Imagine being on a video call and everyone interacting with you are fake?
Which is technically possible but it is not going to be realistic unless all those videos were prerecorded because to process one deep fake video of few second will need few minutes if your hardware is on high end so imagine processing 5 videos on real time that would require pool of graphic resource like a mining pool to get the result that you expect and it won't be same as realistic.
But scammers orchestrate things perfectly and that's how they are pulling money from users, so it is important to be careful and sensible as well.