Here's a challenge (decryption)

I'm not a cryptologist but aren't these things only able to crack if you:

a) have multiple documents with the same key
or
b) brute force

A is out because there's only this document.

B could take a while.

Depends. We have enough information to make a somewhat-intelligent brute force attempt. We know it contains documents, and it's likely that the majority of the documents are the same format. With enough finagling, a file pattern could be discerned from document containers, and make a good start point for decrypting.

It would be rough, that's for sure. But given that we know the encryption format and general contents configuration, that's a pretty good start point.

Turns out it is pretty simple - change the filename to .zip and use your favourite decompression software.

Quote:

Original post by necreia Depends. We have enough information to make a somewhat-intelligent brute force attempt. We know it contains documents, and it's likely that the majority of the documents are the same format. With enough finagling, a file pattern could be discerned from document containers, and make a good start point for decrypting. It would be rough, that's for sure. But given that we know the encryption format and general contents configuration, that's a pretty good start point.

There is no such thing as a "somewhat-intelligent" brute force cryptanalysis. Recognizing a successful decryption would be pretty trivial -- the first several bytes will be one of a few archive headers -- but there are no known attacks on AES-256, known-plaintext or otherwise. If you want to count to 2²⁵⁶, you'll have to start pretty early.

Just thought about something though. Would a distributed hack attempt, like SETI or the folding project be feasible. Everyone takes a segment of brute force entries and then submits the results to a server. Or would there not be enough storage space in the universe to hold the results?

Quote:

Original post by ChurchSkiz Just thought about something though. Would a distributed hack attempt, like SETI or the folding project be feasible. Everyone takes a segment of brute force entries and then submits the results to a server. Or would there not be enough storage space in the universe to hold the results?

The numbers are still too large.

With rainbow tables, the time would be feasible, but one would require yottabytes of storage. This is a bit more than there currently exists.

With bruteforce, the computing time is in the order of 10^50 years. Even if divided over a million or billion computers, it's still too big.

So unless someone comes with a smart attack, it isn't happening.

I have no experience with cryptography at all - where do these numbers come from? Are we looking at solving just the password (assuming it's AES256)? Or are we assuming it's something else and so have to decrypt the entire thing?

If it's just the password, is this of a fixed character length?

(keeping in mind I don't expect any practical shot at this, but it's still interesting)

Quote:

Original post by Chris Reynolds I have no experience with cryptography at all - where do these numbers come from? Are we looking at solving just the password (assuming it's AES256)? Or are we assuming it's something else and so have to decrypt the entire thing? If it's just the password, is this of a fixed character length? (keeping in mind I don't expect any practical shot at this, but it's still interesting)

its not a password in the traditional sense, its a key.
For AES256 its 256 bit long thus there are 2^256 possible keys, in order to know if a key is correct you'd have to attempt to decrypt the document using the key and then analyze the resulting "decrypted" document to determine if you used the correct key or not (If the resulting decrypted document is pure garbage you probably used the wrong key, if it resembles the kind of data you expected (in this case text documents, possibly some images) you probably used the correct key), checking each key takes quite a bit of time and 2^256 is an insanely large number, its around 1.15*10^77 (or 1.150.000.000.000.insert 65 more zeroes here) , if you can process 10 million keys per second you'd still need 3.6*10^62 years to test all keys (Which is far longer than the expected remaining life for the sun).

Basically 256 bit AES encryption is virtually impossible to break using bruteforce attacks. (You could get lucky and hit the correct key early, but the odds of that happening are extremely small)

while i don't know the detail of this AES thing, brute force has come better these days with the advance of GPU.

http://archive.atomicmpc.com.au/forums.asp?s=2&c=9&t=17490

Quote "For example, the search speed for MD5 hashes with GeForce 8800GS is 270 million passwords per second."

And this is in 2008.

Oh wait, something related to AES

http://www.tomshardware.com/news/nvidia-gpu-wifi-hack,6483.html

Quote "The problem here is that most VPNs also use AES encryption which is the same encryption that is employed by WPA2. WPA employs RC4. If this new type of approach at breaching WPA/WPA2 protection actually works with ease, then the majority of VPNs are at risk as well."

if the database does truly contain names and other extra sensitive info, then i'm sure lot's of government hackers are working on it already.

Original post by Sneftel

Quote:

Original post by necreia .... but there are no known attacks on AES-256, known-plaintext or otherwise. If you want to count to 2256, you'll have to start pretty early.

Not sure if that's strictly true: http://www.schneier.com/blog/archives/2009/07/new_attack_on_a.html

Quote:

For example, the search speed for MD5 hashes with GeForce 8800GS is 270 million passwords per second.

An order of magnitude off SimonForsman's example. 3.6 * 10^61 isn't much better than 3.6 * 10^62. Even if the search speed is another few orders of magnitude faster in the mean time since the article, it still isn't feasible.

Quote:

The problem here is that most VPNs also use AES encryption which is the same encryption that is employed by WPA2. WPA employs RC4. If this new type of approach at breaching WPA/WPA2 protection actually works with ease, then the majority of VPNs are at risk as well.

You misunderstand the nature of the attack. They are brute forcing the password, they expect that many networks have keys generated from a password, and many people pick common passwords.

Wikileaks would probably not have been so careless to choose an easily attacked password, if they even generated the key from a password at all. This also assumes that the file is actually aes256. That could be a deliberate move to waste the time of people trying to attack the file, it could be encrypted with another algorithm. It might not even be data, it could be random garbage encrypted in the hope that their attackers wouldn't risk what it might contain.

well if he is bluffing he better have one hell of a poker face!

Quote:

Original post by FableFox brute force [...] GPU

Massively parallel brute force architectures that are much faster than GPUs have existed for decades.
A much bigger problem than the actual parallel computing power when attempting a brute force attack on key sizes upwards of 200 bits is energy. For any computer that runs within the commonly accepted parameters of our universe (i.e. what we currently consider as "facts"), you would need an energy source the size of a supernova for such an attack (not taking into account the energy needed for cooling).

Quote:

Original post by rip-off That could be a deliberate move to waste the time of people trying to attack the file

This would be a rather silly move though, since the people who arguably have a legitimate interest (if you can call it legitimate) will likely break an AES encryption within seconds anyway, regardless of key length.
The NSA had well-known backdoors built into all mainstream ciphers of the past, either by being strongly involved during the design phase or by bribery/blackmail such as in the case of IDEA that came up a couple of years ago.
It is likely, almost guaranteed, that the same that was once true for DES and IDEA holds true for AES as well, why should it be any different. Of course you can take it for granted that none of the people involved in the AES design will ever admit to such a thing, because most people object to the idea of drowning in their bathtub or cutting their ulnar artery after swallowing three dozen tablets in a forest a mile from their house, which are typical things to happen in such a case.

Thus, it doesn't really matter whether you use a 256 bit key or a 16384 bit key, if you're afraid that Some Bad Guys will read your secrets. If you have Some Bad Guys interested in what you're doing, you have a problem on an entirely different scale anyway.
64 bits will prevent your little sister from reading your mail, 128 bits will protect you from every normal person, and what goes beyond that is a lost cause, end of story.
If you look like someone who might have some interesting information, it will be read, and otherwise it doesn't matter anyway (because nobody cares).

The bigger question is why would someone with the means to break AES want to read that file at all (supposed it's genuine in the first place, and not a hoax)? It's not like Wikileaks published or could possibly publish any secrets that the intelligence services did not know in the first place. That is the very nature of a "leak" -- the information, whatever it may be, is no big news to the insiders.

Quote:

Original post by FableFox Oh wait, something related to AES http://www.tomshardware.com/news/nvidia-gpu-wifi-hack,6483.html Quote "The problem here is that most VPNs also use AES encryption which is the same encryption that is employed by WPA2. WPA employs RC4. If this new type of approach at breaching WPA/WPA2 protection actually works with ease, then the majority of VPNs are at risk as well."

Some of those comments look pretty shady. Maybe I'm wrong but doesn't WPA2 change keys every few minutes? So how could you "break it" overnight or over a few days? Don't you have to be able to bruteforce or dictionary attack the password before the key resets? Either it works in one key cycle or it doesn't right?

Quote:

The bigger question is why would someone with the means to break AES want to read that file at all (supposed it's genuine in the first place, and not a hoax)? It's not like Wikileaks published or could possibly publish any secrets that the intelligence services did not know in the first place. That is the very nature of a "leak" -- the information, whatever it may be, is no big news to the insiders.

The threat is to release the key so the public will know whatever it is too.

Quote:

This would be a rather silly move though, since the people who arguably have a legitimate interest (if you can call it legitimate) will likely break an AES encryption within seconds anyway, regardless of key length.

Firstly, it isn't just the NSA who might attack Wikileaks or its members. There are some states who you could say definitely didn't have any direct input into AES selection and their secret services might not be able to break it.

I'm not 100% sure about your assertion about NSA backdoors. I do recall they interfered with DES, but it turned out in the end that the values they changed actually prevented a type of attack that was not publicly known at the time. When I get some time later on I will look into IDEA.

I'm curious as to what it is. Sadly, the password isn't hunter2 either.

Is it all the unreleased cables so far? It says on Wikileaks.ch that 913/251287 diplomatic cables have been released, so I wonder if the encrypted archive contains the other unreleased ones? Or do you think it's something else entirely that's as yet unannounced?

For all we know, it could be a leaked copy of some movie...

Quote:

Original post by rip-off When I get some time later on I will look into IDEA.

I can't find any of the articles now, but there were at least two newspaper articles that I remember in addition to some guy's private website some years ago.
The scandal story was about all Ascom encryption boxes using IDEA having been built to secretly transmit the key along with every message, which the NSA allegedly used for over a decade to read communications of allied nations.

That wouldn't be a backdoor in the cryptosystem, though, it would be a backdoor in the device. Presuming that they used a well-known and peer-reviewed open-source implementation of AES -- which I'd pretty much take as a given -- the potential for that sort of thing is essentially nil.