Does two-factor authentication need to be fixed? Tough criticisms heard this week from researchers about the effectiveness of two-factor authentication, especially as it's used in its token form for one-time passwords and smartcards, suggest advances need to be made to restore its luster as security protection.
Two-factor authentication sounded tarnished enough in the report "Dissecting Operation High Roller" from McAfee and Guardian Analytics that describes how an international crime gang has been targeting bank accounts of businesses and individuals to try and steal millions through unauthorized, fraudulent funds transfers using an automated process tied to remote servers elsewhere. Not only did two-factor authentication tokens for accessing bank accounts not stop the crooks, which had subverted the victim's computers with malware, but the user's commandeered authentication process was actually integrated into the automated flow of criminal processing.
"I'd never seen it anywhere else," says Dave Marcus, director of advanced research and threat intelligence at McAfee, co-author with threat researcher Ryan Sherstobitoff at Guardian Analytics about the discoveries the two security firms made as part of the forensics and investigation into a cybercrime spree that appears to have started last winter as European banks and their customers, primarily, were hit.
The fraudsters in this case designed their account takeover process for optimum exploitation of two-factor information. "They developed a fraud technology that relies on two-factor — it requires the two-factor authentication," Marcus says.
The automated system the crooks came up with takes the credentials of the person logging into the compromised machine and embeds the chip-and-pin information into the automated hacking process to carry out fraudulent funds transfers. "The collection of the token information is part of the fraud process, it's integrated into it," Marcus says.
That's why McAfee and Guardian Analytics made the strong statement they did in their report this week, saying, "The defeat of the two-factor authentication that uses physical devices is a significant breakthrough for fraudsters. Financial institutions must take this innovation seriously, especially considering the technique used can be expanded for other forms of physical security devices."
Marcus is careful to say he's not advising anyone to stop using two-factor authentication or that it's somehow intrinsically broken. "Chip and pin is a solid defense," he says. But he adds the European crime spree all suggests there needs to be some kind of design improvement in two-factor to outwit such wily cybercrime.
Steve Hope, technical director at Winfrasoft, based in the United Kingdom, which has come up with its own two-factor authentication method called PINgrid, agrees it's time for innovative approaches. Although it's not something the firm sees its enterprise customers doing today, it's possible to suggest new approaches to two-factor authentication to address the issues raised.
"Today, two-factor authentication has nothing to do with the transaction," Hope points out, saying the underlying problem may be that it is not directly tied into validation of transactions and the account code, he points out. The two processes are separate today but it should be possible to unite them to ward off sophisticated attacks. But he adds: "malware has the power, at the moment."
Did "Team Prosecco" score a goal against two-factor authentication?
Another debate that has stirred up against two-factor authentication came when cryptographic researchers based in France at the National Institute of Research in Computer Science (INRIA) issued a highly technical paper claiming they've found practical means to speed up attacks on token devices. The paper in which they describe this carries the geeky title "Efficient Padding Oracle Attacks on Cryptographic Hardware."
Calling themselves "Team Prosecco," the group intends to discuss their findings more at the upcoming CRYPTO conference. In saying they could extract encryption keys from tokens such as those from Alladin, Gemalto, RSA SecurID , Safenet and Siemens, the researchers stirred up a hornet's nest of response in some quarters.
RSA, the security division of EMC, ardently rebutted Team Prosecco's findings about the SecurID token, which Team Prosecco said it had narrowed an attack time to 13 minutes. Tokens from other manufacturers were also called vulnerable to attack by Team Proseccor, but attack times were said to be longer, ranging from 21 minutes to 92 minutes.
"This is an alarming claim and should rightly concern customers who have deployed the RSA SecurID 800 authenticator," writes Sam Curry, CTO, in his corporate blog this week. "The only problem is that it's not true. Much of the information being reported overstates the practical implications of the research, and confuses technical language in ways that make it impossible for security practitioners to assess risk associated with the products they use today accurately. The initial result is time wasted by product users and the community at large, determining the facts of the situation." Curry has been reaching out to publications that RSA believes got it wrong, posting comments to that end.
However, some crypto researchers in the U.S. said the claims by the researchers based in France should not be lightly dismissed.
Matthew Green, cryptographer and researcher at Johns Hopkins University, recently wrote in his own blog that there have been "a bad couple of years for the cryptographic token industry" and that the paper out from Team Prosecco could be just the latest bad news.
When asked his views about the paper, Green states, "All of these tokens used a known-vulnerable implementation of the RSA encryption scheme. We've known that this scheme is vulnerable since about 1998. So, in that sense, there's nothing fundamentally novel here." But he says what the researchers have done is, they "showed that these tokens are vulnerable to these known attacks. There's no good reason for this, and the developers should have recognized this as a problem even before the paper was published."
Secondly, the Team Prosecco researchers "hugely sped up the attack and made it practical to attack these token devices. This is a big deal, since the tokens aren't that fast. The new attack can run in just a few minutes, rather than hours or days."
Green says he didn't intend to be "alarmist" about what the attack means since it all "depends on how tokens are used in specific applications. Nonethless, security is not about hoping for the best, it's about planning for the worst."
He concluded businesses that depend on the tokens should be concerned and "take steps to protect themselves and their customers' data."
Ellen Messmer is senior editor at Network World, an IDG publication and website, where she covers news and technology trends related to information security.