Yibo Peng, James Song, Lei Li, Xinyu Yang, Mihai Christodorescu, Ravi Mangal, Corina S. Pasareanu, Haizhong Zheng, Beidi Chen. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Tianyang Zhou, Ziyi Zhang, Haowen Lin, Somesh Jha, Mihai Christodorescu, Kirill Levchenko, Varun Chandrasekaran. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

The major mobile platforms, Android and iOS, have introduced changes that restrict user tracking to improve user privacy, yet apps continue to covertly track users via device fingerprinting. We study the opportunity to improve this dynamic with a case study on mobile fingerprinting that evaluates developers’ perceptions of how well platforms protect user privacy and how developers perceive platform privacy interventions. Specifically, we study developers’ willingness to make changes to protect users from fingerprinting and how developers consider trade-offs between user privacy and developer effort. We do this via a survey of 246 Android developers, presented with a hypothetical Android change that protects users from fingerprinting at the cost of additional developer effort.

This paper presents a large-scale analysis of fingerprinting-like behavior in the mobile application ecosystem. We take a market-based approach, focusing on third-party tracking as enabled by applications' common use of third-party SDKs. Our dataset consists of over 228,000 SDKs from popular Maven repositories, 178,000 Android applications collected from the Google Play store, and our static analysis pipeline detects exfiltration of over 500 individual signals. To the best of our knowledge, this represents the largest-scale analysis of SDK behavior undertaken to date.

Large Language Models’ success in text generation has also made them better at code generation and coding tasks. While a lot of work has demonstrated their remarkable performance on tasks such as code completion and editing, it is still unclear as to why. We help bridge this gap by exploring to what degree auto-regressive models understand the logical constructs of the underlying programs. We propose Counterfactual Analysis for Programming Concept Predicates (CACP) as a counterfactual testing framework to evaluate whether Large Code Models understand programming concepts. With only black-box access to the model, we use CACP to evaluate ten popular Large Code Models for four different programming concepts. Our findings suggest that current models lack understanding of concepts such as data flow and control flow.

Test-time adversarial attacks have posed serious challenges to the robustness of machine-learning models, and in many settings the adversarial perturbation need not be bounded by small $\ell_p$-norms. Motivated by attacks in program analysis and security tasks, we investigate $\textit{relational adversaries}$, a broad class of attackers who create adversarial examples in a reflexive-transitive closure of a logical relation. We analyze the conditions for robustness against relational adversaries and investigate different levels of robustness-accuracy trade-off due to various patterns in a relation. Inspired by the insights, we propose $\textit{normalize-and-predict}$, a learning framework that leverages input normalization to achieve provable robustness. The framework solves the pain points of adversarial training against relational adversaries and can be combined with adversarial training for the benefits of both approaches. Guided by our theoretical findings, we apply our framework to source code authorship attribution and malware detection. Results of both tasks show our learning framework significantly improves the robustness of models against relational adversaries. In the process, it outperforms adversarial training, the most noteworthy defense mechanism, by a wide margin.

Threshold cryptography enables cryptographic operations while keeping the secret keys distributed at all times. Agrawal et al. (CCS'18) propose a framework for Distributed Symmetric-key Encryption (DiSE). They introduce a new notion of Threshold Symmetric-key Encryption (TSE), in that encryption and decryption are performed by interacting with a threshold number of servers. However, the necessity for interaction on each invocation limits performance when encrypting large datasets, incurring heavy computation and communication on the servers.

Widely used payment splitting apps allow members of a group to keep track of debts between members by sending charges for expenses paid by one member on behalf of others. While offering a great deal of convenience, these apps gain access to sensitive data on users’ financial transactions. In this paper, we present a payment splitting app that hides all transaction data within a group from the service provider, provides privacy protections between users in a group, and provides integrity against malicious users or even a malicious server.

Home-based Internet of Things (IoT) devices have gained in popularity and many households became "smart'' by using devices such as smart sensors, locks, and voice-based assistants. Given the limitations of existing authentication techniques, we explore new opportunities for user authentication in smart home environments. Specifically, we design a novel authentication method in IoT-enabled smart homes. We perform initial experiments and a user study leveraging network traffic collected from 8 IoT devices in our university lab. Preliminary results show that our LSTM model achieves a maximum accuracy of 75% in identifying users.

Phones today carry sensitive information and have a great number of ways to communicate that data. As a result, malware that steal money, information, or simply disable functionality have hit the app stores. Current security solutions for preventing undesirable data leaks are mostly high-overhead and have not been practical enough for smartphones. In this paper, we show that simply monitoring just some instructions (only memory loads and stores) it is possible to achieve low overhead, highly accurate information flow tracking. Our method achieves 98% accuracy (0% false positive and 2% false negative) over DroidBench and was able to successfully catch seven real-world malware instances that steal phone number, location, and device ID using SMS messages and HTTP connections.

The sophistication of malicious adversaries is increasing every day and most defenses are often easily overcome by such attackers. Many existing defensive mechanisms often make differing assumptions about the underlying systems and use varied architectures to implement their solutions. This often leads to fragmentation among solutions and could even open up additional vulnerabilities in the system.

Abstract Online trackers compile profiles on users for targeting ads, customizing websites, and selling users’ information. In this paper, we report on the first detailed study of the perceived benefits and risks of tracking-and the reasons behind them-conducted in the context of users’ own browsing histories. Prior work has studied this in the abstract; in contrast, we collected browsing histories from and interviewed 35 people about the perceived benefits and risks of online tracking in the context of their own browsing behavior. We find that many users want more control over tracking and think that controlled tracking has benefits, but are unwilling to put in the effort to control tracking or distrust current tools. We confirm previous findings that users’ general attitudes about tracking are often at odds with their comfort in specific situations. We also identify specific situational factors that contribute to users’ preferences about online tracking and explore how and why. Finally, we examine a sample of popular tools for controlling tracking and show that they only partially address the situational factors driving users’ preferences.We suggest opportunities to improve such tools, and explore the use of a classifier to automatically determine whether a user would be comfortable with tracking on a particular page visit; our results suggest this is a promising direction for future work.

Despite the increasing number of social engineering attacks through web browser applications, detection of socially engineered trojan downloads by enticed victim users remains a challenging endeavor. In this paper, we present TroGuard, a semi-automated web-based trojan detection solution, that notifies the user if the application she downloaded behaves differently than what she expected at download time. TroGuard builds on the hypothesis that in spite of millions of currently downloadable executables on the Internet, almost all of them provide functionalities from a limited set. Additionally, because each functionality, e.g., text editor, requires particular system resources, it exhibits a unique system-level activity pattern. During an offline process, TroGuard creates a profile dictionary of various functionalities. This profile dictionary is then used to warn the user if she downloads an executable whose observed activity does not match its advertised functionality (extracted through automated analysis of the download website). Our experimental results prove the above mentioned premise empirically and show that TroGuard can identify real-world socially engineered trojan download attacks effectively.

Much of the debate surrounding online behavioral advertising (OBA) has centered on how to provide users with notice and choice. An important element left unexplored is how advertising companies' privacy practices affect users' attitudes toward data sharing. We present the results of a 2,912-participant online study investigating how facets of privacy practices---data retention, access to collected data, and scope of use---affect users' willingness to allow the collection of behavioral data. We asked participants to visit a health website, explained OBA to them, and outlined policies governing data collection for OBA purposes. These policies varied by condition. We then asked participants about their willingness to permit the collection of 30 types of information. We identified classes of information that most participants would not share, as well as classes that nearly half of participants would share. More restrictive data-retention and scope-of-use policies increased participants' willingness to allow data collection. In contrast, whether the data was collected on a well-known site and whether users could review and modify their data had minimal impact. We discuss public policy implications and improvements to user interfaces to align with users' privacy preferences.

As more and more systems are controlled by computers and connected to the Internet, a tremendous amount of activity is generated when these systems interact with each other and with their users. The continuous stream of data related to such Internet-connected and computer-controlled systems is known as Big Data and introduces problems at new scales. Computer security and privacy can both benefit and suffer from Big Data, as tools and techniques are made readily available to process security-event streams and as attacks are harder to detect in the huge volume of event data. The BADGERS'12 workshop is a forum for discussing the challenges in and potential solutions for taking advantage of Big Data for improving computer security and privacy. The workshop builds on the successful first edition of BADGERS which was held in conjunction with the EuroSys 2011 conference in Salzburg, Austria [3].

Over the last decade, there has been a significant increase in the number and sophistication of malware-related attacks and infections. Many detection techniques have been proposed to mitigate the malware threat. A running theme among existing detection techniques is the similar promises of high detection rates, in spite of the wildly different models (or specification classes) of malicious activity used. In addition, the lack of a common testing methodology and the limited datasets used in the experiments make difficult to compare these models in order to determine which ones yield the best detection accuracy. In this paper, we present a systematic approach to measure how the choice of behavioral models influences the quality of a malware detector. We tackle this problem by executing a large number of testing experiments, in which we explored the parameter space of over 200 different models, corresponding to more than 220 million of signatures. Our results suggest that commonly held beliefs about simple models are incorrect in how they relate changes in complexity to changes in detection accuracy. This implies that accuracy is non-linear across the model space, and that analytical reasoning is insufficient for finding an optimal model, and has to be supplemented by testing and empirical measurements.

Models based on system calls are a popular and common approach to characterize the run-time behavior of programs. For example, system calls are used by intrusion detection systems to detect software exploits. As another example, policies based on system calls are used to sandbox applications or to enforce access control. Given that malware represents a significant security threat for today's computing infrastructure, it is not surprising that system calls were also proposed to distinguish between benign processes and malicious code.

Fueled by an emerging underground economy, malware authors are exploiting vulnerabilities at an alarming rate. To make matters worse, obfuscation tools are commonly available, and much of the malware is open source, leading to a huge number of variants. Behavior-based detection techniques are a promising solution to this growing problem. However, these detectors require precise specifications of malicious behavior that do not result in an excessive number of false alarms. In this paper, we present an automatic technique for extracting optimally discriminative specifications, which uniquely identify a class of programs. Such a discriminative specification can be used by a behavior-based malware detector. Our technique, based on graph mining and concept analysis, scales to large classes of programs due to probabilistic sampling of the specification space. Our implementation, called Holmes, can synthesize discriminative specifications that accurately distinguish between programs, sustaining an 86% detection rate on new, unknown malware, with 0 false positives, in contrast with 55% for commercial signature-based antivirus (AV) and 62-64% for behavior-based AV (commercial or research).

Graphs are prevalent in many domains such as Bioinformatics, social networks, Web and cyber-security. Graph pattern mining has become an important tool in the management and analysis of complexly structured data, where example applications include indexing, clustering and classification. Existing graph mining algorithms have achieved great success by exploiting various properties in the pattern space . Unfortunately, due to the fundamental role subgraph isomorphism plays in these methods, they may all enter into a pitfall when the cost to enumerate a huge set of isomorphic embeddings blows up, especially in large graphs. The solution we propose for this problem resorts to reduction on the data space . For each graph, we build a summary of it and mine this shrunk graph instead. Compared to other data reduction techniques that either reduce the number of transactions or compress between transactions, this new framework, called Summarize-Mine, suggests a third path by compressing within transactions . Summarize-Mine is effective in cutting down the size of graphs, thus decreasing the embedding enumeration cost. However, compression might lose patterns at the same time. We address this issue by generating randomized summaries and repeating the process for multiple rounds, where the main idea is that true patterns are unlikely to miss from all rounds. We provide strict probabilistic guarantees on pattern loss likelihood. Experiments on real malware trace data show that Summarize-Mine is very efficient, which can find interesting malware fingerprints that were not revealed previously.

Malware detection is a crucial aspect of software security. Current malware detectors work by checking for signatures , which attempt to capture the syntactic characteristics of the machine-level byte sequence of the malware. This reliance on a syntactic approach makes current detectors vulnerable to code obfuscations, increasingly used by malware writers, that alter the syntactic properties of the malware byte sequence without significantly affecting their execution behavior. This paper takes the position that the key to malware identification lies in their semantics. It proposes a semantics-based framework for reasoning about malware detectors and proving properties such as soundness and completeness of these detectors. Our approach uses a trace semantics to characterize the behavior of malware as well as that of the program being checked for infection, and uses abstract interpretation to “hide” irrelevant aspects of these behaviors. As a concrete application of our approach, we show that (1) standard signature matching detection schemes are generally sound but not complete, (2) the semantics-aware malware detector proposed by Christodorescu et al. is complete with respect to a number of common obfuscations used by malware writers and (3) the malware detection scheme proposed by Kinder et al. and based on standard model-checking techniques is sound in general and complete on some, but not all, obfuscations handled by the semantics-aware malware detector.

Malicious software (or malware) has become a growing threat as malware writers have learned that signature- based detectors can be easily evaded by "packing" the malicious payload in layers of compression or encryption. State-of-the-art malware detectors have adopted both static and dynamic techniques to recover the pay- load of packed malware, but unfortunately such techniques are highly ineffective. In this paper we propose a new technique, called OmniUnpack, to monitor the execution of a program in real-time and to detect when the program has removed the various layers of packing. OmniUnpack aids malware detection by directly providing to the detector the unpacked malicious payload. Experimental results demonstrate the effectiveness of our approach. OmniUnpack is able to deal with both known and unknown packing algorithms and introduces a low overhead (at most 11% for packed benign programs).

Malware detection is a crucial aspect of software security. Current malware detectors work by checking for "signatures," which attempt to capture (syntactic) characteristics of the machine-level byte sequence of the malware. This reliance on a syntactic approach makes such detectors vulnerable to code obfuscations, increasingly used by malware writers, that alter syntactic properties of the malware byte sequence without significantly affecting their execution behavior.This paper takes the position that the key to malware identification lies in their semantics. It proposes a semantics-based framework for reasoning about malware detectors and proving properties such as soundness and completeness of these detectors. Our approach uses a trace semantics to characterize the behaviors of malware as well as the program being checked for infection, and uses abstract interpretation to "hide" irrelevant aspects of these behaviors. As a concrete application of our approach, we show that the semantics-aware malware detector proposed by Christodorescu et al. is complete with respect to a number of common obfuscations used by malware writers.

Malware detectors require a specification of malicious behavior. Typically, these specifications are manually constructed by investigating known malware. We present an automatic technique to overcome this laborious manual process. Our technique derives such a specification by comparing the execution behavior of a known malware against the execution behaviors of a set of benign programs. In other words, we mine the malicious behavior present in a known malware that is not present in a set of benign programs. The output of our algorithm can be used by malware detectors to detect malware variants. Since our algorithm provides a succinct description of malicious behavior present in a malware, it can also be used by security analysts for understanding the malware. We have implemented a prototype based on our algorithm and tested it on several malware programs. Experimental results obtained from our prototype indicate that our algorithm is effective in extracting malicious behaviors that can be used to detect malware variants.

Recent research has proposed self-checksumming as a method by which a program can detect any possibly malicious modification to its code. Wurster et al. developed an attack against such programs that renders code modifications undetectable to any self-checksumming routine. The attack replicated pages of program text and altered values in hardware data structures so that data reads and instruction fetches retrieved values from different memory pages. A cornerstone of their attack was its applicability to a variety of commodity hardware: they could alter memory accesses using only a malicious operating system. In this paper, we show that their page-replication attack can be detected by self-checksumming programs with self-modifying code. Our detection is efficient, adding less than 1 microsecond to each checksum computation in our experiments on three processor families, and is robust up to attacks using either costly interpretive emulation or specialized hardware

Existing reputation systems used by online auction houses do not address the concern of a buyer shopping for commodities - finding a good bargain. These systems do not provide information on the practices adopted by sellers to ensure profitable auctions. These practices may be legitimate, like imposing a minimum starting bid on an auction, or fraudulent, like using colluding bidders to inflate the final price in a practice known as shilling.We develop a reputation system to help buyers identify sellers whose auctions seem price-inflated. Our reputation system is based upon models that characterize sellers according to statistical metrics related to price inflation. We combine the statistical models with anomaly detection techniques to identify the set of suspicious sellers. The output of our reputation system is a set of values for each seller representing the confidence with which the system can say that the auctions of the seller are price-inflated.We evaluate our reputation system on 604 high-volume sellers who posted 37,525 auctions on eBay. Our system automatically pinpoints sellers whose auctions contain potential shill bidders. When we manually analyze these sellers' auctions, we find that many winning bids are at about the items' market values, thus undercutting a buyer's ability to find a bargain and demonstrating the effectiveness of our reputation system.

Information about string values at key points in a program can help program understanding, reverse engineering, and forensics. We present a static-analysis technique for recovering possible string values in an executable program, when no debug information or source code is available. The result of our analysis is a regular language that describes a superset of the string values possible at a given program point. We also impart some of the lessons learned in the process of implementing our analysis as a tool for recovering C-style strings in x86 executables.

A malware detector is a system that attempts to determine whether a program has malicious intent. In order to evade detection, malware writers (hackers) frequently use obfuscation to morph malware. Malware detectors that use a pattern-matching approach (such as commercial virus scanners) are susceptible to obfuscations used by hackers. The fundamental deficiency in the pattern-matching approach to malware detection is that it is purely syntactic and ignores the semantics of instructions. In this paper, we present a malware-detection algorithm that addresses this deficiency by incorporating instruction semantics to detect malicious program traits. Experimental evaluation demonstrates that our malware-detection algorithm can detect variants of malware with a relatively low run-time overhead. Moreover our semantics-aware malware detection algorithm is resilient to common obfuscations used by hackers.

In today's interconnected world, malware, such as worms and viruses, can cause havoc. A malware detector (commonly known as virus scanner) attempts to identify malware. In spite of the importance of malware detectors, there is a dearth of testing techniques for evaluating them. We present a technique based on program obfuscation for generating tests for malware detectors. Our technique is geared towards evaluating the resilience of malware detectors to various obfuscation transformations commonly used by hackers to disguise malware. We also demonstrate that a hacker can leverage a malware detector's weakness in handling obfuscation transformations and can extract the signature used by a detector for a specific malware. We evaluate three widely-used commercial virus scanners using our techniques and discover that the resilience of these scanners to various obfuscations is very poor.