3 ms·
With apologies to all, everyone is missing the point, including the author. Spoiler/tl;dr summary of this comment: Soon you will be able to generate a key whil
by PeterWhittaker 12y ago
With apologies to all, everyone is missing the point, including the author.
Spoiler/tl;dr summary of this comment: Soon you will be able to generate a key while standing at the lock, scanning it with the laser scanner built into your mobile device, generating the key with a hand-held 3D printer you bought at the convenience store on the corner.
OK, longer version.
A number of comments dismiss the security threat as not particularly meaningful (give someone your keys, give them an opportunity to mess you up) and they are correct insofar as this very specific threat is concerned. It really is very similar to the traditional "give someone your credit card, they can mess you up" threat.
But think about that threat for a moment: The reality of that threat led to chip-and-pin cards, to CCVs, and to PCI compliance.
That threat was also greatly magnified by the move from the pen-and-paper world to the world of online shopping - and we've had to adjust our threat mitigation strategies appropriately.
The point of the article isn't actually for example that valets require great trust. The author actually misses the real point while describing it quite clearly: It has always been possible to do this, it's just orders of magnitude easier, and cheaper, to do it today, because of the blend of off-the-shelf widely available low-tech technology. That's the point of mentioning KeyMe: It is so easy to build the bits to do this threat that there are reliable commercial services that make money from low-cost apps and unattended kiosks.
And it is going to get much, much, much, much worse, very, very, very soon.
Very soon now you will be able to buy a pocket or at worst back-pack size multifunction replication device, that is, a 3D printer with a scanner built-in, or perhaps a 3D printer that uses your mobile device as a scanner.
With that device, you will be able to duplicate the keys on the way up the stairs.
And soon you will be able to use a commercial hand-held 3D printer with a commercial hand-held laser scanner to generate a key by scanning the lock.
That's just the beginning of tomorrow's threat model.
EDIT: Added a missing word above, and the following thought....
We assess attackers based on motivation and resourcefulness: a motivated, resourceful attacker is always a worry. The key duplication threat used to require a motivated attacker capable to acquiring resources, e.g., the prisoner, who would duplicate a key slowly and laboriously, using low-tech tools. This is a very motivated attacker.
Existing and emerging technologies make the resources widely available and lower the motivation bar dramatically. The key duplication scenario will soon be populated by the equivalent of script kiddies who will perform the threat just because they can, just because it's that easy.
That's the point of the article. That's the threat model we face going forward.
- x1798DE 12y agoIt is very very unlikely that you'll be able to duplicate keys using any kind of laser scanning technology, that's just not how they work. You could probably get the appropriate shape of the keyway, but you wouldn't be able to decode the lock without being able to image through metal, which is simply not going to be a technology built into phones any time soon. I still find it an unlikely threat model, but I would admit that in the unlikely event that a technology allowing you to enter a door just by scanning it and printing up a key were invented and integrated into mobile phones or something else people are carrying around anyway, then that might actually overtake "crowbar"/"breaking the window" as the lowest-cost lowest-risk way of breaking into someone's home. And even then, probably it wouldn't affect all that many people. Plus you'd just have to modify lock technology slightly to fix the problem.
- PeterWhittaker 12y agoYou got me wondering about other ways to scan, such as sonically. I'll bet that right now any kind of sonic scan would be crude and rough - but how long before signal processing can be cleaned up to make it good enough to guesstimate the lengths of the pin segments? Especially if the scan included analyzing against known lock types? This is the sort of problem for which the Internet is ideal: An optical scan to narrow down lock makers and models, load custom signal processing configurations for those likely locks, perform a sonic scan of the lock and collect reflections, distortions, etc., and use the custom config to ease the decoding: Wah-la! A picture of the inside of the lock, with pin segment lengths. Then print 5-10 likely keys on the spot. Picking a lock requires a suspicious body posture and special tools. Scanning a lock this way just requires you be close enough to the door. Checking your messages? Patting your pockets to make sure you have everything? Leaning your bag against the door on your knee to make sure you have everything? Many plausible scenarios for staying close enough long enough, and the tools are COTS, common. Or maybe I "deliver" a "package" to your door, lean it against the door (UPS and others do this all the time), and the top of the package contains the scanning tools. I come back later (seconds, minutes?) after realizing I've made a mistake and take the package back. The more I think about it, the more I cannot help but wonder if the physical lock of the future will not need a proof-of-authorization mechanism to vet the keyholder - at which point we eliminate the keyhole entirely and move to direct identification of authorized entrants. And then of course watch for 0-day vulnerabilities in THAT technology.