No it doesn't confirm that at all.
Currently all smart speakers process on their own device as they come in, and if it hears something it decides is a wake work it sends off a recording to some central servers which process it. All this does is include a short period before the wake word in the data that is dispatched to the server, this is far from "recording every single thing you say."
It certainly increases the amount of things that are sent, and also adds some new threats - you can say the wake word after someone said something "I'm going to kill you", "Alexa" - would result in the threat being recorded. But it does not fundamentally change the nature of the privacy threat of voice assistants beyond that.