Biblio

Zero-day Web attacks are arguably the most serious threats to Web security, but are very challenging to detect because they are not seen or known previously and thus cannot be detected by widely-deployed signature-based Web Application Firewalls (WAFs). This paper proposes ZeroWall, an unsupervised approach, which works with an existing WAF in pipeline, to effectively detecting zero-day Web attacks. Using historical Web requests allowed by an existing signature-based WAF, a vast majority of which are assumed to be benign, ZeroWall trains a self-translation machine using an encoder-decoder recurrent neural network to capture the syntax and semantic patterns of benign requests. In real-time detection, a zero-day attack request (which the WAF fails to detect), not understood well by self-translation machine, cannot be translated back to its original request by the machine, thus is declared as an attack. In our evaluation using 8 real-world traces of 1.4 billion Web requests, ZeroWall successfully detects real zero-day attacks missed by existing WAFs and achieves high F1-scores over 0.98, which significantly outperforms all baseline approaches.

Web technology has evolved to offer 360-degree immersive browsing experiences. This new technology, called WebVR, enables virtual reality by rendering a three-dimensional world on an HTML canvas. Unfortunately, there exists no browser-supported way of sharing this canvas between different parties. As a result, third-party library providers with ill intent (e.g., stealing sensitive information from end-users) can easily distort the entire WebVR site. To mitigate the new threats posed in WebVR, we propose CanvasMirror, which allows publishers to specify the behaviors of third-party libraries and enforce this specification. We show that CanvasMirror effectively separates the third-party context from the host origin by leveraging the privilege separation technique and safely integrates VR contents on a shared canvas.

The detection of malicious HTTP(S) requests is a pressing concern in cyber security, in particular given the proliferation of HTTP-based (micro-)service architectures. In addition to rule-based systems for known attacks, anomaly detection has been shown to be a promising approach for unknown (zero-day) attacks. This article extends existing work by integrating outlier explanations for individual requests into an end-to-end pipeline. These end-to-end explanations reflect the internal working of the pipeline. Empirically, we show that found explanations coincide with manually labelled explanations for identified outliers, allowing security professionals to quickly identify and understand malicious requests.

Well-known approaches to risk analysis suggest considering the level of an information system risk as one frame in a film. This means that we only can perform a risk assessment for the current point in time. This article explores the idea of risk assessment in a future period, as a prediction of what we will see in the film later. In other words, the article presents an approach to predicting a potential future risk and suggests the idea of relying on forecasting the likelihood of an attack on information system assets. To establish the risk level at a selected time interval in the future, one has to perform a mathematical decomposition. To do this, we need to select the required information system parameters for the predictions and their statistical data for risk assessment. This method can be used to ensure more detailed budget planning when ensuring the protection of the information system. It can be also applied in case of a change of the information protection configuration to satisfy the accepted level of risk associated with projected threats and vulnerabilities.

Web browsers display content in the form of HTML, CSS and JavaScript retrieved from the world wide web. The loaded content is subject to the web security model and considered untrusted and potentially malicious. To complicate security matters, Firefox uses the same technologies to render its user interface as it does to render untrusted web content which blurs the distinction between the two privilege levels.Getting interactions between the two correct turns out to be complicated and has led to numerous real-world security vulnerabilities. We study those vulnerabilities to discover common threats and explain how we address them systematically to harden Firefox.

Nowadays, phishing is one of the most usual web threats with regards to the significant growth of the World Wide Web in volume over time. Phishing attackers always use new (zero-day) and sophisticated techniques to deceive online customers. Hence, it is necessary that the anti-phishing system be real-time and fast and also leverages from an intelligent phishing detection solution. Here, we develop a reliable detection system which can adaptively match the changing environment and phishing websites. Our method is an online and feature-rich machine learning technique to discriminate the phishing and legitimate websites. Since the proposed approach extracts different types of discriminative features from URLs and webpages source code, it is an entirely client-side solution and does not require any service from the third-party. The experimental results highlight the robustness and competitiveness of our anti-phishing system to distinguish the phishing and legitimate websites.

Cross-Site Request Forgery (CSRF) is one of the oldest and simplest attacks on the Web, yet it is still effective on many websites and it can lead to severe consequences, such as economic losses and account takeovers. Unfortunately, tools and techniques proposed so far to identify CSRF vulnerabilities either need manual reviewing by human experts or assume the availability of the source code of the web application. In this paper we present Mitch, the first machine learning solution for the black-box detection of CSRF vulnerabilities. At the core of Mitch there is an automated detector of sensitive HTTP requests, i.e., requests which require protection against CSRF for security reasons. We trained the detector using supervised learning techniques on a dataset of 5,828 HTTP requests collected on popular websites, which we make available to other security researchers. Our solution outperforms existing detection heuristics proposed in the literature, allowing us to identify 35 new CSRF vulnerabilities on 20 major websites and 3 previously undetected CSRF vulnerabilities on production software already analyzed using a state-of-the-art tool.

Web applications, as a conventional platform for sensitive data and important transactions, are of great significance to human society. But with its open source framework, the existing security vulnerabilities can easily be exploited by malicious users, especially when web developers fail to follow the secure practices. Here we present a distributed scanning system, FalconEye, with great precision and high performance, it will help prevent potential threats to Web applications. Besides, our system is also capable of covering basically all the web vulnerabilities registered in the Common Vulnerabilities and Exposures (CVE). The FalconEye system is consists of three modules, an input source module, a scanner module and a support platform module. The input module is used to improve the coverage of target server, and other modules make the system capable of generic vulnerabilities scanning. We then experimentally demonstrate this system in some of the most common vulnerabilities test environment. The results proved that the FalconEye system can be a strong contender among the various detection systems in existence today.

The development of the internet and the web makes human activities more practical, comfortable, and inexpensive. So that the use of the internet and websites is increasing in various ways. Public networks make the security of websites vulnerable to attack. This research proposes a Honeypot for server security against attackers who want to steal data by carrying out a brute force attack. In this research, Honeypot is integrated on the server to protect the server by creating a shadow server. This server is responsible for tricking the attacker into not being able to enter the original server. Brute force attacks tested using Medusa tools. With the application of Honeypot on the server, it is proven that the server can be secured from the attacker. Even the log of activities carried out by the attacker in the shadow server is stored in the Kippo log activities.

Web browser warnings should help protect people from malware, phishing, and network attacks. Adhering to warnings keeps people safer online. Recent improvements in warning design have raised adherence rates, but they could still be higher. And prior work suggests many people still do not understand them. Thus, two challenges remain: increasing both comprehension and adherence rates. To dig deeper into user decision making and comprehension of warnings, we performed an experience sampling study of web browser security warnings, which involved surveying over 6,000 Chrome and Firefox users in situ to gather reasons for adhering or not to real warnings. We find these reasons are many and vary with context. Contrary to older prior work, we do not find a single dominant failure in modern warning design—like habituation—that prevents effective decisions. We conclude that further improvements to warnings will require solving a range of smaller contextual misunderstandings.

The complex architecture of browser technologies and dynamic characteristics of JavaScript make it difficult to ensure security in client-side web applications. Browser-level security policies alone are not sufficient because it is difficult to apply them correctly and they can be bypassed. As a result, they need to be completed by application-level security policies. In this paper, we survey existing solutions for specifying and enforcing application-level security policies for client-side web applications, and distill a number of desirable features. Based on these features we developed Guardia, a framework for declaratively specifying and dynamically enforcing application-level security policies for JavaScript web applications without requiring VM modifications. We describe Guardia enforcement mechanism by means of JavaScript reflection with respect to three important security properties (transparency, tamper-proofness, and completeness). We also use Guardia to specify and deploy 12 access control policies discussed in related work in three experimental applications that are representative of real-world applications. Our experiments indicate that Guardia is correct, transparent, and tamper-proof, while only incurring a reasonable runtime overhead.

Web security is a big concern in the current Internet; users may visit websites that automatically download malicious codes for leaking user's privacy information, or even mildly their web browser may help for someone's cryptomining. In this paper, we analyze abusive web resources (i.e. malicious resources and cryptomining) crawled from the Alexa Top 150,000 sites. We highlight the abusive web resources on Alexa ranking, TLD usage, website geolocation, and domain lifetime. Our results show that abusive resources are spread in the Alexa ranking, websites particularly generic Top Level Domain (TLD) and their recently registered domains. In addition, websites with malicious resources are mainly located in China while cryptomining is located in USA. We further evaluate possible counter-measures against abusive web resources. We observe that ad or privacy block lists are ineffective to block against malicious resources while coin-blocking lists are powerful enough to mitigate in-browser cryptomining. Our observations shed light on a little studied, yet important, aspect of abusive resources, and can help increase user awareness about the malicious resources and drive-by mining on web browsers.

Today, maintaining the security of the web application is of great importance. Sites Intermediate Script (XSS) is a security flaw that can affect web applications. This error allows an attacker to add their own malicious code to HTML pages that are displayed to the user. Upon execution of the malicious code, the behavior of the system or website can be completely changed. The XSS security vulnerability is used by attackers to steal the resources of a web browser such as cookies, identity information, etc. by adding malicious Java Script code to the victim's web applications. Attackers can use this feature to force a malicious code worker into a Web browser of a user, since Web browsers support the execution of embedded commands on web pages to enable dynamic web pages. This work has been proposed as a technique to detect and prevent manipulation that may occur in web sites, and thus to prevent the attack of Site Intermediate Script (XSS) attacks. Ayrica has developed four different languages that detect XSS explanations with Asp.NET, PHP, PHP and Ruby languages, and the differences in the detection of XSS attacks in environments provided by different programming languages.

Web applications are now considered one of the common platforms to represent data and conducting service releases throughout the World Wide Web. A number of the most commonly utilised frameworks for web applications are written in PHP. They became main targets because a vast number of servers are running these applications throughout the world. This increase in web application utilisation has made it more attractive to both users and hackers. According to the latest web security reports and research, cross site scripting (XSS) is the most popular vulnerability in PHP web application. XSS is considered an injection type of attack, which results in the theft of sensitive data, cookies, and sessions. Several tools and approaches have focused on detecting this kind of vulnerability in PHP source code. However, it is still a current problem in PHP web applications. This paper describes the popularity of PHP technology among other technologies, and highlight the approaches used to detect the most common vulnerabilities on PHP web applications, which is XSS. In addition, the discussion and the conclusion with future direction of research within this domain are highlighted.

Cross-site scripting (XSS) is a scripting attack targeting web applications by injecting malicious scripts into web pages. Blind XSS is a subset of stored XSS, where an attacker blindly deploys malicious payloads in web pages that are stored in a persistent manner on target servers. Most of the XSS detection techniques used to detect the XSS vulnerabilities are inadequate to detect blind XSS attacks. In this research, we present machine learning based approach to detect blind XSS attacks. Testing results help to identify malicious payloads that are likely to get stored in databases through web applications.

\textbackslashtextbackslashtextitBackground: JavaScript frameworks are widely used to create client-side and server-side parts of contemporary web applications. Vulnerabilities like cross-site scripting introduce significant risks in web applications.\textbackslashtextbackslash\textbackslashtextbackslash \textbackslashtextbackslashtextitAim: The goal of our study is to understand how the security features of a framework impact the security of the applications written using that framework.\textbackslashtextbackslash\textbackslashtextbackslash \textbackslashtextbackslashtextitMethod: In this paper, we present four locations in an application, relative to the framework being used, where a mitigation can be applied. We perform an empirical study of JavaScript applications that use the three most common template engines: Jade/Pug, EJS, and Angular. Using automated and manual analysis of each group of applications, we identify the number of projects vulnerable to cross-site scripting, and the number of vulnerabilities in each project, based on the framework used.\textbackslashtextbackslash\textbackslashtextbackslash \textbackslashtextbackslashtextitResults: We analyze the results to compare the number of vulnerable projects to the mitigation locations used in each framework and perform statistical analysis of confounding variables.\textbackslashtextbackslash\textbackslashtextbackslash \textbackslashtextbackslashtextitConclusions: The location of the mitigation impacts the application's security posture, with mitigations placed within the framework resulting in more secure applications.

The adoption of the HTTPS - i.e. HTTP over TLS - protocol by the Hellenic websites is studied in this work. Since this protocol constitutes a de-facto standard for secure communications in the web, our aim is to identify whether the underlying TLS protocol in popular websites in Greece is properly configured, so as to avoid known vulnerabilities. To this end, a systematic approach utilizing two well-known TLS scanner tools is adopted to evaluate 241 sites of high popularity. The results illustrate that only about half of the sites seem to be at a satisfactory level and, thus, there is still much room for improvement, mainly due to the fact that obsolete ciphers and/or protocol versions are still supported; there is also a small portion - i.e. about 3% of the sites - that do not implement the HTTPS at all, thus posing very high security risks for their users who provide their credentials via a totally insecure channel. We also examined, using an appropriate online questionnaire, whether the users are actually aware of what the HTTPS means and how they check the security of the websites. The outcome of this research shows that much work needs to be done to increase the knowledge and the security awareness of an average Internet user.

Direct access to the system's resources such as the GPU, persistent storage and networking has enabled in-browser crypto-mining. Thus, there has been a massive response by rogue actors who abuse browsers for mining without the user's consent. This trend has grown steadily for the last months until this practice, i.e., CryptoJacking, has been acknowledged as the number one security threat by several antivirus companies. Considering this, and the fact that these attacks do not behave as JavaScript malware or other Web attacks, we propose and evaluate several approaches to detect in-browser mining. To this end, we collect information from the top 330.500 Alexa sites. Mainly, we used real-life browsers to visit sites while monitoring resourcerelated API calls and the browser's resource consumption, e.g., CPU. Our detection mechanisms are based on dynamic monitoring, so they are resistant to JavaScript obfuscation. Furthermore, our detection techniques can generalize well and classify previously unseen samples with up to 99.99% precision and recall for the benign class and up to 96% precision and recall for the mining class. These results demonstrate the applicability of detection mechanisms as a server-side approach, e.g., to support the enhancement of existing blacklists. Last but not least, we evaluated the feasibility of deploying prototypical implementations of some detection mechanisms directly on the browser. Specifically, we measured the impact of in-browser API monitoring on page-loading time and performed micro-benchmarks for the execution of some classifiers directly within the browser. In this regard, we ascertain that, even though there are engineering challenges to overcome, it is feasible and bene!cial for users to bring the mining detection to the browser.

Nowadays, Cross Site Scripting (XSS) is one of the major threats to Web applications. Since it's known to the public, XSS vulnerability has been in the TOP 10 Web application vulnerabilities based on surveys published by the Open Web Applications Security Project (OWASP). How to effectively detect and defend XSS attacks are still one of the most important security issues. In this paper, we present a novel approach to detect XSS attacks based on deep learning (called DeepXSS). First of all, we used word2vec to extract the feature of XSS payloads which captures word order information and map each payload to a feature vector. And then, we trained and tested the detection model using Long Short Term Memory (LSTM) recurrent neural networks. Experimental results show that the proposed XSS detection model based on deep learning achieves a precision rate of 99.5% and a recall rate of 97.9% in real dataset, which means that the novel approach can effectively identify XSS attacks.

We present research on the security of the most popular websites, ranked according to Alexa's top one million list, based on an HTTP response headers analysis. For each of the domains included in the list, we made four different requests: an HTTP/1.1 request to the domain itself and to its "www" subdomain and two more equivalent HTTPS requests. Redirections were always followed. A detailed discussion of the request process and main outcomes is presented, including X.509 certificate issues and comparison of results with equivalent HTTP/2 requests. The body of the responses was discarded, and the HTTP response header fields were stored in a database. We analysed the prevalence of the most important response headers related to web security aspects. In particular, we took into account Strict- Transport-Security, Content-Security-Policy, X-XSS-Protection, X-Frame-Options, Set-Cookie (for session cookies) and X-Content-Type. We also reviewed the contents of response HTTP headers that potentially could reveal unwanted information, like Server (and related headers), Date and Referrer-Policy. This research offers an up-to-date survey of current prevalence of web security policies implemented through HTTP response headers and concludes that most popular sites tend to implement it noticeably more often than less popular ones. Equally, HTTPS sites seem to be far more eager to implement those policies than HTTP only websites. A comparison with previous works show that web security policies based on HTTP response headers are continuously growing, but still far from satisfactory widespread adoption.

As the era of cloud technology arises, more and more people are beginning to migrate their applications and personal data to the cloud. This makes web-based applications an attractive target for cyber-attacks. As a result, web-based applications now need more protections than ever. However, current anomaly-based web attack detection approaches face the difficulties like unsatisfying accuracy and lack of generalization. And the rule-based web attack detection can hardly fight unknown attacks and is relatively easy to bypass. Therefore, we propose a novel deep learning approach to detect anomalous requests. Our approach is to first train two Recurrent Neural Networks (RNNs) with the complicated recurrent unit (LSTM unit or GRU unit) to learn the normal request patterns using only normal requests unsupervisedly and then supervisedly train a neural network classifier which takes the output of RNNs as the input to discriminate between anomalous and normal requests. We tested our model on two datasets and the results showed that our model was competitive with the state-of-the-art. Our approach frees us from feature selection. Also to the best of our knowledge, this is the first time that the RNN is applied on anomaly-based web attack detection systems.

The de facto approach to Web security today is HTTPS. While HTTPS ensures complete security for clients and servers, it also interferes with transparent content-caching at middleboxes. To address this problem and support both security and caching, we propose a new approach to Web security and privacy called GroupSec. The key innovation of GroupSec is that it replaces the traditional session-based security model with a new model based on content group membership. We introduce the GroupSec security model and show how HTTP can be easily adapted to support GroupSec without requiring changes to browsers, servers, or middleboxes. Finally, we present results of a threat analysis and performance experiments which show that GroupSec achieves notable performance benefits at the client and server while remaining as secure as HTTPS.

JavaScript is the source of many security problems, including cross-site scripting attacks and malicious advertising code. Central to these problems is the fact that code from untrusted sources runs with full privileges. Information flow controls help prevent violations of data confidentiality and integrity. This article explores faceted values, a mechanism for providing information flow security in a dynamic manner that avoids the stuck executions of some prior approaches, such as the no-sensitive-upgrade technique. Faceted values simultaneously simulate multiple executions for different security levels to guarantee termination-insensitive noninterference. We also explore the interaction of faceted values with exceptions, declassification, and clearance.

Malicious crowdsourcing forums are gaining traction as sources of spreading misinformation online, but are limited by the costs of hiring and managing human workers. In this paper, we identify a new class of attacks that leverage deep learning language models (Recurrent Neural Networks or RNNs) to automate the generation of fake online reviews for products and services. Not only are these attacks cheap and therefore more scalable, but they can control rate of content output to eliminate the signature burstiness that makes crowdsourced campaigns easy to detect. Using Yelp reviews as an example platform, we show how a two phased review generation and customization attack can produce reviews that are indistinguishable by state-of-the-art statistical detectors. We conduct a survey-based user study to show these reviews not only evade human detection, but also score high on "usefulness" metrics by users. Finally, we develop novel automated defenses against these attacks, by leveraging the lossy transformation introduced by the RNN training and generation cycle. We consider countermeasures against our mechanisms, show that they produce unattractive cost-benefit tradeoffs for attackers, and that they can be further curtailed by simple constraints imposed by online service providers.

Testing and fixing Web Application Firewalls (WAFs) are two relevant and complementary challenges for security analysts. Automated testing helps to cost-effectively detect vulnerabilities in a WAF by generating effective test cases, i.e., attacks. Once vulnerabilities have been identified, the WAF needs to be fixed by augmenting its rule set to filter attacks without blocking legitimate requests. However, existing research suggests that rule sets are very difficult to understand and too complex to be manually fixed. In this paper, we formalise the problem of fixing vulnerable WAFs as a combinatorial optimisation problem. To solve it, we propose an automated approach that combines machine learning with multi-objective genetic algorithms. Given a set of legitimate requests and bypassing SQL injection attacks, our approach automatically infers regular expressions that, when added to the WAF's rule set, prevent many attacks while letting legitimate requests go through. Our empirical evaluation based on both open-source and proprietary WAFs shows that the generated filter rules are effective at blocking previously identified and successful SQL injection attacks (recall between 54.6% and 98.3%), while triggering in most cases no or few false positives (false positive rate between 0% and 2%).