Slashdot Mirror
NSA Building US's Biggest Spy Center
New submitter AstroPhilosopher writes "The National Security Agency is building a complex to monitor and store 'all' communications in a million-square-foot facility. One of its secret roles? Code-breaking your private, personal information. Everybody's a target. Quoting Wired: 'Breaking into those complex mathematical shells like the AES is one of the key reasons for the construction going on in Bluffdale. That kind of cryptanalysis requires two major ingredients: super-fast computers to conduct brute-force attacks on encrypted messages and a massive number of those messages for the computers to analyze. The more messages from a given target, the more likely it is for the computers to detect telltale patterns, and Bluffdale will be able to hold a great many messages. "We questioned it one time," says another source, a senior intelligence manager who was also involved with the planning. "Why were we building this NSA facility? And, boy, they rolled out all the old guys—the crypto guys." According to the official, these experts told then-director of national intelligence Dennis Blair, "You’ve got to build this thing because we just don’t have the capability of doing the code-breaking." It was a candid admission.'"
In american America, people monitor the government.
In soviet America, the government monitors the people.
How many bits should we use for encryption now?
More.
Ceci n'est pas un sig.
How many bits should we use for encryption now?
If you assume peak computing power is doubling ever n years, they you need one more bit every n years to keep ahead.
And of course, whatever you use now will be breakable in the future, if anyone cares to save your messages until computing catches up.
Sheesh, evil *and* a jerk. -- Jade
That's basically what happens today with most protocols like SSL/TLS. For each new connection, the client and server negotiate a new key via public key crypto like RSA. Actually, based on some comments in the article, like needing more "transactions" to help break the encryption, makes me believe the NSA is actually working to break RSA then AES.
Use no encryption and have a sig like mine. Eventually someone gets bored of reading every mundane post and email and puts you on an "ignore" filter.
---
ECHELON is a government program to find words like bomb, jihad, plutonium, assassinate, and anarchy.
First post, never got that before.
You must be using the new FTL neutrino submission system.
Sheesh, evil *and* a jerk. -- Jade
We use our signals intelligence capability to pass the trade secrets of foreign companies on to our own domestic companies; there is plenty of money to be made from being able to decrypt messages that the NSA intercepts.
Palm trees and 8
My understanding is that the best known general cryptanalytic attacks on AES are only marginally better than brute-force. Even AES-128 is essentially unbreakable under any known attacks then, since brute forcing a single AES-128 password is so far beyond feasibility, it's absurd. My understanding is that the best known attacks on AES are side-channel attacks, which require only modest computational resources, but need access to the encrypting machine, and related-key attacks that are only effective for certain small classes of keys.
So we can then assume that NSA has a general attack on AES that makes it many, many orders of magnitude easier to break than the best known published attacks? Or is this more likely to be disinformation spread to make people *think* that AES is broken by NSA? My understanding was that NSA is generally somewhat but not extremely far beyond the academic state of the art these days.
And there have been several reports of FBI and other federal agencies being unable to recover AES-256 encrypted hard drives. So if NSA has the capability to do so even for small numbers of keys using existing computing power, they obviously keep it incredibly restricted and under wraps.
So... this is BS by somebody, right? Either congress is getting BSed into funding stuff that won't do what they're being told it will do, or the public is getting BSed into believing that using encryption is pointless because NSA can real-time decrypt anything, so just don't bother, mmm'kay?
uckfay offway ationalnay ecuritysay agencyway
First off, AES isn't public key, it's just usually used in conjunction with public key. The public key portion of the exchange is used to communicate an AES key (the "shared secret") which is then used for communication moving forward. This is because public key encryption is "expensive" by comparison to block cyphers like AES. Secondly, you don't communicate a passphrase with public key. The passphrase that you're used to using is so that keys can be securely stored and someone that gains access to your key file doesn't get access to your key.
You could potentially communicate a new AES key with every message, which would greatly reduce the chances of a bruce force attack being successfully since most rely on the ability to analyze a large number of blocks that use the same key. That said, if you crack one key you do gain access to every key that followed in the chain.
and even better: send false positives to waste their time.
perhaps the crypto protocols need enhancing to allow fake bullshit messages that can't easily be told from real crypto stuff.
ie, DOS them.
I know, they have lots of power but it IS a war. war on our privacy and its so blatant now, they don't even try to hide their break-in attempts to us, anymore.
the ONLY reason encryption was allowed in the first place was for banking and online 'business'. if there was not this use-case, we would be disallowed encryption entirely.
--
"It is now safe to switch off your computer."
Anyway AES is public key encryption.
AES is a symmetric-key algorithm.
WHO would work for them, I ask you?
decades ago, the people didn't view their government quite the way they do today. some patriotism did exist and people wanted to help their government. *generally*.
today we all see how invasive and evil our government has become. totally 100% lost its way. almost anything it does, it does badly and hurts people, long and short run.
if I was offered a job for the so-called white hats (which I now see as black hats) I'd turn it down. I would not be able to live with myself knowing I'm helping an evil force become more evil and more forceful.
I do realize a lot of people can easily shelve their ethics and see money-making jobs as separate. but I wonder how many people still believe that if they join the government or gov-sponsored jobs, that they are really HELPING things?
too many black marks on the government. working for them could be as bad as working for the old mafias. the people that they do get, I would not trust. they are whores.
--
"It is now safe to switch off your computer."
The one time pad could make a comeback in the form of a one time DVD's or maybe even SD or Micro SD chips. I know, it is not scalable due to the problem of distribution. It is also symmetric in that the same "key" encrypts and decrypts, but it is also immune to brute force since your one time key is equal to or longer than the message length. An interesting variation might be to use an image file that is very long, but completely innocent as a pseudo random key and only have two copies of that exact image. The former Soviet Union used a one time cypher for all of their clandestine agent communications.
How many bits should we use for encryption now?
All of them.
I wonder if that sentence says more than they intended it to. Could it be that the skills of the NSA people are eroding just like the skills at CIA did? I knew that CIA was in trouble - tradecraft-wise - when a COS let an asset into their HQ and he blew half the station to kingdom come. No one would have done that in the old days. Maybe NSA is having the same problem.
Crypto-guys are the "old guys" from a tradecraft point of view. AFAIK, in the NSA, many of the old-guys are involved with developing clever new internal ciphers (so-called classified "suite-A" algorithms). Since many of the "bad-guys" aren't nation states with heavy duty crypto development capablities, they often are using off the shelf stuff like AES/ECDSA (members of the "suite-B" algorithms). Until someone discovers a huge gaping hole backdoor, breaking these "suite-B" algorithms benefit from mostly from brute force (even if you know a few clever tricks that others do not which chops things down an order of magnitude or two). This is pretty much an admission that there is no huge gaping back door in these suite-B algorithms, not that any crypto-tradecraft capability was in trouble.
I find it oddly somewhat comforting that the we have "old-guys" that realize that sometimes the best thing to do is to throw this problem at a box of computers and spend their time on other pursuits. Who knows, this facility might be dedicated to cranking on some clever cracking algorithm that is unknown to the public, all we know it it takes lots of OPS. Isn't surpising to me that cracking these algorithms are hard. As a historical data point, DES was apparently hard for even the NSA to crack so they deliberatly limited the DES key size from the original 64-bits, to the final 56-bit (although the NSA apparently lobbied for a mere 48-bits).
I actually doubt that they are most interested in brute-force codebreaking through the front door except in a few rare situations.
Most of the time, it's massive traffic analysis: searching and analyzing a titanic, dynamically changing graph, nodes are IP addresses and phone numbers of the planet.
Once they find a 'target of interest', then they would usually ask the FBI or other authority just to put a tap on a specific line, or if necessary break in and install a trojan on the target's phone or computer, avoiding front-door code-bashing, which isn't generally feasible in large scale any more.
There are companies (e.g. http://www.conveycomputer.com/) which make highly parallel co-processors from FPGA's which give user-definable vectorized instructions on enormous memory bandwidth.
This is just the thing for the NSA.
There's no way they can crack AES-128 unless there's a hole in the algorithm or they have quantum computing.
Current best practices are:
1) AES-128 to AES-256 for symmetric keys (although AES-256 has its own problems which can sometimes collapse it to AES-128 - these are ameloriated by increasing the key rounds)
2) 2048-bit to 4092-bit for RSA keys (2048 may be breakable by 2030 with conventional computing, 4092-bit will take much longer).
If quantum computing becomes feasible then AES keys will effectively halve in complexity (i.e. AES-128 goes to 64-bit, AES-256 goes to AES-128) and RSA and DSA keys will be useless.
Yeah, that's always the problem, innit? I personally wouldn't mind the NSA reading all of my email if it were, in fact, a sort of protector of good. How can any politician EVER control a beast that knows where every skeleton in every closet is and can protect that information behind armed guards and blast-proof doors? It's a deal with the devil if there ever has been one.
To ensure perfect aim, shoot first and call whatever you hit the target