The video (and even more the comments) suggests that there is something shady going on.

But how else would the phone detect that there is a face? It's the same with Siri, Amazon Echo etc constantly listening (you can't hear the trigger word without listening).

The real test if something shady is going on is monitoring the network traffic if any of these data is transmitted to an Apple server.

I suspect the 5s interval picture has to do with figuring out if anyone is looking at the phone and get a better sense of when to turn off the screen.