A team of researchers have found a way to create incredibly convincing fake videos of people talking.
They trained their neural network on 17 hours of footage of Obama talking, and here are the results...
They can get him to say excerpts from an existing speech with seamless results. They can also feed in speech from other sources, such as the guy later on in the video imitating Obama.
This stuff is incredibly cool, but also a bit scary at the same time!