Forgot your password?
typodupeerror
Security Privacy The Almighty Buck

Webmail and Online Banks Targeted By Phishing Proxies 50

Posted by timothy
from the my-credit-union-won't-even-work-with-mint.com dept.
An anonymous reader writes "Netcraft confirms a recent increase in the number of malicious proxy auto-config (PAC) scripts being used to sneakily route webmail and online banking traffic through rogue proxy servers. The scripts are designed to only proxy traffic destined for certain websites, while all other traffic is allowed to go direct. If the proxy can force the user to keep using HTTP instead of HTTPS, the fraudsters running these attacks can steal usernames, passwords, session cookies and other sensitive information from online banking sessions."
This discussion has been archived. No new comments can be posted.

Webmail and Online Banks Targeted By Phishing Proxies

Comments Filter:
  • Why HTTP? (Score:5, Insightful)

    by AK Marc (707885) on Saturday February 16, 2013 @09:09PM (#42925133)
    Why bother with HTTP? Plenty of malware gets signed certs. If you are messing with malware, change the root certs on the machine (assuming your malware installing the proxy has root), and use HTTPS to www.citibank.com. The user won't know the difference. It'll show up as a valid cert to the right domain, and the proxy can re-encrypt it and use the unencrypted username and password submitted to it. Plenty of corporates do this and have the ability to sniff 100% of employee traffic, even encrypted, because it's all signed and trusted certs, there will be no warnings, though you can inspect the cert for trusted sites, and you'd have to verify DNS or certs for every secure site, which breaks all the usability models. If it takes root to get the malware on in the first place, the hackers screwed up big if they didn't make it work for HTTPS.
    • Re:Why HTTP? (Score:5, Insightful)

      by karnal (22275) on Saturday February 16, 2013 @09:34PM (#42925237)

      Path of least resistance at this point. What's easier, getting a malicious PAC script installed, or getting the same PAC script installed as well as having a user sign off on an invalid certificate?

      Admittedly, getting someone to blindly click "yes" to accept the bad certificate isn't difficult, but if it doesn't pop at all - all the better for the malicious person on the other end.

      • by AK Marc (707885)
        If you've got root, would they have to have the user click again to install the cert?
        • If you've got root, would you bother with HTTP interception at the network level?
          • by AK Marc (707885)
            If you want to intercept data and get it off the computer with no possibility of detection for transmitting it back home, it's great. If your detection software is aimed at blocking unauthorized transmissions, then it's a great attack vector.
    • Re: (Score:2, Insightful)

      by Anonymous Coward

      Why redirect the traffic at all? Why not just use a key logger and grab credentials that way? Most banks and webmail don't use two factor authentication.

      • by AK Marc (707885)
        Well, my bank (Bank of America) makes me click on a photo. So a keylogger would see that I clicked at 40,140 or whatever, but not know what I clicked on. The proxy would be able to capture the submit response, and use that to determine the full login credentials.
        • If it's tailored to bank-credential stealing, taking screenshots on each mouse click isn't exactly out of the question.
        • by AmiMoJo (196126) *

          Couldn't it just capture the screen when you click? That's how the old technique of making you pick from drop-down menus was circumvented.

          • by AK Marc (707885)
            I've never seen any capture that works well for typing "ass" click at the front "P" [end] "word". And there's also the one that counts the cadence. I type my password with a particular rhythm. But a logger could map the cadence and replay it, but none do, so far as I've seen, so the ones that track not just what you type, but how, are safe for the moment.
      • by Mojo66 (1131579)

        Why redirect the traffic at all? Why not just use a key logger and grab credentials that way?

        Because a keylogger would alert anti-virus Software, while a PAC script wouldn't.

      • by allo (1728082)

        password + TAN isn't two-factor?

    • Re:Why HTTP? (Score:4, Insightful)

      by manu0601 (2221348) on Saturday February 16, 2013 @10:45PM (#42925513)

      Why bother with HTTP? Plenty of malware gets signed certs.

      The attack described here does not involve malware. On WPAD requests seen on DHCP or DNS, just inject a WPAD reply with a malicious PAC script and you are done.

  • by Anonymous Coward

    Oh my God. This is serious.

  • I have an issue with the so called, "session cookies."

    While they are a part of online the presence, non of their behavior would be stomached in actual day-to-day life.
    So the issue is that we've got two set's of paradigms. An online one where you can be tracked by default and a real life
    one where you have to be explicitly informed if one is to monitor your every activity.

    Sad, indeed.

    • by fostware (551290) on Saturday February 16, 2013 @10:39PM (#42925485) Homepage

      A session cookie is just like a case number, it may be used to speed up communication between departments or sections of your website. Whenever I'm in a queue, there's always a ticket I hold to identify where I am in the queue, what my wait time is, possibly referenced by their third party SLA/QA company, and it could be tied to my Account Number when I get to the counter. It's stomached in real life, because it brings order to what could be chaos, and makes our lives that little bit simpler.

      Secondly you must be rather naive to think permission is required to monitor your *every* activity. Through various laws, mutual sharing agreements, and straight greed, there's a wealth of information for people to gather and misuse. While they limit "personally identifiable" information, they gather everything they can and assign it their own ID. It then only takes a little homework to link the ID and your real ID together, and its just this last step which is prohibited in these privacy clauses.

      • " Whenever I'm in a queue, there's always a ticket I hold to identify where I am in the queue, "

        In the US, we don't get into queues. Instead, we just stand in line. No numbers, please. We depend on our ferocious looks plus the possibility that we might be carrying a concealed weapon to hold our place in line. In some cases, people depend on their super great looks, plus the possibility that they might be carrying a concealed weapon. No other identification is needed, thank you.

        As for tracking a person

    • Sure. Let's just get tracked in real life too. Or maybe we should avoid the tracking and do away with session cookies, thereby logging you out every time you want to make a forum post on the Web. Neither one sounds good, and HTTP needs to be scrapped anyway. It's good at what it does, and that includes the issue with sessions, which is why sessions are so useful to those with malicious intent.
      • thereby logging you out every time you want to make a forum post on the Web.

        Why would that be? If you are logged in, you stay logged in, cookies or not. But then, if you are logged in you can already be tracked by your user credentials, so all this becomes moot...

        • In order to make a post, you need to be logged in, assuming anonymous/guest posts aren't allowed. How does the server keep track of the fact that you're logged in and who you are? Your time spent on that site is called a session, and the server sets a session cookie to say, "Okay, you've already logged in, so I'll save you the trouble of doing it again." Some sites use that information to reveal things like user preferences, which wouldn't ordinarily be available to a guest user.

          The only alternative I ca

          • How does the server keep track of the fact that you're logged in and who you are?

            By your user name and password.

            Your time spent on that site is called a session, and the server sets a session cookie to say, "Okay, you've already logged in,

            If you are logged in, the server sees your username and password, then why does it need anything else?

            so I'll save you the trouble of doing it again."

            Most (all?) browsers only prompt for a username/password a single time, and then keep in in memory for further requests from the same site and realm.

            The only alternative I can think of is the server responding to your log-in action with "Okay, I got your IP address, so I'll just make a note in my database that you logged in from IP address [IPaddr].

            Why not simply use your login and password?

            • I agree that's perfectly fine, aside from the whole bit about security of saving a password (store it in a secure manner and each time check it against the secure form stored in the user database). How does it save that information in a persistent way that uniquely identifies the computer you're using? That's the role sessions fill. Without the data persistence, refreshing the page would just show you the page as if you weren't logged in. Without the unique identifier (e.g. a session cookie), how does t

              • I agree that's perfectly fine, aside from the whole bit about security of saving a password (store it in a secure manner and each time check it against the secure form stored in the user database). How does it save that information in a persistent way that uniquely identifies the computer you're using?

                I guess, in a hash table keyed by hostname and realm? Probably the first browser did it that way in any case, although nowadays it's probably in a more evolved data structure...

                Without the data persistence, refreshing the page would just show you the page as if you weren't logged in.

                That's why pages are sent with a Vary: Http-Authorization tag. This tell caches, including the one in the browser itself, that they should include authorization in the cache key for the page. This is to avoid that a cached unauthenticated page (or worse: authenticated for a different user...) would show up after the user has been aut

    • by fatphil (181876)
      A session cookie is little different from a train ticket.
      You use it to prove you're someone who's performed some kind of transaction in the past, and it's only valid for a short period of time.

      I have a bigger problem with non-session cookies, and third-party cookies. Those you have to carry around for ever, and you have to accept them from and show them to who-the-hell-knows-whom.
      • Which cookies do you "have to accept"? And, which ones must you keep forever?

        I accept few cookies, almost none of those infamous ever-cookies, and they are almost universally deleted when I close my browser.

        https://addons.mozilla.org/en-US/firefox/addon/betterprivacy/ [mozilla.org] One among several tools that are useful when preventing trackers from tracking you.

        • by fatphil (181876)
          My "accept" was from the viewpoint of the very lowest layers of the HTTP communication. You can't stop them being sent to you (except by not loading the payload they're in the headers for). You might chose to bin them and never return them to their sender, which is what I do. But the " Set-Cookie:" line still polutes the HTTP headers, and there's nothing you can do to stop that. If you're accepting the headers and the payload, which you are, then, as by accepting the whole you've accepted all parts, you've
  • by Anonymous Coward

    With the introduction of LUA, Netcraft confirms that NETBSD is dead because it allows proxy auto-config scripts in the kernal!

  • by retroworks (652802) on Saturday February 16, 2013 @09:46PM (#42925287) Homepage Journal
    In the future, everyone will be a billionaire for 15 minutes, until their ill-gained 15 minute life savings is phished by the next billionaire. The bank account hijack will rotate around and around, shared by everyone in the world, boosting all our credit ratings... momentarily.
    • Thanks for posting that here, where some unscrupulous writer might want to grab the idea to use for a novel. :)

    • Not sure how your bank works, but with mine (multiple banks in different countries), the password only gets you in to view info and some other minor stuff. If you actually want to move money it's a one time only two factor authentication system. Auto proxies aren't much use against this, so I guess I'll be hanging onto my billion dollars a bit longer.
  • DNSSEC would be nice (Score:4, Interesting)

    by Anonymous Coward on Saturday February 16, 2013 @09:52PM (#42925313)

    It'd be nice if one could bypass the various CA's and enforce HTTP Strict Transport Security (HSTS) as well. I could then have an unlimited number of certificates for my domain and sub-domains. I would see that owning the .com or whatever domain would go up in price though since Verisign and others still want their money somehow and someone still signs the root somewhere.

    It'd just be nice to be my own CA for my own domain anyway.

  • by jehan60188 (2535020) on Saturday February 16, 2013 @10:02PM (#42925345)

    avoid banking at work? i always figured that was more secure than at my own home (shared wifi with two room mates- neither seem tech savvy, but you never know.; WPA2 but short password)
    it sounds like if my room mates computers are compromised, i can get phished with the method?

  • by pcjunky (517872) <walterp@cyberstreet.com> on Saturday February 16, 2013 @11:28PM (#42925649) Homepage

    SSL will, if correctly setup, will prevent this. Unless you click through all the warnings your browser shows regarding the sites certificate.

    • SSL will, if correctly setup, will prevent this. Unless you click through all the warnings your browser shows regarding the sites certificate.

      You are aware, I'm sure, that your first sentence is handily negated by the second one?

    • by Anonymous Coward

      As long as one of the hundreds of CAs your browser trusts doesn't make a mistake, sure. But we've seen that some CAs will sign anything. I expect inside jobs at CAs will also become more popular over time.

  • One word: PGP

    All this talk of SSL and signed certs. Band aids. If every person and corp used PGP none of this would even be a problem would it?

A rock store eventually closed down; they were taking too much for granite.

Working...