Google’s SynthID Detector Is Now Public: Here’s What It Can Actually Detect

Google has opened its SynthID Detector to the public, allowing users to check images, videos and audio for AI watermarks. The tool, previously limited to journalists, researchers and media professionals, can now be accessed by anyone in English. SynthID is an invisible watermark embedded into AI-generated content to distinguish it from authentic media.

The detector identifies watermarks from Google and participating companies, such as OpenAI, NVIDIA and Kakao. Apple support is expected "soon." Supported formats include JPG, PNG, WEBP, GIF, MP4, MOV, MP3, WAV and FLAC. SynthID does not determine if content "looks" AI-generated; it searches for the watermark embedded by supported AI systems. The watermarks remain detectable even after common changes like noise, MP3 compression and speed alterations.

Google's SynthID Detector is primarily a user-friendly interface for the existing verification system, which is already integrated into products like Gemini, Search and Chrome, handling over 1 million verification requests daily. While SynthID can't detect every AI-generated content, it's useful for establishing that supported AI was involved in creation.

However, it cannot confirm the entire history of a file or rule out AI generation entirely. Users should treat the result as a signal, not a final verdict, and combine it with other evidence for accurate assessment.