I doubt anyone knows if they actually have this technology for certain, but from a technical perspective, does this seem like something the big agencies would have access to? AES has been out almost 20 years, I would find it unbelievable security agencies wouldn't have cracked it by now. Does anyone have any bearing on this?

  • 1
    yeah, 20 years down, only quintillions more to go...
    – dandavis
    Nov 17, 2017 at 21:40
  • @dandavis that's a naïve attitude given all the zero days we've seen Nov 17, 2017 at 21:42
  • 1
    all evidence says they can't. they can't even get in an iphone after all. they typically don't bother cracking encryption: meta alone is enough for a conviction or drone hit authorization, and they can easily implant exfiltratation aids; they don't need to decrypt targeted subject's data when they catch keys and screens.
    – dandavis
    Nov 17, 2017 at 21:49
  • @dandavis you obviously haven't read anything from vault7, you are categorically incorrect Nov 17, 2017 at 21:50
  • 1
    can you link me to the part about them breaking AES? i'd be very interested.
    – dandavis
    Nov 17, 2017 at 21:54

4 Answers 4


Major government organizations which have the task to help their countries companies and protect the local economy still recommend to use AES. I doubt that they would provide this recommendation if they would assume that other organizations might be able to crack AES.

AES has been out almost 20 years, I would find it unbelievable security agencies wouldn't have cracked it by now.

While the age of the algorithms might suggest that somebody must have cracked it in the meantime it might also be an indicator that the authors have created a very good algorithm.

  • Couldn't it just be a zero day like all the other exploits we've seen as of late? Nov 17, 2017 at 21:43
  • 4
    @script8man You're confusing ciphers and programs. A program is usually large, does several things, and usually contains a very high number of programming errors that make it vulnerable. Cryptography is hugely complex, but in the end a serious block cipher is a small algorithm that's been designed over years, with centuries of research in cryptography behind it, and then studied by other experts who didn't find any significant flaw in it, even though it's simple on the surface. Software is always vulnerable, it's less clear-cut for ciphers.
    – Hey
    Nov 17, 2017 at 21:51
  • A zero day refers to any unpublished exploit; regardless of if it's a cipher, a program, or a padlock Nov 17, 2017 at 21:59
  • 2
    @script8man That was implicit in Ydob Emos's comment, which still holds true. A zero day in a well studied encryption algorithm is far less likely that one in an arbitrary software program.
    – Xander
    Nov 17, 2017 at 22:09
  • 1
    @script8man - there is a much higher chance of an implementation flaw in a common AES toolset than in the underlying algorithm itself. The maths behind AES is viewed to be unbreakable with modern technology. Of course if parameters are chosen badly by buggy software anything encrypted with that software may be breakable.
    – Hector
    Nov 18, 2017 at 0:20

In the (highly trusted) opinion of Bruce Schneier:

My guess is that they can't. That is, they don't have a cryptanalytic attack against the AES algorithm that allows them to recover a key from known or chosen ciphertext with a reasonable time and memory complexity. I believe that what the "top official" was referring to is attacks that focus on the implementation and bypass the encryption algorithm: side-channel attacks, attacks against the key generation systems (either exploiting bad random number generators or sloppy password creation habits), attacks that target the endpoints of the communication system and not the wire, attacks that exploit key leakage, attacks against buggy implementations of the algorithm, and so on. These attacks are likely to be much more effective against computer encryption.


EDIT: Also, this has been asked and answered here.

  • And of course Schneier submitted Twofish to the AES contest, so one would assume that he would speak up about any practical attack on the competitor Rijndael if one was known.
    – Arne Vogel
    Jun 25, 2019 at 6:12

Nobody knows except them. This question can't have an informed answer, except in these cases :

  • they can and an insider leaks their method,
  • they can and an unambiguous case of them breaking it becomes publicly known.

From a technical perspective, you have to consider that AES has been out there as a global Internet standard since 1997, and the public cryptography community still hasn't been able to significantly weaken it. There are known, powerful attacks that can weaken some versions of it, but it's still not expected to be broken for at least a decade. It's probably one of the most-studied ciphers in history, and people who are at least as competent as the NSA's researchers still trust it.

Knowing this, the NSA breaking AES would mean that they are a lot more advanced than the academic community, and there is not necessarily a reason for that. If you ignore the common, misleading image of a secret government agency with so much power that it has access to technology 50 years in advance of everyone, the NSA is just a well-funded spy organization with some good researchers, which surely makes them competent, but not nearly enough to be so far ahead of the whole world.

Your sentence

AES has been out almost 20 years, I would find it unbelievable security agencies wouldn't have cracked it by now.

...makes it sound like breaking a cipher is just a question of throwing enough mathematicians and funding at it, until you inevitably find a decryption algorithm. This is not reality. Advances in cryptography mean that modern ciphers are most likely out of reach of even the most competent cryptographers in the world for now, and breaking them implies finding a vulnerability that we aren't even sure exists.

In short, they may or may not have broken it, but nothing indicates that they have been more successful at it than everyone else.


None of the above contemplates backdoors. No national government actually uses standard algorithms in their above top secret programs, deduce your own reasons.

Little doubt in my mind AES was dirty in 1999 when I rejected it for custom curve ECC developed with Certicom which my Gov customer promptly said was not exportable. Nothing classified here. Just made a commercial choice sitting in the RSA offices with guys in dark suits and no business cards.

  • 1
    Backdooring is not "cracking".
    – schroeder
    Mar 28, 2019 at 7:50
  • 1
    It was rejected because it was not exportable? That has nothing to do with the algorithm being dirty, but with international agreements.
    – schroeder
    Mar 28, 2019 at 8:01
  • No national government actually uses standard algorithms in their above top secret programs, deduce your own reasons. <- Any sources for that?
    – Tom K.
    Mar 28, 2019 at 9:07

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .