TOR stalker safety: SSL MITM attack possible between client and first Tor node?

Discussion in 'privacy technology' started by aqrfz138, Jul 14, 2012.

Thread Status:
Not open for further replies.
  1. aqrfz138

    aqrfz138 Registered Member

    Joined:
    Jul 14, 2012
    Posts:
    3
    Location:
    n
    Tor network chart as I understand it (please correct me if I got it wrong):

    My mouse/keyboard --1--> Tor Browser Bundle <--2--> My ISP <--3--> First Tor node <--4--> more Tor nodes <--5--> Tor exit node <--6--> My SSL enabled website (e.g. bank/e-mail)





    The obvious security issues with the steps, as I understood them so far:

    During step 1: Problems with my computer.
    Either physical ones (planted hardware), built-in issues (such as Absolute Computrace) or malware installed by people with access to my computer or because of online mistakes.

    During step 2 and 3: o_O

    During step 4 and 5: No risk.
    There's no risk since the traffic travels wildly all over the world and is encrypted inside the Tor network, to the nodes as well.

    During step 6: SSL encryptions can get cracked by a stranger using a Man-In-The-Middle Attack.*
    A stranger can get hold of my private information using fake certificates/technical voodoo, in spite of me using SSL/https enabled websites such as my bank or e-mail provider.**





    My question:

    Is it remotely possible for someone, perhaps the people working at my ISP, the government, or the crazy guy I've seen lurking around my garbage cans***, to read my traffic in step 2 or 3 --- despite me using Tor to access an end-to-end SSL encryption enabled website (like a bank login)?

    For instance, by planting a malicious box between me and my ISP (insert an extra instance in step 2) or, for that matter, somewhere at my very ISP if he happens to work there.

    For example, like this:
    My mouse/keyboard --1--> Tor Browser Bundle <--2^modified--> Crazy guy's $40,000 spy-server <--2^modified--> My ISP <--3--> Tor node 1 ... etc

    Would it be, in the example case above, possible to listen in to website traffic?



    Thanks for any help.




    * For an outline of how an SSL Man-in-the-middle (MITM) attack occours, see:
    http://www.windowsecurity.com/articles/Understanding-Man-in-the-Middle-Attacks-ARP-Part4.html

    ** For examples of exit nodes successfully cracking end-to-end SSL encrypted sessions, see:
    https://www.f-secure.com/weblog/archives/00001321.html
    http://www.teamfurry.com/wordpress/2007/11/20/tor-exit-node-doing-mitm-attacks

    *** j/k. I have no crazy guy lurking around my garbage cans.
     
    Last edited: Jul 15, 2012
  2. chronomatic

    chronomatic Registered Member

    Joined:
    Apr 9, 2009
    Posts:
    1,343
    SSL MiTM attacks are possible and Tor doesn't really have anything to do with it. Combating a MiTM attack would be the same whether using Tor or not. Sure, using Tor gives you "more hops" where more things could go wrong, but the MiTM attack would be the same regardless.

    The only way to ensure a MiTM attack is not happening is to verify that the cert of the website is indeed its real cert. The easiest way to do that is to use a tool like convergence, which will check the cert via various perspectives around the Internet.

    And I wouldn't use Tor for banking. I see no point in that. And some banks may have Tor IP's marked as malicious anyway.
     
  3. Tomwa

    Tomwa Registered Member

    Joined:
    Feb 3, 2010
    Posts:
    162
    This. I use TorBrowser as my regular browser (It's a really strict setup) and I wouldn't dream of passing such sensitive information over the Tor network even with SSL.
     
  4. aqrfz138

    aqrfz138 Registered Member

    Joined:
    Jul 14, 2012
    Posts:
    3
    Location:
    n
    Thank you chronomatic for that clear reply. I have a few more questions! :cool:



    To sum this up (please correct me if I'm wrong):
    I need to make sure the SSL certificate used in the session has the correct fingerprints (my browser will NOT tell me if the fingerprints are incorrect).

    In case the fingerprint checks out: This means it is not possible that a MITM attack is in progress and that nobody else is listening in, right or wrong? Regardless of the network (Tor, VPN, ...) I use to connect, the connection must then be secure?

    For it is absolutely impossible to create a situation where the fingerprints of an SSL certificate checks out AND someone else is reading the information passed through the SSL tunnel? For to be able to do that, they'd need to have the private SSL key of the bank/website/secure site.(?)


    Then the following two ways would work to be sure about the SSL certificates:

    1. I go to my bank / call my e-mail provider and ask them what the fingerprints of their SSL certificates are. I write this down on a piece of paper and compare it to whatever my browser tells me (by clicking on the blue/green button in the URL bar) each time I'm presented with one of these important https:// pages.

    2. I use a service/plugin such as Convergence or Perspectives.* (Which is also a question of trust.)



    One more question:
    Does anybody know how it is possible to verify SSL certificates used by applications other than browsers? For example, to see the information regarding my e-mail providers SSL certificate, I just click on the "button" in the left portion of the URL field and I can see the fingerprints, info regarding the certificate etc.

    But how do I verify certificates used by IMAPS in Thunderbird, for example? Or the SSL certificate used to connect to a VPN service? Or the one used in my chat program, in Skype, in ... you name it! As I understood it, all of these things are vulnerable to SSL MITM attacks.

    How do I verify the certificates currently used by the Tor Browser Bundle to connect my PC to the Tor network?**







    P.S. to Tomwa:
    No, I do not use Tor for banking. However, should the fingerprints of the SSL certificate check out when I visit my bank, the connection is just as secure through Tor as through any other network, right?

    If you are stalked by someone, you wouldn't care about what an unknown Tor user or the government will do with your data. The priority becomes revealing as little as possible to the person(s) targeting you.

    Well out safely in the Tor network, it is extremely unlikely that your stalker has the resources to plant MITM attacks between the nodes that you happen to be surfing through that particular day (unless your stalker has government-like capacities, which isn't likely no matter how scary they may want to look like).




    * Convergence: http://convergence.io/
    Perspectives: http://perspectives-project.org/

    ** For others wondering, Tor certificate fingerprints are specified here: https://www.torproject.org/docs/faq.html.en#SSLcertfingerprint
     
    Last edited: Jul 16, 2012
  5. aqrfz138

    aqrfz138 Registered Member

    Joined:
    Jul 14, 2012
    Posts:
    3
    Location:
    n
    Bump.

    Main question: How do I confirm that the Tor client is secure and not the subject of a MITM attack?

    I.e. can I check the certificate fingerprints and SHA1 of the Vidalia client?
     
Loading...
Thread Status:
Not open for further replies.