How does SynthID detect whether a text is watermarked through the API?

I saw a tool for adding watermarks to text and detecting watermarked text at SynthID:用于为 LLM 生成的文本添加水印和检测此类文本的工具  |  Responsible Generative AI Toolkit  |  Google AI for Developers

I learned from the SynthID - Google DeepMind page that there is an entry.

But I cannot verify whether the results on the SynthID - Google DeepMind page are watermarked

Hi @lidashuaiqi ,

There is no public API as of now that verifies SynthID watermark on text content directly. You may need to request access or use internal tools mentioned on DeepMind or Google AI sites.