Research Papers

Artificial Intelligence (AI) poses both significant risks and valuable opportunities for democratic governance. This paper introduces a dual taxonomy to evaluate AI's complex relationship with democracy: the AI Risks to Democracy (AIRD) taxonomy, which identifies how AI can undermine core democratic principles such as autonomy, fairness, and trust; and the AI's Positive Contributions to Democracy (AIPD) taxonomy, which highlights AI's potential to enhance transparency, participation, efficiency, and evidence-based policymaking. Grounded in the European Union's approach to ethical AI governance, and particularly the seven Trustworthy AI requirements proposed by the European Commission's High-Level Expert Group on AI, each identified risk is aligned with mitigation strategies based on EU regulatory and normative frameworks. Our analysis underscores the transversal importance of transparency and societal well-being across all risk categories and offers a structured lens for aligning AI systems with democratic values. By integrating democratic theory with practical governance tools, this paper offers a normative and actionable framework to guide research, regulation, and institutional design to support trustworthy, democratic AI. It provides scholars with a conceptual foundation to evaluate the democratic implications of AI, equips policymakers with structured criteria for ethical oversight, and helps technologists align system design with democratic principles. In doing so, it bridges the gap between ethical aspirations and operational realities, laying the groundwork for more inclusive, accountable, and resilient democratic systems in the algorithmic age.

Large Language Models (LLMs) rapidly reshape modern life, advancing fields from healthcare to education and beyond. However, alongside their remarkable capabilities lies a significant threat: the susceptibility of these models to jailbreaking. The fundamental vulnerability of LLMs to jailbreak attacks stems from the very data they learn from. As long as this training data includes unfiltered, problematic, or 'dark' content, the models can inherently learn undesirable patterns or weaknesses that allow users to circumvent their intended safety controls. Our research identifies the growing threat posed by dark LLMs models deliberately designed without ethical guardrails or modified through jailbreak techniques. In our research, we uncovered a universal jailbreak attack that effectively compromises multiple state-of-the-art models, enabling them to answer almost any question and produce harmful outputs upon request. The main idea of our attack was published online over seven months ago. However, many of the tested LLMs were still vulnerable to this attack. Despite our responsible disclosure efforts, responses from major LLM providers were often inadequate, highlighting a concerning gap in industry practices regarding AI safety. As model training becomes more accessible and cheaper, and as open-source LLMs proliferate, the risk of widespread misuse escalates. Without decisive intervention, LLMs may continue democratizing access to dangerous knowledge, posing greater risks than anticipated.

Visual Analytics (VA) integrates humans, data, and models as key actors in insight generation and data-driven decision-making. This position paper values and reflects on 16 VA process models and frameworks and makes nine high-level observations that motivate a fresh perspective on VA. The contribution is the HDMI Canvas, a perspective to VA that complements the strengths of existing VA process models and frameworks. It systematically characterizes diverse roles of humans, data, and models, and how these actors benefit from and contribute to VA processes. The descriptive power of the HDMI Canvas eases the differentiation between a series of VA building blocks, rather than describing general VA principles only. The canvas includes modern human-centered methodologies, including human knowledge externalization and forms of feedback loops, while interpretable and explainable AI highlight model contributions beyond their conventional outputs. The HDMI Canvas has generative power, guiding the design of new VA processes and is optimized for external stakeholders, improving VA outreach, interdisciplinary collaboration, and user-centered design. The utility of the HDMI Canvas is demonstrated through two preliminary case studies.

The adoption of Artificial Intelligence (AI) in the healthcare service industry presents numerous ethical challenges, yet current frameworks often fail to offer a comprehensive, empirical understanding of the multidimensional factors influencing ethical AI integration. Addressing this critical research gap, this study introduces the Multi-Dimensional Ethical AI Adoption Model (MEAAM), a novel theoretical framework that categorizes 13 critical ethical variables across four foundational dimensions of Ethical AI Fair AI, Responsible AI, Explainable AI, and Sustainable AI. These dimensions are further analyzed through three core ethical lenses: epistemic concerns (related to knowledge, transparency, and system trustworthiness), normative concerns (focused on justice, autonomy, dignity, and moral obligations), and overarching concerns (highlighting global, systemic, and long-term ethical implications). This study adopts a quantitative, cross-sectional research design using survey data collected from healthcare professionals and analyzed via Partial Least Squares Structural Equation Modeling (PLS-SEM). Employing PLS-SEM, this study empirically investigates the influence of these ethical constructs on two outcomes Operational AI Adoption and Systemic AI Adoption. Results indicate that normative concerns most significantly drive operational adoption decisions, while overarching concerns predominantly shape systemic adoption strategies and governance frameworks. Epistemic concerns play a facilitative role, enhancing the impact of ethical design principles on trust and transparency in AI systems. By validating the MEAAM framework, this research advances a holistic, actionable approach to ethical AI adoption in healthcare and provides critical insights for policymakers, technologists, and healthcare administrators striving to implement ethically grounded AI solutions.

Explainable AI Based Diagnosis of Poisoning Attacks in Evolutionary Swarms

Mehrdad Asadi, Roxana Rădulescu, Ann Nowé

Swarming systems, such as for example multi-drone networks, excel at cooperative tasks like monitoring, surveillance, or disaster assistance in critical environments, where autonomous agents make decentralized decisions in order to fulfill team-level objectives in a robust and efficient manner. Unfortunately, team-level coordinated strategies in the wild are vulnerable to data poisoning attacks, resulting in either inaccurate coordination or adversarial behavior among the agents. To address this challenge, we contribute a framework that investigates the effects of such data poisoning attacks, using explainable AI methods. We model the interaction among agents using evolutionary intelligence, where an optimal coalition strategically emerges to perform coordinated tasks. Then, through a rigorous evaluation, the swarm model is systematically poisoned using data manipulation attacks. We showcase the applicability of explainable AI methods to quantify the effects of poisoning on the team strategy and extract footprint characterizations that enable diagnosing. Our findings indicate that when the model is poisoned above 10%, non-optimal strategies resulting in inefficient cooperation can be identified.

The increasing automation of traffic management systems has made them prime targets for cyberattacks, disrupting urban mobility and public safety. Traditional network-layer defenses are often inaccessible to transportation agencies, necessitating a machine learning-based approach that relies solely on traffic flow data. In this study, we simulate cyberattacks in a semi-realistic environment, using a virtualized traffic network to analyze disruption patterns. We develop a deep learning-based anomaly detection system, demonstrating that Longest Stop Duration and Total Jam Distance are key indicators of compromised signals. To enhance interpretability, we apply Explainable AI (XAI) techniques, identifying critical decision factors and diagnosing misclassification errors. Our analysis reveals two primary challenges: transitional data inconsistencies, where mislabeled recovery-phase traffic misleads the model, and model limitations, where stealth attacks in low-traffic conditions evade detection. This work enhances AI-driven traffic security, improving both detection accuracy and trustworthiness in smart transportation systems.

Fairness auditing of AI systems can identify and quantify biases. However, traditional auditing using real-world data raises security and privacy concerns. It exposes auditors to security risks as they become custodians of sensitive information and targets for cyberattacks. Privacy risks arise even without direct breaches, as data analyses can inadvertently expose confidential information. To address these, we propose a framework that leverages differentially private synthetic data to audit the fairness of AI systems. By applying privacy-preserving mechanisms, it generates synthetic data that mirrors the statistical properties of the original dataset while ensuring privacy. This method balances the goal of rigorous fairness auditing and the need for strong privacy protections. Through experiments on real datasets like Adult, COMPAS, and Diabetes, we compare fairness metrics of synthetic and real data. By analyzing the alignment and discrepancies between these metrics, we assess the capacity of synthetic data to preserve the fairness properties of real data. Our results demonstrate the framework's ability to enable meaningful fairness evaluations while safeguarding sensitive information, proving its applicability across critical and sensitive domains.

RAIL in the Wild: Operationalizing Responsible AI Evaluation Using Anthropic's Value Dataset

Sumit Verma, Pritam Prasun, Arpit Jaiswal +1

As AI systems become embedded in real-world applications, ensuring they meet ethical standards is crucial. While existing AI ethics frameworks emphasize fairness, transparency, and accountability, they often lack actionable evaluation methods. This paper introduces a systematic approach using the Responsible AI Labs (RAIL) framework, which includes eight measurable dimensions to assess the normative behavior of large language models (LLMs). We apply this framework to Anthropic's "Values in the Wild" dataset, containing over 308,000 anonymized conversations with Claude and more than 3,000 annotated value expressions. Our study maps these values to RAIL dimensions, computes synthetic scores, and provides insights into the ethical behavior of LLMs in real-world use.

This paper establishes a theoretical foundation for understanding the fundamental limits of AI explainability through algorithmic information theory. We formalize explainability as the approximation of complex models by simpler ones, quantifying both approximation error and explanation complexity using Kolmogorov complexity. Our key theoretical contributions include: (1) a complexity gap theorem proving that any explanation significantly simpler than the original model must differ from it on some inputs; (2) precise bounds showing that explanation complexity grows exponentially with input dimension but polynomially with error tolerance for Lipschitz functions; and (3) a characterization of the gap between local and global explainability, demonstrating that local explanations can be significantly simpler while maintaining accuracy in relevant regions. We further establish a regulatory impossibility theorem proving that no governance framework can simultaneously pursue unrestricted AI capabilities, human-interpretable explanations, and negligible error. These results highlight considerations likely to be relevant to the design, evaluation, and oversight of explainable AI systems.

Machine learning (ML) systems for medical imaging have demonstrated remarkable diagnostic capabilities, but their susceptibility to biases poses significant risks, since biases may negatively impact generalization performance. In this paper, we introduce a novel statistical framework to evaluate the dependency of medical imaging ML models on sensitive attributes, such as demographics. Our method leverages the concept of counterfactual invariance, measuring the extent to which a model's predictions remain unchanged under hypothetical changes to sensitive attributes. We present a practical algorithm that combines conditional latent diffusion models with statistical hypothesis testing to identify and quantify such biases without requiring direct access to counterfactual data. Through experiments on synthetic datasets and large-scale real-world medical imaging datasets, including \textsc{cheXpert} and MIMIC-CXR, we demonstrate that our approach aligns closely with counterfactual fairness principles and outperforms standard baselines. This work provides a robust tool to ensure that ML diagnostic systems generalize well, e.g., across demographic groups, offering a critical step towards AI safety in healthcare. Code: https://github.com/Neferpitou3871/AI-Alignment-Medical-Imaging.

The Open Radio Access Network (Open RAN) is an emerging idea -- transforming the traditional Radio Access Networks (RAN) that are monolithic and inflexible into more flexible and innovative. By leveraging open standard interfaces, data collection across all RAN layers becomes feasible, paving the way for the development of energy-efficient Open RAN architectures through Artificial Intelligence / Machine Learning (AI/ML). However, the inherent complexity and black-box nature of AI/ML models used for energy consumption prediction pose challenges in interpreting their underlying factors and relationships. This work presents an integration of eXplainable AI (XAI) to understand the key RAN parameters that contribute to energy consumption. Furthermore, the paper delves into the analysis of RAN parameters -- \emph{airtime}, \emph{goodput}, \emph{throughput}, \emph{buffer status report}, \emph{number of resource blocks}, and many others -- identified by XAI techniques, highlighting their significance in energy consumption.

This study investigates the extent to which Mel-Frequency Cepstral Coefficients (MFCCs) capture first language (L1) transfer in extended second language (L2) English speech. Speech samples from Mandarin and American English L1 speakers were extracted from the GMU Speech Accent Archive, converted to WAV format, and processed to obtain thirteen MFCCs per speaker. A multi-method analytic framework combining inferential statistics (t-tests, MANOVA, Canonical Discriminant Analysis) and machine learning (Random Forest classification) identified MFCC-1 (broadband energy), MFCC-2 (first formant region), and MFCC-5 (voicing and fricative energy) as the most discriminative features for distinguishing L1 backgrounds. A reduced-feature model using these MFCCs significantly outperformed the full-feature model, as confirmed by McNemar's test and non-overlapping confidence intervals. The findings empirically support the Perceptual Assimilation Model for L2 (PAM-L2) and the Speech Learning Model (SLM), demonstrating that L1-conditioned variation in L2 speech is both perceptually grounded and acoustically quantifiable. Methodologically, the study contributes to applied linguistics and explainable AI by proposing a transparent, data-efficient pipeline for L2 pronunciation modeling. The results also offer pedagogical implications for ESL/EFL instruction by highlighting L1-specific features that can inform intelligibility-oriented instruction, curriculum design, and speech assessment tools.