Hacker News Reader

Comment by glenstein | Hacker News Reader

Could someone ELI5 how this is even conceptually possible? I suppose that is exactly what the ACLU was trying to find out, but I thought something like AES-256 should be impossible in principle to break.

So what type of thing is most likely happening? Are hardware vendors secretly installing back doors? is there some sort of software encryption standard that's compromised? Or some kind of secret breakthrough in computation theory that easily "solves" encryption? Or maybe, bolstered by NSA surveillance, they just know passwords and unlock codes to devices?

Again, I understand that that is the exact mystery here, but I would appreciate knowing if one or more of the possibilities is conceptually more likely than the others.

View on HN

Usually the weakest link in any crypto system is the code surounding it. AES-256 is very solid, but using it correctly can be hard. Apple knows what they are doing but they are still human.

For example if this is a lock screen bypass (and not a bypass of say the start up process before the key is entered), maybe some memory corruption issue allows you to bypass the password check. Or another example, many people use short numeric pins to encrypt their phones. Apple then does a lot of complex stuff to take that and turn it into a high entropy key. Maybe this is an exploit in that part, allowing for a bruteforce attack.

These are all just wild gueses, i have no idea what these capabilities are or for that matter, i dont really know very much about how iphone FDE works.

You can’t turn a low entropy key to a high entropy one by deterministic operations. It’s theoretically impossible.

Yes, what their doing is basically using a randomly generated key per device combined with the user generated key. So, it becomes a question of extracting the hidden key from the device or using the phone to brute force the decryption.

However, without that key an attacker is stuck trying to brute force Apples high entropy key, which causes some confusion.

The algorithms are secure.

* The hdd in a phone that is not powered off is mounted. It takes an exploit in OS to pass the Lock Screen or read the RAM.

* The vast majority of people don’t have long alphanumerical passwords. The short pins can be broken.

* Some manufacturing companies don’t handle encryption keys properly. Synology NAS stores the keys next to the encrypted data for automatic access upon boot (unless you change the default and store the keys on an encrypted USB or enter keys manually upon boot).

* There are sometimes bugs in algorithm implementations, especially if a company roles its own crypto with closed source software.

* Sometimes encryption keys are not properly generated. Bad RNGs were a thing for a long time.

* And of course there could be back doors for law enforcement.

The RNG is the backdoor always. And as said above, how you criple the entropy.

Except for obvious pins, they cannot be brute forced due to the throttling built-into the hw vault.

If you are referring to upper limits on password trials enforced by chips such as TPM, T2 or those in Yubikeys, these are software and hardware — but not fundamental algorithmic — limits.

I wouldn't say an HSM such as a Yubikey is secure against a government agency. But apart from that, you can take out the hard drive and deal directly with random blubs, bypassing software limits.

Of course these measures are effective against the majority of the attackers, but we are talking about FBI here.

> people don’t have long alphanumerical passwords

Do you know how long is enough? Of course 64 chars would be dope but I'm not good at memorizing things. I know it's trainable but I'm lazy too :)

My current password is a bit longer than 10 chars and I wonder if I should upgrade it.

Use an easily memorable personalized passphrase. One that's easy for you to remember and impossible for anyone else. Repeat it in your mind like a mantra for days and it will stay with you decades later.

Personally I started doing this ~2000. I still remember my first one that I gave up meanwhile because I told it to my GF at the time and we broke up meanwhile.

You can easily top 100 characters this way. Though this is not something you want to do on a touchscreen device, it's a pain there with fat fingers.

Unfortunately, Android has a maximum password length of 16 characters.

Yeah I already have a passphrase for my desktop, but for mobile I'm on GrapheneOS and it asks my password frequently enough that I don't want something too painful to enter.

90 bits, clean source, is bare minimum. Better to push up to 128 bits.

A 10-word diceware passphrase is often recommended.

If you choose a password rather than a passphrase, the bare minimum is 15 (90 bits) to 22 (128 bits) characters, upper and lower case and numbers.

Either way, purely random is key here.

Thanks.

They are likely using GreyShift or Cellbrites UFED or similar products. Most of them leverage exploits in the phone that are usable on the lock screen to either enable you to bypass the lock screen or to allow you to have unlimited password guesses then brute force the password.

Thats my guess at least.

Im pretty sure its this now.

https://twitter.com/matthew_d_green/status/13417461712205373...

According to this[1] Twitter thread, it's a matter of convenience outweighing security. Phones keep the encryption keys used for most apps in memory so the apps can do background work while the phone is locked.

[1] https://twitter.com/matthew_d_green/status/13417461712205373...

Most PINs or passwords are so short that they would be easy to brute-force but phones have anti-brute-force mechanisms. So maybe there's a tool that does that.

It is known in some cases and highly probable in others that randomness is compromised systemically in many products and therefore services.

Reply on news.ycombinator.com