People do try, but it isn't the basis for a successful computing platform. Works for block chain (though AMD delivers more flops/$ for simple sha-ing grunt work) and specific workloads where you have time and expertise but not much cash.
As for nVidia banning it, if you've the cash you can buy cards, just becomes painful. nVidia were doing it more because the GPGPU demand was choking the supply to graphics customers, and later dumping obsolete (for crunching) cards on the market en masse, interfering with the mid-tier card sales channel.
The always-listening portion is not. Also “enabled by default” still requires active user consent and agreement to plain English privacy terms. It is a step that cannot be missed when setting up the phone.
"Voice Match", the always-listening functionality of Google Assistant, also needs to be enabled. It requires training to your particular voice in order to make it more difficult for someone else to try to access your information using the feature. When being enabled a warning regarding the potential security implications of doing so is presented.
Occam's razor also needs to be applied. If the explanation involving malice is the simpler option that makes fewer assumptions, then it is more likely.
I bet this has been discussed internally, but nobody cares because they don’t care about UX enough. They don’t understand the impact. The internal organization is fractured into fiefdoms defending internal services (e.g. deployment, AMP) and defending layers (e.g. SRE department), with nobody left responsible for the complete, vertically integrated user experience.