As always, if someone has physical access and unlimited time, no device or computer is safe.
Also, Mailbox.app only supports GMail. Security minded people are obviously not the target market.
(edited to make my point more clear :)
Whether users pick appropriate passwords is another matter entirely.
When the device is locked the file is encrypted and cannot be easily retrieved with a USB cable and a file explorer. An app that does not properly secure its files is readable even when the device is locked.
>if someone has physical access and unlimited time There is no such thing as unlimited resources.
If I had unlimited time I could crack every encrypted message on the planet.
Using DPAPI turns a 30 second hack into an online cracking job. The crypto processor in the iPhone can only check one password every ~80ms and you need the chip with you. An attacker cannot do an offline attack.
I wasn’t suggesting it shouldn’t be. My point is that the article’s headline is overly dramatic: Mailbox.app is not a complete security failure because of one hack that requires physical access. Given that Mailbox only supports GMail, I’d be more worried to put my email in Google’s hands than worrying over someone grabbing my phone out of mine.
“Mailbox is BIG. We are not talking of an average app here!”
Mailbox.app is a free app that has been downloaded a couple of million times, I wouldn’t call it “BIG” yet. It’s very new, it’s still on version 1, so it’s not expected to be perfect.
The problem is that we take mobile devices with us every place we go. So physical access is not difficult to obtain.
This really is a big deal primarily because the developers of Mailbox.app did not take steps to even obfuscate the stored data...which would deter all but the most determined of attackers.
All this stuff can be kludged onto email, but the attitude should be "unless I've taken measures to add security this thing is not secure".
A basic bit of security, especially one that doesn't put any more load on the user (to have to maintain or set up) is a pretty big no-brainer. Raising the bar for a successful hack is also worth doing when the cost is a single line of code and no effort on the user's part.
What's on offer here? 10 minutes extra tamper resistance? For a protocol which is inherently insecure?
Sure, if you're the CEO of some big company and a skilled attacker really wants at your email, this is only a stopgap - but this is also sufficient to stop less proficient attackers entirely. For most people this is all they need.
> "it's more like telling bicycle riders to use anti-puncture tape."
If anti-puncture tape has literally no downsides whatsoever to the bicycle rider's experience, and costs nothing, then yes. Why wouldn't you have it?
Given physical access and "unlimited" time (i.e. no more than a million human lifetimes, say), then certainly an attacker can gain access to the device and make it do what he wants.
However, if the data on the device is securely encrypted, then physical access and (reasonable) time doesn't matter. He won't be able to get at the data.