How good are deep fakes these days? It’s pretty good to get a look at the latest deep fake technologies. I traveled to Las Vegas for the Defcon conference, which brings together tens of thousands of computer hackers and other cybersecurity experts. Each year, the conference consists of several villages covering everything from social engineering to lockpicking to aviation security and, of course, artificial intelligence. In the AI ​​Village, I met Brandon Kovacs, a top red teamer at cybersecurity company Bishop Fox, who used only a gaming computer, a DS LR camera, a green screen and basic studio lighting, as well as a few funny wigs and other props. He and his AI model were able to create some pretty convincing deep fakes. Volunteers were transformed into all sorts of people, from his colleagues to politicians. And I watched as the AI ​​turned me first into Keanu Reeves and then into Jackie Chan. Later, I got the chance to play with the software myself and tweak the face swapping software. This meant that the deep fake’s digital mask fit perfectly over the person being transformed. Kovac says the idea behind the demonstration is to show people that deep fakes exist and to create examples that AI models used for deep fake detection can learn from.