GlyphDec 23, 2022

While you should stop using LastPass in favor of better password manager soon, I think it's important to keep a few things in perspective:

1. This isn't your fault. LastPass fucked up. It was reasonable to trust them, and they betrayed your trust. (Infosec folks: Do not shame people for not knowing this. If we knew and they didn't, that's on us. We should have communicated this better.)

Show threadGlyphDec 23, 2022
2. You are still in a way better position, having used a password manager, then you would have been if you just reused passwords or used some predictable scheme for them. This is NOT some kind of proof that password managers (even cloud password managers) are inherently a bad idea. The alternatives are worse.
Show threadGlyphDec 23, 2022
3. Your passwords are now almost certainly crackable, particularly if you've had an account for a long time. It looks like LastPass has never upgraded the difficulty factor on their KDF, which is very bad. But "crackable" is not the same as "cracked". It is eminently possible to crack a password in a couple of days, but *each* password is going to take at least a few hours on some very high-end hardware; attackers will need to be motivated.
Show threadchrismarget

@glyph

"*each* password is going to take at least a few hours"

This reads to me like you're talking about individual passwords within a user's vault.

If the attackers are working through dictionaries, aren't they attacking the master password (in which case the whole vault is compromised at once)?

Dec 23, 2022 at 9:17pmWeb
Show threadGlyphDec 23, 2022
@chrismarget oh right, they’re all correlated, because it’s one key. Ugh, you’re right. I mean as an expected-value calculation across the whole population the point still stands, but yes.
Show threadGlyphDec 23, 2022
@chrismarget wait no I left it totally ambiguous here, let me pretend I meant “each master password”, there, bow I’m retroactively correct
Show threadchrismargetDec 23, 2022

@glyph I happened to be looking at this a couple of days ago with @stephen0x2dfox. We thought it was interesting that master password key derivation uses the username as a salt.

Seems fine, I guess? The purpose of salt assumes attacker knowledge, so it's not a secret, and guaranteed unique that way.

It felt funny at first though.