[09:38 Fri,18.March 2022 by Thomas Richter] |
It actually took an astonishingly long time, but yesterday it was done: the first DeepFake used instrumentally in a major global political crisis has been produced: a fake video in which Ukrainian President Volodymyr Zelensky calls on his people to surrender, appeared on the official, but hacked Ukrainian news site "Ukraine 24" together with an identical ticker text.
The video, surprisingly for its propagandistic importance, is not a very well-made DeepFake (his face is quite convincing, but the head does not move in unison with the body, the voice sounds mechanical and his accent is fake), but that may not have mattered for such an action, after all, it has been viewed and sharked online several 100,000 thousand times, doing its rounds and sowing doubts (about the authenticity of real Zelensky videos) . However, the Ukrainian government reacted quickly and  Zelensky himself made it clear via video that the surrender statement was a fake - the fake clip was then quickly deleted on all major platforms such as Facebook and YouTube. In this case, the DeepFake was recognized as such pretty quickly. One thing is clear: in order to have a truly resounding effect, a DeepFake must be both well-made and credible, distributed via trustworthy channels, and the original must be deprived of the opportunity to quickly contradict it publicly in a credible manner. But as the DeepFake technology is getting better and better and needs less and less foreknowledge, probably many more DeepFakes will appear in at more or less celebrities, but also private persons, put into the world by ex-partners, competitors, misguided fans or whoever and for whatever reasons. It is interesting to remember in what context DeepFakes first came on the scene around 4 years ago: in porn movies, the faces of the actresses were replaced with those of well-known actresses. Here is the very well-known and well-done Tom Cruise DeepFake What are DeepFakes?DeepFakes - a conglomerate of Deep Learning and Fake - use Deep Learning technologies, which are trained on sample footage of many faces, then swap one face for another in a video. The replaced face then performs the mouth movements and facial expressions of the original face, which then acts like a puppeteer, but with its own special idiosyncrasies (learned via training). In the case of Zelensky, there was in any case enough material of Zelensky&s speeches available to train the DeepFake-Learning algorithm with it. The whole thing can then be supplemented by a simulation of the voice of the person exchanged, which was also done by Deep Learning. Then it is relatively easy to put any words into the mouth of any person, which (depending on the technology used, the expertise of the fakers and the training material) look deceptively real. From the very beginning of the DeepFake technology it was clear and warned that the technique could also be used for *big actions*&. Meanwhile, anyone can easily acquire the relevant expertise and create such fakes by means of some source video footage of the face to be swapped in. And film studios such as Disney rely on DeepFake technology, for example, to bring dead actors (such as Princess Leia in future Star Wars episodes) back to life as cinematically lifelike as possible. Bilder zur Newsmeldung:
 more infos at bei www.vice.comdeutsche Version dieser Seite: DeepFake Video als Kriegspropaganda - vermeintliche Kapitulation von Zelensky |
;)
;)
 |
|
|