Researchers Find Huge Weakness in European Payment Cards
By Jeremy Kirk
PCWorldFeb 12, 2010 5:58 pm PST
Hundreds of millions of payment cards throughout Europe have a flaw that could allow criminals with a stolen card to enter any random PIN to complete a transaction, according to researchers from the University of Cambridge.
The findings, which will be presented at the IEEE Symposium on Security and Privacy in California in May, cast new doubts on chip-and-PIN or EMV cards. The cards contain a microchip that verifies a correct PIN in order to complete a transaction.
European banks hail the system as more secure, as U.S. cards do not have the microchip, which has so far prevented some types of card cloning.
But the Cambridge researchers have found a weakness in the complicated EMV protocol that allows for a man-in-the-middle attack. It essentially tricks the point-of-sale terminal into believing it has received a correct PIN no matter what digits are entered.
The attack requires high-level knowledge of the chip-and-PIN system and some external hardware as the researchers demonstrated on the BBC’s Newsnight program on Thursday.
Nonetheless, “this flaw is really a popper,” said Ross Anderson, professor of security engineering, on Newsnight.
On the program, Anderson’s colleague, Saar Drimer, is shown illustrating the attack at the University of Cambridge’s cafeteria. He wore a backpack containing a laptop computer and an FPGA (field programmable gate array) board. A dummy card is actually inserted into the point-of-sale device, which is wired to the stolen card.
For the demonstration, the researchers used a credit card from HSBC, one from the John Lewis department store, and debit cards from Barclays and The Cooperative Bank. Drimer is seen slyly inserting the dummy card — which has a protruding wire — into the point-of-sale terminal. The transaction is authorized even though Drimer enters an incorrect PIN of “0000.”
The banking industry was informed about two months ago by the researchers of the attack. HSBC and Barclays referred comment to U.K. Payments, a banking trade association.
“We never said chip-and-PIN was completely infallible,” a spokeswoman said Friday. “We firmly believe it [the attack] is not really plausible in a day-to-day environment. They’ve created a convoluted way of committing this fraud.”
The equipment looks complicated but is doing something very simple and could be condensed, said Steven J. Murdoch, one of the researchers.
The PC, which does computational work, could be replaced with a tiny microprocessor. The FPGA, which is used to modify voltages and data rates between the PC and payment cards, could be replaced with much smaller technology. Both of the payment cards could also be modified to communicate wirelessly, which would eliminate the wire shown in their current demonstration, Murdoch said.
In both the latest Newsnight demonstration and in tests last December, the banks did not apparently detect the fraud. “It looks like if you were to do this, you would not be caught,” Murdoch said.
Cambridge researchers have been highly critical in the past of chip-and-PIN, finding numerous technical problems with the specification and criticizing the lack of transparency under which it was developed.
They argue that’s troubling since the banks tend to hold customers responsible for losses for transactions in which their PIN was used, even if the customer says no one else knew the PIN, they say.
“So far, banks have refused to refund such victims because they assert that a card cannot be used without the correct PIN,” according to their paper, entitled “Chip and PIN is Broken.” “This paper shows their claim is false.”
About 730 million chip-and-PIN cards are in use worldwide. Most European countries use the cards, and they are also being introduced in Canada and discussed in the U.S. Murdoch and Anderson are scheduled to speak on Wednesday to the Federal Reserve Bank of Philadelphia about security issues associated with EMV cards.
In addition to Drimer and Anderson, the research was done by Steven J. Murdoch and Mike Bond. More information is available on the Light Blue Touchpaper blog.