An unsolved design problem: False activations in smart speakers

I remember reading that Amazon chose "Alexa" as their wake-up word because it tested well in different activation tests and they already owned trademarks for the word. On the other hand, Google went with a brand focused wake-up word that doesn't work that well for this type of interaction.

Of course, because of this, many TV shows activate Google Assistant when somebody says "Ok, {{word that vaguely resembles the phonetics of the word "google"}}"

How did they manage to oversee this issue? Seems like they compromised the final user experience just to find a way to extend their brand equity to a mundane interaction.