As generative AI continues to blur the line between human and machine-created content, Google has introduced a significant tool to maintain digital trust. The SynthID Detector portal, announced at Google I/O 2025, provides a centralized platform for verifying whether content contains Google's invisible SynthID watermarks.
The technology works by scanning uploaded media for imperceptible watermarks embedded during the AI generation process. When a watermark is detected, the system can pinpoint specific areas of an image, segments of audio or video, or portions of text most likely to contain the watermark. This granular approach helps users understand not just if content was AI-generated, but precisely which parts were created or modified using Google's AI tools.
SynthID has evolved significantly since its 2023 introduction. Initially focused on images, the technology now spans all major content types and works across Google's AI ecosystem, including Gemini for text, Imagen for images, Lyria for audio, and Veo for video. The system has already watermarked over 10 billion pieces of content, demonstrating widespread implementation.
To expand SynthID's reach beyond Google's ecosystem, the company has partnered with NVIDIA to watermark videos generated by their Cosmos microservice and with GetReal Security for third-party verification capabilities. Google has also open-sourced the SynthID text watermarking framework, allowing developers to incorporate the technology into their own models.
Despite these advances, SynthID faces limitations. The system primarily works with content generated by Google's AI tools or those using its watermarking technology. Content from other AI providers like OpenAI's ChatGPT remains undetectable. Additionally, Google acknowledges that SynthID isn't foolproof against extreme manipulations, particularly with text that's heavily rewritten or translated.
The SynthID Detector is currently available to early testers, with journalists, media professionals, and researchers able to join a waitlist for access. As AI-generated content becomes increasingly sophisticated, tools like SynthID represent a crucial step toward maintaining transparency and combating misinformation in the digital landscape.