![](https://azmotech.com/wp-content/uploads/2024/05/1-236-1024x614.webp)
At the Google I/O 2024 event, numerous new AI features were introduced to Google’s suite of apps. Among these additions is an upgrade to SynthID, Google’s existing AI watermarking system. Previously capable of detecting AI-generated images and audio, SynthID is now expanding to support more media formats.
SynthID can now embed watermarks into videos and texts generated by AI. According to Google, these digital watermarks are undetectable to the human eye and do not compromise video quality. The technique involves embedding a watermark into the pixels of every frame of a video.
The company states that the watermarks will remain intact even after modifications to videos. This means that cropping, adding filters, changing frame rates, and adjusting color tones will not affect the detectability of the digital watermark. As AI video generation continues to grow, the purpose of this system is to help viewers identify whether a video is AI-generated.
![](https://azmotech.com/wp-content/uploads/2024/05/2-140.webp)
For text-based content, SynthID will embed a watermark during real-time text generation. The system predicts the next most likely set of text for a sequence and assigns probability scores to possible characters, words, or phrases. This process is repeated throughout the entire text generation, and the combined probability score serves as the watermark.
Availability of Google SynthID
Google has confirmed that SynthID will begin watermarking texts generated on the Gemini mobile app and web app. Additionally, digital watermarks will be embedded in videos generated on Google’s new Veo AI Video Generator. The company has stated that SynthID can also be combined with other AI detection techniques for broader coverage. In an effort to prevent misuse of AI tools, Google intends to open-source SynthID to enable its use across a wider AI ecosystem.
What are your thoughts on Google’s upgraded SynthID system? Do you believe this tool can help curb the spread of misinformation and assist people in making informed decisions about AI-generated content? Share your opinions in the comments below.
0 Comments