I can't speak for the google home, but with the echo there's a lot of work that has gone into making it so it is difficult, on a hardware level, to record without showing some sign of doing so. When the mic is on the lights are on as well and the mute button is a physical hardware disconnect that stops voltage from the mics.
There is also some interesting circuitry around the mic itself- the keyword activation chip essentially has control over the mics- they buffer input until they hear the codeword ('alexa') and then activate the main circuitry and dump the buffer into it. The upside to this is that software can't turn the mic on by itself- the chip needs to help. This isn't completely perfect though as once the software has been given control it can keep that control forever (with the caveat that lights would be on).
It's possible to use technology to our advantage without having to sacrifice privacy, we just have to demand it (and, as we all should have learned from Therac-25, hardware limits are preferable over software).
There is also some interesting circuitry around the mic itself- the keyword activation chip essentially has control over the mics- they buffer input until they hear the codeword ('alexa') and then activate the main circuitry and dump the buffer into it. The upside to this is that software can't turn the mic on by itself- the chip needs to help. This isn't completely perfect though as once the software has been given control it can keep that control forever (with the caveat that lights would be on).
It's possible to use technology to our advantage without having to sacrifice privacy, we just have to demand it (and, as we all should have learned from Therac-25, hardware limits are preferable over software).