A flaw in the protocol used to secure communications over the Internet could have been used to hack Twitter accounts, according to an IBM security researcher.
Last week Anil Kurmus demonstrated how a flaw in the SSL (Secure Sockets Layer) protocol could be used to essentially trick victims into sending Twitter messages that contained their password information. For the flaw to be exploited, a hacker would first have to find a way to get onto the victim’s network, launching what’s known as a man-in-the middle attack, so it would be hard to affect a large number of Twitter users with this technique. The issue was soon patched by Twitter, but it has security experts wondering how many Web sites might suffer from a similar problem.
A consortium of Internet companies has scrambled to fix the SSL issue since Nov. 5, when it was inadvertently made public on a discussion list. But there has been some debate about the seriousness of the flaw. Shortly after the bug was made public, IBM researcher Tom Cross said that, for the most part, major Web applications would not be affected by the issue.
But Cross changed his mind, writing: “Unfortunately, the situation is worse than I thought.”
Webmail applications, in particular, may also be at risk from this attack. And security experts also worry that other applications—databases, for example—may be at risk.
Twitter.com was susceptible to the bug because it did what’s called client renegotiation under SSL. Client renegotiation gives the Web site a way to ask the Twitter user for an SSL certificate after a user is already connected to the site. It’s a useful tool for sites that let users log on using smart cards or for sites that restrict access to a select group of predefined Web surfers, but until the flaw is fixed, client renegotiation also opens the door for SSL attacks.
There are probably many sites such as Twitter that allow client renegotiation simply because it’s built into the SSL protocol and its successor, TLS (Transport Layer Security), said Marsh Ray, one of the PhoneFactor developers who discovered the issue. “A lot of people didn’t realize that they were doing it,” he said.
The good news is that many sites can simply disable it outright, which is apparently what Twitter has done. Twitter did not respond to a message asking for comment on this story.
According to Ray, people should realize that while the SSL flaw is not catastrophic, “this is a serious bug and people need to patch it.”