Hacker Newsnew | past | comments | ask | show | jobs | submitlogin
Police asked 3D printing lab to recreate a dead man’s fingers to unlock phone (fusion.net)
166 points by theandrewbailey on July 21, 2016 | hide | past | favorite | 83 comments


Privacy expires at death. http://www.nytimes.com/2001/11/26/opinion/l-privacy-after-de...

If you want to take something to the grave, you'll have to be more creative than simply assuming nobody is going to go through your stuff when you die.


At least on iPhone Touch ID expires too… no need to die, just wait eight hours.


Similarly on the Nexus fingerprint scanners. There's a 48 hr timeout.

All the reasons it will require a secondary method[1]:

• When your fingerprint isn't recognized after a few tries

• After restarting your device

• After switching to a different user on the device

• After more than 48 hours have passed since you last unlocked using your backup method

[1] https://support.google.com/nexus/answer/6285273?hl=en


Or let the phone's battery die. Passcode is required on cold restart too.


Deadman's switch attached to a thermite pile?


https://www.youtube.com/watch?v=-bpX8YvNg6Y Might be harder than it looks/sounds.


Just "lost" an hour of my life. Thanks! :)


Or you know a password that isn't written down and strong encryption.


Strong encryption when you die is weak encryption a few years later. Thermite is the way to go



And a written down "password".


That's just not permanent or fail-deadly, maybe something like HCl over your documents on death. The thermite might get spread around or fail to ignite.


Fingerprints should not be considered to be a password. They are a username. You can't rotate your fingerprints like you can rotate a password, and what is more you leave your fingerprints on everything you touch.


Totally agree, though I'd extend this to all biometrics. If you can't rotate it, it should only ever be used as an identifier, not a password.


Yup. If I look at my phone right light, I can see the key to unlock the phone on the glass.


Cyanogenmod has a feature that lets you randomize the numbers on the keyboard each time you try to unlock it, if you use a PIN.


can you explain this a little bit more?


The numbers on the keypad appear in a random order so you’re touching different parts of the screen each time, even though you’re entering the same code.


1. Place phone on scanner-bed

2. Open mirror image of fingerprint in photoshop and maximize contrast, crop to fingerprint alone

3. Print finger print mirror image to UV-stencil-type material

4. Etch into UV-material w/ UV printer

5. Pour rubber/plastic-type material over UV etching

6. Remove rubber thumb, apply to fingerprint detector, voila..


Though since you know the shape you'd have to get someone else to try unless it's really clear.


Why kind of phone is this? I believe iPhones don't allow unlocking with TouchID after 24 hours of inactivity, presumably for this reason...


Could just be that the police don't know this, and they'll get a rude surprise when they go and try it.

Not that it's important, but I believe it's 48 hours, and 8 hours if you haven't entered your passcode in the last six days. You also only get five tries, so a finger that is sometimes recognized but not with great reliability would still probably fail.


Correct, rules outlined here: http://www.macworld.com/article/3072181/ios/new-touch-id-rul...

• Not unlocked for 8 hours after not having used passcode for 6 days

• Not unlocked for 48 hours otherwise

• Restarted

• 5 failed fingerprints

• Enrolling new fingerprints


Well, does the OS or the fingerprint TPM handle the timeout?

If the OS handles it (and potentially if the OS files themselves aren't completely and fully encrypted) then you can trick the OS into thinking it hasn't been 24-48 hours - all you need is for the fingerprint TPM to keep returning 'success'. If the TPM handles it then it is slightly harder to do.


IIRC the Secure Enclave handles it.


Important to note as well that the fingerprint sensor unlock is not protected by the 5th amendment like passwords are. [1]

[1]http://www.theatlantic.com/technology/archive/2016/05/iphone...


Which, IMHO, is perfectly reasonable. Your fingerprints are used by the government for /identification/, just like your driver's license. It's not a password. You are required to surrender your fingerprints when you are arrested. It's not compelled speech because it's an identification marker, not a password.

If you do have a runin with the police, just reboot your phone by holding down the power and home button for 10 seconds. Or, if you actually have something to hide, just don't use Touch ID.


I have to ask the morbid question, why couldn't they use the dead man's fingers directly?


In addition to the capacitive sensor, there's an RF sensor that apparently responds to living tissue, but not dead tissue. http://fortune.com/2013/09/16/apple-touch-id-and-the-fear-ha...

I imagine there are some Frankensteinian ways to create the necessary current to activate the RF sensor, but… well, Frankenstein


Yeah, I'd take some of that with a grain of salt:

http://www.theregister.co.uk/2014/09/23/iphone_6_still_vulne... - glue fingerprint replica sufficient

http://www.iphonehacks.com/2016/02/iphone-touch-id-hacked-wi... - PlayDoh, dental paste


In that case, couldn't they just coat the finger with the same stuff they're coating the 3D printed replica with? Actually, how are they getting around the RF issue?


I suspect deformation due to decomposition. Without doing any research I'd say that the decomposition process may rapidly change finger prints enough to no longer be recognized.

Alternatively, and possibly more likely, as a murder victim there's any number of ways his fingerprints could have been obliterated.


My complete guess based is that the body is either already cremated, or in the hands of the family of the deceased. And they wouldn't exactly be jumping at the opportunity to let the police mess with their dead relative.


And they probably want something they could reliably continue to unlock the phone with. Sure, they could just remove all authentication once they're in, but maybe they want to tamper with the phone and its settings as little as possible.


I don't think you can make any changes to the security setting unless you know the passcode.


Quite possibly. I don't think the article mentioned what kind of phone it is.


I was under the impression that a 3D print was overkill, that you could do this much more easily. Have the sensors gotten that much better?


It's not only overkill, it probably doesn't work because the resolution is typically way to big.

http://qz.com/631697/a-regular-inkjet-printer-can-spoof-a-fi...


It should be noted that these are the same researchers in both articles. I would guess that the inkjet method didn't work otherwise they are just ripping off the police department?


It's overkill by a whole dimension; fingerprints are 2D.


No it's not overkill, and no, fingerprints are not 2D. The fingerprint scanners on iPhones (and I believe other phones as well) are capacitive, meaning they rely on the physical ridges in your fingerprint.


Fingerprints aren't 2D, but the sensor is. The sensor does not directly measure the height of the epidermal ridges. Rather, the varying thickness of the epidermic vary the dielectric between the conductive dermis and sensor. So you can replace the ridges with something completely 2 dimensional but with the same dialectic pattern. A normal home printer ought to be able to do that.



From the link:

"The spoof worked on the Samsung and Huawei handsets, but not the Apple and Meizu ones"


indeed the iphone fingerprint does check if the objects are capacitive though, but that's a separate check. the "scan" is still just iirc. a infrared 2d cam

cough graphite powder cough


>fingerprints are 2D

>Fingerprints aren't 2D


I doubt very much that you could do it with a 3d print at all. Even direct casts from fingerprints have proved difficult to use to defeat the scanner. I don't think a 3d print would have the necessary resolution.


You can almost 3d print at the atom level with the right kind of machinery...

Some FDM printers apparently have resolutions in the low 10s of microns.

This laboratory probably isn't using off-the-shelf consumer-grade 3d printer


>I don't think a 3d print would have the necessary resolution.

I bought and run a Projet 3500(http://www.3dsystems.com/3d-printers/professional/projet-350...) at my day job. In the highest resolution mode, which is all we really use, it's 750x750x1600 DPI. You could definitely print an accurate finger with fingerprints on it.


One way I've seen it done is to create a PCB with heavy copper, and then use that to cast gelatin or some other skin analog.

http://torkjel.com/projects/attacking_fingerprint_sensors.pd...

Creating PCBs can be with very short turn around times or even in house.


I agree. Just mold and cast the finger.

A bigger question is, why do they need the fingerprint... The phone has a passcode too.


I'd like to see brainwaves being used as passwords. And the fun accompanying the government forcing suspects to think "unlock"


You'd need one hell of an accurate EEG in similar conditions to do this consistently.

That being said, give it a couple decades


You can detect strong signals with current consumer technology. One easy pattern is the SSVEP[1], which shows up as a spike in the brain signal's FFT at the frequency of a flashing light that the subject is looking at. Interestingly, it also seems to work with binaural audio.

It might be hard to tell peoples' signals apart with the confidence of a strong password, but individuals do respond pretty differently in terms of how long it takes to appear, how quickly it grows, how much of a spike is observed, small offsets, etc.

[1] https://en.wikipedia.org/wiki/Steady_state_visually_evoked_p...


Yeah, SSVEP has it's uses[1], but I don't think it would work well enough to identify various users with any sort of confidence. It would likely be easier to do with showing an image (or series of such) that would present people with a very different emotional responses. This could be loved ones, childhood homes, etc.

If you are curious in this stuff, I am pretty familiar with this stuff, and my startup[2] uses consumer EEG's regularly. Our single largest concern is actually signal correction due to movement / other environment factors[3].

[1] http://synaptitude.me/blog/a-quick-intro-to-ssvep-steady-sta...

[2]http://synaptitude.me/

[3] http://synaptitude.me/blog/using-computer-vision-to-improve-...


Pretty sure a simple suggestion of "don't think of an open lock" would do the trick. Or showing an animation of a lock opening. Even if the required image isn't of a lock, but of a "password image," the suggestion of "don't think of your password image" will probably conjure it up for most.


If that were the case it'd be fairly easy to add duress detection that just defaults the phone to password-only authentication.

Well, unless the cops use neural jammers or something. How cyberpunk.


Probably stuff like ecstasy would work- but that probably produces different patterns too.


Are there any features of brain-related signals that are biometric in nature?!


Just a reminder that fingerprints are not passwords.


Well, I prefer to think of it as "passwords with a narrower threat model". They protect against randos at a party, petty thieves, and even close friends from getting into your data, but not well-funded governments or people willing to cut off your hand.

That may be a good enough for your use case, esp since you don't have to remember it.


Thank you very much for saying that without the "they are usernames" that usually accompanies it. Fingerprints are something different from either, with their own properties, and trying to shoehorn them into the existing username/password system means you'll get things wrong.


Likewise, neither is DNA. We leave both of them everywhere.


This seems very problematic, since of course going forward they'll be doing this with live suspects. Police stations and other law enforcement agencies are known locations. It wouldn't that be hard to write an app that has a database of GPS coordinates for these locations and puts the phone in passcode required mode the moment the phone enters one of these zones.


this is an awesome idea, but the other way around... only allow fingerprint in white listed locations.


Even better - great idea!


Reminds me of the story about the 3D copier that somebody 3D copied, but they had their thumb on the scanner when they copied it, so all descendent copies of that 3d copier had a copy of their thumb on the scanner.


got a link for this absurd story?


Question: Can we trust that the person is dead?

The police aren't providing the dead man's fingers, just a fingerprint scan. Without more, I have to admit the possibility that the police aren't telling the truth, that this is an attempt to unlock a live person's phone. If the person was in custody then there would be no need. They can force his finger onto the phone. Perhaps they have the phone but not the man?


Police asked locksmith to recreate a dead man's key to unlock house.


Maybe if the key was the dead man's finger, and the house contained swathes of the dead man's private life including correspondances, photos, friends/family, all analyzable in mere seconds.

Come to think of it: Police asks 3D printing lab to recreate a dead man's fingers to unlock phone.


>and the house contained swathes of the dead man's private life including correspondances, photos, friends/family,

Based on the tone of your comment, it seemed that you implied that these wouldn't be in someone's house? They certainly would be for many people I know...


Swathes? Maybe not a strong enough word. Do an inventory and compare it to what's in your phone. Even amongst the least technical people I know, some of them have thousands of emails, their facebook correspondances are filled with personal details, their SMS/chat history is a perfectly timestamped history of their past few years, etc.

Yeah there's a few hundred, or even maybe thousands of photos in my mother's house. I'm sure there's a lot of unread magazines, too. All of that, accumulated over decades. I think it's still a safe bet there is more data in her phone would than in the combined homes of the entire neighbourhood.


Worth noting that Mythbusters did an experiment with spoofing fingerprint scanners. They didn't use a 3D printer, but etched a copper plate. It looks like the fingerprint reader technology is not the same, but it seems like the ballistics gel version they made might have similar capacitance to a human finger and could work.

https://www.youtube.com/watch?v=3Hji3kp_i9k

tldw: http://www.discovery.com/tv-shows/mythbusters/mythbusters-da...


This was several years ago when fingerprint scanners were cameras under glass, they are a lot more sophisticated now and harder to fool. These techniques will get you no where on a current Apple device.


IIRC, they also had success on some readers with just a photocopy of a dusted print, with a warm/conductive finger behind the paper.


> ... and the police think there might be clues to who murdered him stored in his phone.

I am always skeptical of these kinds of statements.


Wouldn't this[1] be easier? If they have the print on file, I can't imagine why they'd need to go all 3D-printer-happy.

http://www.extremetech.com/computing/51228-gummy-fingers-foo...


Important privacy matters aside, would this even work? I didn't realize that 3D printing had improved to the resolution of a fingerprint.


Got a high school dropout trying to play McNulty and be clever

oh well


Sounds like a new Pirates of the Caribbean film. "Pirates of the Caribbean: Dead man's fingers"


Url changed from http://www.theverge.com/2016/7/21/12247370/police-fingerprin..., which points to this.

Submitters: please submit original sources, especially when there's an obvious one. This is in the HN guidelines (https://news.ycombinator.com/newsguidelines.html).


[flagged]


Please don't do this here.




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

Search: