Threats against browsers are getting more sophisticated and branching out into such exotic areas as gaming, experts told attendees at RSA Conference 2008.
New attacks from games and virtual-world Web sites can deliver bot-like control of browsers to attackers, said Ed Skoudis, a security consultant with Intelguardians, speaking at RSA. All that's needed is for the infected image of an avatar to appear. "The character walks into view of the screen, and I take over the box," he said.
Compromised browsers can act as a stage to launch further hacking of computers, Skoudis said. An attack could shut off corrupted machines' keyboard and mouse control, making it more difficult to stop. Or a compromised browser could escalate a machine's network privileges, and even change time stamps in registries to mask the attacks from later forensic investigation, he said.
Browser attacks can be layered so an infected site might divert a browser to another site that barrages it with a broad spectrum of attacks, seeking vulnerabilities to take advantage of, said Rahit Dhamankar, head of security research for TippingPoint Technologies.
Such Web-based attacks can even be more effective than individuals banging away at machines, Dhamankar said. At a recent hacking contest, participants tried to compromise laptops running Vista, Mac and Ubuntu Linux operating systems for an entire day without success. The next day those same machines were allowed to browse the Internet and became infected by Web sites they visited, he said.
Phones with browsers are subject to similar hijacking, Dhamankar said, and he has seen vulnerabilities found in specific phones posted for sale on the Internet.
The vulnerabilities extend to applications that plug into or integrate with browsers, such as flash readers. "They become a large attack surface," said Michael Montecillo, an analyst with EMA attending the conference.
Attacks are carefully crafted, Montecillo said. For instance, a criminal seeking to take over the machines of wealthy people might hack the Web site of a well-heeled church in an affluent community so it downloads malware to vulnerable machines that connect with it. "Such a site exploit might go unnoticed for a long time," he said.
Web 2.0 applications that let site viewers upload content give attackers yet another means to distribute malware downloads, Montecillo said.
A download could contain an entire, Java-based IP stack supporting a VPN endpoint, a tool demonstrated by Dan Kaminsky, a penetration tester for IOActive, Skoudis said. That would give attackers unfettered access to other systems within corporate firewalls. Administrative machines that use browser interfaces to view such transactions on user machines then become at-risk. "Now that becomes a script on an administrative user's browser," which increases the potential damage to a business, he said.
Such downloads also could contain browser scripts that enable attackers to scan other systems on machines to find further vulnerabilities to exploit, Skoudis said. The implications can be dire, depending on the business infected. Web-based cash-management applications in banks, for instance, could be at risk if an attacker can execute actions on an administrator's browser, he said.
"There's not a really solid answer for this," Montecillo said. "Protecting the browser against exploits really isn't there yet."
Defenses for these attacks include using software libraries that notify users that sites are known to be dangerous, as well as keeping antivirus software updated to catch malware as it is downloaded, Montecillo said.