Memory Gaffe Leaves Aussie Bank Accounts Open To Theft

mask.of.sanity writes "A researcher has found flaws in the way major Australian banks handle customer login credentials which could allow the details to be siphoned off by malware. He built proof of concept malware to pull unencrypted passwords, account numbers and access credentials from volatile memory of popular web browsers every two hours."

Comments

Careful Reporting These

[Anonymous Coward]

In the 80s, my comp sci partner and I discovered a similar case at Acadia University. We reported it to the head of the computer center. He told us it wouldn't work, it couldn't be done. I left that meeting feeling betrayed. My partner decided to write a proof of concept. He was successful and to prove it logged in as the main admin account. Days later he decided to try it again to see if they still hadn't fixed it or changed the password. They were waiting. He was expelled from Acadia. He was a brilliant honors student.

It's worse these days. They will charge you for cybercrimes, or treason, and sentence you to decades in prison. Or hold you without trial. Be careful when you do the right thing and report these. Just report them, don't "proof of concept" or you could be charged. It's unfair and immoral but it's what they'll do to you, mostly out of their own shame and embarrassment.

Re:Careful Reporting These

[Anonymous Coward]

This is why whenever I expose security flaws I do so anonymously. If it isn't fixed within the first couple days I just make it public knowledge and instigate the first attack myself. They had their fair warning, and now they get the shit storm they deserve.

Re:Careful Reporting These

[darkfeline]

I hear about these kinds of things all the time. It's utter bullshit; they're literally making it more appealing for people to anonymously sell these exploits on the black market. "No, we don't want to know if our software has an exploit. If you've found one, go ahead and sell it to whoever you want, as long as we don't know, it's cool, we can keep deluding ourselves, thanks."

It reminds me of, among other counterproductive measures, media conglomerates pushing oppressive DRM on consumers as if to drive them toward piracy or forcing drug addicts to carry their criminal status with them as if to force them back toward poverty and drug abuse. If an alien race were to monitor us, they'd probably assume we're running some sort of elaborate self-extermination campaign.

Re:Careful Reporting These

[__aaltlg1547]

Whatever his intentions, he broke the law. It's important to remember that you can be prosecuted for breaking the law even if you consider yourself a "white hat." Instead, he should have sent the demonstration code in hardcopy without ever actually intruding on the system he was trying to help improve.

Re:Careful Reporting These

[Sarten-X]

The fact that he did it for a noble cause is irrelevant. What matters is "criminal intent" - whether he intentionally broke the law.

The expelled student intended to gain unauthorized access to the computer system. He knew that the malware he wrote would harvest credentials of other users, and he knew that he wasn't allowed to log in as someone else. Yet he did so anyway. That certainly seems intentional to me, and that's what matters to prosecutors (and college judiciaries).

Re:Careful Reporting These

[Velex]

Yes, but that's not even good enough. You and I both know how these arrogant pinheads work. They have a social status and nothing more. If some damned kid can just show them up, what would that mean about them? Sure we can call the kid a "genius" or a "wiz" and dress him up in other terms to attempt to shield the pinhead's social status, but at the end of the day the fact remains that the pinhead got shown up by a damned kid barely out of diapers.

It seems the only correct answer is to either do nothing or as another poster suggested, sell the information to others.

Now imagine the threat to the social status of a billionaire if some faggot like me found an exploit. Total social status inversion. I'm supposed to be an AIDS-infested nobody who's too incompetent to do anything. Imagine how embarassing that would be to a billionaire to get shown up by an AIDS-infested (presumably) faggot.

It's all about social status. The taller they are, the harder they fall. Fortunately for them, we live in a culture that can shoot the messenger so they don't have to fall.

Those evil hackers, I'll tell ya. They're genius mutant autistic savants from the 9th dimension who never get laid. That's the only reason they can show up our best and brightest.

Re:Careful Reporting These

[__aaltlg1547]

Why don't you climb down off that high horse and join the rest of us in the real world, where breaking the law is a risky move and publishing a cookbook showing how to penetrate somebody else's computer network is considered antisocial behavior?

and now he be researching the side of jail down un

[Joe_Dragon]

and now he can be researching the in side of jail down under hands on.

Already running?

[Anonymous Coward]

You have to be infected first for your credentials to be stolen? Couldn't the hacker just have installed a key logger?

If you can't trust the machine, don't put your sensitive data on the thing.

Re:Already running?

[slashmydots]

And yet regardless, he's not getting in. If I so much as log into my online banking from another computer let alone another state or country, I have to enter multiple security question answers as well. Almost every bank does it that way. If yours doesn't, get a new bank.

Re:Already running?

[You're All Wrong]

So you're saying that if you log in from a new infected machine, your bank obliges you to leak sensitive security information to the keylogger that's been installed there?

Congratulations for feeling all warm and fuzzy from your bank's security measures whilst gaining very little actual security against real threats - that's what they were hoping you'd feel, you're a good customer.

*One time* passwords are the *only* thing that *can't* be re-used. By definition. If your bank does not use them, get a new bank.

Re:Already running?

[Impy the Impiuos Imp]

Based on how this works, I've hashed out a method to spy on the president:

1. Sneak into the White House
2. Hide under the oval office desk.
3. Now the tricky part -- listen to conversations.

Re:Already running?

[gl4ss]

You have to be infected first for your credentials to be stolen? Couldn't the hacker just have installed a key logger?

If you can't trust the machine, don't put your sensitive data on the thing.

well, it sort of matters if you can log back into the bank again with those credentials after you've signed out. that means you're note really signed out.

that is a big deal, actually.

Re:Already running?

[You're All Wrong]

Correct. MITM is a real threat.

However, it's one that has been mostly solved. Never bank with a bank that logs you in over anything but HTTPS POSTs. Do not accept certificates by default. Do not accept CA certificates by default (apart from Honest Achmed - I bought a scooter from him, he's trustworthy). Verify new certificates - check the identities of both parties (site + CA). Do not run javascript or other scripts from arbitrary sites. For paranoia, use NoScript's additional protections for XSRF, etc.

You may call me paranoid. I prefer to call it "safe". (In fact, the only reason I accept certificates is as a promise of protection against MITMs. I do not interpret it as in anyway meaning that I actually trust the site to not send me malicious data, for example.)

How bloody embarrassing!

[beaverdownunder]

Aussie IT is a bit Mickey Mouse all around, sadly -- especially in the banks, oddly (you'd expect a higher standard where billions of dollars are concerned, but no...)

As for the researcher, they didn't actually 'hack' into anything, merely scraped their own computer for data, so I wouldn't expect them to face any problems over revealing the exploit. Probably hasn't won them any friends in the banking sector though...

Re:How bloody embarrassing!

[bloodhawk]

While this isn't exactly shining a pleasant light on the quality of the banks code. It is still very much a storm in a teacup, if you have access to scrape the memory of the computer then you could have gotten access to credentials in a far simpler means such as keylogging. The simple fact is if you can't trust the machine you are using you're already boned and no amount of secure coding from the bank is going to save you. Besides which I believe most of those banks (if not all) do 2 factor auth to transfer funds to accounts you haven't previously transferred too. (at least the 2 of them I use do).

Re:How bloody embarrassing!

[mathew42]

Westpac are reported not to be vulnerable to this hack, but their online banking usernames are a 8 digit number and the password are only six characters. The available characters are [a-z] and [0-9]. This is the login page.

Re:How bloody embarrassing!

[skegg]

But I presume you only get 3 attempts before the account is locked-out. Even 10 attempts would be safe.

Wait, so your machine is already compromised?

[complete loony]

So he's running malware that's sniffing your browsers memory? If your machine is already compromised, there are easier ways to get access to login credentials.

Re:Wait, so your machine is already compromised?

[Anonymous Coward]

so keyboard logging AND screenscraping? Now enough info for the bad guy to get money?

Re:and now he be researching the side of jail down

[Frobnicator]

Sadly, he probably will.

Financial institutions want to keep their vulnerabilities quiet. People who shout them to the world face lawsuits

If you are smart enough to discover a major exploit, also be smart in how you notify them. There are many great security companies who work as middle-men to help submit the bugs to the corporations and at an appropriate time make the information public so it gets fixed.

Going through a security company is free, and means you won't get the big splash on news sites or all the public attention, but it also means you can generally avoid hiring a lawyer, or worse, having he cops knock at your door with warrants.

My bank doesn't seem vulnerable

[jonwil]

My bank uses POST in the login form which means that sniffing memory for URLs (which is what this malware seems to do) wont get you a login.
Plus, in order to actually transfer money to someone you haven't transferred money to before you have to input a second password.

The biggest failing of the bank in question is that it has a 10 char maximum on passwords for some stupid reason.

Re:My bank doesn't seem vulnerable

[chrismcb]

The biggest failing of the bank in question is that it has a 10 char maximum on passwords for some stupid reason.

I've always assumed that anyone that limits the password to an arbitrarily small number, or limits what characters you can use, does so because of incompetence. And so it makes me wonder what other security vulnerabilities there are.

Re:My bank doesn't seem vulnerable

[DarkOx]

I agree its major red flag. Yes there needs to some limit; you don't ever want to take user input of undefined maximum length, but in the case of passwords a sane max is like 255 bytes, which might be a bit shorter than 255 chars if you are running utf8, and is probably still enough if you need to use a two byte character encoding.

When you lengths like 8 or 10 it leads one to assume passwords probably are being stored insecurely; after all if they were hashing passwords like they should be the final storage requirement would not depend on the size of the original password string.

Its like hanging a sign out "Hey pen testers compromise this box, good password list to try on everything else can be found here!"

Re:My bank doesn't seem vulnerable

[WD]

You're joking, right? Please tell me that you don't think you're protected from banking malware because your bank uses POST instead of GET.

horses and barns

[stenvar]

If malware has access to the RAM of another process, the horse has left the barn.

Re: horses and barns

[starsky51]

And if the user is gullible enough to run any exe that they come across, then the sheep is sitting to close to the barbecue.

Umm - all banks worldwide?

[Anonymous Coward]

This would probably affect every single Internet site in existence. And there is no solution, nor can there be

There is a company in Australia selling JavaScript that encrypts form field - I assume this guy is associated to that company & trying to drum up a sale, while hiding the fact they are selling snake oil.

Re:Umm - all banks worldwide?

[Anonymous Coward]

I beg to differ ! This is half the browsers fault and the other half Banks/sites ... The browsers should not store the memory that long, and the sites should atleast use similar coding as to the way in that dudes video with the Jscript encoding. Sure there is other ways to grab those details but in way what your saying sounds the same as "There is a cure for for liver cancer, but don't get THAT because there's other cancers that can kill you!"

Re:Umm - all banks worldwide?

[Anonymous Coward]

I actually do this as well on a site I'm about to release. I use Javascript RSA library from some students at standford (http://www-cs-students.stanford.edu/~tjw/jsbn/). What I do is, hide the signup & login forms if the user has javascript disabled. I create an SSH Private/Public key pair for the user server side and pass the rsa_e & rsa_n modulus (public key) to the Javascript library. When the user exits a particular field such as a password field or more importantly an credit card related field, I use jQuery to convert the field to password type and I also encrypt the data in the field with the generated SSH public key. I also use end to end SSL. The reason I do this is to make Man in the middle attacks on CC Data and login credentials that much more complicated. The encrypted form values are then sent encrypted to the server and then decrypted by my server side code.

The only downside I have found so far is if the form fails, it makes it impossible to pass back validated data so the user doesnt have to re-enter. I have come up with a somewhat elegant solution for that as well.

After reading this, im hoping anything stored via memory will be stored encrypted as well as that is what would be displayed if the form is refreshed.

I'm starting to be sick

[trifish]

I am really starting to be sick of these "security researchers" who don't know that the 1st law of the computer security is:

If malware is running on your computer, it is not your computer anymore.

It follows that no matter what you do, malware will win. Discovering that malware can "siphon" memory is really... uh, groundbreaking.

What makes me even more sick is the incredibly amount of various BlackHat "security conferences" and supposedly geek-oriented media like Slashdot that let those people present this kind of "discoveries" as legitimate, notable, noteworthy, important and new.

I am really, really, sick of you.

Re:and now he be researching the side of jail down

[Architect_sasyr]

Not really a banks fault though - why is the browser hanging on to post'd data after it's been post'd??

Re:and now he be researching the side of jail down

[FireFury03]

Not really a banks fault though - why is the browser hanging on to post'd data after it's been post'd??

So that when you hit refresh on the page, the browser can pop up its usual "you'll need to repost to refresh this page, are you sure?" and do the repost if you tell it to.

Re:and now he be researching the side of jail down

[Architect_sasyr]

The Bendigo at least set those fields to autocomplete off - so should the browser actually be doing that then... or even keeping it for two hours plus.

Limited password length

[140Mandak262Jamuna]

It is really dumb to limit the password to something so small as 8 or 10 char. Or disallowing non-alpha numberics like $ + - @ # % .

But one of the common vulnerabilities is buffer overrun. So they want to limit the read to some fixed number instead of looking for the trailing null, in an unlimited loop. So the right thing to do is set the limit to some moderately large number, like 128, allocate space, write nulls into it and then read the data into that buffer. Why it can't be really big like 1K or 2K? Well, it is possible to pack lots of instructions into a 1K or 2K buffer, and we dont want to provide that much of memory in a user writable space. Of course a well written authenticator will immediately clear every user written buffer as soon as they are done reading.

In reality some UI designer limits the amount of data to be entered limited to the space provided in the edit box in the GUI. By default most screen controls like buttons and edit boxes are sized by the string buffer allocated for it. It is always possible to change the size of the control explicitly, but there are many programmers who are lazy or incompetent and don't use it,

This article lacks key information

[Gumbercules!!]

*All* of those banks insist on two factor authentication for money transfers. I use 2 of them and every single person I know (here in Australia) is either issued an RSA token or has SMS alerts on money transfers (an SMS is sent to you with a code that must be entered before the transfer will take place). So even with the password, you can't transfer money out of an Australian bank.

Re:This article lacks key information

[Gumbercules!!]

Yes, everyone has heard of cases where this has happened however they're few and far between - and generally historic (as in at least 3 - 5 years ago). This is why Telco's are supposed to have passwords on phone accounts, required before you can port. Speaking as someone who regularly ports numbers for customers at work, it's not longer a simple process - you are required to verify ownership of a number before it can be ported. I'm not saying it can't be done but it's not as easy as it was before a high profile case of it happening broke about 3 years back

However my point was, and remains, relating to this article - taking someone's bank password does not automatically mean you can transfer money out of their account. In the case of these banks, it's unlikely someone who had your password alone could even transfer money. The Commonwealth issues RSA tokens, not SMS verification, for example.

Most money fraudulently obtained from Australian banks - in fact the vast, vast majority of it - just uses plain old social engineering (i.e. dating site scams on lonely people or too good to be true purchases from Gumtree or eBay that people still fall for).

Government will steal it anyway

[CuteSteveJobs]

Australian government is now seizing bank accounts by declaring them 'inactive' if they haven't had a transaction in three years. Financial planner found $150K vanished and they also shafted a pensioner who got back from heart surgery to find his account seized. Probably hit other people who won't know yet, or elderly whose relatives won't even know the money is missing. Sure it'll be put to good use refurnishing bureaucrats offices: http://www.couriermail.com.au/news/queensland/brisbane-woman-has-had-more-than-150000-taken-from-bank-account-under-recent-law-changes/story-e6freoof-1226654782499?from=trendinglinks

Re:Government will steal it anyway

[kermidge]

Three years? I wonder how long one must be missing to be declared dead. Seems to me the bank account should wait for probate or the equivalent.

Could I get a copy.

[ralphaostrander]

:) Oh I would have paid.

Focus on the Solution, not the Problem.

[VortexCortex]

It would be great if financial companies were required to make a publicly accessible testing site, in order to qualify for benefits from government, like insurance. The testing site would be a mock-up of the current system. Just copy the code over keep a separate database, it wouldn't have to be large because it won't do the same volume and we don't all need unique accounts. I mean, there is testing and production systems already, right? So, after pushing to production you also push to public testing. This way, I can hack your systems all day and night, and not worry about going to jail for trying out the exploits I think exist -- Some are even just changing URL parameters...

The government insures the banks, but the banks aren't setting up a system where it's easy for folks to test and report vulnerabilities. It should be a no-brainer. You want car insurance? You have to drive safely and get your car inspected, also anyone can report your bad driving or smoking engine via your mandatorily exposed license plate number... Since everyone can't just visibly inspect the live version of the online systems without falling afoul of the law, then we need a mock-up.

I mean, they let me inspect the vault where my safe deposit box is... I don't get to swing a hammer, but at least I can see if the door is made of steel, and the guard is armed and paying attention. We should be able to knock on the digital vault door to ensure it's not wafer thin. I don't trust the bank to put items in my safe deposit box for me, I do that myself. Just because I put my money in the bank, doesn't mean I trust their security practices completely. I don't think we should be trying to hack the live systems because it could cause disruption, but in the current system if we notice a damn exploit we can't even report it. It would be like noticing the guards are just distracted teens with cell phones instead of guns or batons, and that there's a huge hole in the side of the vault with muddy footsteps leading in and out through it, but you'll get thrown in jail if you say anything about it at all!

Back when this online banking thing started I accidentally changed the URL parameters while logged into one of my banks' online portals. I was trying to copy paste the URL field, but ended up changing the digits in my user ID. Suddenly, my account balance was drained!? I brought up a few more account pages and my savings account wasn't just drained, it was GONE!! Wait, no, the name in small print under the company logo wasn't mine! Another users account had been pulled up. Whew.... Oh Shit! I just accessed another users account AND rummaged around looking at all the funds! I immediately logged out. I did not report anything to anyone. I was afraid that accidentally discovering a vulnerability could land me in jail if I reported it, even if I never intended to "hack" anything. It was still accessing an account without permission, a violation of the US's computer fraud and abuse act -- Similar to using a browser that someone is still logged into, you see their "private" social media stuff because they're not logged out; The bar for triggering the CFAA violation is ridiculously low. Since the bank had crappy security, and incompetent web developers I closed my account the next day. When asked why I was leaving their service, I said in the sternest voice I could conjure, "There is a large theoretical hole in the side of your bank's vault, apparently no one can see it but me, and I can't even legally show you where it is." The look on the clerks face was priceless. These banks shouldn't qualify for government benefits, IMO. I mean, security audit? No, that's obviously not working, or you wouldn't have been able to drain any account by changing a number in the URL bar... The public would do a better job for less.

I would have loved to be able to log into a mock-up site. Perform the "exploit", show them what's happening, and give them all the info they need to fix the problem. In TFA, it would be

Put your claws back in, Fix the problem, Move on

[kojimasec]

few people commenting saying that it's no danger since all Aussie banks use 2-factor SMS etc. They seem to think the password is worth nothing, That's fine however i doubt these people actually know how transfer fraud works. Meaning you need the password just as much as you need the SMS-code, And if you have access to the machine or at least password, It increases your chances to be able to port the SIM-CARD. It usually works like this FYI - 1. Got login pass for Bank, even better if they use same for e-mail ( You can delete the money transfer notification ) 2. Depending on access be it E-mail or just PC access remotely chances are you can be crafty enough to get the details needed to port the SIM-CARD 3. DOB, License No., Address etc 4. Go to carrier shopfront request blank sim-card 5. Call carrier saying you lost your phone and you need to transfer sim 6. After 20-45 minutes, Victims phone will lose connectivity which can be combined with a bogus message from attacker warning of network drop-outs 7. Login with harvested pass, SMS security message comes to you. 8. Bobs your uncle. References - http://www.bankwest.com.au/media-centre/media-releases/mobile-phone-porting-new-type-of-scam-to-look-out-for-1292493597511 - http://www.scmagazine.com.au/News/282310,45k-stolen-in-phone-porting-scam.aspx/0 - http://www.flyingpenguin.com/?p=14540 Put your claws back in, and focus on the problem here, If one bank can avoid it they all should.

What about me...

[PuZZleDucK]

... I'm with a credit union? :p