Encryption Works, Right?
So you’ve got a bunch of encrypted data. It’s safe, right? After all, it’s encrypted, which means that nobody can get at it, right? Right?
The answer to that is, very definitely, “Maybe, maybe not, couldn’t say for sure”. The thing is, there are so many contributing factors to whether your data is safe.
Bad KeysIt doesn’t really matter how good your system is, if your password is “password1!” or “admin”, or something else on this list of worst passwords of 2017. Or, heck, this list of the 10,000 most-used passwords from hacked password lists.
Leaky Keys
Even if you’re awesome, and have special super-complicated keys like “correct horse battery staple”, it doesn’t particularly help if you, write it on a post-it. Maybe you’re better than that (please be!) and actually store it somewhere safe. It still might get hacked, and lords, if it does, it can be bad. Just ask the fine folks at Lockheed Martin who got nailed thanks to the super-secret seeds for their RSA devices getting stolen by hackers.
Even if you’re awesome, and have special super-complicated keys like “correct horse battery staple”, it doesn’t particularly help if you, write it on a post-it. Maybe you’re better than that (please be!) and actually store it somewhere safe. It still might get hacked, and lords, if it does, it can be bad. Just ask the fine folks at Lockheed Martin who got nailed thanks to the super-secret seeds for their RSA devices getting stolen by hackers.
Faster Computers
Computers just get more and more powerful over time. Which is just another way of saying “It gets easier and easier to crack encryption over time”. And that doesn’t even begin to get into fun stuff like using GPUs, FPGAs, and the like. Heck, DES was the ne plus ultra of encryption schemes back in 1977, while these days it’s not much better than plain text — and that’s almost completely due to computers getting faster.
Computers just get more and more powerful over time. Which is just another way of saying “It gets easier and easier to crack encryption over time”. And that doesn’t even begin to get into fun stuff like using GPUs, FPGAs, and the like. Heck, DES was the ne plus ultra of encryption schemes back in 1977, while these days it’s not much better than plain text — and that’s almost completely due to computers getting faster.
Broken Encryption
All the above presumes that the encryption schemes you use is actually good. Is it though? Schemes get broken all the time — Wikipedia has an alarmingly long list of broken ciphers (let alone just putting all the broken stuff into a page with subcategories).
This isn’t about old stuff like DES (•), SHA-1 was recommended as recently as 2010, and it was comprehensively broken by the SHAppening in 2015!
All the above presumes that the encryption schemes you use is actually good. Is it though? Schemes get broken all the time — Wikipedia has an alarmingly long list of broken ciphers (let alone just putting all the broken stuff into a page with subcategories).
This isn’t about old stuff like DES (•), SHA-1 was recommended as recently as 2010, and it was comprehensively broken by the SHAppening in 2015!
Loopholed Encryption
A special category created by our special friends at the NSA (Motto: “The Eye Of Sauron Has Nothing On Us”) where they actually introduce encryption standards that you think is secure, but which they have happily back-doored so that they can get at anything. That’s ok, ‘cos they’re the Good Guys, right? And, of course, nobody else can, right? Because data never leaks, right?
Mind you, this happy state of affairs has resulted in people distrusting pretty much anything with an NSA fingerprint on it. Not that they don’t have cause — the Snowden leaks revealed that the NSA had paid RSA to use (the backdoored) Dual_EC_DRBG by default. The problem, of course, is that cryptography is hard, and the NSA has a huge percentage of the people that actually understand Crypto. Sigh. (••)
A special category created by our special friends at the NSA (Motto: “The Eye Of Sauron Has Nothing On Us”) where they actually introduce encryption standards that you think is secure, but which they have happily back-doored so that they can get at anything. That’s ok, ‘cos they’re the Good Guys, right? And, of course, nobody else can, right? Because data never leaks, right?
Mind you, this happy state of affairs has resulted in people distrusting pretty much anything with an NSA fingerprint on it. Not that they don’t have cause — the Snowden leaks revealed that the NSA had paid RSA to use (the backdoored) Dual_EC_DRBG by default. The problem, of course, is that cryptography is hard, and the NSA has a huge percentage of the people that actually understand Crypto. Sigh. (••)
Broken SystemsSystem design is hard, and I mean this literally. It’s pretty surprising that systems actually work, given the scale, scope and sheer complexity of all the components that are involved. The thing is attack-surface tends to scale (geometrically?) with system complexity. (Yes, I made this up — there really is no easy way to measure attack surface, though many have tried). Given this, it’s not surprising that hacks can range from spear-phishing (Yahoo!), to remote script execution (LinkedIn!), to malware embedded in the OS (NSA!).
Broken Processes
It doesn’t matter how secure the actual system is, if the processes that wrap around it are porous. Archiving data into an unprotected S3 bucket (Verizon!)is, well, just asking for it. Or, losing your laptop that has TopSecret™ data (too many people to count).
The point, of course, is that you can’t just rely on the system being secure, you have to make sure that all the stuff that wraps around your system is secure too. Which leads directly to
It doesn’t matter how secure the actual system is, if the processes that wrap around it are porous. Archiving data into an unprotected S3 bucket (Verizon!)is, well, just asking for it. Or, losing your laptop that has TopSecret™ data (too many people to count).
The point, of course, is that you can’t just rely on the system being secure, you have to make sure that all the stuff that wraps around your system is secure too. Which leads directly to
 |
| /via https://xkcd.com/538/ |
People
Can the security of your super-secret system be imperiled by physical threats? Is access to your facilities protected by somebody making $8/hr? And yeah, I get it, you have security cameras, but how far do they go in preventing anything from happening? And it doesn’t even have to be malice, it could just be somebody going for a smoke-break at exactly the wrong time.
Can the security of your super-secret system be imperiled by physical threats? Is access to your facilities protected by somebody making $8/hr? And yeah, I get it, you have security cameras, but how far do they go in preventing anything from happening? And it doesn’t even have to be malice, it could just be somebody going for a smoke-break at exactly the wrong time.
In the end, the real problem with super-special-secret-keys is that the more important the data that you are encrypting with these keys, the more incented BadPeoples™ are to get at them. It’s an arms race, with you having to stay one step ahead, and unlike the BadPeoples™, you actually have other stuff to do (while they are solely focused on getting at you….)
(••) For more about designing block cipher algorithms that pass all the security tests, but which have backdoors in them, check out “Partition-Based Trapdoor Ciphers” by Arnaud Bannier and Eric Filiol a. An associated talk — “By-design Backdooring of Encryption System Can We Trust Foreign Encryption Algorithms?” — by them at BlackHat has easier to digest slides.
Comments