This leak makes me wonder a bit about the security of AES.
Personally I like Blowfish and RC4 .. many think RC4 is weak but I think it’s fine if properly implemented. It’s very easy to code RC4 for use in embedded systems. I love RC4 for its elegance and simplicity. http://ciphersaber.gurus.org/
ECC is what we need to use for public key, it’s what the NSA uses.
I imagine the NSA uses a lot of custom ASIC chips for code breaking...probably made in their own Fab. I bet NSA would be great at Bitcoin mining.
The ability of NSA to decrypt a particular implementation or type of encryption is tested by foreign adversaries by encoding false info with the system and watching to see if the U.S. takes any action based on that info.
Don’t trust anything but open-source encryption products.
For the most critical data I’d recommend the two parties create a truly random set of data using a noise source like brownian noise. Both parties must hold this data and keep it secure. This allows the parties to add a one-time-pad step to their usual encryption routine. The one-time-pad is unbreakable by any method, even when powerful quantum computers come on line they will have no hope of penetrating a one-time-pad system. The big problem with one-time-pad is you are taken back to the bad old days of the key exchange problem...secret data that must be shared by all users, it’s a drag!
Steganography must still be a huge problem for the NSA since there are nearly limitless ways to implement it. Just a few bits inside a huge data set can hold important info...how do you discern this??
In the comments to the article at http://www.schneier.com/blog/archives/2013/09/the_nsa_is_brea.html, Bruce Schneier, the creator of Blowfish and codesigner of Twofish, said in response to a comment:
Commenter: “On the crypto bits in your guardian piece, I found especially interesting that you suggest classic discrete log crypto over ecc. I want to ask if you could elaborate more on that.”
Bruce: I no longer trust the constants. I believe the NSA has manipulated them through their relationships with industry.
In other words, ECC is probably compromised.
Steganography must still be a huge problem ...