Vulnerability Development mailing list archives

Re: Secure Yahoo logins


From: "Steve Bremer" <steveb () nebcoinc com>
Date: Wed, 28 Aug 2002 11:34:21 -0500


If you can perform a mitm attack, there is no doubt you can read the 
traffic since you'll actually be decrypting it.  However, performing the 
mitm attack is the problem here.  With the exception of the recent 
browser flaws, it's not necessarily an easy attack.

You would have to find a way to get their browser to go to your 
machine in the first place.  This would require some sort of ARP 
spoofing (in which case you would need to be on their local network 
segment) or some sort of DNS diversion which would probably 
require you to have control of the necessary DNS servers.  At that 
point, you'd also have to deal with the warning the user would 
receive about an unsigned certificate being used (although this may 
not be difficult since many people click through the warnings).

There may be other (easier) methods as well, but I'm not aware of 
them.

Please enlighten me if I'm wrong.

Steve

On 28 Aug 2002 at 1:36, Nick Jacobsen wrote:

I just love this...  You are telling me that I can't sniff information
from an SSL session using a mitm attack?  the whole point is that you
are in the middle...  i.e. client connects to you and you coneect to
server, therefore the SSL session with the server is between you and
the server, not the client and the server...  you simply pass
everything on to the client as well, acting as the remote server... 
Try using ettercap, then tell me I am wrong...


Nick J.
Ethics Design
nick () ethicsdesign com
ethics () netzero net

----- Original Message -----
From: "David Thiel" <lx () redundancy redundancy org>
To: "Nick Jacobsen" <nick () ethicsdesign com>
Cc: <vuln-dev () securityfocus com>
Sent: Tuesday, August 27, 2002 9:06 PM
Subject: Re: Secure Yahoo logins


On Tue, Aug 27, 2002 at 08:36:40PM -0700, Nick Jacobsen wrote:
it supports SSH(Secure Telnet)

SSH is not even remotely like "Secure Telnet".

and SSL(HTTPS) decryption and sniffing, as

Only if you have the server's keypair.

I guess my main point is that if you are having your users log in
using "secure log in" for the express reason of making it so their
password
cannot
be sniffed, it is pointless, as anyone can STILL sniff it!

There's a higher difficulty level involved with MITM attacks, and
measures can be taken to prevent and/or recognize such attacks. SSL
is not a panacea, but it's a useful layer of security.  The fact
that MITM attacks exist is not proper rationale for abandoning the
use of encryption.





Current thread: