Google has launched its SynthID Detector website globally, allowing anyone with a supported account to check whether images, videos, or audio files carry an imperceptible AI watermark. The tool currently recognizes content generated by Google, OpenAI, Nvidia, and Kakao, while Apple plans to add support soon.
Google made its SynthID Detector publicly available worldwide on Tuesday, moving the tool from limited testing into an open web portal available globally in English. Pushmeet Kohli, VP of Science and Strategic Initiatives at Google DeepMind, announced the expansion to let media teams and the public check files before publication, according to Search Engine Journal. Last year at Google I/O, the tech giant initially provided this testing utility to specific media professionals, journalists, and researchers, as pointed out by TechCrunch.
The website requires users to log in with a Google, Apple, or OpenAI account before uploading files, as CNET and MacRumors point out. Google recommends uploading the highest-quality copy available to get the most accurate result from the detector. The site runs at synthid.com, where users upload a file to check it.
Google Expands AI Watermark Support
The standalone detector supports a wide range of file types for digital media creators, though its underlying technology relies entirely on industry participation. According to reporting by TechCrunch, the supported formats include:
- Images: JPG, JPEG, PNG, BMP, WEBP, AVIF, HEIC, HEIF, TIFF, TIF, and GIF
- Videos: MP4, MOV, and WEBM
- Audio: WAV, MP3, OGG, FLAC, AAC, and M4A
Google originally introduced SynthID in 2023. Since then, the company states it has watermarked over 180 billion images and videos, alongside 240,000 years of audio content
across its generative models, figures echoed by iAfrica and GSMArena. SynthID is utilized to embed watermarks into all media created by the corporation’s Veo, Lyria, and Nano Banana models, alongside generation-capable systems such as Vids, Gemini, Flow, and ProducerAI.
Detector Limitations Define SynthID Reach
While the tool provides a straightforward way to verify participating AI media, analysts note caveats for everyday users. The detector cannot identify content generated by companies outside the SynthID ecosystem, such as Meta, xAI, Stability AI, and Midjourney, nor can it scan the open-weight model ecosystem, as highlighted by iAfrica. OpenAI maintains its own site for checking content, and Microsoft and Meta have their own standards for watermarking and verifying content, though these tools are not infallible and often fail to identify content created by their makers’ models.
MacRumors reports that Google's tool cannot distinguish between AI-created and AI-edited media, meaning an original photograph edited using AI tools will also return positive watermark detections indicating the media was made or edited with participating models.
“This is not a general AI detector. We can only detect media from the above companies, who have adopted SynthID technology.”
Google, via Search Engine Journal
Nigeria Prepares for 2027 Election Cycle
Communications managers have noted practical applications for the public verification tool as major democratic votes approach. Taiwo Kola-Ogunlade, Google’s West Africa Communications and Public Affairs Manager, pointed to upcoming political campaigns in Nigeria as a key use case.