"This feature is dangerous because you cannot trust Apple to only use it for what they say they will use it, hence you cannot trust Apple since they doing something which can be abused"

Using your analogy that would be:

"A knife is dangerous because you cannot trust a criminal wielding a knife to not stab you, hence your friend by definition is a criminal since it wields a knife (she claims she's cutting bread but by definition she's no longer your friend because she broke your trust by wielding a knife which, slippery slope, could be used to stab you if she ever becomes a criminal, but we all know she will because ... the knife...)

all I'm saying is that this whole communication debacle has nothing to do with proving or disproving Apple's trustworthiness: they may be nefarious or not, independently of this feature. This feature doesn't make anything possible that wasn't possible before, and breaks the trust with the users only insofar users misunderstand what it is all about, and apparently that's what's happening.