Information Security News mailing list archives

Attack Exposes ATM Vulnerabilities


From: InfoSec News <isn () c4i org>
Date: Mon, 24 Feb 2003 05:15:06 -0600 (CST)

http://www.eweek.com/article2/0,3959,899796,00.asp

[Also: http://cryptome.org/pacc.htm - WK]


By Dennis Fisher
February 21, 2003 

Two Cambridge University researchers have discovered a new attack on
the hardware security nodules employed by banks that makes it possible
to retrieve customers' cash machine PINs in an average of 15 tries.

The attack takes advantage of a weakness in the cryptographic model
used by many HSMs to encrypt, store and retrieve PINs. The system,
used by many ATMs, reads the customer's account number that is encoded
on the magnetic strip of the ATM card. The software then encrypts the
account number using a secret DES key. The ciphertext of the account
number is then converted to hexadecimal and the first four digits of
it are retained.

Those digits are then put through a decimalization table, which
converts them to a format that's usable on the ATM keypad. By
manipulating the contents of this table, it's possible for an attacker
to learn progressively more about the PIN with each guess. Using
various schemes described in the paper, a knowledgeable attacker could
discover as many as 7,000 PINs in a half hour, the authors say.

The paper, written by Mike Bond and Piotr Zielinski, goes on to say
that typical security countermeasures such as intrusion detection
systems are all but useless against this attack. Many banks have
systems in place that prevent users from trying another PIN once
they've failed three times in a row. These failures generate alerts
within the bank.

But, as the authors point out, an internal attacker "can discover a
PIN without raising the alarm by inserting the attack transactions
just before genuine transactions from the customer which will reset
the count."

The researchers' paper has drawn quite a bit of attention and is now
part of a controversial court case in the U.K. concerning so-called
phantom withdrawals from ATMs. The case concerns a South African
couple that claims someone used their Diners Club card to make 190
withdrawals at ATMs all over the U.K. while they were in South Africa.  
The card's issuer says that's not possible, because their ATM network
is secure, and is suing the couple to recover the nearly $80,000 that
was charged against the card. The couple has refused to pay, according
to court documents filed in the case.

As part of the defense, Bond has been asked to testify about the
ATM-related weaknesses he and Zielinski address in their paper.  
However, the plaintiffs, Diners Club SA Ltd., have asked for a secrecy
order around the testimony of Bond and other security experts, saying
that the publication of the ATM issues described in the paper would
harm their business and open their networks up to attack.

Ross Anderson, a well-known and highly regarded security expert and
Bond's research advisor at Cambridge, wrote a letter to the judge
handling the case, asking him to deny Diners Club's request. Anderson
argues that Bond's and Zielinski's paper—as well as a related one he
and Bond wrote last fall—are based on information in the public domain
and that the order would interfere with research, teaching and, in the
end, make worldwide banking networks less secure.

"In addition to being published material, derived from open sources,
and of crucial importance to the defendants' case, the vulnerabilities
are likely to be crucially important in other cases brought in the
U.K. and elsewhere over disputed ATM transactions," Anderson wrote in
his letter. "Bond plans to incorporate much of this material into his
Ph.D. thesis. It is spectacularly unfair for the applicant to ask you,
in effect, to prohibit Bond from including in his thesis a scientific
discovery that he has already published."

The judge began hearing testimony in the case Thursday. No ruling has
been made on the secrecy order.



-
ISN is currently hosted by Attrition.org

To unsubscribe email majordomo () attrition org with 'unsubscribe isn'
in the BODY of the mail.


Current thread: