I'm not apologizing for them, though the bit in the article talking about uploading your VOICE to Apple servers... how is that shocking. Google did that since DAY ONE with their Android phones. For a while now, places have been handling the voice->text translation on a server and just sending you back the text.
The bit with the names/contacts... it gets a little fuzzy. On one hand, it makes SOME sense. Voice recognition is bad enough with first names, and with last names it's a cluster ####. My last name is the (phonetic) combination of 3 english words so voice->text often records it as WORD WORD WORD instead of LastName. Train it enough, a FIRST LASTNAME said together will know to write it that way instead FIRST WORD WORD WORD.
That being said... the obvious privacy concerns are there. It's possible they'll have the world's largest "who-knows-who" tree available. And who knows what that means.
"Space. It seems to go on and on forever. But then you get to the end and a gorilla starts throwing barrels at you." -Fry, Futurama