Message-ID: <20040304123314.TDIK411419.fep03-mail.bloor.is.net.cable.rogers.com@BillDell>
From: full-disclosure at royds.net (Bill Royds)
Subject: E-mail spoofing countermeasures (Was: Backdoor not recognized by Kaspersky)

 Having a MS record would not eliminate spam coming from users validated on
the sending server, but it would identify the server that it comes from as
"knowing" the sender name. Compromised client boxes would need to use the
ISP mail server to send mail, rather than  spewing it directly, since the
servers allowed on the MS entry for that domain would not include the client
host.
  Either the ISP owing the server blocks spam spew or that ISP gets a
blackhole block that would be very effective. 
   Yesterday I inspected the spam I had in my spam bucket for kinds of
actual senders (last sender on Received header for my ISP). Of 11 spam
messages in the last hour, 9 were from compromised machines sending
directly. If they had to send this stuff through their ISP (comcast,
telstra, swbell etc.), they would  be blocked fairly quickly. The envelope
from address was often Yahoo, so the ISP would block on this as well.
  Requiring MS entries would not block spam or viruses immediately but would
help make RBL lists more effective and prosecution of spammers easier
(easier to trace a registered user of an ISP).

-----Original Message-----
From: full-disclosure-admin@...ts.netsys.com
[mailto:full-disclosure-admin@...ts.netsys.com] On Behalf Of Nick FitzGerald
Sent: March 4, 2004 3:00 AM
To: full-disclosure@...ts.netsys.com
Subject: RE: [Full-Disclosure] E-mail spoofing countermeasures (Was:
Backdoor not recognized by Kaspersky)

"Bill Royds" <broyds@...ers.com> wrote:

<<snippage>>
>   Using authenticated SMTP, this would still allow a different return
> address in headers since envelope from would be user who authenticated to
> SMTP server. But it  would prevent spoofed email (although spam would
still
> arrive, it could be tied to actual sender, allowing things like CAN-SPAM
to
> work).

Wrong.  It would, at best, identify the sending _machine_, not the 
"actual sender".

There is far too much prior art in the Windows malware armory to not be 
aware of how easily an agent program on a "compromised" Windows box can 
steal whatever configuration and authentication data it may need to 
"properly" send mail "just like" the user's preferred MUA.  Just 
because, of late, spam and mass-mailing viruses have used randomized 
From: and SMTP envelope FROM addresses does not mean thay have to 
continue to do so, nor that not doing so will necessarily be less 
effective for them...

These are important considerations to not overlook despite the fact 
that the SPF, etc pushers make a habit of ignoring such.  Further, 
several IRC bot-nets in tens-of-thousands of active bots size range 
have already been found and there are probably several million such 
compromised mnachiens out there waiting for the fateful order to "wake 
up" and answer the call of their "master".

SMTP "sender authentication" is a far less trivial problem to solve 
that the SPF, aller-ID, etc folk would have you believe (and, of 
course, they don't like us pointing out that their preferred 
"solutions" are already doomed to failure).


-- 
Nick FitzGerald
Computer Virus Consulting Ltd.
Ph/FAX: +64 3 3529854

_______________________________________________
Full-Disclosure - We believe in it.
Charter: http://lists.netsys.com/full-disclosure-charter.html

Powered by blists - more mailing lists