10.3 Malicious Code and Security Risks
Malicious code and security risks present some of the more sinister risks to the election process. Malicious code, such as threats that leverage rootkit capabilities,11 has the potential to gain complete and absolute control over a victim's computer system. Likewise, security risks such as adware and spyware pose serious concerns, both in terms of their invasiveness to a user's privacy (in the case of spyware) and their ability to present users with unexpected or undesired information and advertisements (in the case of adware).
We can consider a number of scenarios where well-known classes of malicious code may be tailored specifically to target those participating in an election. Targets may range from candidates and campaign officials to voters themselves. In discussing these risks we begin with what we consider the less serious category of security risks; we then move into the more serious, insidious category of malicious code.
Adware, in its truest form, may not pose an immediate and dire risk to the end user. Once installed, however, its control over a user's Internet experience places it into a strategic position on the end user's computer. Adware has the potential to manipulate a user's Internet experience by displaying unexpected or unwanted advertisements. These advertisements may be displayed on the user's desktop or shown to the user through the web browser as the user visits Internet web sites. These advertisements may appear as pop-up windows, or they may appear as content (ads) that are either overlaid or inserted into existing web pages visited by the user. Both techniques have been used frequently by such well-known adware applications as 180Solution's Hotbar , Gator Corporation's Gator , and WhenU's Save . Adware may be installed by the end user as part of another third-party application, or it may be installed surreptitiously through the exploitation of a software vulnerability in the user's web browser. Chapter 12 discusses adware in more detail.
Adware might be used in numerous ways to influence or manipulate users during the course of an election. In its most innocuous form, adware might simply present the user with advertisements promoting a particular candidate and directing the user to the candidate's web site when clicked. Taking a more deceptive angle, adware might be used to silently replace advertisements for one candidate with another. This may be done directly in the user's browser by manipulating the incoming HTML content before it is rendered or by overlaying a new advertisement on top of an existing advertisement on the user's screen.
Until it is observed in the wild, it is difficult for us to predict the real-world impact that such an adware application might have. It would be important for such an application to be silent and unobtrusive, acting clandestinely to avoid annoying the end user lest its objective backfire. In addition, such an effort may simply help to sway those voters who have not already committed to a particular party or candidate, rather than those voters who have already made their decision.
We have frequently seen adware and spyware traits combined into a single application that both delivers advertising and monitors a user's Internet habits. For the purposes of our discussion, we chose to distinguish between the distinct behaviors of adware and spyware, discussing each separately. Spyware, with its ability to secretly profile and monitor user behavior, presents an entirely new opportunity for the widespread collection of election-related trend data and behavioral information.
When discussing the use of spyware, we can conceive of a number of behaviors that might be collected throughout the course of an election in an attempt to provide insight into voters' dispositions. The most basic tactic would be to monitor the browsing behavior of voters and to collect the party affiliations of the Internet sites most frequently visited by the end user. Even without the installation of spyware on an end user's computer, one web site may silently acquire a history of other web sites that the user has previously visited. This capability has been demonstrated by researchers in the past and can be observed at https://www.indiana.edu/~phishing/browser-recon. This type of data collection may also include the tracking of online news articles that are viewed and online campaign contributions that are made by determining whether a particular URL was visited.
With the addition of spyware on the end user's computer, these information-gathering efforts can be taken a step further. Emails sent and received by the user can be monitored, for example. In our study, we found that all 19 candidates allow a user to subscribe to their campaign mailing lists, from which a user receives regular frequent updates on the campaign's progress. Knowing how many voters have subscribed to a particular candidate's mailing list may provide insight into the overall support levels for that candidate.
Of course, Internet and browsing behavior alone may not be an indicator of a voter's preference, as voters may be just as likely to visit a competing candidate's web sites and subscribe to a competing candidate's mailing list so as to stay informed about that candidate's campaign. Unfortunately, we could find no prior research that examined the correlation between user Internet behavior and party or candidate affiliation. Nevertheless, spyware clearly poses a new risk in terms of the mass accumulation of election-related statistics that may be used to track election trends.
The collection of voter disposition data is certainly not new, as groups such as the Gallup Organization  (known for the Gallup Poll) have been collecting and analyzing user behavior since 1935. What is different in this case is spyware's ability to capture and record user behavior without consent and without the voter's knowledge. Even when a spyware application's behavior is described clearly in an end-user license agreement (EULA), few users either read or understand these complex and lengthy agreements . This changes the landscape dramatically when it comes to election-related data collection.
10.3.3 Malicious Code: Keyloggers and Crimeware
By far one of the most concerning attacks on voters, candidates, and campaign officials alike is that of malicious code infection. Malicious code that is targeted toward a broad spectrum of voters has the potential to cause widespread damage, confusion, and loss of confidence in the election process itself. When we consider the various types of attacks mentioned in this chapter, malicious code—in the form of keyloggers, trojans, and other forms of crimeware—has the potential to carry each of them out with unmatched efficiency. These attacks include the monitoring of user behavior, the theft of user data, the redirection of user browsing, and the delivery of misinformation.
One additional angle for crimeware is the notion of intimidation. Given a threat's presence on a voter's computer, that threat has the potential to collect personal, potentially sensitive information about that individual. This capability may include turning on the computer's microphone and recording private conversations. It may include turning on the computer's video camera and recording activities in the room. It may include retrieving pictures, browser history documents, or copyrighted files from a voter's computer. Perhaps the individual would be turned in to the RIAA if copyrighted music was found on his or her computer. This kind of information gathering creates the potential for an entirely new form of voter intimidation. The collection of such personally sensitive or legally questionable data by a threat might, therefore, allow an attacker to intimidate that individual in an entirely new way. We would, of course, expect and hope that the number of voters who might be intimidated in such a way would be relatively low, but only time will tell whether such speculation becomes reality.
Another form of threat that we have seen in the past involves holding a victim's data hostage until a fee is paid to release it. This possibility was first discussed in . An example of such a threat is Trojan.Gpcoder , which encrypts the user's data, erasing the original information, until this fee is paid. Such a threat may present another new form of intimidation whereby the only way for a user to regain access to his or her personal data is to vote accordingly. Such an attack presents obvious logistical challenges. For example, how is the attacker to know which way the victim voted? The attacker may, however, take comfort in the belief that he or she has intimidated enough of the infected population to make a meaningful difference.
Just as the widespread infection of the populace's computers poses a risk to voters, the targeted, calculated infection of specific individuals' computers is equal cause for concern. A carefully placed targeted keylogger has the potential to cause material damage to a candidate during the election process. Such code may also be targeted toward campaign staff, family members, or others who may be deemed material to the candidate's efforts. Such an infection might potentially result in the monitoring of all communications, including email messages and web site access initiated on the infected computer. This monitoring would give the would-be attacker unparalleled insight into the progress, plans, and disposition of the candidate's campaign, perhaps including new messaging, speeches, and otherwise sensitive information critical to the candidate's campaign.