Papers

Public-key cryptography algorithms have evolved towards increasing computational complexity to hide desired messages, which is accelerating with the development of the Internet and quantum computing. This paper introduces a novel public-key cryptography algorithm that generates ciphertexts by counting the number of elements in randomly extracted subsets from a multiset. After explaining the novel cryptographic concept, the process of mathematically refining it using satisfiability problems is described. The advantages of the proposed algorithm are: first, it is significantly faster than other public-key algorithms; second, it does not require big numbers, making it executable on any devices; and third, it can be easily extended into a public-key cryptosystem using a single public key and multiple private keys while maintaining quantum resistance.

USDT, a stablecoin pegged to dollar, has become a preferred choice for money laundering due to its stability, anonymity, and ease of use. Notably, a new form of money laundering on stablecoins – we refer to as crowdsourcing laundering – disperses funds through recruiting a large number of ordinary individuals, and has rapidly emerged as a significant threat. However, due to the refined division of labor, crowdsourcing laundering transactions exhibit diverse patterns and a polycentric structure, posing significant challenges for detection. In this paper, we introduce transaction group as auxiliary information, and propose the Multi-Task Collaborative Crowdsourcing Laundering Detection (MCCLD) framework. MCCLD employs an end-to-end graph neural network to realize collaboration between laundering transaction detection and transaction group detection tasks, enhancing detection performance on diverse patterns within crowdsourcing laundering group. These two tasks are jointly optimized through a shared classifier, with a shared feature encoder that fuses multi-level feature embeddings to provide rich transaction semantics and potential group information. Extensive experiments on both crowdsourcing and general laundering demonstrate MCCLD's effectiveness and generalization. To the best of our knowledge, this is the first work on crowdsourcing laundering detection.

Stablecoins such as USDT and USDC aspire to peg stability by coupling issuance controls with reserve attestations. In practice, however, transparency remains fragmented across heterogeneous data sources, with key evidence about circulation, reserves, and disclosure dispersed across records that are difficult to connect and interpret jointly. We introduce a large language model (LLM)-based automated framework for bridging cross-domain transparency in stablecoins by aligning issuer disclosures with observable circulation evidence. First, we propose an integrative framework using LLMs to parse documents, extract salient financial indicators, and semantically align reported statements with corresponding market and issuance metrics. Second, we integrate multi-chain issuance records and disclosure documents within a model context protocol (MCP) framework that standardizes LLM access to both quantitative market data and qualitative disclosure narratives. This framework enables unified retrieval and contextual alignment across heterogeneous stablecoin information sources and facilitates consistent analysis. Third, we demonstrate the capability of LLMs to operate across heterogeneous data domains in blockchain analytics, quantifying discrepancies between reported and observed circulation and examining their implications for transparency and price dynamics. Our findings reveal systematic gaps between disclosed and verifiable data, showing that LLM-assisted analysis enhances cross-domain transparency and supports automated, data-driven auditing in decentralized finance (DeFi).

Cybersecurity of space infrastructures is an emerging topic, despite space-related cybersecurity incidents occurring as early as 1977 (i.e., hijacking of a satellite transmission signal). There is no single dataset that documents cyber attacks against space infrastructures that have occurred in the past; instead, these incidents are often scattered in media reports while missing many details, which we dub the missing-data problem. Nevertheless, even “low-quality” datasets containing such reports would be extremely valuable because of the dearth of space cybersecurity data and the sensitivity of space infrastructures which are often restricted from disclosure by governments. This prompts a research question: How can we characterize real-world cyber attacks against space infrastructures? In this paper, we address the problem by proposing a framework, including metrics, while also addressing the missing-data problem by leveraging methodologies such as the Space Attack Research and Tactic Analysis (SPARTA) and the Adversarial Tactics, Techniques, and Common Knowledge (ATT&CK) to “extrapolate” the missing data in a principled fashion. We show how the extrapolated data can be used to reconstruct “hypothetical but plausible” space cyber kill chains and space cyber attack campaigns that have occurred in practice. To show the usefulness of the framework, we extract data for 108 cyber attacks against space infrastructures and show how to extrapolate this “low-quality” dataset containing missing information to derive 6,206 attack technique-level space cyber kill chains. Our findings include: cyber attacks against space infrastructures are getting increasingly sophisticated; successful protection of the link segment between the space and user segments could have thwarted nearly half of the 108 attacks. We will make our dataset available.

The emergence of Large Language Models (LLMs) is rapidly accelerating the development of autonomous multi-agent systems (MAS), paving the way for the Internet of Agents. However, traditional centralized MAS architectures present significant challenges, including single points of failure, vulnerability to censorship, inherent scalability limitations, and critical trust issues. We propose a novel Decentralized Multi-Agent System (DMAS) architecture designed to overcome these fundamental problems by enabling trust-aware, scalable, and censorship-resistant interactions among autonomous agents. Our DMAS features a decentralized agent runtime underpinned by a blockchain-based architecture. We formalize a trust-aware communication protocol that leverages cryptographic primitives and on-chain operations to provide security properties: verifiable interaction cycles, communication integrity, authenticity, non-repudiation, and conditional confidentiality, which we further substantiate through a comprehensive security analysis. Our performance analysis validates the DMAS as a scalable and efficient solution for building trustworthy multi-agent systems.

Smart contracts are a core component of blockchain technology and are widely deployed across various scenarios. However, atomicity violations have become a potential security risk. Existing analysis tools often lack the precision required to detect these issues effectively. To address this challenge, we introduce AtomGraph, an automated framework designed for detecting atomicity violations. This framework leverages Graph Convolutional Networks (GCN) to identify atomicity violations through multimodal feature learning and fusion. Specifically, driven by a collaborative learning mechanism, the model simultaneously learns from two heterogeneous modalities: extracting structural topological features from the contract's Control Flow Graph (CFG) and uncovering deep semantics from its opcode sequence. We designed an adaptive weighted fusion mechanism to dynamically adjust the weights of features from each modality to achieve optimal feature fusion. Finally, GCN detects graph-level atomicity violation on the contract. Comprehensive experimental evaluations demonstrate that AtomGraph achieves 96.88% accuracy and 96.97% F1 score, outperforming existing tools. Furthermore, compared to the concatenation fusion model, AtomGraph improves the F1 score by 6.4%, proving its potential in smart contract security detection.

Large Language Model (LLM)-based agents have demonstrated remarkable capabilities in reasoning, planning, and tool usage. The recently proposed Model Context Protocol (MCP) has emerged as a unifying framework for integrating external tools into agent systems, enabling a thriving open ecosystem of community-built functionalities. However, the openness and composability that make MCP appealing also introduce a critical yet overlooked security assumption – implicit trust in third-party tool providers. In this work, we identify and formalize a new class of attacks that exploit this trust boundary without violating explicit permissions. We term this new attack vector implicit toxicity, where malicious behaviors occur entirely within the allowed privilege scope. We propose LeechHijack, a Latent Embedded Exploit for Computation Hijacking, in which an adversarial MCP tool covertly expropriates the agent's computational resources for unauthorized workloads. LeechHijack operates through a two-stage mechanism: an implantation stage that embeds a benign-looking backdoor in a tool, and an exploitation stage where the backdoor activates upon predefined triggers to establish a command-and-control channel. Through this channel, the attacker injects additional tasks that the agent executes as if they were part of its normal workflow, effectively parasitizing the user's compute budget. We implement LeechHijack across four major LLM families. Experiments show that LeechHijack achieves an average success rate of 77.25%, with a resource overhead of 18.62% compared to the baseline. This study highlights the urgent need for computational provenance and resource attestation mechanisms to safeguard the emerging MCP ecosystem.

This paper studies how multimodal large language models (MLLMs) undermine the security guarantees of visual CAPTCHA. We identify the attack surface where an adversary can cheaply automate CAPTCHA solving using off-the-shelf models. We evaluate 7 representative MLLMs on 18 real-world CAPTCHA task types, measuring single-shot accuracy, success under limited retries, end-to-end latency, and per-solve cost. We further validate our findings through a supplemental external dataset and an adaptive-attacker setting with session memory, while also analyzing the impact of task-specific prompt engineering and few-shot demonstrations on solver effectiveness. We reveal that MLLMs can reliably solve recognition-oriented and low-interaction CAPTCHA tasks at human-like cost and latency, whereas tasks requiring fine-grained localization, multi-step spatial reasoning, or cross-frame consistency remain significantly harder for current models. By examining the reasoning traces of such MLLMs, we investigate the underlying mechanisms of why models succeed/fail on specific CAPTCHA puzzles and use these insights to derive defense-oriented guidelines for selecting and strengthening CAPTCHA tasks. To validate these principles, we present a proof-of-concept by hardening a vulnerable CAPTCHA type using our guidelines. We demonstrate that incorporating fine-grained localization and implicit counting reduces the success rate of state-of-the-art MLLMs from over 95% to 0%, confirming that structural changes can effectively mitigate the threat. We conclude by emphasizing the urgent need for CAPTCHA redesign as MLLM capabilities increasingly threaten existing defenses. Code Availability (https://doi.org/10.5281/zenodo.20406852).

Omni-modal Large Language Models (OLLMs) that process text, images, videos, and audio introduce new challenges for safety and value guardrails in human-AI interaction. Prior guardrail research largely targets unimodal settings and typically frames safeguarding as binary classification, which limits robustness across diverse modalities and tasks. To address this gap, we propose OmniGuard, the first family of omni-modal guardrails that performs safeguarding across all modalities with deliberate reasoning ability. To support the training of OMNIGUARD, we curate a large, comprehensive omni-modal safety dataset comprising over 210K diverse samples, with inputs that cover all modalities through both unimodal and cross-modal samples. Each sample is annotated with structured safety labels and carefully curated safety critiques from expert models through targeted distillation. Extensive experiments on 15 benchmarks show that OmniGuard achieves strong effectiveness and generalization across a wide range of multimodal safety scenarios. Importantly, OmniGuard provides a unified framework that enforces policies and mitigates risks in omni-modalities, paving the way toward building more robust and capable omnimodal safeguarding systems.

This work presents vQFL (vehicular Quantum Federated Learning), a new framework that leverages quantum machine learning techniques to tackle key privacy and security issues in autonomous vehicular networks. Furthermore, we propose a server-side adapted fine-tuning method, ft-VQFL,to achieve enhanced and more resilient performance. By integrating quantum federated learning with differential privacy and quantum key distribution (QKD), our quantum vanguard approach creates a multi-layered defense against both classical and quantum threats while preserving model utility. Extensive experimentation with industry-standard datasets (KITTI, Waymo, and nuScenes) demonstrates that vQFL maintains accuracy comparable to standard QFL while significantly improving privacy guaranties and communication security. Our implementation using various quantum models (VQC, QCNN, and SamplerQNN) reveals minimal performance overhead despite the added security measures. This work establishes a crucial foundation for quantum-resistant autonomous vehicle systems that can operate securely in the post-quantum era while efficiently processing the massive data volumes (20-40TB/day per vehicle) generated by modern autonomous fleets. The modular design of the framework allows for seamless integration with existing vehicular networks, positioning vQFL as an essential component for future intelligent transportation infrastructure.

In today's university environment, wireless connectivity is an essential resource for academic, administrative, and research activities. However, at the National University of the Altiplano of Puno (UNAP), the use of a QR code access system on the institutional Wi-Fi network has generated vulnerabilities related to the lack of individual authentication, user traceability, and access control. Given this situation, this study aims to strengthen the security of the university's wireless network through the application of data analytics, employing descriptive, predictive, and prescriptive approaches to the logs generated by the wireless controller (WLC). The methodology consisted of collecting and processing connection data from users, devices, and daily traffic, analyzing behavioral patterns, and detecting anomalies based on statistical models and machine learning algorithms. The results revealed critical usage peaks between 10:00 and 14:00, as well as anomalous behavior associated with recurring devices and irregular traffic spikes. This allowed for the establishment of dynamic alert thresholds and recommendations for improvements in bandwidth management and authentication. Furthermore, the conclusion states that integrating advanced analytics into the management of university networks not only identifies vulnerabilities and optimizes WiFi service performance, but also advances towards an intelligent, proactive infrastructure aligned with modern institutional cybersecurity standards.

Retrieval-Augmented Generation (RAG) and Supervised Finetuning (SFT) have become the predominant paradigms for equipping Large Language Models (LLMs) with external knowledge for diverse, knowledge-intensive tasks. However, while such knowledge injection improves performance, it also exposes new attack surfaces. Membership Inference Attacks (MIAs), which aim to determine whether a given data sample was included in a model's training set, pose serious threats to privacy and trust in sensitive domains. To this end, we first systematically evaluate the vulnerability of RAG- and SFT-based LLMs to various MIAs. Then, to address the privacy risk, we further introduce a novel, model-agnostic defense framework, Ensemble Privacy Defense (EPD), which aggregates and evaluates the outputs of a knowledge-injected LLM, a base LLM, and a dedicated judge model to enhance resistance against MIAs. Comprehensive experiments show that, on average, EPD reduces MIA success by up to 27.8% for SFT and 526.3% for RAG compared to inference-time baseline, while maintaining answer quality.

The integration of large language models (LLMs) into healthcare IoT systems promises faster decisions and improved medical support. LLMs are also deployed as multi-agent teams to assist AI doctors by debating, voting, or advising on decisions. However, when multiple assistant agents interact, coordinated adversaries can collude to create false consensus, pushing an AI doctor toward harmful prescriptions. We develop an experimental framework with scripted and unscripted doctor agents, adversarial assistants, and a verifier agent that checks decisions against clinical guidelines. Using 50 representative clinical questions, we find that collusion drives the Attack Success Rate (ASR) and Harmful Recommendation Rates (HRR) up to 100% in unprotected systems. In contrast, the verifier agent restores 100% accuracy by blocking adversarial consensus. This work provides the first systematic evidence of collusion risk in AI healthcare and demonstrates a practical, lightweight defence that ensures guideline fidelity.

HOT Protocol provides the infrastructure that allows smart contracts to securely own and manage private keys. The Multi-Party Computation (MPC) Network manages signing keys. By running an MPC node inside a Trusted Execution Environment (TEE), the protocol achieves stronger security guarantees while lowering economic requirements for participants. The NEAR Protocol provides a decentralized and efficient state layer. Key management can be integrated with any smart contract across Stellar, TON, Solana, and EVM-compatible networks.

Traffic classification (TC) plays a critical role in cybersecurity, particularly in IoT and embedded contexts, where inspection must often occur locally under tight hardware constraints. We use hardware-aware neural architecture search (HW-NAS) to derive lightweight TC models that are accurate, efficient, and deployable on edge platforms. Two input formats are considered: a flattened byte sequence and a 2D packet-wise time series; we examine how input structure affects adversarial vulnerability when using resource-constrained models. Robustness is assessed against white-box attacks, specifically Fast Gradient Sign Method (FGSM) and Projected Gradient Descent (PGD). On USTC-TFC2016, both HW-NAS models achieve over 99% clean-data accuracy while remaining within 65k parameters and 2M FLOPs. Yet under perturbations of strength 0.1, their robustness diverges: the flat model retains over 85% accuracy, while the time-series variant drops below 35%. Adversarial fine-tuning delivers robust gains, with flat-input accuracy exceeding 96% and the time-series variant recovering over 60 percentage points in robustness, all without compromising efficiency. The results underscore how input structure influences adversarial vulnerability, and show that even compact, resource-efficient models can attain strong robustness, supporting their practical deployment in secure edge-based TC.

We propose a methodology for falsifying safety properties in robotic vehicle systems through property-guided reduction and surrogate execution. By isolating only the control logic and physical dynamics relevant to a given specification, we construct lightweight surrogate models that preserve property-relevant behaviors while eliminating unrelated system complexity. This enables scalable falsification via trace analysis and temporal logic oracles. We demonstrate the approach on a drone control system containing a known safety flaw. The surrogate replicates failure conditions at a fraction of the simulation cost, and a property-guided fuzzer efficiently discovers semantic violations. Our results suggest that controller reduction, when coupled with logic-aware test generation, provides a practical and scalable path toward semantic verification of cyber-physical systems.

The Apache Software Foundation (ASF) ecosystem underpins a vast portion of modern software infrastructure, powering widely used components such as Log4j, Tomcat, and Struts. However, the ubiquity of these libraries has made them prime targets for high-impact security vulnerabilities, as illustrated by incidents like Log4Shell. Despite their widespread adoption, Apache projects are not immune to recurring and severe security weaknesses. We conduct a historical analysis of the Apache ecosystem to follow the "breadcrumb trail of vulnerabilities" by compiling a comprehensive dataset of Common Vulnerabilities and Exposures (CVEs) and Common Weakness Enumerations (CWEs). We examine trends in exploit recurrence, disclosure timelines, and remediation practices. Our analysis is guided by four key research questions: (1) What are the most persistent and repeated CWEs in Apache libraries? (2) How long do CVEs persist before being addressed? (3) What is the delay between CVE introduction and official disclosure? and (4) How long after disclosure are CVEs remediated? We present a detailed timeline of vulnerability lifecycle stages across Apache libraries and offer insights to improve secure coding practices, vulnerability monitoring, and remediation strategies. Our contributions include a curated dataset covering 24,285 Apache libraries, 1,285 CVEs, and 157 CWEs, along with empirical findings and developer-focused recommendations.

Phishing remains one of the most prevalent online threats, exploiting human trust to harvest sensitive credentials. Existing URL- and HTML-based detection systems struggle against obfuscation and visual deception. This paper presents PhishSnap, a privacy-preserving, on-device phishing detection system leveraging perceptual hashing (pHash). Implemented as a browser extension, PhishSnap captures webpage screenshots, computes visual hashes, and compares them against legitimate templates to identify visually similar phishing attempts. A 2024 dataset of 10,000 URLs (70%/20%/10% train/validation/test) was collected from PhishTank and Netcraft. Due to security takedowns, a subset of phishing pages was unavailable, reducing dataset diversity. The system achieved 0.79 accuracy, 0.76 precision, and 0.78 recall, showing that visual similarity remains a viable anti-phishing measure. The entire inference process occurs locally, ensuring user privacy and minimal latency.

The field of AI Control seeks to develop robust control protocols, deployment safeguards for untrusted AI which may be intentionally subversive. However, existing protocols that rely on weaker monitors to detect unsafe behavior often fail on complex tasks beyond the monitor's comprehension. We develop control protocols based on factored cognition, in which a trusted model decomposes a task, an untrusted model solves each resultant child task in isolation, and the results are reassembled into a full solution. These protocols may improve safety by several means, such as by simplifying the context for monitors, or by obscuring vulnerabilities in the environment. We implement our protocols in the APPS coding setting and red team them against backdoor attempts from an adversarial GPT-4.1 Nano. We find that: (i) Adding factored cognition to a trusted monitoring protocol can boost safety from 41% to 63%; (ii) Safety improves because monitor performance improves; (iii) Factored cognition makes it no harder for capable LLMs to write backdoors in APPS. While our protocols show low usefulness in APPS, they hold promise for more complex tasks.

We provide a survey on the Hidden Subgroup Problem (HSP), which plays an important role in studying the security of public-key cryptosystems. We first review the abelian case, where Kitaev's algorithm yields an efficient quantum solution to the HSP, recalling how classical problems (such as order finding, integer factorization, and discrete logarithm) can be formulated as abelian HSP instances. We then examine the current state of the art for non-abelian HSP, where no general efficient quantum solution is known, focusing on some relevant groups including dihedral group (connected to the shortest vector problem), symmetric groups (connected to the graph isomorphism problem), and semidirect product constructions (connected, in a special case, to the code equivalence problem). We also describe the main techniques for addressing the HSP in non-abelian cases, namely Fourier sampling and the black-box approach. Throughout the paper, we highlight the mathematical notions required and exploited in this context, providing a cryptography-oriented perspective.

As the Deterministic Random Bit Generator (DRBG) serves as a fundamental component in random number generation and cryptographic applications, its performance and security are particularly critical in resource-constrained embedded systems, where memory capacity and computational efficiency are limited. Establishing a high-performance and secure DRBG is therefore an important issue for embedded system design. Furthermore, the National Institute of Standards and Technology (NIST) established the Ascon lightweight cryptographic standard in August 2025, which is suitable for use in resource-limited embedded environments. Therefore, this study revises the DRBG standard and proposes three Ascon-driven constructions: the Ascon-Driven Hash-Based DRBG, the Ascon-Driven keyed-Hash Message Authentication Code (HMAC) DRBG, and the Ascon-Driven Counter-mode (CTR) DRBG. In the experiments, these methods are implemented on a Raspberry Pi platform. The experimental results demonstrate that the proposed approaches achieve higher computational efficiency and lower memory usage compared with existing DRBG implementations, making them suitable for deployment in embedded systems.

Fast time-domain algorithms have been developed in signal processing applications to reduce the multiplication complexity. For example, fast convolution structures using Cook-Toom and Winograd algorithms are well understood. Short length fast convolutions can be iterated to obtain fast convolution structures for long lengths. In this paper, we show that well known fast convolution structures form the basis for design of fast algorithms in four other problem domains: fast parallel filters, fast polynomial modular multiplication, and fast pointwise multiplication in the DFT and NTT domains. Fast polynomial modular multiplication and fast pointwise multiplication problems are important for cryptosystem applications such as post-quantum cryptography and homomorphic encryption. By establishing the equivalence of these problems, we show that a fast structure from one domain can be used to design a fast structure for another domain. This understanding is important as there are many well known solutions for fast convolution that can be used in other signal processing and cryptosystem applications.

Phishing remains a persistent cybersecurity threat; however, developing scalable and effective user training is labor-intensive and challenging to maintain. Generative Artificial Intelligence offers an interesting opportunity, but empirical evidence on its instructional efficacy remains scarce. This paper provides an experimental validation of Large Language Models (LLMs) as autonomous engines for generating phishing resilience training. Across two controlled studies (N=480), we demonstrate that AI-generated content yields significant pre-post learning gains regardless of the specific prompting strategy employed. Study 1 (N=80) compares four prompting techniques, finding that even a straightforward "direct-profile" strategy–simply embedding user traits into the prompt–produces effective training material. Study 2 (N=400) investigates the scalability of this approach by testing personalization and training duration. Results show that complex psychometric personalization offers no measurable advantage over well-designed generic content, while longer training duration provides a modest boost in accuracy. These findings suggest that organizations can leverage LLMs to generate high-quality, effective training at scale without the need for complex user profiling, relying instead on the inherent capabilities of the model.

Large-scale vulnerability notifications (VNs) can help hosting provider organizations (HPOs) identify and remediate security vulnerabilities that attackers can exploit in data breaches or phishing campaigns. Previous VN studies have primarily focused on factors under the control of reporters, such as sender reputation, email formatting, and communication channels. Despite these efforts, remediation rates for vulnerability notifications continue to remain consistently low. This paper presents the first in-depth study of how HPOs process vulnerability notifications internally and what organizational and operational factors influence VN effectiveness. We examine the problem from a different perspective to provide the first detailed understanding of the reasons behind persistently low remediation rates. Instead of manipulating parameters of VN campaigns, we interview hosting providers directly, investigating how they handle vulnerability notifications and what factors may influence VN effectiveness, such as VN awareness and reachability, HPOs' service models, and perceived security risks. We conducted semi-structured interviews with 24 HPOs across shared hosting and web development services, representing varied company sizes and operator roles. Our findings reveal practical insights on VN processing and abuse workflows. While some providers remain hard to reach due to complex infrastructures, most report routinely handling VNs. However, limited remediation often stems from strict responsibility boundaries, where web application issues are seen as the customer's domain. Low hosting fees and high volumes of daily compromises further discourage both proactive and reactive measures. Our findings show that HPOs blame negligent website owners, and prior works on website owners confirms they often undervalue their sites or lack security know-how.

Deepfakes are a type of synthetic media created using artificial intelligence, specifically deep learning algorithms. This technology can for example superimpose faces and voices onto videos, creating hyper-realistic but artificial representations. Deepfakes pose significant risks regarding misinformation and fake news, because they can spread false information by depicting public figures saying or doing things they never did, undermining public trust. In this paper, we propose a method that leverages BLS signatures (Boneh, Lynn, and Shacham 2004) to implement signatures that remain valid after image cropping, but are invalidated in all the other types of manipulation, including deepfake creation. Our approach does not require who crops the image to know the signature private key or to be trusted in general, and it is O(1) in terms of signature size, making it a practical solution for scenarios where images are disseminated through web servers and cropping is the primary transformation. Finally, we adapted the signature scheme for the JPEG standard, and we experimentally tested the size of a signed image.

This paper presents a privacy-preserving protocol for identity registration and information sharing in federated authentication systems. The goal is to enable Identity Providers (IdPs) to detect duplicate or fraudulent identity enrollments without revealing users personal data or enabling cross-domain correlation. The protocol relies on Oblivious Pseudorandom Functions (OPRFs) combined with domain-specific transformations, ensuring that each IdP generates independent pseudonymous identifiers derived from a shared cryptographic service while maintaining full input confidentiality. A central authority maintains a blind registry that records successful and failed identity verifications using only pseudonymous identifiers, allowing global consistency checks without exposing sensitive information or linking users across domains. The proposed construction provides a general and abstract framework suitable for a wide range of federated authentication systems, achieving strong privacy guarantees while supporting effective fraud-prevention mechanisms during identity registration.

Advanced nuclear reactor systems face increasing cybersecurity threats as sophisticated attackers exploit cyber-physical interfaces to manipulate control systems while evading traditional IT security measures. This research presents a comprehensive evaluation of artificial intelligence approaches for cybersecurity protection in nuclear infrastructure, using Argonne National Laboratory's Mechanisms Engineering Test Loop (METL) as an experimental platform. We developed a systematic evaluation framework encompassing four machine learning detection paradigms: Change Point Detection, LSTM-based Anomaly Detection, Dependency Violation analysis, and Autoencoder reconstruction methods. Our comprehensive attack taxonomy includes 15 distinct scenarios targeting reactor control systems, each implemented across five severity tiers to evaluate detection performance under varying attack intensities. The experimental evaluation encompassed 300 rigorous experiments using realistic METL operational data. Change Point Detection emerged as the leading approach with mean AUC performance of 0.785, followed by LSTM Anomaly Detection (0.636), Dependency Violation (0.621), and Autoencoder methods (0.580). Attack detectability varied significantly, with multi-site coordinated attacks proving most detectable (AUC = 0.739) while precision trust decay attacks presented the greatest detection challenge (AUC = 0.592). This work delivers practical performance benchmarks and reference architecture that advance AI-based cybersecurity capabilities for critical nuclear infrastructure, providing essential foundations for operational deployment and enhanced threat response in cyber-physical systems.

Machine learning (ML) has been widely used to analyze API call sequences in malware analysis, which typically requires the expertise of domain specialists to extract relevant features from raw data. The extracted features play a critical role in malware analysis. Traditional feature extraction is based on human domain knowledge, while there is a trend of using natural language processing (NLP) for automatic feature extraction. This raises a question: how do we effectively select features for malware analysis based on API call sequences? To answer it, this paper presents a comprehensive study of investigating the impact of feature engineering upon malware classification.We first conducted a comparative performance evaluation under three models, Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and Transformer, with respect to knowledge-based and NLP-based feature engineering methods. We observed that models with knowledge-based feature engineering inputs generally outperform those using NLP-based across all metrics, especially under smaller sample sizes. Then we analyzed a complete set of data features from API call sequences, our analysis reveals that models often focus on features such as handles and virtual addresses, which vary across executions and are difficult for human analysts to interpret.

This paper analyzes the proliferation of cybersecurity ontologies, arguing that this surge cannot be explained solely by technical shortcomings related to quality, but also by a credibility deficit - a lack of trust, endorsement, and adoption by users. This conclusion is based on our first contribution, which is a state-of-the-art review and categorization of cybersecurity ontologies using the Framework for Ontologies Classification framework. To address this gap, we propose a revised framework for assessing credibility, introducing indicators such as institutional support, academic recognition, day-to-day practitioner validation, and industrial adoption. Based on these new credibility indicators, we construct a classification scheme designed to guide the selection of ontologies that are relevant to specific security needs. We then apply this framework to a concrete use case: the Franco-Luxembourgish research project ANCILE, which illustrates how a credibility-aware evaluation can reshape ontology selection for operational contexts.

Homomorphic secret sharing (HSS) allows multiple input clients to secretly share their private inputs to a function among several servers such that each server can homomorphically compute the function over its share to produce a share of the function's output. In HSS-enabled applications such as secure multi-party computation (MPC), security requires that the output shares leak no more information about the inputs than the function output. Such security is ensured by the context-hiding property of HSS. The typical rerandomization technique achieves context hiding but increases the share size. To address this, we formalize the context-hiding property of HSS for individual functions, examine the context-hiding property of Shamir-based HSS for monomials, and extend the study to polynomials.