4 ms·

Can someone explains to me how Graham claims he can decrypt the intercepted traffic? The proxy communicates securely with the intended website. It's just the br

by halayli 12y ago

Can someone explains to me how Graham claims he can decrypt the intercepted traffic? The proxy communicates securely with the intended website. It's just the browser <-> proxy communication that's vulnerable but that's local on the machine, no ?

shawkinaw 12y ago
I'm wondering the same thing. As far as I understand, the proxy is local to the machine, so HTTPS traffic over Wi-Fi should be past the proxy and therefore encrypted using the real certificate.
- Animats 12y ago
  The installed backdoor certificate is trusted as a root certificate. Its private key is contained in the MITM software, and is now known publicly. So anyone can now create phony certs signed by the backdoor cert, and Lenovo machines accept them as valid. Here is such a page: https://badfish.filippo.io/yes.png https://badfish.filippo.io/yes.png That's an image of the word "Yes" signed with the Superfish certificate. If your browser shows that image without warnings about an invalid cert, the backdoor exists.
  - halayli 12y ago
    Right but this only means you can decrypt data coming from websites using a starfish cert. It doesn't mean you can decrypt your bank traffic because you have this proxy installed which is what Graham is claiming.
    - Animats 12y ago
      Yes, it does mean others can decrypt your bank traffic. Here's how this type of MITM attack works. Situation: user is using laptop in public location with WiFi. Between WiFi device and net is a computer with MITM software. Client laptop requests "https://www.bigbank.com" https://www.bigbank.com". MITM box gets HTTPS request, sees it is for "bigbank.com", and generates a fake cert for that site. It then uses the Superfish root cert to sign the fake cert. MITM box acts as server for that connection and sees the user's traffic in the clear, unencrypted. The Lenovo client laptop sees a valid cert chain descending from the Superfish cert installed by Lenovo. The user sees a green bar and lock icon. MITM box then opens an HTTPS connection to "https://www.bigbank.com" https://www.bigbank.com", and acts as client for that connection. The two connections are connected together as a proxy, so that the user sees what looks like a valid HTTPS connection. The MITM box can log everything, including bank passwords. There's even open source software for doing MITM attacks: https://code.google.com/p/subterfuge/ https://code.google.com/p/subterfuge/
      - halayli 12y ago
        If you have a computer between the user and net, then yes all bets are off because you can generate certs the browser will trust.
        Animats 12y ago
        Only if the root cert store of the user's machine has been tampered with. If you have a valid cert store, you can detect MITM attacks on HTTPS connections.
        halayli 12y ago
        yep. I was referring to superfish's case.
- MichaelGG 12y ago
  The proxy accepts fake certificates so you can MiTM anyone. It does not properly verify the remote site's certificate.
  - halayli 12y ago
    Yes, but MiTM isn't the same as sniffing packets over a wifi in a cafe, which what Graham claimed.
    - MichaelGG 12y ago
      If you can read (sniff) WiFi in a cafe, you can write (MiTM), so the difference isn't really important, is it?
      - halayli 12y ago
        Usually sniffing refers to capturing packets. But yes if you can read and write then you can definitely decrypt the traffic since you can provide the user with a trusted cert.

by halayli 12y ago

shawkinaw 12y ago

I'm wondering the same thing. As far as I understand, the proxy is local to the machine, so HTTPS traffic over Wi-Fi should be past the proxy and therefore encrypted using the real certificate.

Animats 12y ago
The installed backdoor certificate is trusted as a root certificate. Its private key is contained in the MITM software, and is now known publicly. So anyone can now create phony certs signed by the backdoor cert, and Lenovo machines accept them as valid. Here is such a page: https://badfish.filippo.io/yes.png https://badfish.filippo.io/yes.png That's an image of the word "Yes" signed with the Superfish certificate. If your browser shows that image without warnings about an invalid cert, the backdoor exists.
- halayli 12y ago
  Right but this only means you can decrypt data coming from websites using a starfish cert. It doesn't mean you can decrypt your bank traffic because you have this proxy installed which is what Graham is claiming.
  - Animats 12y ago
    Yes, it does mean others can decrypt your bank traffic. Here's how this type of MITM attack works. Situation: user is using laptop in public location with WiFi. Between WiFi device and net is a computer with MITM software. Client laptop requests "https://www.bigbank.com" https://www.bigbank.com". MITM box gets HTTPS request, sees it is for "bigbank.com", and generates a fake cert for that site. It then uses the Superfish root cert to sign the fake cert. MITM box acts as server for that connection and sees the user's traffic in the clear, unencrypted. The Lenovo client laptop sees a valid cert chain descending from the Superfish cert installed by Lenovo. The user sees a green bar and lock icon. MITM box then opens an HTTPS connection to "https://www.bigbank.com" https://www.bigbank.com", and acts as client for that connection. The two connections are connected together as a proxy, so that the user sees what looks like a valid HTTPS connection. The MITM box can log everything, including bank passwords. There's even open source software for doing MITM attacks: https://code.google.com/p/subterfuge/ https://code.google.com/p/subterfuge/
    - halayli 12y ago
      If you have a computer between the user and net, then yes all bets are off because you can generate certs the browser will trust.
      - Animats 12y ago
        Only if the root cert store of the user's machine has been tampered with. If you have a valid cert store, you can detect MITM attacks on HTTPS connections.
        halayli 12y ago
        yep. I was referring to superfish's case.
MichaelGG 12y ago
The proxy accepts fake certificates so you can MiTM anyone. It does not properly verify the remote site's certificate.
- halayli 12y ago
  Yes, but MiTM isn't the same as sniffing packets over a wifi in a cafe, which what Graham claimed.
  - MichaelGG 12y ago
    If you can read (sniff) WiFi in a cafe, you can write (MiTM), so the difference isn't really important, is it?
    - halayli 12y ago
      Usually sniffing refers to capturing packets. But yes if you can read and write then you can definitely decrypt the traffic since you can provide the user with a trusted cert.