Google announced at its annual developers conference the expansion of watermark technology to identify and detect artificial intelligence content.
Each video clip in the VideoFX app is created using the new company model known as Veo, featuring a digital watermark thanks to the SynthID system.
Furthermore, the SynthID system can watermark text created through the Gemini artificial intelligence model.
The company spoke about the new SynthID watermark system and explained that it is now capable of digitally watermarking videos in addition to adding text using artificial intelligence.
The importance of watermarks for AI-generated content is increasing with the proliferation of technology, especially when AI is used for malicious purposes.
Machine learning has been used to spread misleading political information, falsify someone’s statements, and create fake sexual content.
Google announced SynthID last August, adding subtle watermarks to AI-generated content so detection tools can recognize that this content was created by AI.
Due to Veo’s ability to create high-definition long clips better than before, knowing the source of this content is important.
This approach differs from other watermark standards like C2PA, as it involves adding encrypted descriptive data to AI-generated content.
Google allowed the SynthID system to add inaudible watermarks to music created by AI through the Lyria model from DeepMind.
SynthID is just one of several AI assurance measures currently being developed to combat technology misuse, with the Biden administration directing federal agencies to provide guidance on them.
As productive AI models evolve, companies resort to watermarking out of fear that AI might aid in the spread of misleading information.
Watermark systems provide platforms the opportunity to detect AI-created content that may be difficult to distinguish.
