It shouldn't come as a surprise to anyone who is familiar with my opinions on encryption strength that I do not hold 512 bit RSA encryption keys in high regard. To put it mildly.
Unfortunately, I continue to be disappointed by the many sites using such weak keys. But I am even more disappointed by the fact that it turns out there are banks that are still using these keys.
Recently I have become aware of two banks/financial services that are using such weak keys:
- The Israeli Bank Hapoalim's US branch (International site)
- Halifax UK's branded credit card site (e.g. used by Amazon UK MasterCard)
Both of these sites are, as of June 15th, 2007, still using 512 bit RSA keys, and have been doing so for about a year, at least.
I have tried to contact these banks yesterday evening and today, but have so far gotten no response (and I also had trouble submitting the comments to the Israeli bank), and to Amazon UK, who answered quickly that they would forward the information internally.
The problem with 512 bit RSA keys is that an attacker, using (almost literally) a handful of computers, can break them in a couple of weeks, at most.
And when an RSA key is broken the attacker can do a lot of things:
- He can read all previous traffic with the site if he have recorded it (if the site uses Ephemeral Diffie-Hellman keys, then those keys must be broken instead).
- If he is able to intercept current traffic to the site, he can pretend to be the site and read all the data passing back and forth between the site and the users.
- If he is able to pretend to be the site, he can also change the information on the fly, if he wants too.
There is absolutely no way to detect any of these attacks through the SSL/TLS protocol, because the sanctity of the site's secret private key is the fundamental pillar of the protocol. That key is used to agree upon the encryption keys that are used to protect the rest of the
communication.
10 years ago, 512 bit RSA was the highest level of encryption the US government permitted foreign users of US manufactured servers and browsers to use, except for financial services, who could use 1024 bit RSA keys, or better. It was widely assumed that the reason for this limit was that the US government could break 512 bit RSA keys relatively fast. The restriction was lifted in 1999/2000.
The proficiency of RSA key breaking is, as I mentioned a couple of weeks ago, rapidly increasing. At present general 640 bit RSA keys can be broken in 5 months using just 100 computers. But recently a special 1020 bit number was factored (the difficulty of factoring large numbers is what provides RSA with its security) in about 11 months using the same amount of computing power. That means that 1024 bit RSA keys are fast approaching their "consume by" date, if it hasn't already been passed.
Today, I would strongly recommend that secure websites, in particular financial sites, should use 2048 bit RSA keys.
The days of 1024 bit RSA keys are numbered: 1296 days, at most. (See the countdown for the up to date number, on the frontpage).
Update, June 18: Halifax has responded that they are in the process of upgrading the server certificate.
Luckly the Israeli branch of Bank Hapoalim uses 1024bits RSA key?https://login.bankhapoalim.co.il/cgi-bin/poalwwwc?dt=924&nls=HEToo bad their standards support is pretty borked.Hopefully will work better in 9.5…
I knew it was getting close, but I didn’t know it was that close, despite reading your earlier mention on the 1020 bit number.Of course, the reason for my comment is that you mention a countdown… where might I find that?Thanks,MichaelBTW – I read your entries whenever they come out. One of the more enjoyable blogs I read, in a cerebral way 🙂
The countdown is on my frontpage, down on the right side.
I was playing with My Opera last night and noticed countdowns… just didn’t realize they were the same. Cool, thanks.
You said “as of June 15th, 2006” — I suppose you meant “as of June 15th, 2007”.
Ooops! Thank you, elitegeek.