Google has introduced an online portal SynthID Detector, which allows for quick verification of whether an image, video, audio file, or text was created using the company’s generative AI. The tool is based on SynthID technology, which embeds invisible watermarks into content generated by Google models such as Gemini, Imagen, Lyria, and Veo. The watermarks remain even after editing, such as cropping or applying filters.
Users can upload a file to the SynthID Detector portal, after which the system checks it for the presence of a watermark. If such a mark is found, the service shows which specific parts of the content contain it. For audio, separate segments are identified, and for images, specific areas where the likelihood of a watermark is highest.
In the initial phase, SynthID Detector is available to a limited group of testers. After this period, Google will gradually open access to users who register on the waiting list. The company pays special attention to collaborating with journalists, media professionals, and AI researchers to ensure transparency in the use of generative models.
Since the launch of the technology in two thousand twenty-three, over ten billion units of media content have already received SynthID watermarks. According to Google representatives, the implementation of such tools aims to increase trust in AI content and assist users in recognizing materials created using the company’s generative models.
