Many modern cars already have something like this to see if you're falling sleeping while behind the wheel and will start vibrating the seats and steering while sounding alarms to get you alert again.
That being said, it's definitely over engineered compared to a "are you still active" button on the screen.
This isn't about detection and isn't about triggering a sleep mode at all. It's not even about over-engineering for this particular function, because this particular function is not the point. What it's actually about is doing everything they can conceivably think of to monetize engagement tracking, while continuously pointing a camera at your face as you watch movies and adds.
It doesn't make my post a mistake for not accounting for every last thought you had about a different conversation. And if you need it to be clear, by over-engineered I do consider these unnecessary and unwanted.
Can you source this claim? Unless you are part of the engineering or monetization team of this project, I don't understand how you'd know this.