> that it's not universally offensive and reviling to post an AI-generated response to another person

There has even been an advert extoling the behaviour (I forget which phone OS brand it was for, which is a shame as I might like to try avoid buying anything from them): "if a friend asks a question and your phone can answer directly, shouldn't it?", to which my gut reaction is "absolutely fucking not, it shouldn't". My more considered reaction is pretty much exactly the same.

For a start I'd consider it rude and if I found out it was happening reduce my fucking my communication with that person, or call instead if messaging (unless it seems they have a fake them taking calls, no doubt that'll be a thing sooner rather than later). And quite frankly I will reply when I'm good and ready, thankyouverymuch, instant automated reactions plays far too much into the "always available" thing that I don't care for even without AI.

I look forward to future news stories where people with such features turned on suffer PII exfoliation or other hacks when someone finds a prompt injection hole, or just uses the feature as designed because the user has been too open with what they have given it access to and who it should respond to (perhaps by inaction, because the defaults will likely be wide open for "convenience" until there is uproar).