as a practical example, Google has agreed to settle over claims the assistant app was recording without the user’s permission (which would be in the form of the command word) *for advertisement purposes.*
There’s two basic scenarios here:
1) completely true, Google knows it, and the faster they can get the lawsuit over with and out of the news cycle, the better. Instant settlement.
2) okay, well, admittedly sometimes the assistant mishears the command word and starts recording, and this may lead to accidentally recording sensitive conversations without consent, but it’s a bug — it’s not an *intentional plan* to illegally record people in hopes they’ll say something that can be used to target ads.
In the second case, if they didn’t want to settle, they’d have to basically disclose all the source code and every email and chat conversation between every employee who has ever touched these systems to exhaustively prove lack of intent, given the fact that there’s some nonzero number of suspicious recordings. Or… they could just sigh and send everyone gift card money for the inconvenience.
I’m inclined to believe it’s case 2 (just because it doesn’t seem like a very efficient way to be evil), but obviously it could be case 1. The point is just that it’s not PROVEN that it’s case 1.