Here are three critical quotes from an article about DeepFace, the Facebook automatic face recognition system, followed by my explanation of what they really mean

(text in parentheses is my own synthesis or comment):

1. “Appear in a photo taken at a protest march, a gay bar, or an abortion clinic, and your friends might recognize you. But a machine probably won’t - at least for now.”
2. “Once DeepFace identifies your face in one of the 400 million new photos that (you, or anybody who took a picture with you in it) upload every day, “you will get an alert from Facebook telling you that you appear in the picture… You can then choose to blur out your face from the picture to protect your privacy.” Many people, however, are troubled by the prospect of being identified at all - especially in strangers' photographs. Facebook is already using the system, although its face-tagging system only reveals to you the identities of your “friends.”
3. “The intention is not to invade the privacy of Facebook users… but rather to protect it”

What could Facebook really do with DeepFace?

Here are the same quotes above, translated:

• the mechanism above will only “protect your privacy” at the level, and in the way, that is by far the less important: it will not let other Facebook products, er, I mean “users” like you, the possibility to discover that you were indeed in that march, bar, whatever
• whenever Facebook recognizes you in a photograph, it will kindly ask YOU to confirm a) that it’s really you and b) that being associated to that event, place or people would indeed cause trouble to you, or at least embarrassment
• and then, Facebook will promptly store even those confirmations in its database. At immediate disposal, in practice, of every government organization and/or any of the commercial partners, from advertisers to banks, insurance companies or potential employers, with which it may ever enter a partnership
• since bits can be copied forever, even if Facebook closed shop tomorrow, every person or organization who had (or could have in the future) ONE opportunity to copy DeepFace data may be able to figure out, ten or twenty years from now, that yesterday you went to a protest march to support some cause, etc.. even if you changed your mind this morning and will never, ever do it again
• add street cameras to the picture, and Orwell’s Big Brother starts looking like Winnie the Pooh

And the conclusions are…

1. What other human beings (may) discover about you when they personally visit facebook.com is, and will always remain, the very, very, very last of your problems. It’s irrelevant, really. It’s like worrying that a meteor may hit you, while you are standing at night in the middle of a crowded highway
2. what matters is only that Facebook will always know, and in practice be able to share it as it pleases with governments or commercial partners, everything that you put in it
3. (above all) including everything you “report” about other people, while what to tell Facebook should be THEIR decision only
4. So the least you use Facebook, and the least you put there, no matter how “set to private” it is, the better it is for you and your fellow Facebook products, er, I mean “friends”

You can read the whole article from which I took those quotes at sciencemag.org.