67471 research outputs found
Sort by
DAO Decision-Making Simulation for Legislative Consultation: The Case of the Swiss E-ID Law 2019
Get PDFThis paper explores how Decentralized Autonomous Organizations (DAOs) could inform and shape participatory procedures in democratic governance. We apply DAO decision-making, such as rule-based input aggregation, transparent participation, and programmable decision-making, to a real-world case: the legislative development of the Swiss E-ID law, a proposal to establish a digital identity system for secure online authentication for Swiss residents. Using data from the official legislative consultation, we simulate how DAO-inspired mechanisms could have altered the aggregation of input and policy outcomes. Our analysis contributes conceptually and empirically to debates on digital democratic innovations, showing how programmable governance can be used not only to design new institutional forms, but also to critically assess the procedural dynamics of existing ones
Introduction to the Minitrack on Communication, Digital Conversation, and Media Technologies
No full textPrivacy Fragility in Direct-to-Consumer Genetic Testing: Lessons from the 23andMe Journey
Get PDFThis study investigates how structural, legal, and organizational dynamics erode user privacy in Direct-to-Consumer Genetic Testing (DTC-GT), using 23andMe as a critical case. We categorize privacy risks into three harm types: knowledge harms, autonomy and trust harms, and data misuse harms. Our analysis focuses on 23andMe’s 2023 cyberattack, data-sharing practices that changed over time, and the privacy consequences of its 2025 bankruptcy and sale. The breach exposed 6.9 million genetic profiles, many of which were later sold on illicit forums. These privacy risks stem from opaque consent mechanisms, affiliate transfers, and the sale of genetic data in bankruptcy. Regulatory gaps and permissive contracts further enable third-party access, often contradicting consumer-facing assurances. To explain how such vulnerabilities accumulate, we introduce the construct of privacy fragility, which captures how delayed breach responses, irrevocable data permissions, and commercial failure interact to undermine institutional safeguards. We argue these are not isolated failures but interconnected conditions that systematically weaken privacy protections. Our findings support a roadmap of legal, technical, and behavioral mitigations tailored to high-risk platforms. By tracing how privacy protections deteriorate amid shifting business and regulatory pressures, we contribute a scalable framework for evaluating privacy risk in consumer data ecosystems. This framework underscores the urgent need for adaptive governance, particularly in underregulated markets like DTC-GT where data sensitivity is high and user protections remain weak
How to Use Generative Artificial Intelligence in the Research Process: A Modular Course Approach for Early Career Researchers
Get PDFThe integration of generative artificial intelligence (GAI) tools into academic research workflows presents both promise and complexity — particularly for early career researchers (ECRs), who often lack structured guidance on responsible use. This study addresses this gap by designing and evaluating a modular course that supports ECRs in applying GAI systematically and appropriately across key stages of the research process, including literature exploration, hypothesis development, and academic writing. Drawing on the Design Science Research (DSR) Methodology, the course was iteratively developed and assessed through expert interviews and pre- and post-surveys with participants. Expert feedback suggests refinements to pacing and engagement. Quantitative findings indicate increased confidence and frequency of GAI use, especially for literature discovery and scholarly communication. This work contributes to DSR by offering a grounded course concept and actionable guidelines, aimed at advancing GAI literacy in ECRs’ scholarly work, supporting the transparent, responsible integration of GAI into research practice
Refining Neural Network Interpretability through Activation Modification
Get PDFThis research focuses on the problem of how to design real post-hoc modifiable Deep Neural Networks (DNNs) that can achieve or exceed state-of-the-art performance while also providing increased transparency that can help in understanding how predictions made by DNNs were reached. Existing techniques for interpretability are mostly concentrated on inspecting neuron activations as is. Here, we study controlled neuron activation adjustments during inference and examine whether these adjustments can help improve the explainable aspect and generalization of Fully Connected Neural Networks (FCNNs) without retraining. The study introduces three activation method adaptation strategies. All of them introduce a systematic adjustment of neuron activations according to individual activation magnitude, which tends to make the latent feature representation more significant in the inference phase. Experimental results show that the improvement of classification accuracies can be significant on misclassified samples as well as on overall model performance, achieving up to 14% improvements without retraining
Now and Later? Comparing a Nomothetic and Idiographic Analysis of Cybersecurity Fatigue
Get PDFFrom receiving phishing warnings to participating in required training sessions, today’s workers are inundated with a seemingly endless flow of cybersecurity-related activities. For some, this encumbrance can lead to feelings of fatigue, an increased susceptibility to ignore cybersecurity requirements, and engagement in workarounds. However, the extent that cybersecurity fatigue phenomena fluctuate over time remains unclear, despite the notable consequences for organizational risk. In response, we conducted a three-wave, repeated measures survey over a period of seven months, then compared the nomothetic (cross-sectional, between-person) results at wave 1 with the idiographic (longitudinal, within-person) results spanning waves 1-3. Together, the analysis reveals conflicting theoretical conclusions; namely, that only two of the four significant relationships at wave 1 were significant over waves 1-3. We draw on ego-depletion theory as a preliminary theoretical explanation for this finding and highlight paths for future inquiry
Model Evaluation for Radio-Frequency Signal Modulation Classifiers in the Existence of Novel Samples
Get PDFWe present an open-set recognition (OSR) pipeline for radio-frequency (RF) signal modulation that extends the recently proposed varMax uncertainty framework to the communications domain. Leveraging a multi-domain convolutional neural network (CNN) that takes in four representations of an electromagnetic signal in the form of raw In-phase and Quadrature (IQ) values and three transformations, Fast Fourier Transform (FFT), Discrete Cosine Transform (DCT), and polar magnitude, we train on eight signal modulations and treat the ninth modulation as unknown. Per fold, 1000 epochs are used to train each classifier on a unique set of eight modulations. Our approach supports the reliable rejection of unseen modulations on most folds while preserving in-distribution classification accuracy. However, the effectiveness is highly dependent on the makeup of the open and closed sets, and even seemingly inconsequential changes have a significant impact on the model's performance. This study demonstrates the strengths and weaknesses of extending the varMax approach to the RF domain and introduces modifications that improve the model's performance when trained on heterogeneous data
Introduction to the Minitrack on Changing Nature of Work – More Fair Labor Markets and Work Practices through Digital Transformation
No full textAI-Powered Decision Support in Privacy Laws: Benchmarking LLMs on Legal Violation Detection
Get PDFThe integration of Large Language Models (LLMs) into legal practice creates a conflict between performance and data privacy. While proprietary models excel at legal analysis, their cloud-based nature risks breaching regulations, such as the GDPR, by transmitting sensitive data. This study addresses this tension by examining whether locally hosted, open-source LLMs can serve as a viable, privacy-preserving alternative for judicial support. We benchmark five models on classifying GDPR enforcement cases, using chain-of-thought prompting to evaluate accuracy, size, and speed trade-offs. Our findings reveal that while cloud-based models achieve the highest parent-article accuracy (58\%), open-source models offer competitive performance (45\%) while running entirely on-premises. This demonstrates that local models are a potent tool for legal analysis when confidentiality is paramount. \footnote{The code, data, and experimental results will be added to the authors' GitHub repository upon publication.} We conclude by proposing knowledge distillation as a path to creating specialized and secure models, thereby providing a benchmark for developing transparent and auditable AI tools for judicial decision-making
