Michelle L. Mazurek

Yunze Zhao, Yibo Zhao, Yuchen Zhang et al.

LLM inference and serving systems have become security-critical infrastructure; however, many of their most concerning failures arise from the serving layer rather than from model behavior alone. Modern inference engines combine KV cache, batching, prefix sharing, speculative decoding, adapters, and multi-tenant scheduling, creating shared-state behavior that only emerges under realistic concurrent workloads and is missed by standard model, safety, and API tests. We present GRIEF, a greybox fuzzer for LLM inference engines that treats timed multi-request traces as first-class inputs, uses lightweight oracles to detect crashes, hangs, performance pathologies, and silent output corruption, and applies controlled replay with log-probability checks to confirm reproducible serving-layer failures. Across early campaigns on vLLM and SGLang, GRIEF discovers 15 vulnerabilities, 10 confirmed by engine developers, including 2 CVEs, spanning KV-cache isolation failures, cross-request performance interference, and crash or liveness bugs. These results show that concurrency, caching, and state reuse can induce silent cross-request contamination, noisy-neighbor denial of service, and delayed crashes without malformed inputs or explicit server errors, making concurrent serving behavior a first-class security and reliability boundary for LLM infrastructure.

Yang Trista Cao, Lovely-Frances Domingo, Sarah Ann Gilbert et al.

Extensive efforts in automated approaches for content moderation have been focused on developing models to identify toxic, offensive, and hateful content with the aim of lightening the load for moderators. Yet, it remains uncertain whether improvements on those tasks have truly addressed moderators' needs in accomplishing their work. In this paper, we surface gaps between past research efforts that have aimed to provide automation for aspects of content moderation and the needs of volunteer content moderators, regarding identifying violations of various moderation rules. To do so, we conduct a model review on Hugging Face to reveal the availability of models to cover various moderation rules and guidelines from three exemplar forums. We further put state-of-the-art LLMs to the test, evaluating how well these models perform in flagging violations of platform rules from one particular forum. Finally, we conduct a user survey study with volunteer moderators to gain insight into their perspectives on useful moderation models. Overall, we observe a non-trivial gap, as missing developed models and LLMs exhibit moderate to low performance on a significant portion of the rules. Moderators' reports provide guides for future work on developing moderation assistant models.

Omer Akgul, Sai Teja Peddinti, Nina Taft et al.

We present an analysis of 12 million instances of privacy-relevant reviews publicly visible on the Google Play Store that span a 10 year period. By leveraging state of the art NLP techniques, we examine what users have been writing about privacy along multiple dimensions: time, countries, app types, diverse privacy topics, and even across a spectrum of emotions. We find consistent growth of privacy-relevant reviews, and explore topics that are trending (such as Data Deletion and Data Theft), as well as those on the decline (such as privacy-relevant reviews on sensitive permissions). We find that although privacy reviews come from more than 200 countries, 33 countries provide 90% of privacy reviews. We conduct a comparison across countries by examining the distribution of privacy topics a country's users write about, and find that geographic proximity is not a reliable indicator that nearby countries have similar privacy perspectives. We uncover some countries with unique patterns and explore those herein. Surprisingly, we uncover that it is not uncommon for reviews that discuss privacy to be positive (32%); many users express pleasure about privacy features within apps or privacy-focused apps. We also uncover some unexpected behaviors, such as the use of reviews to deliver privacy disclaimers to developers. Finally, we demonstrate the value of analyzing app reviews with our approach as a complement to existing methods for understanding users' perspectives about privacy

Michael Coblenz, Michelle Mazurek, Michael Hicks

Rust is a general-purpose programming language that is both type- and memory-safe. Rust does not use a garbage collector, but rather achieves these properties through a sophisticated, but complex, type system. Doing so makes Rust very efficient, but makes Rust relatively hard to learn and use. We designed Bronze, an optional, library-based garbage collector for Rust. To see whether Bronze could make Rust more usable, we conducted a randomized controlled trial with volunteers from a 633-person class, collecting data from 428 students in total. We found that for a task that required managing complex aliasing, Bronze users were more likely to complete the task in the time available, and those who did so required only about a third as much time (4 hours vs. 12 hours). We found no significant difference in total time, even though Bronze users re-did the task without Bronze afterward. Surveys indicated that ownership, borrowing, and lifetimes were primary causes of the challenges that users faced when using Rust.

Noel Warford, Collins W. Munyendo, Ashna Mediratta et al.

People are frequently required to send documents, forms, or other materials containing sensitive data (e.g., personal information, medical records, financial data) to remote parties, sometimes without a formal procedure to do so securely. The specific transmission mechanisms end up relying on the knowledge and preferences of the parties involved. Through two online surveys ($n=60$ and $n=250$), we explore the various methods used to transmit sensitive documents, as well as the perceived risk and satisfaction with those methods. We find that users are more likely to recognize risk to data-at-rest after receipt (but not at the sender, namely, themselves). When not using an online portal provided by the recipient, participants primarily envision transmitting sensitive documents in person or via email, and have little experience using secure, privacy-preserving alternatives. Despite recognizing general risks, participants express high privacy satisfaction and convenience with actually experienced situations. These results suggest opportunities to design new solutions to promote securely sending sensitive materials, perhaps as new utilities within standard email workflows.

Debjani Saha, Candice Schumann, Duncan C. McElfresh et al.

Bias in machine learning has manifested injustice in several areas, such as medicine, hiring, and criminal justice. In response, computer scientists have developed myriad definitions of fairness to correct this bias in fielded algorithms. While some definitions are based on established legal and ethical norms, others are largely mathematical. It is unclear whether the general public agrees with these fairness definitions, and perhaps more importantly, whether they understand these definitions. We take initial steps toward bridging this gap between ML researchers and the public, by addressing the question: does a lay audience understand a basic definition of ML fairness? We develop a metric to measure comprehension of three such definitions--demographic parity, equal opportunity, and equalized odds. We evaluate this metric using an online survey, and investigate the relationship between comprehension and sentiment, demographics, and the definition itself.

Daniel Votipka, Seth M. Rabin, Kristopher Micinski et al.

Reverse engineering is a complex process essential to software-security tasks such as vulnerability discovery and malware analysis. Significant research and engineering effort has gone into developing tools to support reverse engineers. However, little work has been done to understand the way reverse engineers think when analyzing programs, leaving tool developers to make interface design decisions based only on intuition. This paper takes a first step toward a better understanding of reverse engineers' processes, with the goal of producing insights for improving interaction design for reverse engineering tools. We present the results of a semi-structured, observational interview study of reverse engineers (N=16). Each observation investigated the questions reverse engineers ask as they probe a program, how they answer these questions, and the decisions they make throughout the reverse engineering process. From the interview responses, we distill a model of the reverse engineering process, divided into three phases: overview, sub-component scanning, and focused experimentation. Each analysis phase's results feed the next as reverse engineers' mental representations become more concrete. We find that reverse engineers typically use static methods in the first two phases, but dynamic methods in the final phase, with experience playing large, but varying, roles in each phase. % and the role of experience varies between phases. Based on these results, we provide five interaction design guidelines for reverse engineering tools.

James Parker, Michael Hicks, Andrew Ruef et al.

Typical security contests focus on breaking or mitigating the impact of buggy systems. We present the Build-it, Break-it, Fix-it (BIBIFI) contest, which aims to assess the ability to securely build software, not just break it. In BIBIFI, teams build specified software with the goal of maximizing correctness, performance, and security. The latter is tested when teams attempt to break other teams' submissions. Winners are chosen from among the best builders and the best breakers. BIBIFI was designed to be open-ended; teams can use any language, tool, process, etc. that they like. As such, contest outcomes shed light on factors that correlate with successfully building secure software and breaking insecure software. We ran three contests involving a total of 156 teams and three different programming problems. Quantitative analysis from these contests found that the most efficient build-it submissions used C/C++, but submissions coded in a statically-type safe language were 11 times less likely to have a security flaw than C/C++ submissions. Break-it teams that were also successful build-it teams were significantly better at finding security bugs.

Elissa M. Redmiles, Michelle L. Mazurek, John P. Dickerson

Accurately modeling human decision-making in security is critical to thinking about when, why, and how to recommend that users adopt certain secure behaviors. In this work, we conduct behavioral economics experiments to model the rationality of end-user security decision-making in a realistic online experimental system simulating a bank account. We ask participants to make a financially impactful security choice, in the face of transparent risks of account compromise and benefits offered by an optional security behavior (two-factor authentication). We measure the cost and utility of adopting the security behavior via measurements of time spent executing the behavior and estimates of the participant's wage. We find that more than 50% of our participants made rational (e.g., utility optimal) decisions, and we find that participants are more likely to behave rationally in the face of higher risk. Additionally, we find that users' decisions can be modeled well as a function of past behavior (anchoring effects), knowledge of costs, and to a lesser extent, users' awareness of risks and context (R2=0.61). We also find evidence of endowment effects, as seen in other areas of economic and psychological decision-science literature, in our digital-security setting. Finally, using our data, we show theoretically that a "one-size-fits"-all emphasis on security can lead to market losses, but that adoption by a subset of users with higher risks or lower costs can lead to market gains.

Andrew Ruef, Michael Hicks, James Parker et al.

Typical security contests focus on breaking or mitigating the impact of buggy systems. We present the Build-it Break-it Fix-it BIBIFI contest which aims to assess the ability to securely build software not just break it. In BIBIFI teams build specified software with the goal of maximizing correctness performance and security. The latter is tested when teams attempt to break other teams submissions. Winners are chosen from among the best builders and the best breakers. BIBIFI was designed to be open-ended - teams can use any language tool process etc. that they like. As such contest outcomes shed light on factors that correlate with successfully building secure software and breaking insecure software. During we ran three contests involving a total of teams and two different programming problems. Quantitative analysis from these contests found that the most efficient build-it submissions used CC but submissions coded in a statically-typed language were less likely to have a security flaw build-it teams with diverse programming-language knowledge also produced more secure code. Shorter programs correlated with better scores. Break-it teams that were also build-it teams were significantly better at finding security bugs.