Google opened its SynthID Detector to the public on 7th October, 2026, letting anyone check a file for hidden AI watermarks. The site, synthid.com, works globally but only in English for now.
What Happened?
- Google DeepMind’s SynthID has watermarked over 180 billion images and videos, plus 240,000 years of audio, since its 2023 launch.
- Files made by Google models can be checked, along with output from partners OpenAI, NVIDIA and Kakao, with Apple support coming soon.
- Supported uploads include JPG, PNG, HEIC and GIF images, MP4, MOV and WEBM video, and MP3, WAV and FLAC audio.
- Microsoft and Meta run their own watermark standards. Those tools have often missed content made by their own models.
Until this week, the detector sat with a test group of journalists, media professionals and researchers who got access at Google I/O last year. The public version takes uploads straight from a browser. Google’s announcement says anyone can now “check if an image, video, or audio file was made with AI” from Google or its partners.
Since launching SynthID in 2023 to embed watermarks into AI-generated content, we’ve watermarked more than: 🖼️ 180 billion images and videos 🔊 240,000 years of audio content 🔍 1 million daily verification requests regularly across Search, @GeminiApp, and @GoogleChrome Today, we’re taking the next big step in our industry-wide effort to make AI content more transparent in partnership with @OpenAI, @NVIDIA, Kakao, and soon @Apple. We’re expanding access to SynthID Detector — a verification portal to identify AI-generated content — to everyone around the world in English, so anyone can check if an image, video, or audio file was made with AI from Google or our partners. Try SynthID Detector → https://t.co/0cI7xzG6j4
— News from Google (@NewsFromGoogle) October 7, 2026
Which AI tools leave a SynthID mark?
Google’s Nano Banana image model, Veo and Lyria stamp the watermark on everything they generate. So do apps with generation features, including Gemini, Flow, ProducerAI and Vids.
OpenAI, NVIDIA and Kakao also support SynthID. OpenAI still keeps its own verification page, so anyone checking a suspected ChatGPT image has two places to look.
The new site joins checks Google already built into Search, the Gemini app and Chrome. Google said those built-in tools regularly handle over 1 million requests a day.
Rival watermarks sit outside the partner list
The detector looks for one mark. Microsoft and Meta run separate standards for watermarking and verifying AI content, and neither company appears on Google’s partner list. Watermark checks have also stumbled on their makers’ own work. Meta’s detector missed some of its own cropped AI images earlier this year.
That sets a clear boundary. A match points to a Google or partner model, while a “not detected” result leaves the file’s origin open. Google’s announcement also leaves several practical questions unanswered:
- How well does detection hold up on cropped, compressed or screen-recorded files?
- When will media from Apple’s tools carry a mark the site can read?
- When will languages beyond English arrive?
- Does Google store the files people upload for checking?
Anyone vetting a viral image or clip can upload it to synthid.com before sharing it. Suspected OpenAI output can go through OpenAI’s page too. A negative result should be read as “unknown,” and the file still needs the usual checks on source, date and context.
The bigger change is reach. SynthID checks used to live inside Google’s apps or a press-only pilot. Now any file from anywhere can go into one public page, and every new partner widens what a single upload can confirm. That makes Apple’s promised support the next marker to watch. Until rivals join, SynthID works best as one check among several. The others include AI detector tools that judge content without a watermark.
