Pages:
Author

Topic: TradeHill – Security Update – Round 1 PCI Compliance / Business Verification etc (Read 5157 times)

legendary
Activity: 1008
Merit: 1001
Let the chips fall where they may.
It is only off-topic because Tradehill does not use MD5 Hashing, I can't find what hashing they do at the moment.

However, if the database is compromised somehow, account locks after failed login attempts won't help much. That is why you need to choose a secure (likely hard to remember) password. It doesn't matter how convoluted the hash function is; attackers will have the time to do a dictionary attack on their own machines.

That said, (salted) hashing of the passwords is better than storing them in clear-text. It means that most users have time to change their passwords once they learn about the breach. Hopefully Tradehill won't have such a breach. Smiley
sr. member
Activity: 1344
Merit: 264
bit.ly/3QXp3oh | Ultimate Launchpad on TON
Edit: looks like you still have to have knowledge of both messages to generate a collision.
I think you actually have to be able to control both messages to generate a collision - that's actually the definition of one. In order to be able to generate a second message that gives the same hash as an existing message you need a preimage attack, and I don't think those are practical against MD5 yet.

I think both of you have gotten a bit off topic here and missed one of the finer points.

Collisions don't matter here since Tradehill will lock your account if you try to login too many times.
sr. member
Activity: 1344
Merit: 264
bit.ly/3QXp3oh | Ultimate Launchpad on TON
I'd like to see the site log you out after x amount of time of inactivity.
I've rebooted my system several times and have yet to be prompted for a new password when I go to the site.


We've received feedback from users that love not being logged out and more that would prefer the additional security.
We've evaluated the situation and decided to implement logout due to inactivity. Security trumps laziness  Grin
We're coding it in as I write this and it should be live today after extensive testing.


Good man! 
hero member
Activity: 686
Merit: 564
Edit: looks like you still have to have knowledge of both messages to generate a collision.
I think you actually have to be able to control both messages to generate a collision - that's actually the definition of one. In order to be able to generate a second message that gives the same hash as an existing message you need a preimage attack, and I don't think those are practical against MD5 yet.
legendary
Activity: 1008
Merit: 1001
Let the chips fall where they may.
Well's just the thing: MtGox did salt (AFAIK) and I -did- have a good password and it still bombed, mostly because I believe they only used 1 iteration of MD5.

MD5 hashes are no longer cryptographically secure. If you were indeed using an old password hashed with MD5, the attacker could have generated a collision without guessing your password. However, it is usually easier to guess the password. If you generated the password yourself without using a random number generator, your password may not be as strong as you think it is.

Edit: looks like you still have to have knowledge of both messages to generate a collision.
sr. member
Activity: 420
Merit: 250
Thanks Jared.

I sent an email just now to you.

You should have a new password in your inbox of the email account that you used.
We've responded by email but let me know if there is any confusion.

-Jered
hero member
Activity: 1148
Merit: 501
Thanks Jared.

I sent an email just now to you.
sr. member
Activity: 420
Merit: 250
I currently can't login to tradehill.

Not sure what my password is, and there's no password recovery feature. 

Send us an email

[email protected]

We'll get it taken care of right now.

-Jered
sr. member
Activity: 420
Merit: 250
If you have a long complex password and we hash / salt it that should be sufficient.

Well's just the thing: MtGox did salt (AFAIK) and I -did- have a good password and it still bombed, mostly because I believe they only used 1 iteration of MD5.

That's why I mention PBKDF2 (http://en.wikipedia.org/wiki/PBKDF2, RFC 2898--http://tools.ietf.org/html/rfc2898).  Hell, you couldn't rely on BlackBerry encryption for a while, as ElcomSoft found out that RIM only used one iteration of AES256.  (Apple's iOS uses 10,000 iterations, IIRC).

I'm no security / crypto expect by any means, but I think I got most of that right.  I'm more worried about my BitCoins at TradeHill than I am, say, about my regular bank and USD because of the pseudo-anonymous nature of BTC.

Sorry to be such a pain!

I should have said "properly hashed". MD5 won't cut it.
I agree with you on being more concerned with your Bitcoins than your USD. Not only is it pseudo-anonymous it's non reversible.
The USD we hold is a lot easier to take care of. The Bitcoins get a lot more time put in to securely managing them.

-Jered
hero member
Activity: 1148
Merit: 501
I currently can't login to tradehill.

Not sure what my password is, and there's no password recovery feature. 
hero member
Activity: 836
Merit: 1007
"How do you eat an elephant? One bit at a time..."
TradeHill – Security Update – Round 1 (PCI Compliance)

Immediately after the Mt Gox hack and database leak was announced we shut down our site to provide adequate time for users to reset their passwords. We noticed there were considerable attempts to brute force accounts that had the same user name on Mt Gox and TradeHill. In response we installed a captcha system and auto locked out accounts with too many failed login attempts. To the best of our knowledge this was 100% effective and have not received one email concerning a compromised account on TradeHill.com   

TradeHill is proud to announce that our first round of security upgrades is complete.
We will be continuing to release updates regarding our security and upgrades to TradeHill.com

TradeHill is now PCI Compliant.

We have completed and passed a security audit by Trust Guard the leading online 3rd party website verification service. Trust Guard has searched our site for over 43,000 known vulnerabilities including SQL injection, XSS and many more and performed an ASV certified scan.  This can be verified with the Trust Guard seal on our main page before you log in (when logged in it goes away to avoid clutter).

Our site will be scanned daily for new vulnerabilities and if detected they will be taken care of immediately.

Additionally we have had our corporate contact information (US address and phone numbers) verified to confirm that we are operating in the United States as well as Chile.

User privacy is a very serious issue.
We have updated our privacy policy and are now compliant with:


The Federal Trade Commission Fair Information Practices.

The California Online Privacy Protection Act.

The Childrens Online Privacy Protection Act.

The Privacy Alliance guidelines.

The CAN-SPAM Act.



We believe that this is the bare minimum that an exchange should be operating at.

PCI scanning and putting a seal on your website from Trust Guard, Verisign or McAffe doesn't make you immune to all attacks but it is one step towards a safer exchange and something we should have done a long time ago. 

We are continuing to improve our security and will release updates as information becomes available. At the moment our source code and procedures are being verified by a 3rd party as well and we are working with top names in the security business. We will be happy to release their findings when they are complete.

We are also implementing dual authentication and other security features which will be  announced soon.
+1
full member
Activity: 265
Merit: 100
If you have a long complex password and we hash / salt it that should be sufficient.

Well's just the thing: MtGox did salt (AFAIK) and I -did- have a good password and it still bombed, mostly because I believe they only used 1 iteration of MD5. I lost some of my money. Sorry to be such a pain!
sr. member
Activity: 420
Merit: 250
How does Tradehill feel about the fact that people are spamming the general message board with ads for their service?

We've removed all the email spammers referral codes.
We don't think you should spam the boards with codes either and I believe the mods are putting a stop to that.
I could be wrong there. If you have it in your sig and you're happy with TradeHill and you want to talk about it that's fine.

If the mods want to ban referral codes in sigs that's fine and I can understand it.
I believe the bulk of the people would continue to say good things without referral codes.
I have an inbox full of positive feedback and they haven't tried to slip me a referral code.

I just ask if the mods are going to take an aggressive stance on anything they do so fairly.
There were a lot of posts claiming TradeHill was hacked after the Gox data was leaked and they were based on absolutely nothing.
We dealt with this by answering questions and being available for our users.

To sum it up, no one likes spam, be respectful. If you spam it all over the forums we'll take it away like we do on email.

-Jered
sr. member
Activity: 490
Merit: 250
How does Tradehill feel about the fact that people are spamming the general message board with ads for their service?
hero member
Activity: 560
Merit: 500
Next seal to get, both TH and CBX is the BBB seal. I know some dont like that seal and I can see some things about I dont like aswell. But its yet another seal of approval from a respected institution. The BBB seal is not easy to get.

BBB might not be the best seal to have...
http://today.msnbc.msn.com/id/43528394
http://abcnews.go.com/Blotter/business-bureau-best-ratings-money-buy/story?id=12123843
http://www.ketv.com/r/25776787/detail.html
sr. member
Activity: 420
Merit: 250
Based on feedback over the last several hours we've increased the time out from 10 minutes to 30 minutes.
It should also start over every time you visit a new page.
sr. member
Activity: 420
Merit: 250
I was wondering how much safety would work this:
1) Cascade ciphering.
2) Dividing the final hash in two or more parts.
3) Storing the different parts of the hashes in different servers.

Such an exotic configuration would confuse any low level attacker who simply thinks about dumping databases.
There is some security through obscurity here, but tactically obscurity is always an ally.

And even if the attacker manages to match the hashes, brute forcing would be painfully slow.

We're exploring all reasonable options.
Splitting hashes up would make it extremely secure.
I'm not an expert on security though so we've hired someone who is.

If you have a long complex password and we hash / salt it that should be sufficient.
If your password is short / common words etc it's not even safe from more basic attacks.
A lot depends on the end user and their habits. We can always require longer / more complex passwords
but some users are going to be upset if they can't use "boobookitty" for their password.
donator
Activity: 714
Merit: 510
Preaching the gospel of Satoshi
I was wondering how much safety would work this:
1) Cascade ciphering.
2) Dividing the final hash in two or more parts.
3) Storing the different parts of the hashes in different servers.

Such an exotic configuration would confuse any low level attacker who simply thinks about dumping databases.
There is some security through obscurity here, but tactically obscurity is always an ally.

And even if the attacker manages to match the hashes, brute forcing would be painfully slow.
sr. member
Activity: 420
Merit: 250
10 minutes of inactivity now causes a logout.

You should make it optional to not get logged out. That way both groups are happy.

That's the plan. When we've got more time to things like that we will.

For now the coders are working on things like
the API that's about to launch (I want to say tomorrow, it's working fine)
and focused on high priority items.
sr. member
Activity: 392
Merit: 251
10 minutes of inactivity now causes a logout.

You should make it optional to not get logged out. That way both groups are happy.
Pages:
Jump to: