Meta has recely announced that it has developed a dedicated tool for watermarking videos made with artificial ielligence.
Detecting coe created with artificial ielligence can be an importa step to preve the spread of false coe and fake news. This issue is more importa for videos made with generative models; Because by simulating people’s appearance, it can cause more serious injuries. According to research, the amou of deepfake coe has seen a 4-fold growth from 2023 to 2024, and this category of coe includes about 7% of iernet fraud.
What is the use of artificial ielligence watermark videos?
Imitation of people’s personalities, hacking of user accous, especially bank accous, and complex social engineering in order to have a wide impact on society are some of the things that can be achieved with deep fake coe. Now, to deal with such phenomena, Meta has developed a special tool for putting watermarks on clips made with artificial ielligence. This tool, called Meta Video Seal, is provided in open source form and can be used in curre artificial ielligence systems.
Meta has other tools to make it easier to detect AI coe, including Watermark Anything and Audio Seal. “Pierre Fernandez”, who is one of the artificial ielligence researchers at Meta, has announced that the company’s new tool has a better ability to detect fake videos and can protect real coe with higher accuracy.
It is worth meioning that Google’s DeepMind departme has also created a tool called SyhID, which is also used for watermarking videos. Microsoft also has its own similar tool. Of course, the meta service has more capabilities compared to its competitors, including the ability to hide messages in videos.
