What happened
Google has launched a new public website for SynthID, its tool for identifying AI-generated media. As reported by TechCrunch, the site allows anyone to upload an image, video, or audio file to check for digital watermarks. The tool was previously in a limited preview for researchers and journalists.
Google's own generative models — including Nano Banana, Veo, and Lyria — already use SynthID to watermark their output. The tool is also integrated into the Gemini app and Google Chrome.
How the room's reading it
The move is seen as part of a broader industry push for content provenance standards. While Google has brought partners like OpenAI, Nvidia, and Kakao on board with SynthID, other major players — notably Microsoft and Meta — are pursuing their own watermarking technologies. This fragmentation worries developers building cross-platform applications, as there's no single source of truth for verification.
Practitioners also remain sceptical of the reliability of these tools. Reports frequently highlight that detectors often fail to identify AI-generated content, sometimes even from the models they were designed to track. This makes building dependable systems on top of them a challenge.
Sailfish's take
We see this less as a security feature and more as a user-interface primitive. The unreliability of current detection methods means you can't build a robust moderation system on SynthID — it will miss things, and that's a critical failure. We've shipped enough content-centric apps to know that relying on a fallible, third-party detector for enforcement is a recipe for trouble.
Instead, we'd use this as a transparency tool. The useful application here isn't automatically blocking content, but rather adding a simple 'likely AI-generated' label. It's a signal for users, not a gate for your platform.