FRAMINGHAM, Mass. – Google’s Chrome browser contains a critical vulnerability that under certain circumstances allows attackers to plant malware on a Windows PC, claims a security company.
According to Slovenia-based Acros Security, Google would not categorize the bug as a vulnerability, and instead called it a “strange behavior that [they] should consider changing.”
The vulnerability, said Mitja Kolsek, Acros’ CEO, is one of a string in Windows programs that relies on an attack strategy variously dubbed “DLL load hijacking,” “binary planting” and “file planting.”
The attack jumped into public view in August 2010 when HD Moore, the creator of the Metasploit penetration hacking toolkit and chief security officer at Rapid7, found dozens of vulnerable Windows applications. Moore’s report was followed by others, including several from Kolsek and Acros.
Many Windows applications don’t call DLLs, or dynamic link libraries, using a full path name, but instead use only the filename, giving hackers a way to trick an application into loading a malicious file with the same title as a required DLL. If attackers can dupe users into visiting malicious Web sites or remote shared folders, or get them to plug in a USB drive — and in some cases con them into opening a file — they can hijack a PC and plant malware on it.
Microsoft Corp., for instance, has provided 17 security updates in the last 13 months to fix DLL load hijacking problems, most recently earlier this month .
The newest, however, affects Chrome, the browser that — because of its “sandbox” technology that isolates the browser from the rest of the system — most security experts believe is the safest of the top five.
Chrome’s sandbox doesn’t protect against this DLL load hijacking, Acros said last week in a lengthy write-up of the vulnerability.
The silver linking? Hackers have to have the stars perfectly aligned to exploit the bug, said Acros.
For the vulnerability to be exploited, Chrome must be set to use a search engine other than Google’s, which is, not surprisingly, the default for Google’s browser. Acros confirmed that an attack can be successfully launched against Chrome when users set Yahoo or Bing as the browser’s preferred search site.
The user must have not visited a secure website — one whose URL starts with HTTPS — before the attack begins, added Acros, and must also be duped into trying to load a file and thus have the “Open” dialog box onscreen when the attack initiates.
That was all too much for Google [Nasdaq: GOOG]. In a thread on the Chromium bug tracker a Google developer said, “We’re not treating this as a security bug [because] the preconditions to exploit this are too stretched.”
Later in the thread — started by Kolsek on Sept. 21 when he reported Acros’ findings — the same developer added, “The implausibility of actual exploitation [means] we want to treat this as ‘strange behavior that we should consider changing’ rather than a vulnerability.”
Acros didn’t entirely disagree.
“This is hard to dispute,” Acros said of the required preconditions, and the likelihood hackers would steer for exploits that had a higher probability of succeeding. “[But] as security researchers we consider any ‘feature’ that allows silent downloading of remote code and its execution on user’s computer without warnings a vulnerability.”
Acros raised an interesting question. “How much social engineering is too much?” the company asked in its analysis of the flaw.
That debate isn’t new: Microsoft regularly downgrades the seriousness of vulnerabilities when it decides that “user interaction” — tricking people to do some of an attacker’s job — is involved.
Acros recommended that Chrome users set a secure site — Gmail is one — as their home page to stymie such attacks. While Acros didn’t spell out other options, users can also protect themselves by leaving Google as the browser’s default search engine.