Google is making SynthID, a watermark it uses itself to mark AI content, open-source available to developers.
Generative AI tools are getting better and better at creating texts, images and even videos. Scared that soon you won’t be able to tell the difference between what’s made by a human, and what’s made by AI? Google wants to keep the distinction alive with SynthID, a tool it is making open source starting today, as announced at X.
SynthID is a watermark that marks content produced by AI as such and can be applied to text, images and video. Google first demonstrated the technology last August. The version Google is now making available is designed for AI-written text.
Invisible
The watermark is inserted into the metadata so that it is invisible to the naked eye, but systems can detect it. This is done during the writing process. Google makes the case that its watermark is stronger than alternatives and that it also does not compromise quality, accuracy, creativity and speed. With shorter, factual texts, the tool would have more trouble than with long texts.
read also
Google makes watermark for AI text open-source
SynthID is already integrated into Gemini today, and now Google is making the watermark available for developers to integrate into their AI systems. Google also uses SynthID to mark images and videos created by its own AI.