Slashdot Mirror
CCC Says Apple iPhone 5S TouchID Broken
hypnosec writes with word that the Chaos Computer Club claims to have "managed to break Apple's TouchID using everyday material and methods available on the web. Explaining their method on their website, the CCC hackers have claimed that all they did was photograph a fingerprint from a glass surface, ramped up the resolution of the photographed fingerprint, inverted and printed it using thick toner settings, smeared pink latex milk or white woodglue onto the pattern, lifted the latex sheet, moistened it a little and then placed it on the iPhone 5S's fingerprint sensor to unlock the phone." Update: 09/22 21:32 GMT by T :Reader mask.of.sanity adds a link to a video of the hack.
Remember that a hacker won't know which of 5 fingers the owner uses, so that's another layer of security
fingerprint identification is fundamentally and irredeemably broken. no other authentication method leaves copies of itself all over the place.
everything else is an arms race between verifying it is a finger and pretending to be a finger.
Snowden and Manning are heroes.
I was with you until you said "sheeple".
Instead of using a fingerprint, use a Nipple print!
You know what? I really love the sound of your voice. ... And there's this one word. I've always loved the sound of this word. ... I would really like to hear you say the word ..."passport".
The cops will have copies of all 10 fingers, and will be able to add this technique to their fourth and fifth amendment circumvention strategies.
Has anyone else verified that the suppose hack really does work? Isn't a bit premature to claim Apple is lying off a single youtube video?
No. It wouldn't matter. No matter what they did there would always be the next thing they could have just done. How do we know that the phone wasn't programmed to unlock with the second guys fingerprint? How do we know they didn't edit the video? etc, ad infinitum. What makes it highly believable is none of that. It is the reputation of the Chaos Computer Club that makes it believable. They aren't about to sacrifice a reputation it took them more than 30 years to build, especially for essentially no gain. If it was an unknown group I'd say maybe they are looking for 15 minutes of fame. But this is the CCC we are talking about here.
Guns don't kill people; Physics kills people! - John Lithgow as Dick Solomon on Third Rock From The Sun
Fingerprints are good because they replace ZERO security. Most people don't PIN lock their phones. Finger Print lock is too convenient not to use.
It is meant as a deterrent to common thieves, and works well as such. A robber isn't going to grab your phone, ask for a nice clear print, and then run home to his laser printer and latex (and you could remote wipe the device in the mean time anyway).
If its the government you're worried about...well, if they have physical access to your device they probably have you in custody and can compel you to unlock it anyway, or just use existing forensic tools and warrants to get what they want. Even then we're talking about the unlikely scenario of you being arrested and having anything more interesting on your phone than funny cat pictures.
I'm trying to imagine a "real world" scenario where TouchID is less secure than a 4 digit passcode or no security at all...and I got nothing.
- "Scientia non habet inimicum nisp ignorantem"
sounds really trivial to break. I can see all kinds of kids doing this.
Known vector. Gummy-bear attack.
The core issue is that you leave copies of your authenticator EVERYWHERE. It's as if you dropped 85% accurate copies of your smartcard on every item you touched - with random 15% damage to the material - and a card reader designed for 15% error in reads.
Any such scheme is going to be subject to this kind of impersonation or gaming. This is why biometrics are always a bad ID choice. Also, the A/D conversion is low-entropy, among other problems.
There's a false assumption, that because I can uniquely identify another person with 99.999% accuracy, based on your sound, shape and appearance, that therefore this is the best way a machine should do so. It is a falsehood that is reinforced by a misleading intuitive perception. The core issue concerns the questions related to what constitutes "identity" and an "authentication factor" in systems. Neither of these correlate to actual persons or their real-world characteristics in a unique and meaningful way, that is not also subject to spoofing, injecting or revocation DoS.
"Flyin' in just a sweet place,
Never been known to fail..."
you mean, besides just holding your hand against the sensor? As, if they have your phone, they probably also have you...
How about you jailbreak the phone, and use a PIN to unlock it normally, BUT you customize the reader, so if certain of your fingers get held against the sensor --- it triggers a "disable power off function" and "start wipe device" command.
sounds really trivial to break. I can see all kinds of kids doing this.
Known vector. Gummy-bear attack.
The core issue is that you leave copies of your authenticator EVERYWHERE. It's as if you dropped 85% accurate copies of your smartcard on every item you touched - with random 15% damage to the material - and a card reader designed for 15% error in reads.
Any such scheme is going to be subject to this kind of impersonation or gaming. This is why biometrics are always a bad ID choice. Also, the A/D conversion is low-entropy, among other problems.
There's a false assumption, that because I can uniquely identify another person with 99.999% accuracy, based on your sound, shape and appearance, that therefore this is the best way a machine should do so. It is a falsehood that is reinforced by a misleading intuitive perception. The core issue concerns the questions related to what constitutes "identity" and an "authentication factor" in systems. Neither of these correlate to actual persons or their real-world characteristics in a unique and meaningful way, that is not also subject to spoofing, injecting or revocation DoS.
Let's say you get your grubby hands on an iPhone 5S and are immediately overcome by an irresistible urge to crack it open.
1) Getting the victim to pose his finger for a 2400dpi photo is not an option so you'd have to bag the device and dust it for prints since you'll probably need to make the prints more visible. I suppose you could get the hang of that in about half an hour if you are a novice with a print dusting sets you bought online.
2) Find a good thumb print. There is no guarantee that the print on the button sensor surface is any good nor is there a certainty that there is a usable print anywhere on the phone. I suppose you could monitor your victim and steal some of his drinking glasses and coffee cups but that means 'trivial' goes out the window right there.
3) For the sake of argument let's say you get 1 and 2 right and find a good print on the sensor surface or somewhere else on the phone, eliminating the need to poke around stealing coffee cups and drinking glasses. You now have still have to do what it says in the article and the photo processing, printing and latex covering that sounds like quite a bit more than 10 minutes of work, especially if you have never done it before.
That does not sound exactly trivial to me. Trivial is faking your way past Google's face recognition-login feature with a picture of the phone's owner. You could conceivably do that by borrowing his phone, snapping a picture of him with your iPad and using the image in the iPad to log into his phone... Ooops! somebody already went and did that and it looks like a 20 second operation. Going through the above procedure to defeat the fingerprint scanner takes what? A hour? The average pick-pocket would probably not bother and the time it takes to crack phones this way with no guarantee of reward would make it un-economcal for criminal bands to crack phones on a large scale (in the hope of finding account numbers or dirty pictures for a blackmailing, ... or whatever) which means that this is way better security than no passcode at all. If you are carrying data valuable enough to make it worth while to go through this exercise to retrieve it you should put a 20 character password on your iPhone or consider putting the data on an IronKey in stead. And yes I know the NSA can probably pull this off in 10 minutes or less but if you have the NSA after you:
a) They probably have more efficient ways to get into your device than stealing it and hacking it by lifting your greasy fingerprints.
b) You have bigger things to worry about than somebody reading your e-mail... like getting snatched and sent to a secret jail for a course of water-boarding, or being on the shortlist for a drone strike.
Only to idiots, are orders laws.
-- Henning von Tresckow
You mean like the android face unlock that can be defeated by a photo of the user? (at least you don't leave your photo on the glass surface of the phone when you put it down...)
Let's face it though, unless companies are willing to spend a fair amount more on these biometric sensors, they'll always be trivial to hack, there are good fingerprint readers (that actually don't use the prints, but subdermal tissue) but they cost a lot more than the ones taht are defeated in such trivial ways..
I'm still looking for the retraction from all those people who posted to the original fingerprint reader on iphone thread last week saying this wasn't a simple fingerprint reader on the iphones and wouldn't be susceptible to this form of attack...
Based on their respective histories, a sensible person would probably trust CCC over Apple.
Yes, I agree. No idea why this was modded "troll". There is a decent history to show that.
CCC:
Did this before. They lifted the fingerprints of the German minister of Interior from a water glass and turned it into a little stamp so you can place him now at any crime scene. (The hack was actually to show just how idiotic government use of biometric data is).
Apple:
I of course don't want to say anything negative against this good company, but some people might say that they have a history of over-hyping things.
It's a capacitative scanner. Whether you like it or not, that's not imaging the surface layer of skin, but the complexity of what's behind it.
You're correct that it doesn't image the surface layer, but wrong about it getting what's behind the skin. Capacitive sensors obtain an image of, essentially, the back side of the skin. The ridges are there, but no other subdermal structure is visible, and the ridges are the same ones visible on the surface, so a surface image (e.g. a skin-oil negative), provides a fine panel from which to construct a usable fake finger.
FWIW, I used to build biometric authentication systems, especially fingerprint stuff. I did security analyses of fingerprint scanners (optical and capacitive) for Visa, wrote the Linux kernel driver for the AuthenTec scanner, and a bunch of other stuff over 10-year period. I've never designed them and don't claim to fully understand the physics (though I've consulted extensively with people who do), but I've worked with them, a lot, and I know very well what they do and do not do.
Note to ACs: I usually delete AC replies without reading them. If you want to talk to me, log in.