Papers

Watermarks are an essential tool for identifying AI-generated content. Recently, Christ and Gunn (CRYPTO '24) introduced pseudorandom error-correcting codes (PRCs), which are equivalent to watermarks with strong robustness and quality guarantees. A PRC is a pseudorandom encryption scheme whose decryption algorithm tolerates a high rate of errors. Pseudorandomness ensures quality preservation of the watermark, and error tolerance of decryption translates to the watermark's ability to withstand modification of the content. In the short time since the introduction of PRCs, several works (NeurIPS '24, RANDOM '25, STOC '25) have proposed new constructions. Curiously, all of these constructions are vulnerable to quasipolynomial-time distinguishing attacks. Furthermore, all lack robustness to edits over a constant-sized alphabet, which is necessary for a meaningfully robust LLM watermark. Lastly, they lack robustness to adversaries who know the watermarking detection key. Until now, it was not clear whether any of these properties was achievable individually, let alone together. We construct pseudorandom codes that achieve all of the above: plausible subexponential pseudorandomness security, robustness to worst-case edits over a binary alphabet, and robustness against even computationally unbounded adversaries that have the detection key. Pseudorandomness rests on a new assumption that we formalize, the permuted codes conjecture, which states that a distribution of permuted noisy codewords is pseudorandom. We show that this conjecture is implied by the permuted puzzles conjecture used previously to construct doubly efficient private information retrieval. To give further evidence, we show that the conjecture holds against a broad class of simple distinguishers, including read-once branching programs.

The rise of space AI is reshaping government and industry through applications such as disaster detection, border surveillance, and climate monitoring, powered by massive data from commercial and governmental low Earth orbit (LEO) satellites. Federated satellite learning (FSL) enables joint model training without sharing raw data, but suffers from slow convergence due to intermittent connectivity and introduces critical trust challenges–where biased or falsified updates can arise across satellite constellations, including those injected through cyberattacks on inter-satellite or satellite-ground communication links. We propose OrbitChain, a blockchain-backed framework that empowers trustworthy multi-vendor collaboration in LEO networks. OrbitChain (i) offloads consensus to high-altitude platforms (HAPs) with greater computational capacity, (ii) ensures transparent, auditable provenance of model updates from different orbits owned by different vendors, and (iii) prevents manipulated or incomplete contributions from affecting global FSL model aggregation. Extensive simulations show that OrbitChain reduces computational and communication overhead while improving privacy, security, and global model accuracy. Its permissioned proof-of-authority ledger finalizes over 1000 blocks with sub-second latency (0.16,s, 0.26,s, 0.35,s for 1-of-5, 3-of-5, and 5-of-5 quorums). Moreover, OrbitChain reduces convergence time by up to 30 hours on real satellite datasets compared to single-vendor, demonstrating its effectiveness for real-time, multi-vendor learning. Our code is available at https://github.com/wsu-cyber-security-lab-ai/OrbitChain.git

The widespread use of big data across sectors has raised major privacy concerns, especially when sensitive information is shared or analyzed. Regulations such as GDPR and HIPAA impose strict controls on data handling, making it difficult to balance the need for insights with privacy requirements. Synthetic data offers a promising solution by creating artificial datasets that reflect real patterns without exposing sensitive information. However, traditional synthetic data methods often fail to capture complex, implicit rules that link different elements of the data and are essential in domains like healthcare. They may reproduce explicit patterns but overlook domain-specific constraints that are not directly stated yet crucial for realism and utility. For example, prescription guidelines that restrict certain medications for specific conditions or prevent harmful drug interactions may not appear explicitly in the original data. Synthetic data generated without these implicit rules can lead to medically inappropriate or unrealistic profiles. To address this gap, we propose ContextGAN, a Context-Aware Differentially Private Generative Adversarial Network that integrates domain-specific rules through a constraint matrix encoding both explicit and implicit knowledge. The constraint-aware discriminator evaluates synthetic data against these rules to ensure adherence to domain constraints, while differential privacy protects sensitive details from the original data. We validate ContextGAN across healthcare, security, and finance, showing that it produces high-quality synthetic data that respects domain rules and preserves privacy. Our results demonstrate that ContextGAN improves realism and utility by enforcing domain constraints, making it suitable for applications that require compliance with both explicit patterns and implicit rules under strict privacy guarantees.

Underground mining operations depend on sensor networks to monitor critical parameters such as temperature, gas concentration, and miner movement, enabling timely hazard detection and safety decisions. However, transmitting raw sensor data to a centralized server for machine learning (ML) model training raises serious privacy and security concerns. Federated Learning (FL) offers a promising alternative by enabling decentralized model training without exposing sensitive local data. Yet, applying FL in underground mining presents unique challenges: (i) Adversaries may eavesdrop on shared model updates to launch model inversion or membership inference attacks, compromising data privacy and operational safety; (ii) Non-IID data distributions across mines and sensor noise can hinder model convergence. To address these issues, we propose FedMining–a privacy-preserving FL framework tailored for underground mining. FedMining introduces two core innovations: (1) a Decentralized Functional Encryption (DFE) scheme that keeps local models encrypted, thwarting unauthorized access and inference attacks; and (2) a balancing aggregation mechanism to mitigate data heterogeneity and enhance convergence. Evaluations on real-world mining datasets demonstrate FedMining's ability to safeguard privacy while maintaining high model accuracy and achieving rapid convergence with reduced communication and computation overhead. These advantages make FedMining both secure and practical for real-time underground safety monitoring.

Nested virtualization is now widely supported by major cloud vendors, allowing users to leverage virtualization-based technologies in the cloud. However, supporting nested virtualization significantly increases host hypervisor complexity and introduces a new attack surface in cloud platforms. While many prior studies have explored hypervisor fuzzing, none has explicitly addressed nested virtualization due to the challenge of generating effective virtual machine (VM) instances with a vast state space as fuzzing inputs. We present NecoFuzz, the first fuzzing framework that systematically targets nested virtualization-specific logic in hypervisors. NecoFuzz synthesizes executable fuzz-harness VMs with internal states near the boundary between valid and invalid, guided by an approximate model of hardware-assisted virtualization specifications. Since vulnerabilities in nested virtualization often stem from incorrect handling of unexpected VM states, this specification-guided, boundary-oriented generation significantly improves coverage of security-critical code across different hypervisors. We implemented NecoFuzz on Intel VT-x and AMD-V by extending AFL++ to support fuzz-harness VMs. NecoFuzz achieved 84.7% and 74.2% code coverage for nested virtualization-specific code on Intel VT-x and AMD-V, respectively, and uncovered six previously unknown vulnerabilities across three hypervisors, including two assigned CVEs.

In the EU, the General Data Protection Regulation and the ePrivacy Directive mandate consent for the use of personal data for the purpose of behavioural advertising and tracking technologies. However, the ubiquity of consent banners has led to widespread consent fatigue and questions about the effectiveness of these mechanisms in protecting data subjects' data. To simplify digital laws and make the EU more competitive, the EU Commission recently proposed the Digital Omnibus, introducing a new Article 88b GDPR to express data subjects' choices in a technical way. While the Digital Omnibus is under legislative negotiation, California residents and residents of other US states can already exercise their rights via Global Privacy Control (GPC), a privacy signal to automatically broadcast a legally binding opt-out request to websites. In light of the Digital Omnibus, we evaluate to which extent GPC can be adapted to the EU legal framework to reduce consent banners, mitigate consent fatigue, and improve data protection for EU users. GPC is based on a technical specification, currently being standardised at the World Wide Web Consortium. By sending a GPC signal, data subjects can express their refusal or withdrawal of consent under the GDPR to the use of their personal data for cross-context ad targeting and, in some cases, to express their objection under the GDPR against the use of their data for such purposes. Our evaluation identifies friction between the GPC specification and current EU data protection law. In the longer term, it would be possible for the EU legislator to amend EU laws, as proposed in the current Digital Omnibus, in such a way that internet users can use automated signals to express choices about personal data use and online tracking. In the shorter term, websites and companies who conduct online tracking can already honour GPC.

With the rise of large language models, service providers offer language models as a service, enabling users to fine-tune customized models via uploaded private datasets. However, this raises concerns about sensitive data leakage. Prior methods, relying on differential privacy within device-cloud collaboration frameworks, struggle to balance privacy and utility, exposing users to inference attacks or degrading fine-tuning performance. To address this, we propose PrivTune, an efficient and privacy-preserving fine-tuning framework via Split Learning (SL). The key idea of PrivTune is to inject crafted noise into token representations from the SL bottom model, making each token resemble the $n$-hop indirect neighbors. PrivTune formulates this as an optimization problem to compute the optimal noise vector, aligning with defense-utility goals. On this basis, it then adjusts the parameters (i.e., mean) of the $d_\chi$-Privacy noise distribution to align with the optimization direction and scales the noise according to token importance to minimize distortion. Experiments on five datasets (covering both classification and generation tasks) against three embedding inversion and three attribute inference attacks show that, using RoBERTa on the Stanford Sentiment Treebank dataset, PrivTune reduces the attack success rate to 10% with only a 3.33% drop in utility performance, outperforming state-of-the-art baselines.

Despite advancements in machine learning for security, rule-based detection remains prevalent in Security Operations Centers due to the resource intensiveness and skill gap associated with ML solutions. While traditional rule-based methods offer efficiency, their rigidity leads to high false positives or negatives and requires continuous manual maintenance. This paper proposes a novel, two-stage hybrid framework to democratize ML-based threat detection. The first stage employs intentionally loose YARA rules for coarse-grained filtering, optimized for high recall. The second stage utilizes an ML classifier to filter out false positives from the first stage's output. To overcome data scarcity, the system leverages Simula, a seedless synthetic data generation framework, enabling security analysts to create high-quality training datasets without extensive data science expertise or pre-labeled examples. A continuous feedback loop incorporates real-time investigation results to adaptively tune the ML model, preventing rule degradation. This proposed model with active learning has been rigorously tested for a prolonged time in a production environment spanning tens of thousands of systems. The system handles initial raw log volumes often reaching 250 billion events per day, significantly reducing them through filtering and ML inference to a handful of daily tickets for human investigation. Live experiments over an extended timeline demonstrate a general improvement in the model's precision over time due to the active learning feature. This approach offers a self-sustained, low-overhead, and low-maintenance solution, allowing security professionals to guide model learning as expert “teachers”.

Hackers may create malicious solidity programs and deploy it in the Ethereum block chain. These malicious smart contracts try to attack legitimate programs by exploiting its vulnerabilities such as reentrancy, tx.origin attack, bad randomness, deligatecall and so on. This may lead to drain of the funds, denial of service and so on . Hence, it is necessary to identify and prevent the malicious smart contract before deploying it into the blockchain. In this paper, we propose an ML based malicious smart contract detection mechanism by analyzing the EVM opcodes. After balancing the opcode frequency dataset with SMOTE algorithm, we transformed opcode frequencies to the binary values (0,1) using an entropy based supervised binning method. Then, an explainable AI model is trained with the proposed binary opcode based features. From the implementations, we found that the proposed mechanism can detect 99% of malicious smart contracts with a false positive rate of only 0.01. Finally, we incorporated LIME algorithm in our classifier to justify its predictions. We found that, LIME algorithm can explain why a particular smart contract app is declared as malicious by our ML classifier based on the binary value of EVM opcodes.

We propose a new secret communication scheme over the bosonic wiretap channel. It uses readily available hardware such as lasers and direct photodetectors. The scheme is based on randomness extractors, pulse-position modulation, and Reed-Solomon codes and is therefore computationally efficient. It is secure against an eavesdropper performing coherent joint measurements on the quantum states it observes. In the low-photon-flow limit, the scheme is asymptotically optimal and achieves the same dominant term as the secrecy capacity of the same channel.

The evolving of digital transformation and increased use of technology comes with increased cyber vulnerabilities, which compromise national security. Cyber-threats become more sophisticated as the technology advances. This emphasises the need for strong risk mitigation strategies. To define strong and robust cybersecurity, policies requires an integrated approach of balancing technical expertise with public input. This paper aims to explore strategies used to balance technical expertise and public input to develop effective and robust cybersecurity policies. It also studied how the effective integration of technical expertise with public input is critical to developing effective strategies and resilient cybersecurity frameworks that strengthens national security. A lack of a holistic approach and collaborative efforts to cybersecurity can hinder the effectiveness of cybersecurity policies. This paper followed a systematic literature review with bibliometric analysis using the PRISMA methodology to explore how technical expertise and public input can be integrated to guide cybersecurity policy making. The thematic analysis identified five important themes in developing effective cybersecurity policies, these key themes are: Multi-Stakeholder Involvement and Human Centric Approaches (MSI & HCA), Governance and Policy Frameworks (GPF), Technical Infrastructure (TI), Evaluation and Compliance (EC), and Legal Rights and Sovereignty (LRS). The synthesis shows that there is no adequate exploration of collaborative efforts which undermines the effectiveness of the cybersecurity policies. The findings suggest that inclusive, flexible governance strategies that integrate public input at every stage are necessary for future cybersecurity policy research and practice, which must shift away from a primarily technical and legal perspective.

Sensitive citizen data, such as social, medical, and fiscal data, is heavily fragmented across public bodies and the private domain. Mining the combined data sets allows for new insights that otherwise remain hidden. Examples are improved healthcare, fraud detection, and evidence-based policy making. (Multi-party) delegated private set intersection (D-PSI) is a privacy-enhancing technology to link data across multiple data providers using a data collector. However, before it can be deployed in these use cases, it needs to be enhanced with additional functions, e.g., securely delivering payload only for elements in the intersection. Although there has been recent progress in the communication and computation requirements of D-PSI, these practical obstacles have not yet been addressed. This paper is the result of a collaboration with a governmental organization responsible for collecting, linking, and pseudonymizing data. Based on their requirements, we design a new D-PSI protocol with composable output functions, including encrypted payload and pseudonymized identifiers. We show that our protocol is secure in the standard model against colluding semi-honest data providers and against a non-colluding, possibly malicious independent party, the data collector. It, hence, allows to privately link and collect data from multiple data providers suitable for deployment in these use cases in the public sector.

Vulnerability code-bases often suffer from severe imbalance, limiting the effectiveness of Deep Learning-based vulnerability classifiers. Data Augmentation could help solve this by mitigating the scarcity of under-represented vulnerability types. In this context, we investigate LLM-based augmentation for vulnerable functions, comparing controlled generation of new vulnerable samples with semantics-preserving refactoring of existing ones. Using Qwen2.5-Coder to produce augmented data and CodeBERT as a classifier on the SVEN dataset, we find that our approaches are indeed effective in enriching vulnerable code-bases through a simple process and with reasonable quality, and that a hybrid strategy best boosts vulnerability classifiers' performance. Code repository is available here : https://github.com/DynaSoumhaneOuchebara/LLM-based-code-augmentation-Generate-or-Refactor-

Large Language Models (LLMs) have been integrated into many applications (e.g., web agents) to perform more sophisticated tasks. However, LLM-empowered applications are vulnerable to Indirect Prompt Injection (IPI) attacks, where instructions are injected via untrustworthy external data sources. This paper presents Rennervate, a defense framework to detect and prevent IPI attacks. Rennervate leverages attention features to detect the covert injection at a fine-grained token level, enabling precise sanitization that neutralizes IPI attacks while maintaining LLM functionalities. Specifically, the token-level detector is materialized with a 2-step attentive pooling mechanism, which aggregates attention heads and response tokens for IPI detection and sanitization. Moreover, we establish a fine-grained IPI dataset, FIPI, to be open-sourced to support further research. Extensive experiments verify that Rennervate outperforms 15 commercial and academic IPI defense methods, achieving high precision on 5 LLMs and 6 datasets. We also demonstrate that Rennervate is transferable to unseen attacks and robust against adaptive adversaries.

In the case of upgrading smart contracts on blockchain systems, it is essential to consider the continuity of upgrades and subsequent maintenance. In practice, upgrade operations often introduce new vulnerabilities. Existing static analysis tools usually only scan a single version and are unable to capture the correlation between code changes and emerging risks. To address this, we propose an Upgradeable Smart Contract Security Analyzer, USCSA, which uses Abstract Syntax Tree (AST) difference analysis to assess risks associated with the upgrade process and utilizes large language models (LLMs) for assisted reasoning to achieve high-confidence vulnerability attribution. We collected and analyzed 3,546 cases of vulnerabilities in upgradeable contracts, covering common vulnerability categories such as reentrancy, access control flaws, and integer overflow. Experimental results show that USCSA achieves a precision of 92.26%, a recall of 89.67%, and an F1-score of 90.95% in detecting upgrade-induced vulnerabilities. As a result, USCSA provides a significant advantage to improve the security and integrity of upgradeable smart contracts, offering a novel and efficient solution for security auditing on blockchain applications.

Sensitive information leakage in code repositories has emerged as a critical security challenge. Traditional detection methods that rely on regular expressions, fingerprint features, and high-entropy calculations often suffer from high false-positive rates. This not only reduces detection efficiency but also significantly increases the manual screening burden on developers. Recent advances in large language models (LLMs) and multi-agent collaborative architectures have demonstrated remarkable potential for tackling complex tasks, offering a novel technological perspective for sensitive information detection. In response to these challenges, we propose Argus, a multi-agent collaborative framework for detecting sensitive information. Argus employs a three-tier detection mechanism that integrates key content, file context, and project reference relationships to effectively reduce false positives and enhance overall detection accuracy. To comprehensively evaluate Argus in real-world repository environments, we developed two new benchmarks, one to assess genuine leak detection capabilities and another to evaluate false-positive filtering performance. Experimental results show that Argus achieves up to 94.86% accuracy in leak detection, with a precision of 96.36%, recall of 94.64%, and an F1 score of 0.955. Moreover, the analysis of 97 real repositories incurred a total cost of only 2.2$. All code implementations and related datasets are publicly available at https://github.com/TheBinKing/Argus-Guard for further research and application.

Cyber-physical systems (CPSs) are used extensively in critical infrastructure, underscoring the need for anomaly detection systems that are able to catch even the most motivated attackers. Traditional anomaly detection techniques typically do `one-off' training on datasets crafted by experts or generated by fuzzers, potentially limiting their ability to generalize to unseen and more subtle attack strategies. Stopping at this point misses a key opportunity: a defender can actively challenge the attacker to find more nuanced attacks, which in turn can lead to more effective detection capabilities. Building on this concept, we propose Evo-Defender, an evolutionary framework that iteratively strengthens CPS defenses through a dynamic attacker-defender interaction. Evo-Defender includes a smart attacker that employs guided fuzzing to explore diverse, non-redundant attack strategies, while the self-evolving defender uses incremental learning to adapt to new attack patterns. We implement Evo-Defender on two realistic CPS testbeds: the Tennessee Eastman process and a Robotic Arm Assembly Workstation, injecting over 600 attack scenarios. In end-to-end attack detection experiments, Evo-Defender achieves up to 2.7% higher performance than state-of-the-art baselines on unseen scenarios, while utilizing training data more efficiently for faster and more robust detection.

Device identifiers like the International Mobile Equipment Identity (IMEI) are crucial for ensuring device integrity and meeting regulations in 4G and 5G networks. However, sharing these identifiers with Mobile Network Operators (MNOs) brings significant privacy risks by enabling long-term tracking and linking of user activities across sessions. In this work, we propose a privacy-preserving identifier checking method in 5G. This paper introduces a protocol for verifying device identifiers without exposing them to the network while maintaining the same functions as the 3GPP-defined Equipment Identity Register (EIR) process. The proposed solution modifies the PEPSI protocol for a Private Set Membership (PSM) setting using the BFV homomorphic encryption scheme. This lets User Equipment (UE) prove that its identifier is not on an operator's blacklist or greylist while ensuring that the MNO only learns the outcome of the verification. The protocol allows controlled deanonymization through an authorized Law Enforcement (LE) hook, striking a balance between privacy and accountability. Implementation results show that the system can perform online verification within five seconds and requires about 15 to 16 MB of communication per session. This confirms its practical use under post-quantum security standards. The findings highlight the promise of homomorphic encryption for managing identifiers while preserving privacy in 5G, laying the groundwork for scalable and compliant verification systems in future 6G networks.

In todays digital world, protecting sensitive data is very essential. Steganography hides the existence of secret data instead of its content, providing better security for multimedia communication. This paper proposes a new technique for hiding audio files inside images using the Least Significant Bit (LSB) method optimized by the Harris Hawks Optimization (HHO) algorithm. HHO is a nature-inspired metaheuristic that imitates the hunting behavior of Harris hawks to find optimal pixel positions for embedding data. The proposed method is evaluated using Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), and Mean Square Error (MSE). Experimental results show that HHO achieves better image quality, robustness, and embedding capacity compared to existing methods.

Neural models for vulnerability prediction (VP) have achieved impressive performance by learning from large-scale code repositories. However, their susceptibility to Membership Inference Attacks (MIAs), where adversaries aim to infer whether a particular code sample was used during training, poses serious privacy concerns. While MIA has been widely investigated in NLP and vision domains, its effects on security-critical code analysis tasks remain underexplored. In this work, we conduct the first comprehensive analysis of MIA on VP models, evaluating the attack success across various architectures (LSTM, BiGRU, and CodeBERT) and feature combinations, including embeddings, logits, loss, and confidence. Our threat model aligns with black-box and gray-box settings where prediction outputs are observable, allowing adversaries to infer membership by analyzing output discrepancies between training and non-training samples. The empirical findings reveal that logits and loss are the most informative and vulnerable outputs for membership leakage. Motivated by these observations, we propose a Noise-based Membership Inference Defense (NMID), which is a lightweight defense module that applies output masking and Gaussian noise injection to disrupt adversarial inference. Extensive experiments demonstrate that NMID significantly reduces MIA effectiveness, lowering the attack AUC from nearly 1.0 to below 0.65, while preserving the predictive utility of VP models. Our study highlights critical privacy risks in code analysis and offers actionable defense strategies for securing AI-powered software systems.

The Model Context Protocol (MCP) has emerged as the de facto standard for connecting Large Language Models (LLMs) to external data and tools, effectively functioning as the "USB-C for Agentic AI." While this decoupling of context and execution solves critical interoperability challenges, it introduces a profound new threat landscape where the boundary between epistemic errors (hallucinations) and security breaches (unauthorized actions) dissolves. This Systematization of Knowledge (SoK) aims to provide a comprehensive taxonomy of risks in the MCP ecosystem, distinguishing between adversarial security threats (e.g., indirect prompt injection, tool poisoning) and epistemic safety hazards (e.g., alignment failures in distributed tool delegation). We analyze the structural vulnerabilities of MCP primitives, specifically Resources, Prompts, and Tools, and demonstrate how "context" can be weaponized to trigger unauthorized operations in multi-agent environments. Furthermore, we survey state-of-the-art defenses, ranging from cryptographic provenance (ETDI) to runtime intent verification, and conclude with a roadmap for securing the transition from conversational chatbots to autonomous agentic operating systems.

Retrieval-Augmented Generation (RAG) systems enhance LLMs with external knowledge but introduce a critical attack surface: corpus poisoning. While recent studies have demonstrated the potential of such attacks, they typically rely on impractical assumptions, such as white-box access or known user queries, thereby underestimating the difficulty of real-world exploitation. In this paper, we bridge this gap by proposing MIRAGE, a novel multi-stage poisoning pipeline designed for strict black-box and query-agnostic environments. Operating on surrogate model feedback, MIRAGE functions as an automated optimization framework that integrates three key mechanisms: it utilizes persona-driven query synthesis to approximate latent user search distributions, employs semantic anchoring to imperceptibly embed these intents for high retrieval visibility, and leverages an adversarial variant of Test-Time Preference Optimization (TPO) to maximize persuasion. To rigorously evaluate this threat, we construct a new benchmark derived from three long-form, domain-specific datasets. Extensive experiments demonstrate that MIRAGE significantly outperforms existing baselines in both attack efficacy and stealthiness, exhibiting remarkable transferability across diverse retriever-LLM configurations and highlighting the urgent need for robust defense strategies.

Connected vehicles represent a key enabler of intelligent transportation systems, where vehicles are equipped with advanced communication, sensing, and computing technologies to interact not only with one another but also with surrounding infrastructures and the environment. Through continuous data exchange, such vehicles are capable of enhancing road safety, improving traffic efficiency, and ensuring more reliable mobility services. Further, when these capabilities are integrated with advanced automation technologies, the concept essentially evolves into connected and autonomous vehicles (CAVs). While connected vehicles primarily focus on seamless information sharing, autonomous vehicles are mainly dependent on advanced perception, decision-making, and control mechanisms to operate with minimal or without human intervention. However, as a result of connectivity, an adversary with malicious intentions might be able to compromise successfully by breaching the system components of CAVs. In this paper, we present an attack-tree based methodology for evaluating cyber security vulnerabilities in CAVs. In particular, we utilize the attack-defense tree formulation to systematically assess attack-leaf vulnerabilities, and before analyzing the vulnerability indices, we also define a measure of vulnerabilities, which is based on existing cyber security threats and corresponding defensive countermeasures.

Medical Large Language Models (LLMs) are increasingly deployed for clinical decision support across diverse specialties, yet systematic evaluation of their robustness to adversarial misuse and privacy leakage remains inaccessible to most researchers. Existing security benchmarks require GPU clusters, commercial API access, or protected health data – barriers that limit community participation in this critical research area. We propose a practical, fully reproducible framework for evaluating medical AI security under realistic resource constraints. Our framework design covers multiple medical specialties stratified by clinical risk – from high-risk domains such as emergency medicine and psychiatry to general practice – addressing jailbreaking attacks (role-playing, authority impersonation, multi-turn manipulation) and privacy extraction attacks. All evaluation utilizes synthetic patient records requiring no IRB approval. The framework is designed to run entirely on consumer CPU hardware using freely available models, eliminating cost barriers. We present the framework specification including threat models, data generation methodology, evaluation protocols, and scoring rubrics. This proposal establishes a foundation for comparative security assessment of medical-specialist models and defense mechanisms, advancing the broader goal of ensuring safe and trustworthy medical AI systems.

At ASIACRYPT 2018, a digital attack based on linear least squares was introduced for a variant of the learning with errors (LWE) problem which omits modular reduction known as the integer learning with errors problem (ILWE). In this paper, we present a theoretical and experimental study of the effectiveness of the attack when applied directly to small parameter ILWE instances found in popular digital signature schemes such as CRYSTALS-Dilithium which utilize rejection sampling. Unlike other studies which form ILWE instances based on additional information obtained from side-channel attacks, we take a more direct approach to the problem by constructing our ILWE instance from only the obtained signatures. We outline and introduce novel techniques in our simulation designs such as modular polynomial arithmetic via matrices in $\mathbb{R}$, as well as algorithms for handling large sample sizes efficiently. Our experimental results reinforce the proclaimed security of signature schemes based on ILWE. We additionally discuss the implications of our work and digital signatures as a whole in regards to real-world applications such as in Intelligent Transportation Systems (ITS).

Security Operations Centers face massive, heterogeneous alert streams under minute-level service windows, creating the Alert Triage Latency Paradox: verbose reasoning chains ensure accuracy and compliance but incur prohibitive latency and token costs, while minimal chains sacrifice transparency and auditability. Existing solutions fail: signature systems are brittle, anomaly methods lack actionability, and fully cloud-hosted LLMs raise latency, cost, and privacy concerns. We propose AIDR, a hybrid cloud-edge framework that addresses this trade-off through constrained information-density optimization. The core innovation is gradient-based compression of reasoning chains to retain only decision-critical steps–minimal evidence sufficient to justify predictions while respecting token and latency budgets. We demonstrate that this approach preserves decision-relevant information while minimizing complexity. We construct compact datasets by distilling alerts into 3-5 high-information bullets (68% token reduction), train domain-specialized experts via LoRA, and deploy a cloud-edge architecture: a cloud LLM routes alerts to on-premises experts generating SOAR-ready JSON. Experiments demonstrate AIDR achieves higher accuracy and 40.6% latency reduction versus Chain-of-Thought, with robustness to data corruption and out-of-distribution generalization, enabling auditable and efficient SOC triage with full data residency compliance.

Adversaries (hackers) attempting to infiltrate networks frequently face uncertainty in their operational environments. This research explores the ability to model and detect when they exhibit ambiguity aversion, a cognitive bias reflecting a preference for known (versus unknown) probabilities. We introduce a novel methodological framework that (1) leverages rich, multi-modal data from human-subjects red-team experiments, (2) employs a large language model (LLM) pipeline to parse unstructured logs into MITRE ATT&CK-mapped action sequences, and (3) applies a new computational model to infer an attacker's ambiguity aversion level in near-real time. By operationalizing this cognitive trait, our work provides a foundational component for developing adaptive cognitive defense strategies.

As AI agents increasingly operate in complex environments, ensuring reliable, context-aware privacy is critical for regulatory compliance. Traditional access controls are insufficient because privacy risks often arise after access is granted; agents may inadvertently compromise privacy during reasoning by messaging humans, leaking context to peers, or executing unsafe tool calls. Existing approaches typically treat privacy as a binary constraint, overlooking nuanced, computation-dependent requirements. Furthermore, Large Language Model (LLM) agents are inherently probabilistic, lacking formal guarantees for security-critical operations. To address this, we introduce AgentCrypt, a three-tiered framework for secure agent communication that adds a deterministic protection layer atop any AI platform. AgentCrypt spans the full spectrum of privacy needs: from unrestricted data exchange (Level 1), to context-aware masking (Level 2), up to fully encrypted computation using Homomorphic Encryption (Level 3). Unlike prompt-based defenses, our approach guarantees that tagged data privacy is strictly preserved even when the underlying model errs. Security is decoupled from the agent's probabilistic reasoning, ensuring sensitive data remains protected throughout the computational lifecycle. AgentCrypt enables collaborative computation on otherwise inaccessible data, overcoming barriers like data silos. We implemented and validated it using LangGraph and Google ADK, demonstrating versatility across architectures. Finally, we introduce a benchmark dataset simulating privacy-critical tasks to enable systematic evaluation and foster the development of trustworthy, regulatable machine learning systems.

CapsuleFS (CFS) is the first filesystem to integrate multi-credential functionality within a POSIX-compliant framework, utilizing DataCapsule as the storage provider. This innovative system is established based on the Global Data Plane in the area of edge computing. Our comprehensive design and implementation of CFS successfully fulfill the objective of providing a multi-credential Common Access API. The architecture of CFS is methodically segmented into three integral components: Firstly, the DataCapsule server, tasked with the storage, dissemination, and replication of DataCapsules on the edge. Secondly, the middleware, a crucial element running in a Trusted Execution Environment responsible for the enforcement and management of write permissions and requests. Finally, the client component, which manifests as a POSIX-compliant filesystem, is adaptable and operational across many architectures. Experimental evaluations of CFS reveal that, while its read and write performances are comparatively modest, it upholds a high degree of functional correctness. This attribute distinctly positions CFS as a viable candidate for application in real-world software development scenarios. The paper also delineates potential future enhancements, aimed at augmenting the practicality of CFS in the landscape of software development.

Uganda's rapid digital transformation, supported by national strategies such as Vision 2040 and the Digital Transformation Roadmap, has expanded reliance on networked services while simultaneously increasing exposure to sophisticated cyber threats. In resource-constrained settings, commonly deployed rule-based intrusion detection systems lack the adaptability and ethical safeguards needed to address evolving attack patterns, leading to undetected breaches and excessive blocking of legitimate traffic. This study proposes an Agentic Artificial Intelligence (AAI) framework that integrates reinforcement learning, an explicit ethical governance layer, and human oversight to deliver adaptive and trustworthy cybersecurity. A CPU-optimized simulation environment was developed using a five-node network topology that mirrors key elements of Uganda's critical digital infrastructure and generates both benign and malicious traffic, including phishing, ransomware, and distributed denial-of-service attacks. A Q-learning agent, operating within clearly defined ethical constraints and subject to human auditability, was trained and evaluated against a traditional rule-based baseline. The AAI framework achieved a 100 percent detection rate, zero false positives, and full ethical compliance, compared with 70 percent detection and 15 percent false positives for the baseline system. These results demonstrate that agentic, ethically governed reinforcement learning can substantially improve cybersecurity effectiveness and fairness in CPU-only, resource-constrained environments, offering a practical pathway for operationalizing responsible AI in Uganda's national cybersecurity strategy.