Security shocker: 75% of US bank websites have flaws
Insecure by design
The vast majority of US bank websites jeopardize the security of their online customers by including design flaws that expose passwords and are susceptible to tampering by attackers, researchers say.
In a paper titled "Analyzing Web sites for user-visible security design flaws," researchers from the University of Michigan found 75 percent of bank sites surveyed had at least one such design flaw. The report was presented Friday at the Symposium on Usable Privacy and Security meeting at Carnegie Mellon University.
"To our surprise, design flaws that could compromise security were widespread and included some of the largest banks in the country," said Atul Prakash, a professor in the university's Department of Electrical Engineering and Computer Science, who initiated the study. Doctoral students Laura Falk and Kevin Borders also participated.
The flaws aren't bugs, but rather features built into the design of the sites. They include:
- Placing secure login boxes on insecure pages, i.e. pages that aren't protected by secure sockets layer. That allows passwords to be intercepted through man-in-the-middle attacks.
- Putting contact information and security notices on insecure pages. This makes it easy for scammers to change addresses and phone numbers listed on the page.
- Not making it clear when the website is redirecting customers to a page outside the bank's domain. As a result, customers don't know whether to trust the site.
- Allowing inadequate user IDs and passwords. Sites frequently allowed email addresses as user IDs and didn't require strong passwords.
- Emailing sensitive information. This included passwords and statements.
The report was based on the examination of websites for 214 financial institutions. The study was conducted in 2006, so it's possible the designs have been cleaned up. But we doubt it. ®
John @ Some problems are solved.. - Biometrics
John, you missed something when you read the information.
The use of biometrics creates two immediate attack vectors: print simulations (fake fingerprints a la Chaos Computer Club - which is incidentally a repeat of an article they did before) and "borrowed" fingers.
Both those vectors are countered by NAMING the fingers. When you personalise the token (i.e. make it yours by recording the biometrics) you are asked to name your fingers. Say you choose "FRED". Only you know which finger the device needs when it asks for finger "F", "R", "E" and "D" - and the accpting party can dynamically select how many fingers you need to show correctly before it allows you access.
The token will lock up if you try too much and, depending on the security levels of the channels, may even null some certificates so you'd have to go back to the issuing Bank to get them re-enabled, even if the token itself is unlocked again.
This is a Swiss product - they're a little bit more thorough than you may be used to.
ANother example of that is storing the fingerprints - they're stored inside the EAL 4+ rated crypto chip. They're not leaving the token, which means no hugely dangerous central database to hack either. You, as user, control access to your electronic identities, the way it should be.
The great thing about requiring long, strong and complicated passwords is that the average user does not have a photographic memory and therefore writes their unmemorisable password on a post-it note and sticks it on the monitor.
Paris because she likes it... long strong and complicated too.
@Ian Michael Gumby
>>>When I read the headline, it said US Bank then the article talks about US Banks.
>>>As in multiple financial institutions. There is a bank called US Bank, so you can see why there is a little bit of confusion.
"US bank" != "US Bank"
The headline says "75% of US bank websites..." Note the lack of capitalisation on the word "bank", indicating that it is not part of a proper noun.
"I helped my Uncle Jack off a horse"
"I helped my uncle jack off a horse"