Particle.news

Download on the App Store

Google Launches SynthID Detector to Verify AI-Generated Media

The new portal allows users to detect watermarked AI-generated content across text, images, audio, and video, but faces limitations in scope and circumvention risks.

Image
google AI generative ai AI images Google DeepMind SynthID Detector AI audio SynthID
Image

Overview

  • SynthID Detector, announced at Google I/O 2025, enables users to upload files and check for imperceptible watermarks embedded by Google's AI tools.
  • The portal supports detection across multiple media types, including text, images, audio, and video, marking a significant expansion of SynthID's capabilities since its 2023 launch.
  • Google has partnered with NVIDIA and GetReal Security to extend SynthID adoption beyond its own platforms, aiming to build a broader ecosystem of content verification tools.
  • More than 10 billion pieces of media have been watermarked with SynthID to date, but its detection is limited to media created with compliant tools and is vulnerable to circumvention, especially in text.
  • SynthID Detector is currently available to early testers via a waitlist, with broader access planned to help combat deepfakes and AI-driven misinformation.