Having a microphone/speaker in people's houses seems like a great way to collect a ton of data & guide their behavior.
The devices listen for their wake word before ever transmitting data to the cloud. (This is also because the sheer amount of bandwidth needed for always-on would be un-economical even for Amazon).
No clear idea what that data is, though. There's at least a small chance it isn't benign.
As/when it's put on my institution's open repository I'll post a link. Might be a few weeks though.
While absence is hard to prove, an intentional or unintentional data collection beyond what is stated in their policies is highly unlikely.
Hell, don't even send the sound, just send a weekly report of events.