The functions of auto-checking video recordings for the presence of elements generated using the so-called “deep fake” technologies may appear in test mode in social media players and browsers as early as 2022, according to the Director of the SafeNet regional engineering centre of the National Technology Initiative, Denis Kuvikov, on Monday, May 2.
The expert noted that developers of intelligent video analysis services are constantly improving the efficiency of algorithms, increasing the amount of data for training systems and looking for new areas of technology application. At the same time, the widespread use of “deep fake” videos involves improving the methods of their recognition.
Deep fake technologies are used in mobile apps and programs for photo or video editing. A user can change the appearance of people or add an element to a frame, specifically for the creation of fake materials, with the use of such technologies. Kubikov noted that it is almost impossible to determine the traces of video processing by the eye.
However, the expert believes that since such programs already exist on the market, the next stage of development should be the appearance of browser extensions and functions in video players of social networks that allow analyzing the authenticity of recordings.
“It’s a matter of the shortest possible time. A function of this kind may appear in test mode as early as 2022, it is likely that in less than six months we will see the first products,” he said.