I'm really confused about exactly what "AI" is, and so is the rest of the world, apparently.
If my understanding is correct, this doesn't require AI, It requires a renderer, a photo, and a script. the software to sync the lips isn't AI, the software to show emotion isn't AI,
AI would come in if you had it write the script, and the use of Machine Learning would help it do so.
The renderer in the hands of humans writing scripts to create deep fakes is probably a lot more dangerous than a complete AI character based on a real person for at least the next 5 years or so, and it will probably only take a few of those before the general public learns that video evidence is worthless and stops trusting video altogether. I'm guessing this is not the path to chaos.
Chaos comes when we give AI the instruction to protect us, and the means to do it.... Then we're done for!