Every word is a wake word, but most of the time it just wakes up to record the word in a database with a location and timestamp and then goes back to sleep.

Then it sends that information off in the form of a “software update check” or a “user feedback telemetry ping”.

There’s no need to stream the audio. We’ve had computers capable of turning audio into text since the 90s. Classic MacOS 8 could do it.

Theyre always listening. The thing that gets loat in conversation is essentially how much of it is being stored, if its being sent back and the security of the stored data.

Its part of the reason why Apple goes out of its way to try to explain how it wants to handle "Apple Intelligence" in a data safe manner so that it doesnt break any security regulations.

The problem meta has in its hands is they got caught recording entire parts of conversations, rather than have some like 1 second buffer listening to a keyword that gets constantly overwritten, to then activate recording the rest of the coversation.