[Cryptography] Hope Apple Fights This!

Jerry Leichter leichter at lrw.com
Thu Feb 25 09:58:07 EST 2016


> from https://www.washingtonpost.com/news/the-switch/wp/2016/02/24/apple-is-working-to-make-iphones-even-tougher-to-hack-into/ <https://www.washingtonpost.com/news/the-switch/wp/2016/02/24/apple-is-working-to-make-iphones-even-tougher-to-hack-into/>
> 
> "Fearing that the government may be able to order it to bypass security features in newer-model phones, Apple engineers have quietly begun working on fixes to flaws in their phones that allow the firm to update software without knowing a user’s password..."
Let's step back and look at the general picture here.

Defenses - cryptographic an otherwise - are built in response to a threat model. (Well, the competent ones anyway.)  Up until a couple of years ago, "the USA government is listening in" was not part of the threat model of American corporations or individuals.  It really wasn't really in the threat model of most corporations and individuals in the world.  (Sure, those involved in illegal transactions, or the governments of those on the outs with the US - and to a limited degree even US allies - worried about this; but the former do not represent a large commercial market, and the latter don't rely on commercial software.)

Then came the Snowden revelations - though in hindsight there were plenty of hints earlier.  The Google security guys, to give one example in which some of their comments were reported, were furious to discover that their own government was tapping their lines.  The threat model expanded - and increasing all sorts of internal connections moved to encryption.

We distinguish between passive and active attacks.  Most of the NSA attacks on network infrastructure were "passive" in that they just listened - though of course they were very "active" in putting themselves in a position to "just listen".  But some of the attacks were "active" in sending packets to redirect connections to themselves - and the threat model had to be expanded again.  The response has been growing efforts to encrypt *everything*.

Apple (and presumably Android makers, though it's not entirely clear who the Feds go to) soon faced another "attack" from their own government:  Growing demands to gain access to the contents of locked iPhones.  This is an even more "active" attack, requiring Apple to actually do some work.  Another thread model expansion - and another response:  Encrypt the phones with a key inaccessible even to Apple.

So the government has expanded the threat model yet again - demanding that Apple write code to disable various protections.  The legal side of this will get played out, but there are already reports out of Apples response:  Make it impossible to update the OS without unlocking the phone.  You could argue that they should have done this to begin with - and frankly I'm a bit surprised they didn't.  It does add complexity to the process of recovering a phone that for software or hardware reasons has become non-responsive, but that's life.

So far, Apple (and others) have decided *not* to respond technically to demands for copies of information uploaded to their servers.  Apple could certainly encrypt the uploads.  It would have to do so using a key unavailable to them, which makes sharing across devices more complicated:  There would have to be a common key across all the devices sharing the software *distinct from* the key used to log in to the Apple account, since that's revealed to Apple every time you log in.  But it could certainly be done.  (Most on-line backup services already provide the option to use a private - never revealed to the service - key.)

It'll be interesting to see whether Apple and others offering cloud services move in this direction.  It would well and truly piss off the LE people who've gotten used to getting this information easily.
                                                        -- Jerry


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.metzdowd.com/pipermail/cryptography/attachments/20160225/c3e0c324/attachment.html>


More information about the cryptography mailing list