everyday users
For most people, the practical change is simple: checking suspicious AI media is now easier and more centralized. Instead of asking a model to interpret whether something might be synthetic, users can go straight to Google’s detector and see whether a supported watermark is present.
That does not make the result definitive. A negative check does not prove a file is real, and a positive check only indicates that a supported watermark was found. But in a web filled with AI-generated images, videos, and audio clips, that kind of fast triage is increasingly valuable.
Source: Original report
Was this helpful?
Explore more: AI Automation Services More AI & Automation Tech News
Last Modified: October 7, 2026 at 10:31 pm
0 views
