Ask HN: What security mechanisms have been compromised?
1 comments
The first thing to understand is that the NSA hasn't (or hasn't proved to have) undermined fundamental security technology. There is no indication of a magic attack that breaks AES, just like no magic attack existed for DES before it. What has been disclosed is the massive scale upon which 2 things have happened:
1) The NSA's comprehensive subversion of information technology vendors into making it easy for the NSA to gain access to data. 2) The scale to which the NSA has been allowed to siphon that data from the most optimal places for an attacker.
One of the first things I point my developers at when they ask about crypto is http://www.daemonology.net/blog/2009-06-11-cryptographic-rig.... Nothing in there has changed. However, you probably want to assume that if you're using, say, the Microsoft CrypoAPI or Java JCE that you cannot be assured that a mechanism to access your data isn't present.
1) The NSA's comprehensive subversion of information technology vendors into making it easy for the NSA to gain access to data. 2) The scale to which the NSA has been allowed to siphon that data from the most optimal places for an attacker.
One of the first things I point my developers at when they ask about crypto is http://www.daemonology.net/blog/2009-06-11-cryptographic-rig.... Nothing in there has changed. However, you probably want to assume that if you're using, say, the Microsoft CrypoAPI or Java JCE that you cannot be assured that a mechanism to access your data isn't present.
So all the attack vectors the NSA has used to date are based upon backdoors (voluntary or otherwise) to existing vendors?
Something I've been confused and/or mislead about is that if root/CA ssl certificates are compromised, then all derivative ssl certificates are also compromised. Is this true? Are these some of the companies the NSA has 'backdoors' with?
Sorry, I'm not even sure about the terminology here -.-
Something I've been confused and/or mislead about is that if root/CA ssl certificates are compromised, then all derivative ssl certificates are also compromised. Is this true? Are these some of the companies the NSA has 'backdoors' with?
Sorry, I'm not even sure about the terminology here -.-
Yes, what we know is that the NSA has compromised many vendors, including it seems some of the CA vendors. So it's possible that a CA could issue the NSA a certificate that could fool you into thinking an NSA surveillance node was a valid site, or that a bit of software was written by a legitimate vendor and is safe to run. The devil is in the details, of course...if the goal is to have a valid, signed Microsoft key, you really need the Microsoft CA to sign it, but 1) the vast majority of people don't check so close is often good enough, and 2) I personally think Microsoft would cave faster than Verisign, so...
And you are correct: if a CA has disclosed their root signing certificate, then the NSA could issue seeming legitimate certificates for pretty much anything.
And you are correct: if a CA has disclosed their root signing certificate, then the NSA could issue seeming legitimate certificates for pretty much anything.
Many articles say, "If you are not an expert in cryptography, don't try." Well, I'm not an expert, and I don't want to fake it. I want to understand to the best of my ability, and make informed decisions.
I am a general developer, and I'm one of many who needs help understanding how to apply recent news to my development practices and production environments.
Can someone summarize what general developers should be thinking about moving forward, that we haven't been considering before (i.e. not password management).