3 ms·

The article focuses on social risks, but isn't this just a gaping MITM hole? If we're talking about text messages, they're not encrypted.

• gosho 3y ago
  Telegram itself is not encrypted, unless you use the somewhat obscure "secret chat" feature, which few people if any use
  • Nextgrid 3y ago
    This has nothing to do with Telegram chats, it’s about sending verification SMSes (either for new user number verification or 2FA).
    • lxgr 3y ago
      It has a lot to do with it: Since Telegram is not encrypted and additionally stores all chat history server-side, a compromised login process has a much higher impact since it also compromises past message content, not just those received after the breach.
• Nextgrid 3y ago
  It could be that the login process is two-way, first establish a session and get a token, then a 2FA code is sent which needs to be presented alongside that session secret, thus an attacker merely having a 2FA code (without the associated session token) can’t use it.
  • lxgr 3y ago
    Maybe there is an optional way to enable multifactor auth for Telegram, but in my experience it isn’t the default at least: Receiving and entering the SMS-OTP is everything I need for full account access, including to historic messages.
    • Nextgrid 3y ago
      Yes, my point is that internally the client could be first obtaining a session token and then requesting a 2FA code, and then submitting both. This means an MITM that only has the code wouldn’t be able to complete the flow.
      • lxgr 3y ago
        How is the session token distributed? What prevents the attacker from just requesting one themselves?
        • Nextgrid 3y ago
          The attacker can of course initiate a new login flow (thus getting a session token), but this is rate-limited, would invalidate the previous flow, might alert the user (if the user is logged in on other devices, they get notifications when someone tries to login) and relies on this MITM being picked to send the 2FA code to the user again (they might only use one sender per receiver to thwart this attack, or use IP/fingerprint heuristics to avoid pairing potential attackers & victims together).
          • lxgr 3y ago
            Sure, I hope they take at least that precaution. It’s still a much too high risk in my view, given the sensitivity of a long-term historical communications log that’s on the other side of that authentication. I’m also not sure that these are effective deterrents: The people trying to brute-force my Instagram password certainly don’t care about sending me a warning email and locking my account once per day, for example.
            • Nextgrid 3y ago
              Obviously if you want high security you can enable an actual password on your account. I’m not convinced this is more insecure than sending it though conventional means - telecoms are well and truly compromised and don’t give a shit about it. If you’re gonna get dubious security anyway, at least you can save some money by not having to pay for it. Locking your Insta account means they’re effectively only getting a couple attempts per day - the rate-limit is working at thwarting targeted attacks towards your account. The reason it keeps going on anyway is because they aren’t targeted attacks and are hoping to win the lottery and get into any account by sheer luck (just to spam - this is not a targeted attack and they’d have no interest in your chat history).