FBI Refuses to Say Whether It Bought iPhone Unlocking Tech 'GrayKey'
motherboard.vice.com
motherboard.vice.com
I mean, $30,000 for the "unlimited" one-year license? Seriously, why wouldn't they purchase it. If local and state police agencies have bought this device, I think we can all safely assumed the FBI has as well.
Hell, I wouldn't be surprised to hear that the FBI recommended the purchase of this device to local and/or state agencies.
Like they need any recommendations, they have cases open :). Not all cases are a middle aged guy, minding his own business that gets his phone searched at JFK. A lot of crimes are thefts, drugs, sex and evidence is in the phone, they have search warrants but can't access it. Or couldn't. They are law enforcement conventions and business that cater to them, so they already know. Not to mention Twitter, FB forums and so on.
I'm with Apple, trying to make this unbreakable, but if cops figure it out, and they have warrants, kudos to them. Maybe Apple already does, but they should be buying all zero-days out there. If they don't sell for $1 million, offer them $5 million. Or $15 million and still no one at Apple would have to skip a meal to cut costs https://finance.yahoo.com/quote/AAPL/key-statistics?p=AAPL
Well, there's that part.
Apparently lots of my tax dollars are going towards purchasing a temporary license for a temporary solution to a permanent problem that isn't really a problem at all?
Though I must admit, one day those things are going to look great in some museum. They tell a really interesting story. Hopefully the price tag is included on the exhibit description.
It's every single department, in every single city, in every single state.
That's thousands, if not tens of thousands, of departments buying these when they'll all be useless as soon as Apple patches the exploit.
Seems like a whole lot of wasted money. But who knows, they might just catch enough "big bad Marijuana dealers" to offset the cost.
That's what the LEO really means by cold cases. Nothing important. Just more drug busts.
People dying because of drug overdoses seems like a pretty big deal to me, but then again (thank God) I've never had a family member or close friend succumb to a drug addiction.
If you don't believe that a solution works, then further effort on a non-functioning solution isn't a big deal, even if you feel the problem is a problem.
Depending on where you live, maybe 16 cents on the dollar is going to “defense.” And I’d imagine that $15k per license is a tiny fraction of that.
https://www.cbpp.org/research/federal-budget/policy-basics-w...
The FBI wants backdoors, doesn't get backdoors - as far as we know at least - and uses illegal devices to bruteforce mobile phones. There's no grey area here, either you respect people's privacy or not. Circumventing the essence of law one way or another sends a strong signal and it is the wrong kind of signal IMHO.
I also don't trust these agencies will be held accountable when they abuse these devices.
Sally Suburb has a warrant for unpaid tickets? Better check her phone while we have her!
I recall a certain incident, which I'm sure wasn't isolated. I believe it was in California somewhere, where police essentially saw obtaining and trading nudes they would obtain from peoples' phones as a game.
Using the threat of force to obtain nude images of women you've pulled over sure has a rapey feel, to me. Going on and forwarding those images to your friends sounds an awful lot like revenge porn as well. Apparently the police in that area don't agree, though.
I don't believe any of those officers received any actual punishment beyond the standard "take a paid vacation while we let this blow over in the media".
I know, I know, it was just a few of those mythical and rare "bad apples".. but until the good apples who supposedly outnumber the bad start cleaning house, and stop standing up for those "bad apples" to protect the image of their "brotherhood", trust will continue to be an issue.
This is the problem with the "bad apples" phrase. It seems to me people forget the line is "Bad apples spoil the bunch" it's categorically NOT "A few bad apples sure are annoying as heck" as if merely acknowledging that bad apples exist is enough to signal you're doing something about the problem.
Police Departments across the country seem to have a hard time getting this, and as do many of their supporters to the point of "it's a few bad apples" becoming an utterly worthless phrase anymore, IMO.
This seems to be a software exploit if it requires not opening up the iPhone. There are more sophisticated hardware techniques (one was "decapping" the chip and reading the data out so you can try passcodes elsewhere), but I believe Apple's also finding mitigations for those as well.
Yes but it certainly is exploiting a major hardware(?) flaw in the Secure Enclave Processor and/or the secure boot chain. The SEP is supposed to hold the decryption key, seeded at manufacturing time and unreadable at a hardware level, and block brute forcing with a chip-enforced time delay. And the whole system including the SEP is supposed to have a secure boot chain so only Apple can authorize new firmware.[1]
The FBI could break into the 5c because the brute forcing was only a weaker software limitation at that point. But since the A7 the SEP was supposed to handle it.
[1] https://www.apple.com/business/docs/iOS_Security_Guide.pdf
From the look of this device, you just plug in the lightning cable and never open up the phone.
I guess I can't speak to how the SEP implements its security features, i.e. which pieces are physically burned into hardware or just implemented in firmware with a secure boot chain. But either way this exploit points to a major flaw in the SEP, bigger than your average OS vuln.
I've read a few theories of how that might work, from timing attacks to carefully-sequenced reboots of the phone to try to confuse the lockout mechanism.
My money is on a set of vulns similar to what CTS-Labs found in AMD’s secure coprocessors last month.[1][2]
[1] https://thehackernews.com/2018/03/amd-processor-vulnerabilit...
[2] https://blog.trailofbits.com/2018/03/15/amd-flaws-technical-...
The fact that the company producing the GreyKey has an executive who is a security researcher and used to work for Apple is such an obvious, damning piece of evidence. I don't understand why this connection hasn't been more widely reported and investigated.
While the other two are suspicious, I don't think the last one is. If you're good enough to come up with a clean room implementation than I don't see any foul play here.
On paper sure there's merit to it, in practice it's probably an exercise in futility while the exploit is un-patched and GreyKey is useful to law enforcement.
> This section does not prohibit any lawfully authorized investigative, protective, information security, or intelligence activity of an officer, agent, or employee of the United States, a State, or a political subdivision of a State, or a person acting pursuant to a contract with the United States, a State, or a political subdivision of a State.
So it would probably go nowhere in court
Each device has a unique 256-bit AES key called the "UID", and a programmable "device group ID" called the "GID".
The UID is "fused" and the GID "compiled" into the Application Processor and Secure Enclave during manufacturing, but no software or firmware can access them. The firmware can only see results of encryption and decryption, and the keys are accessible only to the AES engine's silicon. They are not available via JTAG or other debugging interfaces.
On some later chips the Secure Enclave generates the UID itself.
Apart from the UID and GID, the Secure Enclave can also generate new keys using a RNG. See also: Krypton[1].
(see page 12)
Passcodes are "entangled" with the device's UID, so brute-force attempts must be done using the Secure Enclave (or with an electron microscope?).
Each attempt has an iteration count calibrated for 80ms, which would mean an average of ~11 hours to brute force a 6-digit pin[2].
iOS also has longer delays for multiple attempts; 1 minute after 5 attempts, 5 minutes after 6, 15 minutes from 7-8, and 1 hour for each attempt after 9. The paper later mentions that devices with the Secure Enclave will enforce the longer delays, including after reboots, but this doesn't seem to to be the case for GrayKey.
(see page 15)
GrayKey claims to crack an iPhone (with 4-digit pincode?) in around ~2 hours, but more than 3 days for 6-digit pincodes. Which might work out to ~1s per guess?[3].
If you use a alphanumeric passcode, or a custom numeric code, you likely don't have to worry about these unlockers.
A random 10-digit pin will take an average of 12 years 6 months to crack[4].
[0] https://www.apple.com/business/docs/iOS_Security_Guide.pdf
[1] https://krypt.co
[2] 6-digit pin, 80ms/guess: 1e6 * 80 / 1000 / 60 / 60 / 2 = 11h 7m
[3] 4-digit pin, 1s/guess: 1e4 * 1000 / 1000 / 60 / 60 / 2 = 1h 23m
[3] 6-digit pin, 1s/guess: 1e6 * 1000 / 1000 / 60 / 60 / 2 = 5d 18h 53m
[4] 10-digit pin, 80ms/guess: 1e10 * 80 / 1000 / 60 / 60 / 24 / 365 / 2 = 12Y 8M 6d
Does HN censor potential security disclosures?
All I said was it was probably using techniques like voltage and timing analysis for instance as described here:
https://www.coursera.org/learn/hardware-security/lecture/2Ug...
Here's another article from the same source about the FBI using 'classified' tools: https://motherboard.vice.com/en_us/article/7xdxg9/fbi-hackin...
Apple, announces $10 million bounty to reveal exploit. I guess within hours they'll have it, probably from GrayKey engineers (might be hard to claim given NDAs).
But if you have a strong enough password it won't work against you. The exploit only lets them try to guess your password more quickly. It doesn't let them bypass your password.
iOS11 seems to have almost purposeful security weakenesses. I'm willing to give Apple the benefit of the doubt here, but only if they fix whatever flaws these guys and Cellebrite are using to break into iOS11 iPhones.
Both those decryption devices seem to rely on iOS11 so it must a new change, which means it shouldn't be too hard for Apple to figure out which one of its recent changes caused this weakness in security.
I'm fine with the FBI going about the letter of the law with a warrant to obtain evidence.
I'm also fine with Apple hardening their devices against such attacks, eventually the "GrayKey" technology will become commodity and we'll need to protect our information from casual thieves.
However, I take a strong stance against mass surveillance and obtaining (any/all, including metadata) information without due process. This includes getting pulled over and having police search my phone, or as a US citizen, re-entering my country and having my phone searched.
Can you expand on this?
> Both those decryption devices seem to rely on iOS11
Why do you say this? The article says
> based on GrayKey marketing material, that the tool can unlock even the iPhone X, Apple’s most recent phone, as well as devices running iOS 11, the latest Apple mobile operating system.