Phishing Tips and Techniques

Tackle, Rigging, and How & When

to Phish

Peter Gutmann
University of Auckland

Background
Phishing is currently the most widespread financial threat
on the Internet
• Phishing sites increased at 28% / month in 2004-2005
• It’s a $xB / year industry
• We know that it works
• We don’t know why it works
– “Users are idiots” isn’t a reason
Why does it work?
• What are the threats?
• Where are the weak points in our defences?
Why can’t users get security right?
Users are idiots

• Developers build security applications

• Users apply them incorrectly
• Users are idiots
• QED

Why can’t users get security right? (ctd)

Waitaminnit, they can’t all be idiots…
Look at all these idiots driving the wrong way down the
motorway, I’m the only one going in the right direction
If everyone’s getting it wrong, there’s a problem with the
design / implementation
Why can’t users get security right? (ctd)
Developers have created a pile of security widgets for
browsers and similar applications
• Padlock icon
• ‘https://’ indicator
• Coloured URL bar
• Certificate warnings
• (Optional) security toolbars
None of this was ever really tested on users
If users don’t understand it, it doesn’t exist
— Security HCI prime directive

Why can’t users get security right? (ctd)

When it was finally tested (ten years after it was
introduced), the results were disturbing
• 65% ignored the padlock
• 59% paid no attention to the ‘https://’ indicator
• 77% didn’t notice the address bar colouring
– Of those who noticed it only two understood its significance
• When presented with a certificate warning dialog, 68%
immediately clicked ‘OK’ without reading the dialog
– Just one user was able to explain what they’d just done
Why can’t users get security right? (ctd)
Another study found that not a single user checked the
certificate when checking site validity
Yet another study found that only 18% of users could
identify an unprotected (SSL vs. no SSL) site
And yet another study found that […]

OK, we get the picture, it doesn’t work!

User Conditioning
“We can fix security problems with better user education”
• We’ve been educating (conditioning) users for years…
• DNS errors, transient network outages, 404 errors, ASP
problems, Javascript warnings, missing plugins, temporary
server outages, incorrect or expired certificates, MySQL
backend problems (any slashdotted site), …
• In all cases the solution is to click “OK”/”Cancel” or to try
again later until it works
• Users have become conditioned to applying this solution to all
computer/network problems
Network attacks exhibit identical symptoms to the above
• We’re trying to detect attacks with a close to 100% false
positive rate!
User Conditioning (ctd)
The following dialog pops up the first time the user
searches ebay for dog food

• No context for this: Could be a banking PIN or a dogfood

query
• Note the “go away and don’t bother me again” checkbox
– Even the dialog’s designers admit that it’s just an
annoyance

User Conditioning (ctd)

Effect of this dialog on users

• The actual text is irrelevant, this is what the user absorbs

• A Win95 beta dialog told users “in order to demonstrate our
superior intellect, we will now ask you a question you cannot
answer”
– Again, the dialog’s designers knew exactly what they were
asking of users
User Conditioning (ctd)
An entire generation’s computing experience is built
around clicking ‘OK’ to error messages that they don’t
understand
• In a standard HCI context, this would be just moaning about
bad UI design
• In a security HCI context, this is phishing’s primary attack
vector
• You can take this principle to the bank — and the phishers are

User Conditioning Example

Large banking site
• Certificate had expired, leading to browser warnings for
anyone who used the site
• Just one single user out of 300 turned away
Hotmail does this all the time, you just wait awhile and it works
again
— User comment
User Conditioning Example (ctd)
Government
site used to
make multi-
thousand-
dollar property
tax payments
• No-one was
deterred by a
large red cross and warning text indicating that the certificate
was invalid
• All the security mechanisms were working exactly as designed

User Conditioning Example

SSL security depends entirely on user handling of
certificates
• Leaves it up to the user to make the final decision
Security =
• AES-256 +
• RSA-2048 +
• SHA-1 +
• User judgement call
You can see why the attackers are going for phishing
Phishing Tip
Invalid certificates don’t bother users
• Create your own CA with any name that you want
• Use your CA to issue certificates for any web site you want
• More on this later

User Conditioning Example

Financial institutions are actively training their users to
ignore certificate-based security indicators
User Conditioning Example
They practice what they preach…

User Conditioning Example (ctd)

This is a genuine Bank of America site

• It just happens to be at reo.com
User Conditioning Example (ctd)
Bank emails are indistinguishable from phishing emails
Customers should understand that Citibank will never send e-
mails to customers to verify personal and/or account
information […] It is important you disregard and report e-mails
which […] request any customer information — including your
ATM PIN or account details
— Citibank Australia
Go to URL and […] to identify and authenticate yourself, enter
(a) your card number (b) your ATM PIN (c) your account
number
— Citibank Australia email, November 2006

User Conditioning Example (ctd)

It had all the classic signs. It was an e-mail asking the
customer to go to a Web site and enter their ATM or credit
card number, their ATM PIN and their account number. It then
asked them to enter some answers to security questions such
as their mother’s maiden name and create a username and
password
— Bronwyne Edwards, SMS Management & Technology
User Conditioning Example (ctd)
The bank couldn’t see a problem
Are you guys on crack?
— Paraphrased journalist inquiry to Citibank
These are all online banking customers and are used to
receiving e-mails from us. I don’t believe we have contradicted
[our security policy]
— Citibank Australia spokesperson

User Conditioning Example (ctd)

A profusion of domain names serve to confuse users
• citibank.com = Citibank
• citibank-verify.4t.com = Not Citibank
– No problem to
obtain a legitimate
certificate for
4t.com

• accountonline.com = Citibank again

– Citibank uses six more domain names
User Conditioning Example (ctd)
The Citibank namespace alone includes
citibank-america.com, citibank-credicard.com, citibank-credit-
card.com, citibank-credit-cards.com, citibank-account-
updating.com, citibank-creditcard.com, citibank-loans.com,
citibank-login.com, citibank-online-security.com, citibank-
secure.com, citibank-site.com, citibank-sucks.com, citibank-
update.com, citibank-updateinfo.com, citibank-updating.com,
citibankaccount.com, citibankaccountonline.com,
citibankaccounts.com, citibankaccountsonline.com, and
citibankbank.com
Most of these are highly questionable
• citibank-account-updating.com is owned by Ms. Evelyn
Musa (Nigeria), [email protected].

User Conditioning Example (ctd)

Even legitimate domains are a mess
Hanscom Federal Credit Union (serving the massive
Hanscom air force base, a tempting target) uses
www.hfcu.org, locator.hfcu.org, ask.hfcu.org,
calculators.hfcu.org, www.loans24.net,
hfcu.mortgagewebcenter.com, secure.andera.com,
secure.autofinancialgroup.com, hffo.cuna.org,
www.cudlautosmart.com, www.carsmart.com,
reorder.libertysite.com, www.ncua.gov, www.lpl.com,
anytime.cuna.org, usa.visa.com, and www.mycardsecure.com
Phishing Tip
Target US financial institutions
• They have the worst online security practices of any banks
anywhere
– Users are heavily conditioned towards accepting these poor
security practices
• Second-worst are UK banks
– Heise Security/UK found that six of the nine largest UK
banks were trivially vulnerable to frame spoofing/cross-site
scripting
– Phishing sites were indistinguishable from the real thing
– Two banks subsequently fixed their pages
– Only one of the fixes actually worked

Phishing Tip (ctd)

Avoid Australasian/European financial institutions
• Second-best are Australasian banks
• Best are European banks
– PIN calculators, smart cards, TANs (one-time per-
transaction PINs), …
– Don’t bother with these unless you really know what you’re
doing
Results of User Conditioning
SecuritySpace survey found that 58% of all SSL
certificates were invalid (expired, self-signed, unknown
CA, incorrect domain, etc)
• Most people only see the valid certs from big sites, so this
problem isn’t very visible
Browser vendors can’t afford to fix this any more
• The majority of web sites would break
• “Microsoft is using its monopoly position to force people to go
with commercial CAs”
• “Firefox/Opera/Safari can’t access site X, MSIE can. Firefox/
Opera/Safari is broken”

Results of User Conditioning (ctd)

2005 study found that invalid SSL certificates had no effect
whatsoever on people visiting a web site
• Effect of certificates was indistinguishable from placebo
Because most users dismiss certificate verification error
messages, SSL provides little real protection against MITM
attacks
— Security study
Results of User Conditioning (ctd)
SSL certificates provide honesty-box security
• Use a $495 Verisign certificate
– People will come to your site
• Use a $9.95 budget CA certificate
– People will come to your site
• Use a $0 self-signed certificate
– People will come to your site
• Use an expired or invalid certificate
– People will come to your site
• Use no certificate at all, just a disclaimer saying that you’re
secure
– People will come to your site

Results of User Conditioning (ctd)

Study found that users treated a site with no certificates as
being less secure than one with an invalid certificate
• Users assumed that the mere presence of a certificate (even if it
was invalid) made the site legitimate
Expired safety certificate in a lift/elevator doesn’t mean
that it’s unsafe to use, merely that the operators forgot to
get a new one
• How many people even look at these sorts of certificates?
• How many people check that they match the building/elevator
they’re in, and are currently valid?
Results of User Conditioning (ctd)
This is worse than placebo!
Users actually behaved less insecurely when interacting with
the site that was not SSL-secured
— Security study

Phishing Tip
Using a self-signed certificate gets you more respect than
not using a certificate at all
• More on this later
In 2005 alone, 450 “secure phishing” attacks were recorded
• Self-signed certificates
– Taking advantage of the “any certificate means the site is
good” mindset
• XSS, frame injection, …
• Genuine certificates issued to soundalike domains
– Fake site: visa-secure.com
– Real Visa sites: verifiedbyvisa.com,
visabuxx.com, …
How Users Make Decisions
Standard economic decision-making model assumed that
someone making a decision
• Weighs up a set of alternatives
• Chooses the best one
US DoD sponsored research into improving battlefield
decision-making
• Found that users making a decision
– Generate options one at a time, without ever comparing any
two
– Reject approaches that don’t work
– Take the first one that does
This is termed the singular evaluation approach

How Users Make Decisions (ctd)

Singular evaluation is
used when
• User is under pressure
– Computer users
wanting to do their
job automatically
fall into the “under
pressure” category
• Conditions are dynamic,
no time to perform
detailed analysis
• Users have little basis for analysing/comparing choices
– “certificate is for a different domain”  “eddies in the
time/space continuum”
How Users Make Decisions (ctd)
Singular evaluation approach is used constantly when
dealing with computers
• Saves time and effort when dealing with pointless popup
dialogs
The web browsing model encourages this poke-and-hope
approach
• If you make a mistake click “Back” and try again
• Satisficing, an approach that both satisfies and suffices
Web users are deeply immersed in a singular-evaluation
environment

How Users Make Decisions (ctd)

If humans didn’t use singular evaluation, they’d never get
anything done
• Attempts to computerise singular evaluation (a.k.a. “common
sense”) lead to programs that had to grind through millions of
implications to find a solution
• AI researchers call this the frame problem
• In humans, it’s a disorder called somatising catatonic
conversion
How Users Make Decisions (ctd)
Singular evaluation isn’t a bug, it’s what allows humans to
function
• Researchers performed an experiment where users were told to
carefully evaluate a site
• Found that users spend “absurd amounts of time” trying to
verify its legitimacy
– Experiment had to be aborted
• False positive rate rose to 63%
– If you look hard enough, you’ll always find something
suspicious

Phishing Tip
This is not grumbling about idiot users, this is an
immutable law of nature
• You cannot ignore, avoid, or “educate” users out of this
• This behaviour is not the exception, it’s the environment
This isn’t going to be patched in a hurry
• You can’t “solve” this human problem  target it as much as
possible
• Sales people already know about forcing people into singular
evaluation mode: “call in the next 10 minutes”, “offer ends
Monday”, “try our exclusive …”, …
Automatic Processes and Habituation
Controlled processes
• Slow
• Costly in mental effort
• Provide a great deal of flexibility
Automatic processes
• Quick
• Little mental effort
• Acting on autopilot
Novice vs.experienced driver
• Changing gears, checking the rear-view mirror, looking left
and right at intersections is slow and manual or quick and
automatic

Automatic Processes and Habituation (ctd)

Humans are creatures of habit
• Automatic processes are triggered by certain stimuli
• Very hard to stop
• User’s aren’t consciously aware of what they’re doing
– “Did I lock the front door?”
– “Did I leave the iron on?”
Automatic Processes and Habituation (ctd)
Once users become habituated into certain behaviour, it’s
almost impossible to break this conditioning
• Microsoft found that so
many users reflexively
closed the Windows
automatic updates dialog
that they converted it to
nagware to prevent it
from being bypassed
• Users just treated the
security update dialog
as another piece of
popup noise to be clicked away

Automatic Processes and Habituation (ctd)

This was noticed a century ago by Gestalt psychologists
• Users resist attempts to change their behaviour even in the face
of evidence that what they’re doing is wrong
• Gestalt psychologists called this phenomenon “Einstellung”
Software vendors have tried to work around this
• Tip-of-the-day
• MS Office paperclip
– OK, so that didn’t work…
Consequences of Habituation
Users are required to endlessly authenticate themselves

• Browsers make no distinction between security levels

– Firefox master password could be unlocking your bank
account or your knitting patterns

Consequences of Habituation (ctd)

Even the legitimate application’s requests for passwords
are incomprehensible to any normal human

• “Eddies in the time/space continuum! Gimme your password”

Consequences of Habituation (ctd)
Users are habituated into entering their password for
everything they do
• 2006 study found that 96% of users re-use passwords across
sites
• Most users don’t understand the consequences of password
cross-pollination, and don’t protect low-value accounts much
Roll on biometrics!
• Far more vulnerable than password authentication

Phishing Tip
Try an indirect phish for a low-value site
• Who cares about my password for knitting patterns?
• (Not too necessary yet since direct phishing is still so easy)
Try phished credentials at high-value sites
• Hotmail ID at Paypal, Bank of America, Wachovia, …
Reject the first few passwords that the user enters
• Automatic process conditioning: Did I enter the password for
the right site?
• Users are so accustomed to entering passwords that they’ll
switch to other ones thinking that they used the wrong one
Phishing Tip
Try for the backup password (password hint)
• Many accounts have two passwords, the standard one and a
(very weak) backup
• These are uniformly terrible
– “What’s your dog’s maiden name?”

Real or Fake?
Humans are very bad at generating testable hypotheses
• People will try to confirm their hypothesis  confirmation bias
• People are more likely to accept an invalid but plausible
conclusion than a valid but implausible one
How do you check whether a site is for real?
• Enter your username and password
• If it lets you in, it’s real
(If security people had bothered to implement password
authentication properly, this would be a valid test)
• TLS-PSK provides mutual authentication of client and server
• Have the technology fit the user, not the other way round
Real or Fake? (ctd)
Extreme case of rationalisation: Patients whose brain
hemispheres had been physically separated (corpus
callosotomy)
• Tell one half of the brain to do something
• Ask the other half why it’s doing it
• Patients always had an explanation, even though the left half
literally didn’t know what the right half was doing

Real or Fake? (ctd)

You can experience this yourself through visual “blind
spot” tests
• Brain invents stuff to fill the blind spot where the optic nerve
enters the retina
Other examples of the mind making things up
• Confabulation across saccades
• Filling in words in sentences that have been obliterated by a
noise like a cough
• (Many others, this is a fun topic for experimental
psychologists)
Real or Fake? (ctd)
Bank site located in an unexpected place
www.ssl-yahoo.com is a subdirectory of Yahoo, like
mail.yahoo.com
sign.travelocity.com.zaga-zaga.us must be an
outsourcing site for travelocity.com
Sometimes the company has to register a different name
[www.mytargets.com] from its brand. What if
target.com has already been taken by another company?
Sometimes I go to a website and it redirects me to another
address
I have been to other sites that use IP addresses
All of these are real responses from users in security
studies

Phishing Tip
People want to believe what they see

• Create a good enough copy of a site and it won’t matter if it’s

hosted in Romania
Make it easy to “confirm” the site’s authenticity
• Inability to create a testable hypothesis means that
counterindications are never explored
The Watermark Fallacy
Financial institutions have invested a great deal in anti-
counterfeiting technology
• Intaglio printing
• Watermarks
• See-through register
• Rely on the difficulty of replicating physical artifacts
People assume that complexity implies authenticity
• “No-one would be able to replicate this Flash animation”
• Assume that the digital world follows physical copying rules

The Watermark Fallacy (ctd)

Sometimes not even complexity is required
• Holdover from the pre-computer age
• “Security by letterhead” was relatively robust when printing
was hard
• Internet storefronts don’t correspond to physical storefronts
Exploit the watermark fallacy
• Copy and prominently feature Flash, animated graphics, …
Be careful with too-literal copies
• Need to adjust ephemeral information like dates on copied
pages
Phishing Tip
As long as the site looks plausible, this will work

• Surely no-one would bother creating an entire fake site, would

they?

Phishing Tip (ctd)

Federal Financial Institutions Examination Council
(FFIEC) required that banks introduce two-factor auth
• Banks redefined “two-factor” to “twice as much one-factor”
Example: Bank of America uses SiteKey two-stage signon
• Users are directed to an
intermediate page that
contains a personalised
graphic
• Serves no useful purpose
since the initial page still
isn’t protected
Phishing Tip (ctd)
But wait… there’s more…

The SiteKey message is training users to ignore SSL

security indicators in favour of (ineffective) SiteKey
ones!

Phishing Tip (ctd)

This pretend two-factor authentication is actually worse
than the original purely one-factor form
Phishing Tip (ctd)
To bypass SiteKey-type “security”
• Use MITM to fetch the graphic from the real site
• Display a broken-image icon on the intermediate page
• Display “Your security settings have prevented this image from
being displayed”
Independent evaluation found that even the simplest of
these (the last one) convinced 92% of users that the site
was genuine
• Gimmick security mechanisms of this type are rarely tested by
vendors

Phishing Tip (ctd)

These gimmick mechanisms have a net negative effect on
security
• Mechanism provides no value itself, but degrades other
security mechanisms
In any case malware isn’t even bothered by this
• Trojans include “grabbers” that hook into the Javascript engine
and bypass AJAX-based additional “authentication”
The Watermark Fallacy Reloaded
Other factors that convince users that a site is genuine…
Correct spelling and clean layout
• (Poor English still works against victims whose 2nd/3rd
language is English)
Personalisation
• “Love, John Smith” rather than “Sincerely, Mgr., Accounts
Receivable”
Simple unprotected URLs trump complex SSL URLs
• http://www.attuniversalcard.com rated
significantly higher in user testing than
https://www.accountonline.com/-
View?docId=Index&siteId=AC&langId=EN

The Watermark Fallacy Reloaded (ctd)

Pseudo-personalisation of communications
• Use geolocation to get the (US) Zip code
– Some malware already does this
• Display the credit card prefix
– 4828-****-****-****
– Prefixes are shared across large numbers of users
– Targeting specific banks allows you to use the banks’ card
prefix
Legal disclaimers and copyright notices
• “Phishers don’t need legal disclaimers, so they won’t include
them in their messages”
The Watermark Fallacy Reloaded (ctd)
Out-of-band verification channels
• Display a phone number to call for a safety check
No-one calls it, they all assume that someone else will
check
• In any case, phishers have set up their own IVR systems that
mimic the banks’ ones

The Watermark Fallacy Reloaded (ctd)

This phenomenon is so common that it has a name: the
bystander effect
• The more bystanders, the less chance of any one individual
taking any action
– 85% with one bystander, 62% with two, 32% with five
• On the Internet, the bystander count is the entire world
The “someone else’s problem” fallacy, also found in OSS
security software
• “I won’t trust it unless there’s source code available, but I’ll
assume that someone else has checked it”
• Security holes have persisted in OSS security apps for years
until they were found, often by chance
Privacy Seals: Security Chicken Soup
Theory: Sites apply for Better Business Bureau-style
certification
• Guarantees that they meet certain minimal requirements
• Certification is withdrawn if they fail an audit
Practice: Anyone can get a seal
• Too many organisations selling them
• For some the only thing they demonstrate is that money
changed hands (coughTRUSTecough)
– TRUSTe basic seal merely confirms that the site has a
security policy of some form
– “Our policy is to hand over all your private data to the
Russian mafia” would qualify

Privacy Seals: Security Chicken Soup (ctd)

Common among scammers and fly-by-night traders
• So widely used that it has its own name, “seal abuse”
I can get you any result you like / Whats it worth to you?
Verified by Visa, Diners, MasterCard, Verisign Secure
Site, Better Business Bureau (BBB), various medical
certifications (for dodgy pharmaceuticals), etc
Phishing sites can’t have their seals revoked
• Variation of the watermark fallacy
Privacy Seals: Security Chicken Soup (ctd)
Users seem to like seeing them on web pages
• One study found that 42% of users said they’d trust a site more
if they saw a trust mark
– 49% if it was one that they recognised
Chicken soup, can’t hurt to add them to your phishing site

Phishing Tip
The Simon Says Problem
Users are expected to change their behavior in the absence
of a stimulus
• This is very, very hard to do
In web browsers, the absence of a (tiny) padlock is
expected to change the user’s behaviour
• The Hamming weight of the security indicator is close to zero

The Simon Says Problem (ctd)

A usability test of the IE6 SP2 security warning strip found
that not one user noticed its presence
User reaction to Vista UAC colours: “There are different
colours?”

• Now try and figure out what the colours signify

The Simon Says Problem (ctd)
In another test during usability evaluation of spreadsheet
software, no-one noticed a flashing message saying
“There is a $50 bill taped to the bottom of your chair.
Take it”
In a test carried out by psychologists in 1999, only 43% of
viewers noticed a person in a gorilla suit prancing around
during a basketball game

The Simon Says Problem (ctd)

The ability to focus on a single target and sort out relevant
details from the noise is what makes it possible for
humans to function
• Human senses filter light and sound to a manageable level
• Selective attention processes provide further filtering
– Cocktail party phenomenon
• Forgetting discards non-useful information
The Simon Says Problem (ctd)
Humans have, as a part of their evolution, learned to focus
on what’s important
• Flashing lights
• Snakes, tigers, wolves
• Used-car salesmen
A small padlock or blue bar isn’t important, and isn’t
noticed

Phishing Tip
Don’t worry about the MSIE 6 SP2 security ribbon and
similar “phishing” indicators
• Most users simply won’t notice them
• The few that notice them won’t know what they signify
• Security toolbars aren’t installed by default
• 39% of users of various anti-phishing toolbars were fooled by
phishing sites even after being told that they were part of a
phishing study
US financial institutions are working hard to train users to
ignore these indicators anyway
Brand Power
CAs have attempted to introduce “high-assurance”
certificates
• High assurance that you’ll be charged more for them
Most users don’t even know what a CA is
• Term is only defined for locale = xx-geek
• No users know all of the 100-150 CAs hardcoded into their
browsers
The most insignificant mainstream brand has more market
presence than the most significant CA brand
• More people recognise Visa as a trusted CA than Verisign
• Verisign is the world’s largest CA
• Visa isn’t a CA at all

Phishing Tip
Create your own CA belonging to a major brand
• Use that CA to issue site certificates for the brand
• Do you want to trust https://www.visa.com, certified
by the Visa CA?
– Of course I do, it’s Visa!
Phishing Blacklists
Added to both MSIE 7 and Firefox 2
Implementation of “enumerating badness”
• No.2 on Marcus Ranum’s “Six Dumbest Ideas in Computer
Security”
• Actually a special case of default-allow, the No.1 dumbest idea
To sidestep this, just avoid the blacklist
• 0-day phish
– Anti-Phishing Working Group reports that the average
phishing site lifetime is 5 days
– Spammers are already using sites with 6-hour lifetimes
• Reverse proxy via a botnet
– Try blacklisting 10,000 constantly-changing IP addresses

Phishing Blacklists (ctd)

“But it works with virus scanners!”
• Virus scanners only have to find a virus in 100K files on a hard
drive
• Even then, the most popular scanners have an 80% miss rate
(AusCERT)
• Virus writers test their malware against the market leaders to
make sure that it’s not detected
I had a class full of students this semester […] the second
assignment was to write a virus that would pass the anti-virus
software, and all of them did by the following week
— Matt Blaze, 2004 Security Protocols workshop
Phishing Blacklists (ctd)
Phishing blockers have to detect a site among 1+ billion
constantly-changing Internet-connected machines
• You’d have to monitor every machine all of the time and be
able to blacklist them in close to real-time

Phishing Tip
Nothing to worry about
• Just make sure that your site isn’t around long enough to be
blacklisted
• Many sites are already doing this anyway
Like WW2 German superguns
• Working on it diverts resources away from solving the real
problem
Why can’t users get security right (revisited)
Users are idiots
Security people are wierdos
• Go directly against millennia of evolutionary conditioning
• No normal person would ever handle a user interface the way
that security people do
Security people design these interfaces assuming that
they’ll be used the way that they would use them
• At least one user study on PKI un-usability was greeted with
disbelief by security people
• It couldn’t possibly be this hard to use!

Why can’t users get security right (revisited)

What the developers wrote
Why can’t users get security right (revisited)
What the users read

Summary of Phishing Tips

Create your own CA for a well-known brand
• Use brand power to your advantage
Certify your phishing site using this CA
• Users are more likely to fall for your phish if you have any
kind of certificate
Use an indirect phish (low-value credentials at high-value
site)
• Users don’t understand password cross-pollination
Make it as close to the real thing as possible
• Take advantage of confirmation bias/inability to generate
testable hypotheses
Summary of Phishing Tips (ctd)
Copy and feature Flash, animated graphics, …
• Leverage the watermark fallacy
Preferentially target US financial institutions
• Worst security of financial institutions anywhere
Use US banking disclaimers about lack of security
indicators
• US banks have done a lot of user conditioning for you
Don’t sweat the small stuff (padlocks, security ribbons,
other indicators)
• No-one notices these anyway. Make the Simon Says problem
work for you

Summary of Phishing Tips (ctd)

Use short-lived sites/reverse proxies via botnets to avoid
blacklists
If users don’t understand it, it doesn’t exist
• Look for studies showing poor usability of security features
Remember, you only need a 1% success rate for a
successful phish
• The defenders need a 100% success rate
More Information
Slides available from my home page,
http://www.cs.auckland.ac.nz/~pgut001
Lengthy discussion of problems and countermeasures
available from the same site,
http://www.cs.auckland.ac.nz/-
~pgut001/pubs/usability.pdf