Overtaking Google Desktop

This paper describes an innovative attack methodology against Google Desktop which enables a malicious individual to achieve not only remote, persistent access to sensitive data, but full system control as well. This outcome is the result both of the integration between the Google.com Web site and Google Desktop, and Google Desktop's failure to properly encode output containing malicious or unexpected characters.

This represents a significant real world example of a new generation of computer attacks. These attacks take advantage of Web application vulnerabilities and the increasing power of the Web browser. Their purpose is to remotely access private information. Unlike traditional computer penetration attacks, there is no need for binary code to be injected.

In the attack described in this whitepaper, the malicious logic acts as a parasite, using JavaScript code to control Google Desktop functionality. The attacker covertly hijacks confidential information from the system, while evading current information protection systems, such as anti-virus software and firewalls.

The attack also emphasizes the danger of the integration between desktop applications and Web based applications, as this opens an aperture for a malicious attacker to escalate his/her privileges by crossing from the Web environment to the desktop application environment.

In this paper we describe the methodology of attack and provide a valid use case. We include a description of the basic technique and some theoretical outcomes. Finally, we provide fix
recommendations that are appropriate for Google Desktop, as well as for other Web based applications.