Jump to content

Password security


Farelf

Recommended Posts

Split from another topic

Now that I have read that Linkedin emailaddresses and passwords are obtained by russian hackers, I recognize my three spammed emailaddresses from using at Linkedin. Fits with the location of the spamvertised websites.

Yes, for others who might be wondering:

http://nakedsecurity.sophos.com/2012/06/06...ake-action-now/

and confirmation:

http://blog.linkedin.com/2012/06/06/linked...ds-compromised/

Link to comment
Share on other sites

Wonder if like Yahoo email they are actually going to be bothered. Yahoo don't give a toss

Already they're showing more concern than Yahoo ever did. I abhor Linkedin actually (their unsolicited "updates" because some well-intentioned person has included your details as a contact - AAAaaghh!), but then I have no need for professional networks. I'm sure it has its good points for those who do. I'm sure flagellation, hair shirts and mobile/cell 'phones are good for some folk as well, but I digress.

Back to the case in hand ...

This site undertakes a search of the compromised Linkedin password SHA-1 hashes file:

http://things.h-i-r.net/lipass.html

I don't know anything about the author or his assertions so no independent assurances from me on the security of it all. The assumption is that the "easy to guess" passwords have been cracked already and the full hash in the list replaced with a truncated version. The above page checks both full and truncated versions. The O/P or anyone else fairly sure the account name and password has been compromised can check with little further risk (horses and stable doors etc.). Alternatively, as the author suggests, use some other hash utility (it is easy to find stand-alone ones) to generate the hash value and input both the hash and truncated hash in turn without explicitly trusting the security of the site.

Link to comment
Share on other sites

This site undertakes a search of the compromised Linkedin password SHA-1 hashes file:

http://things.h-i-r.net/lipass.html

I don't know anything about the author or his assertions so no independent assurances from me on the security of it all. The assumption is that the "easy to guess" passwords have been cracked already and the full hash in the list replaced with a truncated version. The above page checks both full and truncated versions. The O/P or anyone else fairly sure the account name and password has been compromised can check with little further risk (horses and stable doors etc.). Alternatively, as the author suggests, use some other hash utility (it is easy to find stand-alone ones) to generate the hash value and input both the hash and truncated hash in turn without explicitly trusting the security of the site.

Linkedin use SHA-1 which hackers have a crack for so no guessing needed

Link to comment
Share on other sites

Linkedin use SHA-1 which hackers have a crack for so no guessing needed

SHA-1 has a mathematical weakness but as far as I know, no actual/real world cracks demonstrated. A critical weakness in a hash is evidenced by a "collision", which is to say two different "words" having the same hash, but that may typically be evidenced in a "partial" hash, not the full thing (but sufficient to indicate a potential crack in the world of cryptography). That was shown for the MD5 hash some time ago (which is why SHA-1 came to be preferred) but I haven't heard of even MD5 actually being cracked (not to say it hasn't). Any evidence for a full crack of SHA-1?

grc.security newsgroup has a note from one member who used a "high entropy" random character generator (no, I'm not exactly sure what that means either but it implies high resistance to brute-force replication) for his password and the hash of it is in the file, but not the truncated hash - indicating (as far as anyone knows) the hackers had stolen it but hadn't cracked it. He changed his password of course but that seems to indicate no world-changing SHA-1 crack is in play. Just a sample of one, some supposition involved, but the best I have for my world view so far, cards on the table.

I understand the "cracking" process used by these Russian hackers is to guess the password (many are easy if some other account details are known or derived) and compare the SHA-1 hash for the guesse(s) to the file value hash. Since, supposedly, Linkedin used "unsalted" hashes (no pseudo-random value added) this will result in a useful number of hits for modest effort. The hashes of those they have supposedly cracked are replaced in their file by a truncated hash, possibly as an indicator for other hackers (can't really call them cryptographers) working on the "project" to move on to another.

Bottom line, if your truncated hash is on the list, you've been taken (definitely), if your un-truncated hash is on the list, maybe not yet, in any event, if either is there, CHANGE your password. If neither is there CHANGE your password (but with a lighter heart).

My opinion.

Link to comment
Share on other sites

SHA-1 has a mathematical weakness but as far as I know, no actual/real world cracks demonstrated.

[snip]

Bottom line, if your truncated hash is on the list, you've been taken (definitely), if your un-truncated hash is on the list, maybe not yet, in any event, if either is there, CHANGE your password. If neither is there CHANGE your password (but with a lighter heart).

My opinion.

Read an article here which reckons 6.5 million Linkedin SHA-1 passwords have been cracked?

Links to a cached google of hackers site where hackers are claiming they have!

More

http://www.computerworld.com/s/article/922...kedIn_passwords

Always pays to get a secure password one can remember

I always recommend use the 1st letter of your name in capital (George)

G

The number from your post box 206

an = sign

THEN your Supersecreteasytorememberpassword

This case the password becomes

G206=Supersecreteasytorememberpassword

This is equal to 304 bit security take well over 300 years to crack

Better still check out a password saver that holds your password on a USB key check my sig

Not all password sites allow the use of a "=" sign

Link to comment
Share on other sites

Yes, they are/were claiming 236,578 out of 6,458,020 passwords effectively duplicated in clear text, that is yielding a hash matching one in the stolen file of hashes in each case. Very impressive however it was achieved** but a long way from the 6,458,020 out of 6,458,020 that a true cryptographic breakthrough crack of the SHA-1 hash algorithm would yield. Maybe this is just semantics - "crack" means/implies different things to different people - I've been taking about it as if it means "break wide open" which is extreme. Anyway, it doesn't matter, no-one should trust a password which has a matching hash on the list (even more so, a truncated hash belonging to one of those 236,578). And, just to be sure, it would be time to change the password even if one's own hash is not on the file, that's just first principles in play.

**I haven't verified the site safety but looks okay to me, see http://crackstation.net/hashing-security.htm for some probable methods. [edit - HA! I just "cracked" a couple myself (verifying through the checking-site link in my earlier post 80969[/snapback]) - sad to say mickeymouse your Linkedin account is almost definitely compromised, one of the 236,578, and you too georgewashington.]

Yes, you are quite correct about long (but personally memorable) passphrases/passwords being the way to go.

http://xkcd.com/936/ as usual puts it in memorable context.

Unique, maximum entropy passwords (of any length through truncation or combination - a different set each time the page is refreshed) are always available from

https://www.grc.com/passwords.htm

... but they're a bit hard to remember, those are definitely a "set and forget" proposition.

Potentially, the problem isn't just Linkedin passwords for those who re-use passwords (as I fear many would, who use a "personally memorable passphrase", assuming all accounts allow long passwords which some don't) - Linkedin is just the weakest link in the current situation, exposing any other accounts, using the same password elsewhere, no matter how secure their files might be, the strength of the base hash algorithm and whether or not they use best-practice "salting" (Linkedin used none at all - the weakest link). Once again Randall Munroe illustrates the proposition (sort of):

http://xkcd.com/792/

Needless to say, anyone using the same password in Linkedin and elsewhere is vulnerable in those other accounts as well.

Link to comment
Share on other sites

[Much snipped]

Yes, you are quite correct about long (but personally memorable) passphrases/passwords being the way to go.

Needless to say, anyone using the same password in Linkedin and elsewhere is vulnerable in those other accounts as well.

Excellent post thanks

Why I use a password saver and never keep my passwords on my computer gives not needed to be remembered passwords like

s]426Txz[4$YG_};?7h??U:2r<2 (your ISP need to accept such characters as well as length, check)

My USB "pen-drive is encrypted' and locked by memorable password. I also use "Thunderbird portable" on same USB

Link to comment
Share on other sites

And great that Linkedin are sending compromised members password-change information by email! Presumably at least the first e-mail will contain a phone number. I wonder how many calls they'll get along the lines of "...ende my motherze mayden nayme was Gherrison...".

Link to comment
Share on other sites

  • 2 weeks later...
...A critical weakness in a hash is evidenced by a "collision", which is to say two different "words" having the same hash, but that may typically be evidenced in a "partial" hash, not the full thing (but sufficient to indicate a potential crack in the world of cryptography). That was shown for the MD5 hash some time ago (which is why SHA-1 came to be preferred) but I haven't heard of even MD5 actually being cracked (not to say it hasn't). ...
Just to add (my knowledge being 'way behind the game) that a collision with the full MD5 hash was demonstrated as early as 2004 (I had forgotten):

http://www.x-ways.net/md5collision.html

Feed each of those hexcode blocks in turn into a hex editor and save as a .txt file and you will have two different files with the same MD5 (but not other) hashes - viz

CRC32 79868bb6

MD5 79054025255fb1a26e4bc422aef54eb4

SHA-1 a34473cf767c6108a5751a20971f1fdfba97690a

CRC32 9fd6208a

MD5 79054025255fb1a26e4bc422aef54eb4

SHA-1 4283dd2d70af1ad3c2d5fdc917330bf502035658

Something not often seen!! Still a long way from a crack as such but times move on, the discovery of the Flame malware earlier this year revealed, in part, a "collision attack" (first demonstrated in 2008) - on the MD5 hash used in a Microsoft Terminal Server Licensing Service certificate - to produce a counterfeit certificate.

Still a long way short of what I would call a comprehensive crack but every bit as effective as one for some purposes. And, according to some, the reason behind the enforced MS update on Windows 7, Vista, XP etc. installations last "update Tuesday" (any "ask-before-updating" settings in user Windows Update settings ignored, most users would probably not have noticed):

Improvements made to version 7.6.7600.256 of the Windows Update Agent

•Hardened Windows Update infrastructure so that the Windows Update client will only trust files signed by a new certificate that is used solely to protect updates to the Windows Update client.

•Strengthened the communication channel used by Windows Update Client to protect it in a similar way.

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

×
×
  • Create New...