Comment Re:Is this technically impossible - no. (Score 5, Informative) 191
Very likely, if I can read my mail, so can he. It's only logical.
The fact that an organization acts as a conduit for delivering messages does not necessitate that they have the ability to read the contents of those messages. The one does not follow from the other. It may be likely that the two go hand-in-hand, but by no means is it logical that they would do so.
The various white papers and other security documents Apple has released over the last year or two make it clear that they claim they do not hold the private keys necessary to decrypt their users' data. Those private keys reside on the devices of the users, with unique keys being generated for each device and unique copies of the data being maintained separately for each device. For instance, in the case of iMessages, here's how Apple claims they work:
1) I type up an iMessage to send to another Apple user and press Send.
2) My device queries Apple's servers for the public key(s) of the recipient, which could be numerous if they've configured iMessages to arrive on multiple devices.
3) My device creates and encrypts one copy of the message for each device, using the public key that is specific to each device for the copy going to it.
4) My device signs the copies using its private key.
5) The iMessage is sent to Apple, who then forwards it and immediately deletes it, unless they can't deliver it, in which case it'll stay queued for up to 7 days.
6) The recipient's device verifies the signature against my public key and then decrypts the message using its own private key.
Assuming the system works as described, Apple shouldn't have access to the content of the messages. Whether or not you believe that it works as described is a matter of how much faith you put in corporations and/or the governments that might be compelling them to insert backdoors. For instance, there are trivial ways that they can circumvent their own systems to gain access to messages, without having to compromise the private keys at all. The easiest way I can imagine would be to simply provide the public key of a wiretapping device in addition to the other keys in step #2 above. Unless you're sniffing your own traffic to ensure that you're sending EXACTLY what you're expecting to send, you'd never notice that you've sent out an extra copy of the message, and would be entirely unaware that it had landed on a government agent's device as well.
But again, it isn't logical that they would have that sort of access. "Likely", given the state of things? Sure. But logical? By no means. Again, the one does not follow from the other. Particularly so in the case of Apple, since their money comes from hardware sales, not from monetizing the user's information, so it's in their best interests to make those devices as secure to use as possible.