Credit Card Fraud
The Contactless Generation
Kristin Paget
Chief Hacker, Recursion Ventures [email protected] @KrisPaget
WHAT'S COMING UP?
Contactless payments
What is EMV? How does NFC fit in? Threat vectors Shielding inadequacy Live fraud demo (x2!) GuardBunny
CONTACTLESS PAYMENTS
EMV: EuroPay, Mastercard, Visa
JCB and AmEx joined later Europay bought by MasterCard in 2002 Defines standards for next-gen payments
Contactless in USA Chip and Pin in Europe Same standard, different communications
NFC is a superset of Contactless
Same over-the-air protocol, additional security
DO YOU HAVE A CONTACTLESS CARD?
You might be surprised... Two universal symbols aren't always present Other symbols are brand-specific
NFC AND CONTACTLESS PAYMENT ?
NFC supports EMV-style contactless payment
We BELIEVE keys are stored securely
In the NFC chip on the phone
Software reversing SHOULD NOT allow key recovery
NFC application on the phone must be active
NFC is off when the screen is off (for Google Wallet) PIN number required to unlock the NFC app
With settable timeout
Explicit lock after use is possible
Other than this, NFC is IDENTICAL to EMV
Arguably more secure, arguably just as vulnerable
CONTACTLESS SECURITY
Cards are secure
JCOP smartcards are used
Readers are secure
Again, secure microcontrollers and protected keys
Protocol is secure
Strong encryption (?)
Secure in this context means:
Cost of attack is larger than potential fraud gains Keys can ALWAYS be extracted given adequate budget
IS THE PROTOCOL SECURE?
Maybe, maybe not. There doesn't appear to be mutual auth.
http://nosedookie.blogspot.com/2011/06/reading-chasevisa-paypass-credit-cards.html Read EMV cards from a non-EMV reader! Do we get all the info? Not sure yet.
Some data is available Some encryption is present More work is needed.
LEGACY PAYMENT INFRASTRUCTURE
Payment terminals expect a credit card number
As well as other info: Customer name, CVV or other check digits
Terminals always assume mag-stripes are used
Encryption is not supported
Contactless payment readers have to work with this, so
A secure terminal... ...speaks a secure protocol... ...to a secure device... ...and outputs a plaintext card number
CONTACTLESS FRAUD VECTOR
Contactless readers are widely available
Around $100 on various sites
Let the reader handle whatever crypto is there
Completely transparent to the terminal
Harvest the card number
Data is output via serial port
Write card data to magstripe
Use magstripe as a payment card
DOES THAT REALLY WORK?
DEMO 1: Making a payment
CONTACTLESS FRAUD LIMITATIONS
Contactless check digits change
Unique check digits per-transaction
Check digits are only used once
If re-presented, disable RFID token
Check digits follow a sequence
If sequence is broken, disable RFID token
Check digits are different than magstripe
If check digits don't match, disable RFID token
Some cards (AmEx) use different numbers
One card number for magstripe, different number for RFID
DO THE PROTECTIONS WORK?
Conducting multiple contactless transactions
Easy! Read the card multiple times. Only takes a few seconds per read
Old-style card fraud:
One magstripe good for multiple transactions
New-style card fraud:
Multiple contactless cards, one transaction each
Contactless skimming is far easier than magstripe
Card never needs to leave the victims pocket
MULTIPLE TRANSACTIONS
Demo 2: Read many times
UPPING THE CONTACTLESS ANTENNA
High-power readers are possible
Contactless range is typically 3-5 inches That's using milliWatts of RF power
Contactless operates at 13.56MHz There's a Ham band at 14MHz
Slightly out-of-band amplifiers will work nicely
High power is easy to obtain
Antennas and receivers are harder
Theoretical range limit: At least tens of feet
CONTACTLESS DEFENCES
Passive shields or metallic wallets:
Only reduce the signal strength This will not block a high-powered reader
We lab-tested a dozen different passive shields
Reported for a large consumer magazine
Significant inconsistency across samples, RFID bands No shielding standards exist
FIPS 201 is commonly cited, which simply says: an electromagnetically opaque sleeve or other technology to protect against any unauthorized contactless access to information
http://csrc.nist.gov/publications/fips/fips201-1/FIPS-201-1-chng1.pdf (page 8)
PASSIVE SHIELDS
Shield type
Attenuation (dB)
0 -10
-20
-30 -40 -50 -60 -70
900MHz 13.56MHz 125KHz
-80
PASSIVE SHIELDS: CONCLUSIONS
No single product stood out as The Best
Different leaders in all 3 bands
Crumpling can raise or lower performance
Could even depend on the RF band in use
LOTS of variation on the market
@13.56MHz -50dB between best and worst!
(That's 100,000x for non-radio folks)
Lack of standards mean lack of consistency
Recommend NONE of these products
SHIELDING FAILURES
Demo 3: 125KhZ
GUARD BUNNY : A BETTER SHIELD
Passive shields don't work.
Too unpredictable, can be overpowered
What about active shields? GuardBunny has no CPU or memory
LOWER-power than the tag
It generates similar modulation to the RFID tag
The reader can't tell us apart
More power in, more power out!
VERY hard to overpower.
ACTIVE SHIELDING
Demo 4:
GuardBunny
CAN YOU HAVE ONE?
Currently made of discrete SMDs on PCB
Much more expensive than RFID tags :(
Next step: ASIC production
Will make it cheaper & even lower-power
Forecast: 6-9 months
Happy to talk to engineers or fab owners (Or anyone else who can help us speed that up!)
QUESTIONS?
[email protected] @KrisPaget
