Google debuts SynthID Detector tool for flagging AI-generated content, but it’s far from perfect
Google's SynthID Detector is now public — but it can't reliably spot AI video yet.
Why it matters
Google is widening access to its AI-content detection tool, a move with real implications for enterprises managing synthetic media risk and content moderation. The catch: detection accuracy degrades sharply on edited or manipulated content, and it only identifies material created by Google, OpenAI, and Nvidia models — a significant blind spot for practitioners building moderation stacks.
The key facts
13 to knowSynthID Detector launched as public web tool
Can identify AI-generated and AI-edited images, video, and audio
Works on content created by Google, OpenAI, and Nvidia models only
Detection accuracy is 'far from perfect' per headline; specific benchmarks not disclosed in excerpt
Watermark-based detection approach (implied by SynthID name)
No pricing or rate-limit data disclosed
No regional restrictions mentioned
Google SynthID Detector now publicly available via web interface
Detects AI-generated and AI-edited images, video, and audio
Works on content from Google, OpenAI, and Nvidia models
Accuracy described as 'far from perfect' by reporting outlet
No independent benchmark data or false-positive/false-negative rates disclosed
Watermarking-based detection limits utility to Google-originated content
The story so far
Earlier coverage of this storyline
- Google’s new SynthID website can identify AI-generated mediaTechCrunch AI
- This story
Go to the source
SiliconAnglesiliconangle.com
Publisher excerpt: Google LLC is making its SynthID identification tools for artificial intelligence-created and AI-edited videos and images more widely available with the launch of a new service today that everyone can use. The web-based SynthID Detector can identify images, video and audio clips that have been…