A security flaw in Microsoft Corp.’s Internet Explorer (IE) Web browser can completely undermine the supposedly watertight Secure Sockets Layer (SSL) standard for securing online transactions and e-commerce, researchers said Tuesday.
IE’s implementation of SSL contains a vulnerability which allows what is described as an active, undetected, man-in-the-middle attack, where no dialogs are shown and no warnings are given.
Security researcher Mike Benham said the problem is that IE fails to check the Basic Constraints of certificates signed by intermediate Certificate Authorities (CAs). That means that as far as IE is concerned, anyone with a signed certificate for any domain can generate a certificate for any other domain, which will appear to be signed by a valid CA.
Describing the flaw, Internet security Web site Hideaway.net said: “Spoofing a trusted Web site is thus a trivial exploit; when combined with session hijacking, a man-in-the-middle attack is quite feasible. This destroys the whole purpose of SSL certificates in the first place.”
Benham said that IE 5 and IE 5.5 are totally vulnerable to this kind of exploit, and IE 6 is vulnerable under most circumstances.
“I would consider this to be incredibly severe,” Benham said in a newsgroup thread. “Any of the standard connection hijacking techniques can be combined with this vulnerability to produce a successful man in the middle attack. Since no warnings are given and no dialogs are shown, the attacker has effectively circumvented all security that an SSL certificate provides.”
Microsoft, which is currently investigating the vulnerability report, was not as quick to call the exploit trivial.
The scenario described by Benham would be difficult to exploit since it would require a man-in-the-middle attack, something a Microsoft spokesperson called, in an e-mail to the IDG News Service, “technically difficult, temporary, and (requiring) favourable network topography.”
The attack is also not as anonymous as Benham charges, as it requires a valid certificate and the CA that had issued the certificate would have a record of to whom it had been sold, the spokesperson said.
Lastly, if the user were to inspect the certificate, they would find that it was from someone they hadn’t heard of and should therefore be suspicious, she added.
Microsoft has given no indications that it plans to fix this flaw, and Benham said his experience showed it would be difficult to get Microsoft to address the issue.
“After seeing that, I don’t feel like wasting time with the Microsoft PR department,” Benham said.
Microsoft has long been an advocate of so-called “responsible disclosure,” meaning that researchers ought to give their vulnerability findings to vendors and wait until a patch has been released before disclosing their information. The policy has created controversy in the security research community, with some arguing that better security is achieved through full disclosure, the immediate publication of vulnerability information. In such circles, responsible disclosure is often derided as “security through obscurity.”
Sticking to the responsible disclosure line, the spokesperson said that “only Microsoft can investigate at a source code level; only Microsoft can build a patch, if needed.”
“We’re very concerned that publishing a report in this fashion could cause users to be concerned and apprehensive; if (Benham) had handled it correctly, we all would be in a better position to understand the real scope, and remediation (of the vulnerability),” she said.