A few days after appearing in the Chinese segment of the AppStore, the Zao application became the leader in downloads for iOS. It is not available outside the country, but in China itself it is very popular, albeit overshadowed by a number of scandals. If you look at absolute indicators, then for the last three days more deepfake video has been generated in the world than in the entire history of the development of this technology.
Virtually every video created with Zao is a masterful deepfake. The application draws the user's face instead of the faces of the original characters in any videos. With its help, you can add yourself to any film fragment, news story, TV show, historical chronicle, etc. The result looks very reliable, especially considering that the initial data is static photographs, on the basis of which a dynamic picture is generated.
Moreover, a single image of a person is enough for the application to create many different deepfakes. The AI that lies at the heart of Zao's work raises many questions from experts. There are serious concerns that something much more serious is behind the presentation of the entertainment service. The user agreement for the first version explicitly stated that the application can handle the received data as it pleases and without the permission of consumers. This caused outrage and the rules were quickly edited.
However, China's main social platform, WeChat, has blocked Zao. It is very likely that the same will be done in the West, if the authors of the application begin to distribute it in other countries. According to one of the versions, behind the project may be an attempt by the Chinese authorities to optimize facial recognition technology in order to teach it to resist possible countermeasures. Who, if not a deepfake developer in this area, knows best how it works and how to trick AI? It remains only to conduct several hundred million tests, with which users who are keen on the new toy are doing an excellent job.