It could use array microphone to detect that the sound originates from the driver's seat (in addition to using it for filtering out not-from-driver's-seat sounds).
I mean that's cool, if it actually works reliably. Automation tech enthusiasts are willing to be impressed by 99% reliability. To get buy-in from regular people, it needs to just work better. To be clear, I don't know how reliable this actually is now, but I'd be willing to bet that it's not reliable enough.
Both could be possible. It belongs in an R&D lab unless and until it's good enough. If, after some time and dollar threshold, it's still not good enough there might be a case for spending those resources elsewhere.