> A significant chunk of onboarding is driving the message through that if you abuse your powers as an engineer to access personal user data, you will be insta-fired without warning.
I think focusing on a rogue employee as the threat it missing the mark. The real threat is the ability for huge corporations to legitimize institutional access to data ("business reasons"), and to fight the legal fight if they are caught doing something wrong. Moral hazards of this mode of failure arguably has much more cumulative societal damage (e.g. Cambridge Analytica).
In other words, I worry less about an evil employee reading my messages than a scaled up corporate machinery milking all my data for dollars that feeds the machine back and also can fight any fight at the courts, DC and the public eye.