The keynote session of Google I/O 2025 held on Tuesday showcased significant advancements in artificial intelligence (AI) technology. Among the highlights was the introduction of a new tool designed to enhance transparency regarding AI-generated content. Named SynthID Detector, this verification portal is currently under testing and is capable of detecting and identifying multimodal AI content produced using Google’s AI models. This innovative technology is designed to differentiate between audio, video, images, and text, enabling users to easily determine whether a piece of content is authentically human-created or synthetic.
Google Tests SynthID Detector Verification Platform
SynthID was initially revealed in 2023 as a cutting-edge tool that embeds an imperceptible watermark in content, which cannot be removed or altered. The following year, Google took a step further by open-sourcing the technology for text watermarking, making it available to businesses and developers. This invisible watermark becomes apparent when examined with specialized software. Currently, Google is assessing the functionality of the SynthID Detector, which offers users a quick method to verify if media is AI-generated.
In a recent blog post, Google affirmed that the portal is an essential tool for fostering transparency in the swiftly changing landscape of generative media. With the introduction of advanced AI models like Veo 3 and Imagen 4, which can create hyperrealistic visuals, concerns regarding deepfakes have escalated considerably.
While initiatives such as the Coalition for Content Provenance and Authenticity (C2PA) standard have been established to help AI companies disclose their AI-generated content, these measures are not entirely foolproof. Sophisticated watermarking technologies serve to shield users and organizations from the risks associated with misinformation and harmful synthetic content.
According to Google, the SynthID Detector is user-friendly. Individuals can upload any media they suspect might have been created with Google’s AI tools, and the SynthID Detector will scan the submitted media for any embedded SynthID watermark. The results will indicate whether a watermark is detected and will highlight the specific portions of the content that are presumably AI-generated. It is important to note that this tool is exclusive to media generated by Google’s AI products.
A key advantage of SynthID is its ability to integrate an imperceptible watermark without degrading media quality while ensuring that it cannot be removed or modified. Google is currently providing access to early testers, with plans to expand availability later this year. Journalists, media professionals, and researchers can sign up for a waitlist to secure early access to the SynthID Detector.
