Google Launches Public Site To Check If a Photo or Video Is AI-Made

Google has opened the SynthID Detector site to everyone, not just to journalists and researchers as before. SynthID is Google's system for watermarking AI-generated media, introduced in 2023. The site checks photos, videos, and audio and shows whether an AI model made them. It can recognize more than just Google's own models, such as the Nano Banana image generator and the Veo video generator. It also flags content from the maker of ChatGPT, OpenAI, from chipmaker Nvidia, and from the Korean company Kakao. All three support SynthID. Google says the service already handles a million requests like this per day.

Checks like this will likely become routine rather than a niche tool for journalists. For anyone dealing with AI-made video or avatars, a watermark check is useful as a first filter. It should not be treated as a final verdict, though. Even Meta's own detection system sometimes fails to recognize content made by its own models. For now, a single watermark check should not be trusted as the last word.