Peer-reviewed publications

2019

LATEX GLOVES: Protecting Browser Extensions from Probing and Revelation Attacks
Alexander Sjösten, Steven Van Acker, Pablo Picazo-Sanchez, Andrei Sabelfeld
» Proceedings of the 26th Annual Network and Distributed System Security Symposium (NDSS 2019)
» Abstract:
Browser extensions enable rich experience for the users of today’s web. Being deployed with elevated privileges, extensions are given the power to overrule web pages. As a result, web pages often seek to detect the installed extensions, sometimes for benign adoption of their behavior but sometimes as part of privacy-violating user fingerprinting. Researchers have studied a class of attacks that allow detecting extensions by probing for Web Accessible Resources (WARs) via URLs that include public extension IDs. Realizing privacy risks associated with WARs, Firefox has recently moved to randomize a browser extension’s ID, prompting the Chrome team to plan for following the same path. However, rather than mitigating the issue, the randomized IDs can in fact exacerbate the extension detection problem, enabling attackers to use a randomized ID as a reliable fingerprint of a user. We study a class of extension revelation attacks, where extensions reveal themselves by injecting their code on web pages. We demonstrate how a combination of revelation and probing can uniquely identify 90% out of all extensions injecting content, in spite of a randomization scheme. We perform a series of large-scale studies to estimate possible implications of both classes of attacks. As a countermeasure, we propose a browser-based mechanism that enables control over which extensions are loaded on which web pages and present a proof of concept implementation which blocks both classes of attacks.

2018

Raising the Bar: Evaluating Origin-wide Security Manifests
Steven Van Acker, Daniel Hausknecht, Andrei Sabelfeld
» Proceedings of the 34th Annual Computer Security Applications Conference (ACSAC 2018)
» Abstract:
Defending a web application from attackers requires the correct configuration of several web security mechanisms for each and every web page in that web application. This configuration process can be difficult and result in gaps in the defense against web attackers because some web pages may be overlooked. In this work we provide a first evaluation of the standard draft for an origin-wide security configuration mechanism called the "origin manifest". The mechanism raises the security level of an entire web origin at once while still allowing the specification of web security policies at the web page level. We create prototype implementations of the origin manifest mechanism for both the client-side and server-side, and provide security officers with an automated origin manifest learner and generator to aid them with the configuration of their web origins. To resolve potential collisions of policies defined by the web origin with policies defined by web pages we formalize the comparison and combination of web security policies and integrate it into our prototype implementation. We evaluate the feasibility of the origin manifest mechanism with a longitudinal study of popular websites to determine whether origin manifest files are stable enough to not require frequent reconfiguration, and perform performance measurements on the Alexa top 10,000 to determine the network traffic overhead. Our results show that the origin manifest mechanism can effectively raise the security level of a web origin while slightly improving network performance.

2017

Measuring Login Webpage Security
Steven Van Acker, Daniel Hausknecht, Andrei Sabelfeld
» The Security Track at the ACM Symposium on Applied Computing (SEC@SAC 2017)
» Abstract:
Login webpages are the entry points into sensitive parts of web applications, dividing between public access to a website and private, user-specific, access to the website resources. As such, these entry points must be guarded with great care. A vast majority of today’s websites relies on text-based username/password pairs for user authentication. While much prior research has focused on the strengths and weaknesses of textual passwords, this paper puts a spotlight on the security of the login webpages themselves. We conduct an empirical study of the Alexa top 100,000 pages to identify login pages and scrutinize their security. Our findings show several widely spread vulnerabilities, such as possibilities for password leaks to third parties and password eavesdropping on the network. They also show that only a scarce number of login pages deploy advanced security measures. Our findings on open-source web frameworks and content management systems confirm the lack of support against the login attacker. To ameliorate the problematic state of the art, we discuss measures to improve the security of login pages.
Discovering Browser Extensions via Web Accessible Resources
Alexander Sjösten, Steven Van Acker, Andrei Sabelfeld
» Proceedings of the 7th ACM Conference on Data and Application Security and Privacy (CODASPY 2017)
» Abstract:
Browser extensions provide a powerful platform to enrich browsing experience. At the same time, they raise important security questions. From the point of view of a website, some browser extensions are invasive, removing intended features and adding unintended ones, e.g. extensions that hijack Facebook likes. Conversely, from the point of view of extensions, some websites are invasive, e.g. websites that bypass ad blockers. Motivated by security goals at clash, this paper explores browser extension discovery, through a non-behavioral technique, based on detecting extensions' web accessible resources. We report on an empirical study with free Chrome and Firefox extensions, being able to detect over 50% of the top 1,000 free Chrome extensions, including popular security- and privacy-critical extensions such as AdBlock, LastPass, Avast Online Security, and Ghostery. We also conduct an empirical study of non-behavioral extension detection on the Alexa top 100,000 websites. We present the dual measures of making extension detection easier in the interest of websites and making extension detection more difficult in the interest of extensions. Finally, we discuss a browser architecture that allows a user to take control in arbitrating the conflicting security goals.

2016

Data Exfiltration in the Face of CSP
Steven Van Acker, Daniel Hausknecht, Andrei Sabelfeld
» Proceedings of the 11th ACM on Asia Conference on Computer and Communications Security (AsiaCCS 2016)
» Abstract:
Cross-site scripting (XSS) attacks keep plaguing the Web. Supported by most modern browsers, Content Security Policy (CSP) prescribes the browser to restrict the features and communication capabilities of code on a web page, mitigating the effects of XSS.
This paper puts a spotlight on the problem of data exfiltration in the face of CSP. We bring attention to the unsettling discord in the security community about the very goals of CSP when it comes to preventing data leaks.
As consequences of this discord, we report on insecurities in the known protection mechanisms that are based on assumptions about CSP that turn out not to hold in practice.
To illustrate the practical impact of the discord, we perform a systematic case study of data exfiltration via DNS prefetching and resource prefetching in the face of CSP.
Our study of the popular browsers demonstrates that it is often possible to exfiltrate data by both resource prefetching and DNS prefetching in the face of CSP. Further, we perform a crawl of the top 10,000 Alexa domains to report on the cohabitance of CSP and prefetching in practice. Finally, we discuss directions to control data exfiltration and, for the case study, propose measures ranging from immediate fixes for the clients to prefetching-aware extensions of CSP.
JavaScript Sandboxing: Isolating and Restricting Client-Side JavaScript
Steven Van Acker, Andrei Sabelfeld
» Foundations of Security Analysis and Design VIII (FOSAD VIII 2016)
» Abstract:
Today’s web applications rely on the same-origin policy, the primary security policy of the Web, to isolate their web origin from malicious client-side JavaScript. When an attacker can somehow breach the same-origin policy and execute JavaScript code inside a web application’s origin, he gains full control over all available functionality and data in that web origin. In the JavaScript sandboxing field, we assume that an attacker has the ability to execute JavaScript code in a web application’s origin. The goal of JavaScript sandboxing is to isolate the execution of certain JavaScript code and restrict what functionality and data is available to it. In this paper we discuss proposed JavaScript sandboxing systems divided into three categories: JavaScript sandboxing through JavaScript subsets and rewriting systems, JavaScript sandboxing using browser modifications and JavaScript sandboxing without browser modifications.

2015

Password meters and generators on the web: From large-scale empirical study to getting it right
Steven Van Acker, Daniel Hausknecht, Wouter Joosen, Andrei Sabelfeld
» Proceedings of the 5th ACM Conference on Data and Application Security and Privacy (CODASPY 2015)
» Abstract:
Web services heavily rely on passwords for user authentication. To help users chose stronger passwords, password meter and password generator facilities are becoming increasingly popular. Password meters estimate the strength of passwords provided by users. Password generators help users with generating stronger passwords. This paper turns the spotlight on the state of the art of password meters and generators on the web. Orthogonal to the large body of work on password metrics, we focus on getting password meters and generators right in the web setting. We report on the state of affairs via a large-scale empirical study of web password meters and generators. Our findings reveal pervasive trust to third-party code to have access to the passwords. We uncover three cases when this trust is abused to leak the passwords to third parties. Furthermore, we discover that often the passwords are sent out to the network, invisibly to users, and sometimes in clear. To improve the state of the art, we propose SandPass, a general web framework that allows secure and modular porting of password meter and generation modules. We demonstrate the usefulness of the framework by a reference implementation and a case study with a password meter by the Swedish Post and Telecommunication Agency.
Isolating and Restricting Client-Side JavaScript
Steven Van Acker
» PhD Text
» Abstract:
In today's web applications, no one disputes the important role of JavaScript asa client-side programming language. JavaScript can turn the Web into a lively, dynamic and interactive end-user experience. Unfortunately, JavaScript canalso be used to steal sensitive information and abuse powerful functionality. Sloppy input validation can make a web application vulnerable, allowingmalicious JavaScript code to leak into a web application's JavaScript executionenvironment, where it leads to unintended code execution. An otherwise secure web application may intentionally include JavaScript froma third-party script provider. This script provider may in turn serve untrustedor even malicious JavaScript, leading to the intended execution of untrustedcode. In both the intended and unintended case, untrusted JavaScript ending up inthe JavaScript execution environment of a trusted web application, gains accessto sensitive resources and powerful functionality. Web application securitywould be greatly improved if this untrusted JavaScript could be isolated and itsaccess restricted. In this work, we first investigate ways in which JavaScript code can leak into thebrowser, leading to unintended JavaScript execution. We find that, due to badinput validation, malicious JavaScript code can be injected into a JavaScriptexecution environment through both browser plugins and browser extensions. Next, we review JavaScript sandboxing systems designed to isolate and restrictuntrusted JavaScript code and divide them into three categories, discussingtheir advantages and disadvantages: JavaScript subsets and rewriting systems,JavaScript sandboxing through browser modifications and JavaScript sandboxingsystems without browser modifications. We further research the last twocategories, developing and evaluating a prototype of each.

2014

Monkey-in-the-browser: malware and vulnerabilities in augmented browsing script markets
Steven Van Acker, Nick Nikiforakis, Lieven Desmet, Frank Piessens, Wouter Joosen
» Proceedings of the 9th ACM symposium on Information, computer and communications security (AsiaCCS 2014)
» Abstract:
With the constant migration of applications from the desktop to the web, power users have found ways of enhancing web applications, at the client-side, according to their needs.
In this paper, we investigate this phenomenon by focusing on the popular Greasemonkey extension which enables users to write scripts that arbitrarily change the content of any page, allowing them to remove unwanted features from web applications, or add additional, desired features to them. The creation of script markets, on which these scripts are often shared, extends the standard web security model with two new actors, introducing novel vulnerabilities.
We describe the architecture of Greasemonkey and perform a large-scale analysis of the most popular, community-driven, script market for Greasemonkey. Through our analysis, we discover not only dozens of malicious scripts waiting to be installed by users, but thousands of benign scripts with vulnerabilities that could be abused by attackers. In 58 cases, the vulnerabilities are so severe, that they can be used to bypass the Same-Origin Policy of the user's browser and steal sensitive user-data from all sites. We verify the practicality of our attacks, by developing a proof-of-concept exploit against a vulnerable user script with an installation base of 1.2 million users, equivalent to a "Man-in-the-browser" attack.

2013

Bitsquatting: Exploiting bit-flips for fun, or profit?
Nick Nikiforakis, Steven Van Acker, Wannes Meert, Lieven Desmet, Frank Piessens, Wouter Joosen
» Proceedings of the 22nd international conference on World Wide Web (WWW 2013)
» Abstract:
Over the last fifteen years, several types of attacks against domain names and the companies relying on them have been observed. The well-known cybersquatting of domain names gave way to typosquatting, the abuse of a user's mistakes when typing a URL in her browser's address bar. Recently, a new attack against domain names surfaced, namely bitsquatting. In bitsquatting, an attacker leverages random bit-errors occurring in the memory of commodity computers and smartphones, to redirect Internet traffic to attacker-controlled domains.
In this paper, we report on a large-scale experiment, measuring the adoption of bitsquatting by the domain-squatting community through the tracking of registrations of bitsquatting domains targeting popular web sites over a 9-month period. We show how new bitsquatting domains are registered daily and how attackers are trying to monetize their domains through the use of ads, abuse of affiliate programs and even malware installations. Lastly, given the discovered prevalence of bitsquatting, we review possible defense measures that companies, software developers and Internet Service Providers can use to protect against it.

2012

JSand: complete client-side sandboxing of third-party JavaScript without browser modifications
Pieter Agten, Steven Van Acker, Yoran Brondsema, Phu H Phung, Lieven Desmet, Frank Piessens
» Proceedings of the 28th Annual Computer Security Applications Conference (ACSAC 2012)
» Abstract:
The inclusion of third-party scripts in web pages is a common practice. A recent study has shown that more than half of the Alexa top 10000 sites include scripts from more than 5 different origins. However, such script inclusions carry risks, as the included scripts operate with the privileges of the including website.
We propose JSand, a server-driven but client-side JavaScript sandboxing framework. JSand requires no browser modifications: the sandboxing framework is implemented in JavaScript and is delivered to the browser by the websites that use it. Enforcement is done entirely at the client side: JSand enforces a server-specified policy on included scripts without requiring server-side filtering or rewriting of scripts. Most importantly, JSand is complete: access to all resources is mediated by the sandbox.
We describe the design and implementation of JSand, and we show that it is secure, backwards compatible, and that it performs sufficiently well.
You are what you include: large-scale evaluation of remote javascript inclusions
Nick Nikiforakis, Luca Invernizzi, Alexandros Kapravelos, Steven Van Acker, Wouter Joosen, Christopher Kruegel, Frank Piessens, Giovanni Vigna
» Proceedings of the 2012 ACM conference on Computer and Communications Security (CCS 2012)
» Abstract:
JavaScript is used by web developers to enhance the interactivity of their sites, offload work to the users' browsers and improve their sites' responsiveness and user-friendliness, making web pages feel and behave like traditional desktop applications. An important feature of JavaScript, is the ability to combine multiple libraries from local and remote sources into the same page, under the same namespace. While this enables the creation of more advanced web applications, it also allows for a malicious JavaScript provider to steal data from other scripts and from the page itself. Today, when developers include remote JavaScript libraries, they trust that the remote providers will not abuse the power bestowed upon them.
In this paper, we report on a large-scale crawl of more than three million pages of the top 10,000 Alexa sites, and identify the trust relationships of these sites with their library providers. We show the evolution of JavaScript inclusions over time and develop a set of metrics in order to assess the maintenance-quality of each JavaScript provider, showing that in some cases, top Internet sites trust remote providers that could be successfully compromised by determined attackers and subsequently serve malicious JavaScript. In this process, we identify four, previously unknown, types of vulnerabilities that attackers could use to attack popular web sites. Lastly, we review some proposed ways of protecting a web application from malicious remote scripts and show that some of them may not be as effective as previously thought.
Exploring the ecosystem of referrer-anonymizing services
Nick Nikiforakis, Steven Van Acker, Frank Piessens, Wouter Joosen
» International Symposium on Privacy Enhancing Technologies Symposium (PETS 2012)
» Abstract:
The constant expansion of the World Wide Web allows users to enjoy a wide range of products and services delivered directly to their browsers. At the same time however, this expansion of functionality is usually coupled with more ways of attacking a user’s security and privacy. In this arms race, certain web-services present themselves as privacy-preserving or privacy-enhancing. One type of such services is a Referrer-Anonymizing Service (RAS), a service which relays users from a source site to a destination site while scrubbing the contents of the referrer header from user requests. In this paper, we investigate the ecosystem of RASs and how they interact with web-site administrators and visiting users. We discuss their workings, what happens behind the scenes and how top Internet sites react to traffic relayed through such services. In addition, we present user statistics from our own Referrer-Anonymizing Service and show the leakage of private information by others towards advertising agencies as well as towards ‘curious’ RAS owners.
FlashOver: Automated discovery of cross-site scripting vulnerabilities in rich internet applications
Steven Van Acker, Nick Nikiforakis, Lieven Desmet, Wouter Joosen, Frank Piessens
» Proceedings of the 7th ACM Symposium on Information, Computer and Communications Security (AsiaCCS 2012)
» Abstract:
Today’s Internet is teeming with dynamic web applications visited by numerous Internet users. During their visits, typical Web users will unknowingly use tens of Rich Internet Applications like Flash banners or media players. For HTML-based web applications, it is well-known that Cross- site Scripting (XSS) vulnerabilities can be exploited to steal credentials or otherwise wreak havoc, and there is a lot of research into solving this problem. An aspect of this problem that seems to have been mostly overlooked by the academic community, is that XSS vulnerabilities also exist in Adobe Flash applications, and are actually easier to exploit because they do not require an enclosing HTML ecosystem. In this paper we present FlashOver, a system to automatically scan Rich Internet Applications for XSS vulnerabilities by using a combination of static and dynamic code analysis that reports no false positives. FlashOver was used in a large-scale experiment to analyze Flash applications found on the top 1,000 Internet sites, exposing XSS vulnerabilities that could compromise 64 of those sites, of which six are in the top 50.

2011

Webjail: Least-privilege integration of third-party components in web mashups
Steven Van Acker, Philippe De Ryck, Lieven Desmet, Frank Piessens, Wouter Joosen
» Proceedings of the 27th Annual Computer Security Applications Conference (ACSAC 2011)
» Abstract:
In the last decade, the Internet landscape has transformed from a mostly static world into Web 2.0, where the use of web applications and mashups has become a daily routine for many Internet users. Web mashups are web applications that combine data and functionality from several sources or components. Ideally, these components contain benign code from trusted sources. Unfortunately, the reality is very different. Web mashup components can misbehave and perform unwanted actions on behalf of the web mashup's user.
Current mashup integration techniques either impose no restrictions on the execution of a third-party component, or simply rely on the Same-Origin Policy. A least-privilege approach, in which a mashup integrator can restrict the functionality available to each component, can not be implemented using the current integration techniques, without ownership over the component's code.
We propose WebJail, a novel client-side security architecture to enable least-privilege integration of components into a web mashup, based on high-level policies that restrict the available functionality in each individual component. The policy language was synthesized from a study and categorization of sensitive operations in the upcoming HTML 5 JavaScript APIs, and full mediation is achieved via the use of deep aspects in the browser.
We have implemented a prototype of WebJail in Mozilla Firefox 4.0, and applied it successfully to mainstream platforms such as iGoogle and Facebook. In addition, microbenchmarks registered a negligible performance penalty for page load-time (7ms), and the execution overhead in case of sensitive operations (0.1ms).
Exposing the lack of privacy in file hosting services
Nick Nikiforakis, Marco Balduzzi, Steven Van Acker, Wouter Joosen, Davide Balzarotti
» Proceedings of the 4th USENIX conference on Large-scale exploits and emergent threats (LEET 2011)
» Abstract:
File hosting services (FHSs) are used daily by thousands of people as a way of storing and sharing files. These services normally rely on a security-through-obscurity approach to enforce access control: For each uploaded file, the user is given a secret URI that she can share with other users of her choice. In this paper, we present a study of 100 file hosting services and we show that a significant percentage of them generate secret URIs in a predictable fashion, allowing attackers to enumerate their services and access their file list. Our experiments demonstrate how an attacker can access hundreds of thousands of files in a short period of time, and how this poses a very big risk for the privacy of FHS users. Using a novel approach, we also demonstrate that attackers are aware of these vulnerabilities and are already exploiting them to get access to other users’ files. Finally we present SecureFS, a client-side protection mechanism which can protect a user’s files when uploaded to insecure FHSs, even if the files end up in the possession of attackers.

2010

ValueGuard: Protection of native applications against data-only buffer overflows
Steven Van Acker, Nick Nikiforakis, Pieter Philippaerts, Yves Younan, Frank Piessens
» International Conference on Information Systems Security (ICISS 2010)
» Abstract:
Code injection attacks that target the control-data of an application have been prevalent amongst exploit writers for over 20 years. Today however, these attacks are getting increasingly harder for attackers to successfully exploit due to numerous countermeasures that are deployed by modern operating systems. We believe that this fact will drive exploit writers away from classic control-data attacks and towards data-only attacks. In data-only attacks, the attacker changes key data structures that are used by the program’s logic and thus forces the control flow into existing parts of the program that would be otherwise unreachable, e.g. overflowing into a boolean variable that states whether the current user is an administrator or not and setting it to “true” thereby gaining access to the administrative functions of the program. In this paper we present ValueGuard, a canary-based defense mechanism to protect applications against data-only buffer overflow attacks. ValueGuard inserts canary values in front of all variables and verifies their integrity whenever these variables are used. In this way, if a buffer overflow has occurred that changed the contents of a variable, ValueGuard will detect it since the variable’s canary will have also been changed. The countermeasure itself can be used either as a testing tool for applications before their final deployment or it can be applied selectively to legacy or high-risk parts of programs that we want to protect at run-time, without incurring extra time-penalties to the rest of the applications.