The use of artificial intelligence (AI) is becoming more and more widespread, causing various problems such as students not writing specific assignments themselves. There are several tools that have been developed to detect the use of AI like this. Today Google also launched SynthID Text which can not only detect generative AI generated works but also allows watermarks to be placed on AI works so that they can be detected by other detection systems.
Google is now open-sourcing SynthID Text to encourage other generative AI developers to use their watermarking system. At this point SynthID Text can be accessed through Hungging Face and has been integrated into Gemini.
Watermarks allowed into the text will still be preserved if part of the AI-generated text is edited or rewritten. However, it is not very effective if the AI text has been translated, the text is short or the answer is in the form of facts.
In August last year, Google launched SynthID which embeds watermarks on AI-generated images. The two digital watermarks embedded into these AI-generated works cannot be detected by humans but can be detected using specialized software.