Despite the conventional wisdom that proactive security is superior to reactive security, we show that reactive security can be competitive with proactive security as long as the reactive defender learns from past attacks instead of myopically overreacting to the last attack. Our game-theoretic model follows common practice in the security literature by making worst case assumptions about the attacker: we grant the attacker complete knowledge of the defender's strategy and do not require the attacker to act rationally. In this model, we bound the competitive ratio between a reactive defense algorithm (which is inspired by online learning theory) and the best fixed proactive defense. Additionally, we show that, unlike proactive defenses, this reactive strategy is robust to a lack of information about the attacker's incentives and knowledge.

Many browser-based attacks can be prevented by using separate browsers for separate web sites. However, most users access the web with only one browser. We explain the security benefits that using multiple browsers provides in terms of two concepts: entry-point restriction and state isolation. We combine these concepts into a general app isolation mechanism that can provide the same security benefits in a single browser. While not appropriate for all types of web sites, many sites with high-value user data can opt in to app isolation to gain defenses against a wide variety of browser-based attacks. We implement app isolation in the Chromium browser and verify its security properties using finite-state model checking. We also measure the performance overhead of app isolation and conduct a large-scale study to evaluate its adoption complexity for various types of sites, demonstrating how the app isolation mechanisms are suitable for protecting a number of high-value Web applications, such as online banking.

We propose a formal model of web security based on an abstraction of the web platform and use this model to analyze the security of several sample web mechanisms and applications. We identify three distinct threat models that can be used to analyze web applications, ranging from a web attacker who controls malicious web sites and clients, to stronger attackers who can control the network and/or leverage sites designed to display user-supplied content. We propose two broadly applicable security goals and study five security mechanisms. In our case studies, which include HTML5 forms, Referer validation, and a single sign-on solution, we use a SAT-based model-checking tool to find two previously known vulnerabilities and three new vulnerabilities. Our case study of a Kerberos-based single sign-on system illustrates the differences between a secure network protocol using custom client software and a similar but vulnerable web protocol that uses cookies, redirects, and embedded links instead.

Cross-site scripting flaws have now surpassed buffer overflows as the world's most common publicly-reported security vulnerability. In recent years, browser vendors and researchers have tried to develop client-side filters to mitigate these attacks. We analyze the best existing filters and find them to be either unacceptably slow or easily circumvented. Worse, some of these filters could introduce vulnerabilities into sites that were previously bug-free. We propose a new filter design that achieves both high performance and high precision by blocking scripts after HTML parsing but before execution. Compared to previous approaches, our approach is faster, protects against more vulnerabilities, and is harder for attackers to abuse. We have contributed an implementation of our filter design to the WebKit open source rendering engine, and the filter is now enabled by default in the Google Chrome browser.

Cross-site scripting defenses often focus on HTML documents, neglecting attacks involving the browser's content-sniffing algorithm, which can treat non-HTML content as HTML. Web applications, such as the one that manages this conference, must defend themselves against these attacks or risk authors uploading malicious papers that automatically submit stellar self-reviews. In this paper, we formulate content-sniffing XSS attacks and defenses. We study content-sniffing XSS attacks systematically by constructing high-fidelity models of the content-sniffing algorithms used by four major browsers. We compare these models with Web site content filtering policies to construct attacks. To defend against these attacks, we propose and implement a principled content-sniffing algorithm that provides security while maintaining compatibility. Our principles have been adopted, in part, by Internet Explorer 8 and, in full, by Google Chrome and the HTML 5 working group.

DNS rebinding attacks subvert the same-origin policy of browsers, converting them into open network proxies. Using DNS rebinding, an attacker can circumvent organizational and personal firewalls, send spam email, and defraud pay-per-click advertisers. We evaluate the cost effectiveness of mounting DNS rebinding attacks, finding that an attacker requires less than $100 to hijack 100,000 IP addresses. We analyze defenses to DNS rebinding attacks, including improvements to the classic “DNS pinning,” and recommend changes to browser plug-ins, firewalls, and Web servers. Our defenses have been adopted by plug-in vendors and by a number of open-source firewall implementations.

Cross-Site Request Forgery (CSRF) is a widely exploited web site vulnerability. In this paper, we present a new variation on CSRF attacks, login CSRF, in which the attacker forges a cross-site request to the login form, logging the victim into the honest web site as the attacker. The severity of a login CSRF vulnerability varies by site, but it can be as severe as a cross-site scripting vulnerability. We detail three major CSRF defense techniques and find shortcomings with each technique. Although the HTTP Referer header could provide an effective defense, our experimental observation of 283,945 advertisement impressions indicates that the header is widely blocked at the network layer due to privacy concerns. Our observations do suggest, however, that the header can be used today as a reliable CSRF defense over HTTPS, making it particularly well-suited for defending against login CSRF. For the long term, we propose that browsers implement the Origin header, which provides the security benefits of the Referer header while responding to privacy concerns.

As wireless networks proliferate, web browsers operate in an increasingly hostile network environment. The HTTPS protocol has the potential to protect web users from network attackers, but real-world deployments must cope with misconfigured servers, causing imperfect web sites and users to compromise browsing sessions inadvertently. ForceHTTPS is a simple browser security mechanism that web sites or users can use to opt in to stricter error processing, improving the security of HTTPS by preventing network attacks that leverage the browser's lax error processing. By augmenting the browser with a database of custom URL rewrite rules, ForceHTTPS allows sophisticated users to transparently retrofit security onto some insecure sites that support HTTPS. We provide a prototype implementation of ForceHTTPS as a Firefox browser extension.

DNS rebinding attacks subvert the same-origin policy of browsers and convert them into open network proxies. We survey new DNS rebinding attacks that exploit the interaction between browsers and their plug-ins, such as Flash and Java. These attacks can be used to circumvent firewalls and are highly cost-effective for sending spam e-mail and defrauding pay-per-click advertisers, requiring less than $100 to temporarily hijack 100,000 IP addresses. We show that the classic defense against these attacks, called "DNS pinning," is ineffective in modern browsers. The primary focus of this work, however, is the design of strong defenses against DNS rebinding attacks that protect modern browsers: we suggest easy-to-deploy patches for plug-ins that prevent large-scale exploitation, provide a defense tool, dnswall, that prevents firewall circumvention, and detail two defense options, policy-based pinning and host name authorization.

We propose an abstract model of business processes for the purpose of (i) evaluating privacy policy in light of the goals of the process and (ii) developing automated support for privacy policy compliance and audit. In our model, agents that send and receive tagged personal information are assigned organizational roles and responsibilities. We present approaches and algorithms for determining whether a business process design simultaneously achieves privacy and the goals of the organization (utility). The model also allows us to develop a notion of minimal exposure of personal information, for a given process. We investigate the problem of auditing with inexact information and develop methods to identify a set of potentially culpable individuals when privacy is breached. The audit methods draw on traditional causality concepts to reduce the effort needed to search audit logs for irresponsible actions.

Digital music players protect songs by enforcing licenses that convey specific rights for individual songs or groups of songs. For licenses specified in industry, we show that deciding whether a license authorizes a sequence of actions is NP-complete, with a restricted version of the problem solvable efficiently using a reduction to maximum network flow. The authorization algorithm used in industry is online, deciding which rights to exercise as actions occur, but we show that all online algorithms are necessarily non-monotonic: each allows actions under one license that it does not allow under a more flexible license. In one approach to achieving monotonicity, we exhibit the unique maximal set of licenses on which there exists a monotonic online algorithm. This set of well-behaved licenses induces an approximation algorithm by replacing each license with a well-behaved license. In a second approach, we consider allowing the player to revise its past decisions about which rights to exercise while still ensuring compliance with the license. We propose an efficient algorithm based on linear logic, with linear negation used to revise past decisions. We prove our algorithm monotonic, live, and sound with respect to the semantics of licenses

Contextual integrity is a conceptual framework for understanding privacy expectations and their implications developed in the literature on law, public policy, and political philosophy. We formalize some aspects of contextual integrity in a logical framework for expressing and reasoning about norms of transmission of personal information. In comparison with access control and privacy policy frameworks such as RBAC, EPAL, and P3P, these norms focus on who personal information is about, how it is transmitted, and past and future actions by both the subject and the users of the information. Norms can be positive or negative depending on whether they refer to actions that are allowed or disallowed. Our model is expressive enough to capture naturally many notions of privacy found in legislation, including those found in HIPAA, COPPA, and GLBA. A number of important problems regarding compliance with privacy norms, future requirements associated with specific actions, and relations between policies and legal standards reduce to standard decision procedures for temporal logic.

The advent of inexpensive digital image sensors and the ability to create photographs that combine information from a number of sensed images are changing the way we think about photography. In this paper, we describe a unique array of 100 custom video cameras that we have built, and we summarize our experiences using this array in a range of imaging applications. Our goal was to explore the capabilities of a system that would be inexpensive to produce in the future. With this in mind, we used simple cameras, lenses, and mountings, and we assumed that processing large numbers of images would eventually be easy and cheap. The applications we have explored include approximating a conventional single center of projection video camera with high performance along one or more axes, such as resolution, dynamic range, frame rate, and/or large aperture, and using multiple cameras to approximate a video camera with a large synthetic aperture. This permits us to capture a video light field, to which we can apply spatiotemporal view interpolation algorithms in order to digitally simulate time dilation and camera motion. It also permits us to create video sequences using custom non-uniform synthetic apertures.