Hacker News Reader

Comment by threatofrain | Hacker News Reader

> But the data also contained sensitive text messages, such as two-factor codes and other security messages, which may have allowed anyone viewing the data to gain access to a person’s online accounts. Many of the messages we reviewed contained codes to access online medical services to obtain, and password reset and login codes for sites including Facebook and Google accounts.

> The data also contained usernames and passwords of TrueDialog’s customers, which if used could have been used to access and impersonate their accounts.

View on HN

Hence why 2FA tokens and reset links should have a short window of validity, and why shallow information such as knowing account name, or address, or mothers maiden name, should not be used for sensitive purposes.

It is considered a security best practice to use a 2fa app (like Authy), or a physical token instead of SMS. This way when the token generator is lost you at least notice it in a reasonable timeframe.

For critical systems where 2fa is enabled I also do a "simulated" device loss, where I go trough all the steps at least once that I would have to do in case of a real device loss (fetching backup codes, revoking token, resetting password, adding new token).

This way I do not have to constantly worry about losing critical items, because I'm prepared for the worst and I can be calm.

It also helps to have all your data encrypted at rest with Bitlocker or Veracrypt (hard to enforce this rule on yourself for pendrives, but oh well..).

Thanks! That (simulated device loss followed by all the steps you mentioned) is something I plan to do every now and then but so far I've never gotten around to actually doing it. Encouraged by your post think I'll finally do it (some day, hopefully sooner rather than later).

This brings up some interesting technical questions: how long is too long and what is a deep question for identity?

Something like a minute or two is probably the upper bound. Look at TOTP timers for a sane default (often like 30 seconds), and add a bit more for SMS delivery delays. You can always try to send a new code when the old one expires.

But also: basically the entire security world has been recommending against SMS 2-factor for years, because it's so incredibly easy to steal access. Don't use SMS if at all possible, don't have it as a backup (because "a backup option" == "an option" == "you are only as strong as your weakest link"), etc. Avoid it entirely.

One time I did a "forgot password" reset on an old email account. Apparently young me thought it was a good idea to choose the 'pick your own question' thing and the question I chose was "What?"

...to this day I still don't remember what the answer was.

The dangers of being clever. Briefly used a given LastPass, used another setup for several years, then wanted to go back to that old LastPass for something.

The LastPass password hint I'd set:

> Don't forget your LastPass password.

It wasn't some oblique hint. It was an FU from 20-something me to 30-something me that I distinctly remembered sending as soon as it saw it.

That made me smile :)

Depending on how young, there’s a decent chance the answer was some variation of “chicken butt”

Hah!

I have an 11 year old. You're sooo right about the chance that's it...

I always give non-sense answers to security questions and store the question and answer in my password manager.

I did that for my old college student self service account in 2004. The question was slightly better: "Who's yo daddy?"

I couldn't guess who my daddy was in 2004. Wasn't my real dad nor was it me ¯\_(ツ)_/¯

Your imaginary friend's mother's maiden name.

Which also happens to be an extremely rare maiden name - i.e. not Smith but "Vortangula" or "Brinderbass".

> Vortangula Brinderbass

Thank you for the new D&D character name.

Hence why SMS is not a valid part of any authentication system that should be secure.

This article massively overhypes the breach. 2FA and password resets have a very short window of validity. The database contained historical messages and did not operate in real-time.

How do you know it didn’t operate in real-time? The article doesn’t contain enough information to know.

If it wasn’t real time then how would short lived auth solutions even work?

I understand they mean the database

Perhaps, but you're also making a lot of assumptions about the security practices of companies that send you info via SMS.

To your point: I can count on two hands the number of companies I've encountered that iterate HOTP on use rather than on issuance.

...which means there are bound to be a few stale but still active SMS codes lingering in there from people who attempted but did not complete authentication e.g. because they entered the wrong number or didn't have access to the number they attempted to use when signing in. Services impacted are any which allow for users to authenticate with _just_ SMS HOTP and which don't expire unused codes. That number is unfortunately high enough for me to think that this is equatable to a small credential breach.

Reply on news.ycombinator.com