Google’s SynthID Detector Is Now Public: Here’s What It Can Actually Detect
Google's SynthID Detector now lets anyone check if images, videos, and audio carry AI watermarks. Here's what it actually catches — and what it misses.
Why it matters
Google opens its AI watermark detection tool publicly, shifting from closed beta to public availability. For enterprises and practitioners using or producing synthetic media, this changes the transparency and verification landscape — though detection remains probabilistic and incomplete.
The key facts
13 to knowSynthID Detector now public (date: October 8, 2026)
Detects watermarks in images, videos, and audio
Tool limited to detecting Google's own SynthID watermarks, not third-party AI generation
Detection is probabilistic, not binary verification
No independent benchmark provided for false positive/negative rates
Source does not disclose regional availability or API/integration details
Watermark methodology relies on imperceptible digital signatures embedded during generation
SynthID Detector is now publicly available
Can detect images, videos, and audio with Google SynthID watermarks
Cannot verify AI-generated content that lacks watermarks
Watermark detection is Google-specific (only detects SynthID, not competitor watermarks)
Published October 8, 2026
Tool addresses AI authentication and provenance verification use cases
The story so far
Earlier coverage of this storyline
Go to the source
TechRepublictechrepublic.com
Publisher excerpt: Google opens SynthID Detector to the public, letting users check images, videos and audio for AI watermarks. Here's what the tool can and cannot verify.