There are videos of people using their smart devices to get stuck in a loop of asking each other “what was that/can you repeat that?”

They probably interact more through zeroconf/avahi/mdns than speakers and mics. I think both devices have some implementation of blue tooth low energy network mapping. They know about the other smart devices in your home.

There was a Superbowl commercial for Alexa that sounded funny. Every time they said Alexa it sounded robotic. I guess they scrambled the voice just enough for the real units to not respond. Also, the first time they said it… It was subtle, but they said “Lexa” and not “Alexa.”

South Park did an episode about this like ten years ago.

I think it’s possible. At least in the past I could ask Google Assistant on my phone to “talk to Siri”.

Although I couldn’t test it as I don’t have an iPhone.