Infostealer malware infects devices worldwide and harvests their most sensitive contents: credentials, browser sessions, private keys, and access certificates. Yet its impact on victims remains difficult to study without an ethical, legal, and curated research dataset. To close this gap, we build a privacy-preserving pipeline that turns illicitly sourced infostealer logs into a reproducible research artifact, minimizing sensitive data while preserving measurement utility, and use it to construct a dataset of 170,298 victims from logs of multiple infostealer families. Analyzing these victims, we find that the most compromised services mirror the world's most popular platforms, with gaming and entertainment services strongly overrepresented. Within the sample we identify compromised credentials for high-value organizations, including law-enforcement domains, government and military services, and all eight Ivy League universities, as well as substantial exposure of security-critical infrastructure and of financial, remote-access, and development platforms. Victims also show widespread credential reuse and significant revictimization risk, overlapping with phishing and ransomware victim populations. We release the first anonymized victim-level infostealer dataset under controlled access to enable ethical, privacy-preserving, and reproducible research on information security and victim behavior.
Transformer-based architectures have significantly advanced the generation of complex symbolic sequences, yet a significant gap remains in achieving fine-grained, interpretable control over discrete signal attributes. This paper investigates the mechanistic interpretability of the Multitrack Music Transformer (MMT) and proposes a framework for deterministic attribute modulation without retraining to bridge this gap via inference-time activation steering. Utilizing the Difference-in-Means (DiffMean) methodology, we isolate latent directions for signal attributes, specifically Pitch and Duration, within the residual stream. We validate the Linear Representation Hypothesis in this domain, achieving high correlation between steering magnitude and attribute shift. To address the inherent feature entanglement in multi-attribute steering, we introduce a Dual Steering framework utilizing Gram-Schmidt Orthogonalization. Experimental results demonstrate that this geometric decoupling reduces conceptual interference and signal degradation compared to naive vector addition, enabling independent deterministic control even against strong autoregressive conditioning.
Entity Resolution constitutes a fundamental task for transforming heterogeneous Semantic Web data into high-quality Linked Data. Recent approaches leverage Large Language Models (LLMs), achieving strong performance but incurring substantial computational and financial costs. Existing methods operate either in unsupervised settings using very large models, or in supervised settings requiring labeled data, creating a gap between efficiency and effectiveness. To address this, we investigate Knowledge Distillation to transfer knowledge from large, effective Teachers to smaller, efficient Students without requiring gold-standard annotations. We introduce DistillER, the first framework that bridges this gap spanning three dimensions: (i) Data Selection, which identifies informative training subsets; (ii) Knowledge Elicitation, which explores single- and multi-teacher settings across LLMs; and (iii) Distillation Algorithms, which compare supervised fine-tuning and reinforcement learning strategies. Our experimental evaluation shows that supervised fine-tuning on noisy LLM-generated labels consistently outperforms alternative distillation strategies, while also enabling the generation of high-quality explanations. Finally, we benchmark DistillER against state-of-the-art supervised and unsupervised ER approaches, demonstrating clear gains in effectiveness.
Adversarial attacks pose a significant threat to data-driven systems, and researchers have devoted considerable effort to studying them. Despite its economic relevance, credit card fraud detection has received comparatively little attention. To address this gap, we propose a novel threat model that highlights the limitations of existing attacks and motivates new approaches. We introduce Fraud-RLA, an adversarial attack against credit card Fraud Detection Systems that leverages Reinforcement Learning to evade detection. Fraud-RLA is designed to maximize the amount stolen by optimizing the exploration-exploitation trade-off while requiring substantially less prior knowledge than competing methods. Our experiments on a realistic Fraud Detection System show that Fraud-RLA is effective, even under the severe limitations imposed by our threat model.
Beyond simplistic 3D visualisations, archaeologists, as well as cultural heritage experts and practitioners, need applications with advanced functionalities. Such as the annotation and attachment of metadata onto particular regions of the 3D digital objects. Various approaches have been presented to tackle this challenge, most of which achieve excellent results in the domain of their application. However, they are often confined to that specific domain and particular problem. In this paper, we present ART3mis - a general-purpose, user-friendly, interactive textual annotation tool for 3D objects. Primarily attuned to aid cultural heritage conservators, restorers and curators with no technical skills in 3D imaging and graphics, the tool allows for the easy handling, segmenting and annotating of 3D digital replicas of artefacts. ART3mis applies a user-driven, direct-on-surface approach. It can handle detailed 3D cultural objects in real-time and store textual annotations for multiple complex regions in JSON data format.