Google has introduced a new tool designed to verify whether content was created using its artificial intelligence systems. The feature, known as SynthID Detector, was unveiled during the company’s recent Google I/O event.
According to a statement from Google DeepMind’s Pushmeet Kohli, this tool functions as a “verification portal” capable of identifying AI-generated media that has been watermarked by Google’s proprietary SynthID system. It supports a range of formats including images, audio, text, and video.
The technology behind SynthID allows watermarks to be subtly embedded into digital content generated by Google’s AI models such as Gemini, Imagen, Lyria, and Veo. These marks are designed to be imperceptible to the human eye or ear but can be picked up by SynthID Detector.
Here’s how it works: Users can upload media created using Google AI, and the portal scans the file to detect SynthID watermarks. If a watermark is found, it visually highlights the segments where the embedded data is most likely present. This applies to all supported media types—images, audio clips, videos, and text.
In its initial phase, the tool is being made available to a group of early adopters. Kohli mentioned that feedback from these initial testers will help refine the tool before broader public access is granted through a waitlist system.
