2

I've heard about ssl strip in the past that allowed for man in the middle attacks to take place against browsers, but those vulnerabilities were patched long time ago. So my question is...are there still any form of attacks that can be used to bypass ssl encryption on sites and perform man in the middle attacks?

Also what are the precautions or preventive measures we can take to guard against such form of attacks?

2 Answers 2

3

Sslstrip is not an attack against SSL. It has never been. It is an attack against HTTP (not HTTPS): it looks for points where some HTTP (hence unprotected) Web site tries to switch to HTTPS, and prevents this from happening. Thus there is no SSL to begin with.

To counter sslstrip (which still works, by definition), Web sites must:

  1. make SSL usage explicit: the human user MUST see the https:// URL in his browser's URL bar, with the padlock icon and so on;
  2. educate users into expecting such explicit SSL usage, and bailing out in case the expected SSL does not happen.

There are no known MitM attacks against properly used SSL. "Proper usage" includes, in particular, a user who does not click through warnings about invalid server certificates. It always comes down to the user.

2
  • You may want to add something about Strict Transport Security as a countermeasure too. Commented Aug 27, 2013 at 20:25
  • Thanks, so according to what you're saying there's no way to sniff my account credentials with a site locally if I'm using https / ssl right?
    – irenicus09
    Commented Aug 28, 2013 at 3:44
0

There are other vectors for MITM in SSL. For instance:

  1. Using a weak cipher with weak hashing function (think DES with MD2, gasp...)

  2. Using an outdated SSL version (eg SSL 2.0, which for some weird reason some sites support even now).

  3. Using a client or server that does not do 1 or 2, but allow doing so (attacker can cause downgrade attacks and force the parties to use a weaker cipher/SSL algorithm, then cracking from there)

These allow an attacker to perform MITM by cracking the encryption or protocol itself, whilst still remaining transparent to the user or server.

And then, there are impersonation attacks with a SSL cert validated by a valid CA. The attacker gets a CA to sign his cert for the target site (which he doesn't own), then sits in the middle of the client and server, impersonating as the server to the client). However, this attack is rare and requires quite a bit of resources, such as a nation state forcing a CA to sign its MITM cert, or by compromising the CA itself (see the DigiNotar incident).

You must log in to answer this question.

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