[Full-Disclosure] AOL's Billion SPAM March on Cyberspace

From: Jason Coombs (jasoncat_private)
Date: Sun Mar 16 2003 - 22:54:50 PST

  • Next message: Ben Laurie: "[Full-Disclosure] [ADVISORY] Timing Attack on OpenSSL"

    Aloha, Lonnie.
    
    Your article: "ISPs Seek Bigger Mallet To Eliminate Spammers" caught my
    attention.
    http://www.theledger.com/apps/pbcs.dll/section?Category=COLUMNISTS0203
    
    I'm an information security and computer forensics expert with detailed
    technical knowledge of SPAM and the technology employed by spammers.
    Recently I authored a report on SPAM delivery via AOL -- where a spammer
    gains access to the Internet for the purpose of delivering SPAM to other
    people elsewhere on the Internet. Considering the topic of your recent
    article for The Ledger, I thought you'd be interested in reading this
    report.
    
    AOL is being ridiculous when they suggest that their billion SPAM march on
    cyberspace does any good whatsoever. In fact, AOL is being downright
    deceptive in their assertion that blocking inbound SPAM on behalf of their
    subscribers who use @aol.com e-mail addresses is a virtue: AOL blocks SPAM
    sent to their subscribers without AOL's permission (paid 'advertisements'
    are sent to AOL subscribers with AOL's full support) but AOL does NOT block
    SPAM that AOL users send to people who use OTHER ISP's e-mail services.
    
    AOL may as well capture those billion SPAM messages and relay them to
    non-AOL subscribers because this is exactly what the end-result is of AOL's
    alleged attempts to curtail SPAM. AOL has positioned themselves to be a
    facilitator of SPAM transmission to non-AOL subscribers while simultaneously
    trumpeting their technical triumph over SPAM that originates elsewhere on
    the Internet and is destined for an AOL subscriber's mailbox.
    
    Your readers would be interested to know that anyone with an AOL account can
    send SPAM to any other AOL account and AOL will NOT block it. On the other
    hand, some ISPs are now blocking ALL e-mail that originates from AOL because
    of these very issues.
    
    Sincerely,
    
    Jason Coombs
    jasoncat_private
    
    --
    
    A Report on SPAM Blackholes, Blocking/Filtering, and AOL
    
    For the last month I have purposefully used AOL for SMTP server mail relay
    in order to analyze the real-world impact of blackhole lists. AOL not only
    does not block outbound SMTP from dialup customers, they operate a
    transparent proxy farm that intercepts all outbound SMTP traffic and
    intentionally relays this traffic on to its intended recipient (but not its
    intended SMTP relay point -- you can configure ANY remote IP address as your
    SMTP server and AOL's proxy farm will still do your delivery for you based
    on the MX records present in the destination domain, you need not find an
    open mail relay to exploit nor set up authorized/authenticated SMTP service
    with any third-party service provider in order to relay SPAM through dial-up
    AOL Internet service).
    
    The results have been quite interesting. To summarize, only a few of my
    outbound e-mails have been blocked by blackhole sites in the last month. All
    e-mail sent to mailing lists such as bugtraq has gone through successfully.
    Every rejected message has been returned to me with an explanation (thank
    you, blackhole-enabled servers, your deterministic failure mode made this
    experiment possible because I didn't have to worry about whether my e-mail
    simply disappeared silently and could take corrective action to see that my
    recipient received my message through other channels).
    
    The most interesting failure I encountered was to my own domains. For e-mail
    service we use a third-party service provider, the same provider who does
    our Web hosting on Linux-based servers running Ensim (www.ensim.com). By
    default our service provider refuses all inbound mail delivery based on a
    blocking filter rule (not a blackhole service). This blocking filter
    considers ALL e-mail from AOL to be SPAM and refuses it. This isn't just
    e-mail relayed from a dial-up address block, this is ALL AOL e-mail. No user
    of AOL was able to send e-mail to our domains until we requested that
    inbound filtering be disabled.
    
    It's also interesting to note that my Reporting-MTA FQDN
    "mail.jasoncoombs.com" does NOT have an A record or a PTR or any other DNS
    records associated with it. This doesn't bother AOL's SMTP proxy farm, and
    it likewise did not bother a single SMTP server that relayed or received my
    e-mail during this test.
    
    My conclusion is that blackhole servers and filtering are a terrible way to
    deal with the problem of SPAM. Few people actually benefit from these
    techniques. They introduce unnecessary deterministic failure rather than
    unnecessary nondeterministic failure therefore they offer their own helpful
    work-around instructions. By informing the sender that their e-mail did not
    go through, they automatically ("oracle"-like) produce a comprehensive list
    of target domains that are protected by blackhole services -- a list that
    any spammer would use to relay SPAM from a different point of presence.
    
    Blackhole-enabled services should switch to a non-deterministic failure mode
    that silently kills e-mail delivery. This would have a far greater effect,
    and it would prevent spammers from easily discovering the extent to which
    their SPAM is being automatically discarded systematically. It does not
    appear to be a *good thing* for Non-Delivery Reports (NDR) to be generated
    by blackhole-protected SMTP servers because it turns them into blackhole
    oracles.
    
    And the oracle says very few people use blackhole services.
    Filtering/blocking is a superior solution, and it's very interesting to note
    that AOL is getting filtered/blocked in its entirety by many SMTP servers.
    This causes only minor problems in the real world because NDR are reliably
    generated by AOL SMTP servers and delivered reliably via the AOL mail
    network to the AOL subscriber who attempted mail delivery to a
    filtered/blocked node. Anyone worth communicating with offers a Web site or
    published telephone number that can easily be discovered by senders who find
    themselves blocked, and the net impact is short-term DoS that reroutes
    communication to more reliable failover routes.
    
    ISPs who offer SMTP relay services for end-users need to disclose a
    comprehensive list of domains to which subscribers cannot send e-mail
    because of blocking/filtering or blackhole lists.
    
    ISPs who offer SMTP servers for domain-holders to receive inbound e-mail
    should disclose their blocking/filtering rules that they implement.
    
    Either SMTP-based e-mail service is understood to be extremely unreliable
    over the public Internet, or it is a part of the network's critical
    infrastructure and as such should be operated without filtering or blocking
    of any kind. There are other ways to solve the problem of SPAM than to rely
    on existing flawed technical methods. If we are going to continue to
    implement flawed technical methods, then we need to implement them in
    nondeterministic failure mode so it becomes perfectly clear that NOBODY can
    be certain that their e-mail has been delivered successfully unless they
    request and receive a return receipt or human-originated ACK.
    
    Sincerely,
    
    Jason Coombs
    jasoncat_private
    
    --
    
    The original message was received at Tue, 4 Feb 2003 02:10:06 -0500 (EST)
    from root@localhost
    
       ----- The following addresses had permanent fatal errors -----
    <jasoncat_private>
    
       ----- Transcript of session follows -----
    ... while talking to mail.science.org.:
    >>> MAIL From:<REMOVED>
    <<< 550 5.7.1 Access denied
    554 <jasoncat_private>... Service unavailable
    
    Final-Recipient: RFC822; jasoncat_private
    Action: failed
    Status: 5.0.0
    Remote-MTA: DNS; mail.science.org
    Diagnostic-Code: SMTP; 550 5.7.1 Access denied
    Last-Attempt-Date: Tue, 4 Feb 2003 02:10:13 -0500 (EST)
    --
    Reporting-MTA: dns; rly-ip04.mx.aol.com
    Arrival-Date: Thu, 6 Feb 2003 02:49:00 -0500 (EST)
    
    Final-Recipient: RFC822; <REMOVED>
    Action: failed
    Status: 5.0.0
    Remote-MTA: DNS; <REMOVED>
    Diagnostic-Code: SMTP; 550 5.7.1 Mail from 64.12.138.8 refused by blackhole
    site bl.spamcop.net
    Last-Attempt-Date: Thu, 6 Feb 2003 02:49:37 -0500 (EST)
    --
    Reporting-MTA: dns;mail.jasoncoombs.com
    Received-From-MTA: dns;win2kdev
    Arrival-Date: Thu, 16 Jan 2003 12:38:21 -1000
    
    Final-Recipient: rfc822; <REMOVED>
    Action: failed
    Status: 5.7.1
    Diagnostic-Code: smtp;550 5.7.1 Mail from 172.194.98.240 refused by
    blackhole site dynablock.wirehub.net
    --
    Reporting-MTA: dns;mail.jasoncoombs.com
    Received-From-MTA: dns;win2kdev
    Arrival-Date: Tue, 14 Jan 2003 06:04:02 -1000
    
    Final-Recipient: rfc822; <REMOVED>
    Action: failed
    Status: 5.0.0
    Diagnostic-Code: smtp;554 Service unavailable; [172.191.178.164] blocked
    using relays.osirusoft.com
    
    
    _______________________________________________
    Full-Disclosure - We believe in it.
    Charter: http://lists.netsys.com/full-disclosure-charter.html
    



    This archive was generated by hypermail 2b30 : Sun Mar 16 2003 - 11:09:05 PST