Hacker News new | past | comments | ask | show | jobs | submit login
NYC law enforcement has been able to crack iPhones in-house since Jan 2018 (9to5mac.com)
32 points by traderjane 13 days ago | hide | past | web | favorite | 9 comments

My inner skeptic says if you have enough money and demand for phone decryption that is using a low-entropy key; your best bet is to take a dump of the phone NVRAM and feed it to inhouse iphone emulator, brute-forcing the key in parallel, without any lockouts. Their product may be an ios emulator.

The Secure Enclave should protect against that type of attack. And why would you assume it is a low entropy key? Also, my iPhone is autocorrecting Secure Enclave to capitalize it for some reason.

You should probably read this before believing your “inner skeptic” https://www.apple.com/business/docs/site/iOS_Security_Guide....

There's an individualized (unique per device) embedded cryptographic key in each Apple processor, emulator solves nothing on that front.

A quick web search yields no information about how Cellebrite might be able to achieve this. Does anybody have any inside knowledge? Has anybody played with one of these devices?

Keep in mind that any inside knowledge leaked to the public is also knowledge available to Apple, which can use it to either find and patch the bugs Cellebrite is exploiting, or at least build defense mechanisms that mitigate the effect of exploits. Even short of detailed technical information, simple things like “how long does it take?” or “how many times does the device reboot?” or “what does the USB cable look like?” can still help narrow down which software components (or even hardware components) to look at. So Cellebrite is well-motivated to avoid any leaks.

That said, it is most likely “just” a computer that talks to the phone over USB and uses a chain of software exploits. It has to take over first the application processor, then the Secure Enclave, in order to bypass the passcode entry rate limit. The rate limit is key. iOS encrypts user data based on the passcode, so without guessing the passcode there’s no way to get at the data short of breaking AES. But you can guess. Passcodes default to 6 digits, creating only 1 million possibilities, low enough to bruteforce. Even if the bruteforcing has to be done on-device (because the passcode is tangled with device-specific keys managed in hardware), and even if repetitive crypto operations are added to make the key derivation take longer (not sure), key derivation can’t take too long or it would negatively impact the user experience, and 1 million is just really low. The only way to make a 6-digit passcode secure is to make the nth access attempt take (exponentially) longer than the first, a restriction that can’t be done with pure crypto but requires some trusted software to enforce. Which can be hacked.

That said, if you’re planning to do something sketchy, you can set a long non-numeric passcode and you’ll probably be immune to whatever Cellebrite is doing. (Unless they’re recovering the passcode itself from… somewhere. It’s not supposed to be stored, but bugs are possible.)

Source: iOS Security Guide

if you are ridin' dirty just stay off the phones.

US police are not going to use a literal wrench. They have tried to jail people until they reveal the passcode, but even that has been subject to constitutional challenges. Much simpler to be able to get access without permission. Also, in some cases the owner of the device has not been apprehended; perhaps they don’t even know the police have their phone, and the police would like to keep it that way. Or the owner could be dead, as in the San Bernardino case.

The Chicago police have been known to have black sites, so I wouldn't put a metaphorical wrench past the realm of possibility.

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact