What is SynthID?
By the Remove AI Watermarking team · Updated
Short answer: SynthID is an invisible watermark from Google DeepMind that is embedded in content made by Google's AI tools, including Imagen images, Veo video, Lyria audio and some Gemini text. People can't see or hear it; Google's SynthID Detector can read it.
Key facts
- Made by Google DeepMind; first released in 2023.
- Used across Google's generative models for images, video, audio and some text.
- Designed to survive cropping, compression, filters and resizing.
- Checked with Google's own SynthID Detector, not with metadata tools.
How it differs from metadata labels
Metadata labels like C2PA live beside the content and vanish when the file is re-saved. SynthID lives inside the pixels, sound or word choices, so it usually survives. That also means a metadata reader like our AI Detector can't see it.
How to check an image for SynthID
Use Google's SynthID Detector. For C2PA Content Credentials and other labels stored in the file, use our free AI Detector.
Try it free, right in your browser
Check a file's AI labels