The once-beloved romsite, Emuparadise, has suffered a data breach. It seems a few months before the site had announced it would be removing all warez, account information of over 1 million users of its users had been breached. Haveibeenpwned, a website dedicated to tracking compromised accounts, has just reported that Emuparadise was involved in such an event. The breach took place in April 2018, though it seems this was only revealed now, as those who have accounts on the Emuparadise forums have been receiving emails this morning from Haveibeenpwned denoting a security issue. 1,131,299 registered accounts have been affected. As always, whenever these data breaches occur, it’s wise to check if you were part of the leaked accounts, and to change your passwords immediately if so.

Emuparadise: In April 2018, the self-proclaimed "biggest retro gaming website on earth", Emupardise suffered a data breach. The compromised vBulletin forum exposed 1.1 million email addresses, IP address, usernames and passwords stored as salted MD5 hashes. The data was provided to HIBP by dehashed.com.

Compromised data: Email addresses, IP addresses, Passwords, Usernames

Source

 

[AkitoTheHedgy]

Ah god damnit, I got caught in this one.
And this, ladies and gentlemen, is why we use a password manager. Every site, no matter how trustworthy or secure, needs its own password. I recommend Bitwarden if you're still looking for one.

Well I started using Bitwarden.... And almost lost my accounts.

 

[PrincessLillie]

Well I started using Bitwarden.... And almost lost my accounts.

Almost lost your accounts? How? I've had literally no issues with it.

 

[AkitoTheHedgy]

Almost lost your accounts? How? I've had literally no issues with it.

Forgot the master password in 10 minutes. thank GOD Roblox and GBATemp had password recoveries.
Really don't care about VRV since I started to use it today.

 

[PrincessLillie]

Forgot the master password in 10 minutes. thank GOD Roblox and GBATemp had password recoveries.

This isn't really something to blame on Bitwarden, but more of yourself for being forgetful. This is also why you make the master password something that's easy to remember, but not too insecure (or just write it down somewhere).

 

[AkitoTheHedgy]

This isn't really something to blame on Bitwarden, but more of yourself for being forgetful. This is also why you make the master password something that's easy to remember, but not too insecure (or just write it down somewhere).

Yeah thats what I just did.

 

[Paulsar99]

Ouch! Talk about kicking someone when they're already down.

 

[Pipistrele]

Yeah thats what I just did.

Then how did you forget it?

 

[ChibiMofo]

Oh God no! Hackers in Kyrgyzstan now also have my passwords for Suprnova, LokiTorrent, isoHunt, and KickassTorrents!

 

[CheeseMan13]

Damn, now they have access to my 10minutemail throwaway

 

[MaxiBash]

Yes, I have been...

Pwned.​

 

[Ericthegreat]

FUCKIN PIRITES GIT GOT WHAT THEY DESERE!!!!!!!!!!! -tumbler/nintendolife probably

 

[AkitoTheHedgy]

Yes, I have been...

Pwned.​

F. Hope you can get all your data back. (wait...)

 

[gudenau]

This is why on any site like this, you should be using throwaway accounts with no relation to your usual identity. Sign up with 10minutemail and use a deliberately shit password - it's not like you should be using the forums on such an obviously skeezy place.



Emuparadise still hosts copyrighted content. They just pretend not to. There's a userscript available via an easy google that rewrites ROM links to directly reference their backend server. So no, unless they actually break said script, they're not safe to link here.

Still bad practice for them to use that hash.

 

[chaoskagami]

Still bad practice for them to use that hash.

Not debating that. Anyone using MD5 in this day and age as a cryptographic hash is a moron. It's about as useful as a CRC32 at this point.

 

[FAST6191]

Why is it bad here though?

If you are using for HMAC or something real then yeah that is bad if you hope to have something like security against a determined actor.

For a salted password then it makes little odds from what I can see.

Am I missing something here?

General idea behind it all.
Storing passwords is bad.
Hashing a password and storing that is better. That way only someone with the password should be able to generate the right hash when the password is entered -- the whole one way and unable to be determined thing.
As it is possible to generate a hash for every combination of characters up to so many characters, all the words in the dictionary, all the most common passwords, all the common words with the obvious e and 3 substitution (and the rest) and put this in one giant table of a fair few gigabytes you can actually do a reverse lookup to see if you have a matching thing relatively easily for what most people have as passwords.
To that end if you add a bit of random data to the password and hash both the pass and the random data (a so called salt) these basic hashes no longer work.
If you use the same salt for every user you can make a new rainbow table for the dictionary+... stuff again with this salt and do the lookup. To that end sites will often try to use unique salts for every user. A popular one being the millisecond that the user joined as it is something they are not going to be able to control and is essentially random.
The hash involved if it is vaguely cryptographically rated (and for all the flaws and ability to tickle a supercomputer to generate collisions then MD5 still kind of is) then not a lot of odds in it. SHA256 or whatever the kids will be suggesting is not that much more computationally expensive and rainbow tables can still be made. The extra space for the larger hashes and generation time is not so much as to render them not viable for these attack vectors.

It is still not complete as you can do things like grab the table, find the salt for an admin account and generate a rainbow table for that salt and hash and do an admin takeover but as far as dumping users and having their passwords readily generated it is not going to make a lot of odds here. About as bad as it gets would be if said MD5 was used to force a collision where a collision is less likely on a better method but given the dictionary approach being the main approach used/feared/anticipated I am still not seeing it.

 

[chaoskagami]

Why is it bad here though?

If you are using for HMAC or something real then yeah that is bad if you hope to have something like security against a determined actor.

For a salted password then it makes little odds from what I can see.

Am I missing something here?

General idea behind it all.
Storing passwords is bad.
Hashing a password and storing that is better. That way only someone with the password should be able to generate the right hash when the password is entered -- the whole one way and unable to be determined thing.
As it is possible to generate a hash for every combination of characters up to so many characters, all the words in the dictionary, all the most common passwords, all the common words with the obvious e and 3 substitution (and the rest) and put this in one giant table of a fair few gigabytes you can actually do a reverse lookup to see if you have a matching thing relatively easily for what most people have as passwords.
To that end if you add a bit of random data to the password and hash both the pass and the random data (a so called salt) these basic hashes no longer work.
If you use the same salt for every user you can make a new rainbow table for the dictionary+... stuff again with this salt and do the lookup. To that end sites will often try to use unique salts for every user. A popular one being the millisecond that the user joined as it is something they are not going to be able to control and is essentially random.
The hash involved if it is vaguely cryptographically rated (and for all the flaws and ability to tickle a supercomputer to generate collisions then MD5 still kind of is) then not a lot of odds in it. SHA256 or whatever the kids will be suggesting is not that much more computationally expensive and rainbow tables can still be made. The extra space for the larger hashes and generation time is not so much as to render them not viable for these attack vectors.

It is still not complete as you can do things like grab the table, find the salt for an admin account and generate a rainbow table for that salt and hash and do an admin takeover but as far as dumping users and having their passwords readily generated it is not going to make a lot of odds here. About as bad as it gets would be if said MD5 was used to force a collision where a collision is less likely on a better method but given the dictionary approach being the main approach used/feared/anticipated I am still not seeing it.

The problem is that passwords are expected to be run through a cryptographically secure hash, e.g. one with minimal to no collisions, and an algorithm in which recovering the plaintext is not feasible. MD5 fails both; generating collisions is easy, and the relative time it takes to recover the preimage is far, far lower than equivalent hashes like ripemd160 or sha1.

But to make this more simple - there are several hashing pools that attempt to recover passwords from these types of public breach data sets, and a 14-character preimage of a MD5 can be found in seven minutes by 16 high end GPUs. How much processing power do you think a hashing pool has? Hint: a lot more than 16 high-end GPUs.

 

[FAST6191]

So that is the determining admin passwords thing covered but rainbow tables, which are the main thrust of this sort of thing, are about the same expense. I am not advocating for using it, just that in this scenario it does not seem as bad as using it for signing certs, HMAC or things that I would slap the person that used MD5 for.

 

[chaoskagami]

So that is the determining admin passwords thing covered but rainbow tables, which are the main thrust of this sort of thing, are about the same expense. I am not advocating for using it, just that in this scenario it does not seem as bad as using it for signing certs, HMAC or things that I would slap the person that used MD5 for.

No. Those numbers are not rainbow tables, but brute force.