At this year’s Google I/O, the company introduced the SynthID Detector—a new tool designed to identify content generated with Google’s AI models. Developed by Google DeepMind, the detector provides a simple verification portal for confirming whether an image, video, audio file, or piece of text has been created or modified using Google AI.
In a blog post announcing the tool, Pushmeet Kohli, head of research at Google DeepMind, describes SynthID Detector as a way to help establish transparency in AI-generated content, especially as AI becomes more embedded in creative and commercial workflows.
What Is SynthID?
SynthID is a digital watermarking system embedded into content generated by Google’s suite of AI tools—including Gemini, Imagen, Lyria, and Veo. These watermarks are invisible to the human eye or ear, but can be detected by the SynthID Detector system.
The new tool scans uploaded content for these embedded watermarks and highlights which sections are most likely to contain SynthID signals. This can help verify whether content was created with Google AI assistance, a growing concern in media, marketing, and legal circles.
How the SynthID Detector Works
According to Kohli, the SynthID Detector supports the following:
- Images: Highlights specific areas likely embedded with SynthID.
- Audio: Pinpoints precise segments containing a watermark.
- Video: Flags sections and frames where SynthID signals are found.
- Text: Detects if language content was generated or modified by tools like Gemini.
The tool is currently available to a select group of early testers, and Google plans to expand access in phases. Those interested can sign up for a waitlist to try the tool when broader testing begins.
Google sees SynthID as part of a larger push to build trust and traceability in AI-generated content. With misinformation and deepfakes on the rise, there’s increasing demand for tools that can help distinguish between human-created and AI-created content.
Kohli emphasizes that Google will “take learnings from this cohort of professionals and work to implement content transparency more broadly.” That means SynthID Detector could eventually be integrated into content moderation pipelines, media platforms, and digital asset management systems.
Will People Use It?
That remains to be seen. While early testers will help shape how SynthID performs in real-world scenarios, mass adoption depends on accuracy, ease of use, and industry trust. If the tool proves effective, it could become an essential resource for journalists, educators, legal teams, and content creators trying to verify the origin of digital assets.
