At its recent I/O event, Google introduced a new tool: SynthID Detector. This tool helps users verify if content was created using Google’s AI models. It supports all media: images, audio, video, and text.
SynthID Detector works by checking for hidden digital watermarks. These watermarks are embedded in files created by tools like Gemini, Imagen, Lyria, and Veo.
Although invisible to the eye, they are detectable by the portal.
What SynthID Detector Does
SynthID Detector scans content and highlights the parts most likely to be AI-generated. It pinpoints watermarked image areas, audio segments, video frames, and text likely produced by AI.
How It Works
According to Pushmeet Kohli, Head of Research at Google DeepMind, the tool acts as a “verification portal.”
In each verification process, it gives added details more than a simple yes or no. It pinpoints the watermarked areas:
- For images, it highlights areas that carry the watermark.
- For videos, it checks individual frames for signs of AI.
- For audio, it flags time segments with watermarks.
- For text, it spots and highlights likely AI-written sections.
Why Bother with This Tool
AI-generated content is everywhere. People use AI tools to write, draw, compose music, and edit videos.
As a result, it’s harder to know if a piece of content came from a human or a machine. However, a tool like SynthID Detector can help rebuild trust in digital media by providing clear distinctions between human and AI content.
Plus, it supports creators and media outlets by protecting their voice. Recently, there have been a surge in the use of deepfakes to mislead and misinform.
A tool like this can simplify the process of AI content identification, thereby preventing the spread of false information.
Who Can Use It?
Right now, the tool is available to early testers. However, others can join a waitlist to access it later. Google plans to learn from this first group and improve the tool based on feedback.
And eventually, more users will gain access.
