Overview
- Google made the SynthID Detector publicly available on Wednesday, allowing anyone to upload media to synthid.com for a watermark check after earlier limited access for journalists and researchers.
- The detector reads invisible SynthID watermarks embedded in images, video and audio and recognizes marks from Google models plus partner providers such as OpenAI, NVIDIA and Kakao.
- The tool does not inspect text, does not find other vendors’ proprietary marks like Anthropic’s, and a negative result does not prove a file was not generated or altered by AI.
- Users must sign in with a Google, Apple or OpenAI/ChatGPT account to use the service, which raises privacy and inclusion questions for people who prefer not to register.
- Google has integrated SynthID reading into Search, Gemini and Chrome and says Apple support is coming, but multiple competing watermark standards and the ability to corrupt or remove marks mean no single universal detector exists and regulators are forcing more machine-readable labeling.