Skip to main content
7 events
when toggle format what by license comment
Sep 7, 2013 at 16:52 comment added Nemo @Morty: That graphic is the NYT's, not the NSA's. The actual documents released so far, though redacted, strongly suggest a practical cryptanalytic attack against TLS if nothing else. That means one or more of: the protocol, RC4, 3DES, AES, RSA, DH, or ECDH. I would put my money on ECDH (with NIST curves). Integers and finite fields have been exhaustively and publicly studied for centuries; elliptic curves merely for decades, with most of the attention from cryptographers. Plus NSA has been pushing ECC pretty hard.
Sep 7, 2013 at 16:39 comment added Nemo @PaŭloEbermann: Obviously, "avoid CBC mode" is implied by my assumption. I am talking about CTR, GCM, OCB, etc.
Sep 7, 2013 at 15:35 comment added Paŭlo Ebermann In CBC mode (which is still quite common), the IV only "protects" your first plaintext block. Transmitting it encrypted doesn't help much.
Sep 7, 2013 at 10:30 comment added Morty Given that NSA enumerates the different systems, VPN, SSH m.m. one could speculate that what they have found is particular to those systems. After all, if they had broken e.g. RSA, the insecurity of systems based on this would follow. Also, many of these systems can use different underlying ciphers, so it wouldn't make sense to just write "VPN", you would have to write "VPN with RSA" if RSA was what you had broken.
Sep 7, 2013 at 9:48 comment added Richie Frame People who are concerned about NIST curves should use Shamus Standard Curves instead, or generate their own. If there is an IV concern, negotiating 2 keys is not the answer, rather use a different mode of operation or encrypt the IV with the primary key using a random prepad.
Sep 6, 2013 at 18:38 history edited Nemo CC BY-SA 3.0
added 709 characters in body
Sep 6, 2013 at 18:32 history answered Nemo CC BY-SA 3.0