Also a reminder that apple really does seem to care about privacy, in ways the average user isn’t likely to understand / appreciate.
We normally do this by riding the credit card rails....which requires the full address...
This is privacy in fact prettty strong one, like with any security measure usability and availability may suffer however it’s on you to mitigate those risks by leveraging backups which on OSX is fairly easy to do so with time machine, FileVault and cloud backups which are also encrypted.
If Apple can not offer a recovery service it is also likely that most adversaries will not be able to recover the data which is a pretty darn good thing to have.
You can look at something like the headphone jack and say, "adding a jack doesn't prevent anyone from using wireless headphones." But if they get rid of the chip, then it does prevent people from using it.
Short of selling two separate models with different configs, is there any way they could allow the customer to choose?
Apple have taken the decision here that abandoning the ability to recover data from failed machines is worth the guaranteed security floor of it not being possible to disable strong encryption, and honestly I find it hard to disagree.
Obviously if you're worried about nation state attackers you wouldn't want to do that in a million years. But if your primary threat model is that you're an average boring home user suffering data loss because you don't backup and dropped coffee on your Macbook it would be a solution.
From apple’s perspective, hardrive encryption keys aren’t something the average user is going to care about, but files are. The fact that this drives iCloud storage is probably a happy accident ;)
Why? This is something completely up to the manufacturer.
As for forcing HW encryption of hard drives, other companies already tried (WD My Book Duo) and it didn't land well with users.
Until their Macbook dies and they take it to Apple for servicing. At which point they'll care a lot.
I'd hate to be an Apple Genius. They don't half get thrown under the bus by corporate decisions sometimes.
The iPhone outsells the MacBook by more than 10 to 1 and works exactly the same way. I think they'll be OK.
With Bitlocker unless you backup the recovery key in the cloud or manually there is no way to recover it unless you are going to attempt to compromise the TPM this is no different than FileVault/TimeMachine backups which do exactly that however with the added assurance that at least at the writing of this comment the T2 chip cannot be easily compromised or accesses outside of normal operating parameters.
Also for a security prospective it's not good, the main principle of cryptography is that the only secret to decrypt something should reside in the user, not in the obscurity of the algorithm or the hardware, so a perfect encryption is an encryption where when you turn on you laptop you have to insert a 30 character strong password that only you know, like LUKS on Linux, no obscure hardware, no proprietary software. The laptop broke, take out the SSD, put it in another Linux computer, insert you secret password, and recover the data.
This system is NOT secure, if someone finds a flaw that permits to extract the key from the T2 chip, and it's only a matter of time and resources before someone does it, your data is no more protected, they can decrypt and extract what they want. We must not consider secure every encryption system that doesn't depend on a secret that only the user knows.
Let me tell you a secret, there is nothing truly secure in the world. Only things which haven’t had a vulnerability found for it.
> It's not different from the Intel ME
The important sense that matters is Intel ME has been found vulnerable (highly vulnerable in fact). The T2 hasn’t. Should we be skeptical of black boxes, most assuredly! But calling it equivalent to something found to be flawed in serious ways is a bit premature at best, or grossly negligent at worse.
> the main principle of cryptography is that the only secret to decrypt something should reside in the user
I don’t know about you, but the “main principle of cryptography” is to prevent unauthorized parties from knowing the contents of the data to me. I guess we are allowed to make up any statement that furthers our cause though when that suits us.
> if someone finds a flaw that permits to extract the key from the T2 chip, and it's only a matter of time and resources before someone does it, your data is no more protected, they can decrypt and extract what they want
Your password is part of the process, the extra stuff done in the T2 is just that, extra. So extracting from the T2 gives you nothing without the user’s password, needed to unlock the secret encryption key for the drive (assuming FileVault is enabled[0], and if you didn’t enable it, then why do you have an expectation of security at all?).
My VCR is pretty secure. CRT monitor, too. Even my headphones are nicely secure.
>The important sense that matters is Intel ME has been found vulnerable (highly vulnerable in fact). The T2 hasn’t.
There's your fallacy. Just because the T2 hasn't doesn't mean it can't or won't. You even just said it. 'Only things which haven’t had a vulnerability found for it.'. The problem is that proprietary embedded hardware of a dynamic nature leads to irreversible possibilities of exploitation at a hardware level. ME is a very, very good reason to not trust.
>I don’t know about you, but the “main principle of cryptography” is to prevent unauthorized parties from knowing the contents of the data to me.
Cryptography is not about obfuscation layers on top of cryptography. Cryptographic security is a math, not a logic. Anything further is about hiding intentions making the cost of entry to finding exploitation higher. The cost of entry makes it hard for white hats to find exploits before the market does.
>Your password is part of the process, the extra stuff done in the T2 is just that, extra.
Or is it? Have you looked at the firmware? you nor I can, so you cannot claim what it's actually doing. But we do know that it's a proprietary chip capable of lots of fun.
Also user passwords are trivial to get, and there's always the possibility that the chip is storing the user password itself. Again, can't know, that's up for the black hats now.
I don't think your VCR, CRT monitor, or headphones are nearly as secure as you think they are. You might not care about their vulnerability to a sufficiently-motivated actor, but someone else might.
”In the paper, Van Eck reports that in February 1985 a successful test of this concept was carried out with the cooperation of the BBC. Using a van filled with electronic equipment and equipped with a VHF antenna array, they were able to eavesdrop from a "large distance". There is no evidence that the BBC's TV detector vans actually used this technology, although the BBC will not reveal whether or not they are a hoax.
Van Eck phreaking and protecting a CRT display from it was demonstrated on an episode of Tech TV's The Screen Savers on December 18, 2003.”
What happens when a sufficient power surge comes in and it catches fire and burns down your house. Still secure?
> CRT monitor, too.
You sure? There are known viruses than can destroy some old CRTs. Still secure?
> Even my headphones are nicely secure.
Sure, so is my wind up grandfather clock too. How is that relevant?
If it ran open-source software that wouldn't make it secure. Proprietary software means it can't be independently audited for security, but whether it can be audited != whether it's actually secure.
> Also for a security prospective it's not good, the main principle of cryptography is that the only secret to decrypt something should reside in the user, not in the obscurity of the algorithm or the hardware
Your "main principle" is incorrect, but you are correct that obscurity of algorithm is no defense. However the T2 does not rely on obscurity of algorithm. It relies on having a private key that nobody besides the T2 has access to (not the user, not Apple, nobody) and it combines that key with the one derived from your password. This means decrypting any of that data requires two things:
1. The user password, and
2. Cooperation from the T2 chip
This is strictly more secure than just relying on the password alone.
> This system is NOT secure, if someone finds a flaw that permits to extract the key from the T2 chip, and it's only a matter of time and resources before someone does it, your data is no more protected, they can decrypt and extract what they want.
This is incorrect. You still need the user password. If someone manages to extract the key from the T2 then they've reduced the problem back to what you'd have without the T2; namely, you need the user password to decrypt anything.
I’m not a security expert and I couldn’t possibly tell you whether the T2 is secure even if I was because it’s an opaque blob like you said.