Papers

High-dimensional malware datasets often exhibit feature redundancy, instability, and scalability limitations, which hinder the effectiveness and interpretability of machine learning-based malware detection systems. Although feature selection is commonly employed to mitigate these issues, many existing approaches lack robustness when applied to large-scale and heterogeneous malware data. To address this gap, this paper proposes CAFE-GB (Chunk-wise Aggregated Feature Estimation using Gradient Boosting), a scalable feature selection framework designed to produce stable and globally consistent feature rankings for high-dimensional malware detection. CAFE-GB partitions training data into overlapping chunks, estimates local feature importance using gradient boosting models, and aggregates these estimates to derive a robust global ranking. Feature budget selection is performed separately through a systematic k-selection and stability analysis to balance detection performance and robustness. The proposed framework is evaluated on two large-scale malware datasets: BODMAS and CIC-AndMal2020, representing large and diverse malware feature spaces. Experimental results show that classifiers trained on CAFE-GB -selected features achieve performance parity with full-feature baselines across multiple metrics, including Accuracy, F1-score, MCC, ROC-AUC, and PR-AUC, while reducing feature dimensionality by more than 95%. Paired Wilcoxon signed-rank tests confirm that this reduction does not introduce statistically significant performance degradation. Additional analyses demonstrate low inter-feature redundancy and improved interpretability through SHAP-based explanations. Runtime and memory profiling further indicate reduced downstream classification overhead. Overall, CAFE-GB provides a stable, interpretable, and scalable feature selection strategy for large-scale malware detection.

Financial Generative Pre-trained Transformers (FinGPT) with multimodal capabilities are now being increasingly adopted in various financial applications. However, due to the intellectual property of model weights and the copyright of training corpus and benchmarking questions, verifying the legitimacy of GPT's model weights and the credibility of model outputs is a pressing challenge. In this paper, we introduce a novel zkFinGPT scheme that applies zero-knowledge proofs (ZKPs) to high-value financial use cases, enabling verification while protecting data privacy. We describe how zkFinGPT will be applied to three financial use cases. Our experiments on two existing packages reveal that zkFinGPT introduces substantial computational overhead that hinders its real-world adoption. E.g., for LLama3-8B model, it generates a commitment file of $7.97$MB using $531$ seconds, and takes $620$ seconds to prove and $2.36$ seconds to verify.

As digital threats continue to grow, organizations must find ways to enhance security while protecting user privacy. This paper explores how artificial intelligence (AI) plays a crucial role in achieving this balance. AI technologies can improve security by detecting threats, monitoring systems, and automating responses. However, using AI also raises privacy concerns that need careful consideration.We examine real-world examples from the healthcare sector to illustrate how organizations can implement AI solutions that strengthen security without compromising patient privacy. Additionally, we discuss the importance of creating transparent AI systems and adhering to privacy regulations.Ultimately, this paper provides insights and recommendations for integrating AI into healthcare security practices, helping organizations navigate the challenges of modern management while keeping patient data safe.

Stealing attacks pose a persistent threat to the intellectual property of deployed machine-learning systems. Retrieval-augmented generation (RAG) intensifies this risk by extending the attack surface beyond model weights to knowledge base that often contains IP-bearing assets such as proprietary runbooks, curated domain collections, or licensed documents. Recent work shows that multi-turn questioning can gradually steal corpus content from RAG systems, yet existing attacks are largely heuristic and often plateau early. We address this gap by formulating RAG knowledge-base stealing as an adaptive stochastic coverage problem (ASCP), where each query is a stochastic action and the goal is to maximize the conditional expected marginal gain (CMG) in corpus coverage under a query budget. Bridging ASCP to real-world black-box RAG knowledge-base stealing raises three challenges: CMG is unobservable, the natural-language action space is intractably large, and feasibility constraints require stealthy queries that remain effective under diverse architectures. We introduce RAGCrawler, a knowledge graph-guided attacker that maintains a global attacker-side state to estimate coverage gains, schedule high-value semantic anchors, and generate non-redundant natural queries. Across four corpora and four generators with BGE retriever, RAGCrawler achieves 66.8% average coverage (up to 84.4%) within 1,000 queries, improving coverage by 44.90% relative to the strongest baseline. It also reduces the queries needed to reach 70% coverage by at least 4.03x on average and enables surrogate reconstruction with answer similarity up to 0.699. Our attack is also scalable to retriever switching and newer RAG techniques like query rewriting and multi-query retrieval. These results highlight urgent needs to protect RAG knowledge assets.

Realistic network traffic simulation is critical for evaluating intrusion detection systems, stress-testing network protocols, and constructing high-fidelity environments for cybersecurity training. While attack traffic can often be layered into training environments using red-teaming or replay methods, generating authentic benign background traffic remains a core challenge – particularly in simulating the complex temporal and communication dynamics of real-world networks. This paper introduces TempoNet, a novel generative model that combines multi-task learning with multi-mark temporal point processes to jointly model inter-arrival times and all packet- and flow-header fields. TempoNet captures fine-grained timing patterns and higher-order correlations such as host-pair behavior and seasonal trends, addressing key limitations of GAN-, LLM-, and Bayesian-based methods that fail to reproduce structured temporal variation. TempoNet produces temporally consistent, high-fidelity traces, validated on real-world datasets. Furthermore, we show that intrusion detection models trained on TempoNet-generated background traffic perform comparably to those trained on real data, validating its utility for real-world security applications.

Hallucinations in Large Language Models (LLMs) – generations that are plausible but factually unfaithful – remain a critical barrier to high-stakes deployment. Current detection methods typically rely on computationally expensive external retrieval loops or opaque black-box LLM judges requiring 70B+ parameters. In this work, we introduce [Model Name], a hybrid detection framework that combines neuroscience-inspired signal design with supervised machine learning. We extract interpretable signals grounded in Predictive Coding (quantifying surprise against internal priors) and the Information Bottleneck (measuring signal retention under perturbation). Through systematic ablation, we demonstrate three key enhancements: Entity-Focused Uptake (concentrating on high-value tokens), Context Adherence (measuring grounding strength), and Falsifiability Score (detecting confident but contradictory claims). Evaluating on HaluBench (n=200, perfectly balanced), our theory-guided baseline achieves 0.8017 AUROC. BASE supervised models reach 0.8274 AUROC, while IMPROVED features boost performance to 0.8669 AUROC (4.95% gain), demonstrating consistent improvements across architectures. This competitive performance is achieved while using 75x less training data than Lynx (200 vs 15,000 samples), 1000x faster inference (5ms vs 5s), and remaining fully interpretable. Crucially, we report a negative result: the Rationalization signal fails to distinguish hallucinations, suggesting that LLMs generate coherent reasoning for false premises ("Sycophancy"). This work demonstrates that domain knowledge encoded in signal architecture provides superior data efficiency compared to scaling LLM judges, achieving strong performance with lightweight (less than 1M parameter), explainable models suitable for production deployment.

Virtualization is widely adopted in cloud systems to manage resource sharing among users. A virtualized environment usually deploys a virtual switch within the host system to enable virtual machines to communicate with each other and with the physical network. The Open vSwitch (OVS) is one of the most popular software-based virtual switches. It maintains a cache hierarchy to accelerate packet forwarding from the host to virtual machines. We characterize the caching system inside OVS from a security perspective and identify three attack primitives. Based on the attack primitives, we present three remote attacks via OVS, breaking the isolation in virtualized environments. First, we identify remote covert channels using different caches. Second, we present a novel header recovery attack that leaks a remote user's packet header fields, breaking the confidentiality guarantees from the system. Third, we demonstrate a remote packet rate monitoring attack that recovers the packet rate of a remote victim. To defend against these attacks, we also discuss and evaluate mitigation solutions.

Large language models (LLMs) exhibit powerful capabilities but risk memorizing sensitive personally identifiable information (PII) from their training data, posing significant privacy concerns. While machine unlearning techniques aim to remove such data, they predominantly depend on access to the training data. This requirement is often impractical, as training data in real-world deployments is commonly proprietary or inaccessible. To address this limitation, we propose Data-Free Selective Unlearning (DFSU), a novel privacy-preserving framework that removes sensitive PII from an LLM without requiring its training data. Our approach first synthesizes pseudo-PII through language model inversion, then constructs token-level privacy masks for these synthetic samples, and finally performs token-level selective unlearning via a contrastive mask loss within a low-rank adaptation (LoRA) subspace. Extensive experiments on the AI4Privacy PII-Masking dataset using Pythia models demonstrate that our method effectively removes target PII while maintaining model utility.

The use of algorithms is increasing across various fields such as healthcare, justice, finance, and education. This growth has significantly accelerated with the advent of Artificial Intelligence (AI) technologies based on Large Language Models (LLMs) since 2022. This expansion presents substantial challenges related to accountability, ethics, and transparency. This article explores the potential of the Digital Object Identifier (DOI) to identify algorithms, aiming to enhance accountability, transparency, and reliability in their development and application, particularly in AI agents and multimodal LLMs. The use of DOIs facilitates tracking the origin of algorithms, enables audits, prevents biases, promotes research reproducibility, and strengthens ethical considerations. The discussion addresses the challenges and solutions associated with maintaining algorithms identified by DOI, their application in API security, and the proposal of a cryptographic authentication protocol.

Large Language Model (LLM)-based question-answering systems offer significant potential for automating customer support and internal knowledge access in small businesses, yet their practical deployment remains challenging due to infrastructure costs, engineering complexity, and security risks, particularly in retrieval-augmented generation (RAG)-based settings. This paper presents an industry case study of an open-source, multi-tenant platform that enables small businesses to deploy customised LLM-based support chatbots via a no-code workflow. The platform is built on distributed, lightweight k3s clusters spanning heterogeneous, low-cost machines and interconnected through an encrypted overlay network, enabling cost-efficient resource pooling while enforcing container-based isolation and per-tenant data access controls. In addition, the platform integrates practical, platform-level defences against prompt injection attacks in RAG-based chatbots, translating insights from recent prompt injection research into deployable security mechanisms without requiring model retraining or enterprise-scale infrastructure. We evaluate the proposed platform through a real-world e-commerce deployment, demonstrating that secure and efficient LLM-based chatbot services can be achieved under realistic cost, operational, and security constraints faced by small businesses.

The adoption of Electric Vehicles (EVs) is happening at a rapid pace. To ensure fast and safe charging, complex communication is required between the vehicle and the charging station. In the globally used Combined Charging System (CCS), this communication is carried over the HomePlug Green PHY (HPGP) physical layer. However, HPGP is known to suffer from wireless leakage, which may expose this data link to nearby attackers. In this paper, we examine active wireless attacks against CCS, and study the impact they can have. We present the first real-time Software-Defined Radio (SDR) implementation of HPGP, granting unprecedented access to the communications within the charging cables. We analyze the characteristics of 2,750 real-world charging sessions to understand the timing constraints for hijacking. Using novel techniques to increase the attacks' reliability, we design a robust wireless Man-in-the-Middle evaluation framework for CCS. We demonstrate full control over TLS usage and CCS protocol version negotiation, including TLS stripping attacks. We investigate how real devices respond to safety-critical MitM attacks, which modify power delivery information, and found target vehicles to be highly permissive. First, we caused a vehicle to display charging power exceeding 900 kW on the dashboard, while receiving only 40 kW. Second, we remotely overcharged a vehicle, at twice the requested current for 17 seconds before the vehicle triggered the emergency shutdown. Finally, we propose a backwards-compatible, downgrade-proof protocol extension to mitigate the underlying vulnerabilities.

Graph neural network (GNN) have demonstrated exceptional performance in solving critical problems across diverse domains yet remain susceptible to backdoor attacks. Existing studies on backdoor attack for graph classification are limited to single target attack using subgraph replacement based mechanism where the attacker implants only one trigger into the GNN model. In this paper, we introduce the first multi-targeted backdoor attack for graph classification task, where multiple triggers simultaneously redirect predictions to different target labels. Instead of subgraph replacement, we propose subgraph injection which preserves the structure of the original graphs while poisoning the clean graphs. Extensive experiments demonstrate the efficacy of our approach, where our attack achieves high attack success rates for all target labels with minimal impact on the clean accuracy. Experimental results on five dataset demonstrate the superior performance of our attack framework compared to the conventional subgraph replacement-based attack. Our analysis on four GNN models confirms the generalization capability of our attack which is effective regardless of the GNN model architectures and training parameters settings. We further investigate the impact of the attack design parameters including injection methods, number of connections, trigger sizes, trigger edge density and poisoning ratios. Additionally, our evaluation against state-of-the-art defenses (randomized smoothing and fine-pruning) demonstrates the robustness of our proposed multi-target attacks. This work highlights the GNN vulnerability against multi-targeted backdoor attack in graph classification task. Our source codes will be available at https://github.com/SiSL-URI/Multi-Targeted-Graph-Backdoor-Attack.

Homomorphic encryption (HE) enables arithmetic operations to be performed directly on encrypted data. It is essential for privacy-preserving applications such as machine learning, medical diagnosis, and financial data analysis. In popular HE schemes, ciphertext multiplication is only defined for two inputs. However, the multiplication of multiple inputs is needed in many HE applications. In our previous work, a three-input ciphertext multiplication method for the CKKS HE scheme was developed. This paper first reformulates the three-input ciphertext multiplication to enable the combination of computations in order to further reduce the complexity. The second contribution is extending the multiplication to multiple inputs without compromising the noise overhead. Additional evaluation keys are introduced to achieve relinearization of polynomial multiplication results. To minimize the complexity of the large number of rescaling units in the multiplier, a theoretical analysis is developed to relocate the rescaling, and a multi-level rescaling approach is proposed to implement combined rescaling with complexity similar to that of a single rescaling unit. Guidelines and examples are provided on the input partition to enable the combination of more rescaling. Additionally, efficient hardware architectures are designed to implement our proposed multipliers. The improved three-input ciphertext multiplier reduces the logic area and latency by 15% and 50%, respectively, compared to the best prior design. For multipliers with more inputs, ranging from 4 to 12, the architectural analysis reveals 32% savings in area and 45% shorter latency, on average, compared to prior work.

Quantum channel capacities are fundamental to quantum information theory. Their definition, however, does not limit the computational resources of sender and receiver. In this work, we initiate the study of computational quantum capacities. These quantify how much information can be reliably transmitted when imposing the natural requirement that en- and decoding have to be computationally efficient. We focus on the computational two-way quantum capacity and showcase that it is closely related to the computational distillable entanglement of the Choi state of the channel. This connection allows us to show a stark computational capacity separation. Under standard cryptographic assumptions, there exists a quantum channel of polynomial complexity whose computational two-way quantum capacity vanishes while its unbounded counterpart is nearly maximal. More so, we show that there exists a sharp transition in computational quantum capacity from nearly maximal to zero when the channel complexity leaves the polynomial realm. Our results demonstrate that the natural requirement of computational efficiency can radically alter the limits of quantum communication.

The rapid growth of Internet of Things (IoT) devices has increased the scale and diversity of cyberattacks, exposing limitations in traditional intrusion detection systems. Classical machine learning (ML) models such as Random Forest and Support Vector Machine perform well on known attacks but require retraining to detect unseen or zero-day threats. This study investigates lightweight decoder-only Large Language Models (LLMs) for IoT attack detection by integrating structured-to-text conversion, Quantized Low-Rank Adaptation (QLoRA) fine-tuning, and Retrieval-Augmented Generation (RAG). Network traffic features are transformed into compact natural-language prompts, enabling efficient adaptation under constrained hardware. Experiments on the CICIoT2023 dataset show that a QLoRA-tuned LLaMA-1B model achieves an F1-score of 0.7124, comparable to the Random Forest (RF) baseline (0.7159) for known attacks. With RAG, the system attains 42.63% accuracy on unseen attack types without additional training, demonstrating practical zero-shot capability. These results highlight the potential of retrieval-enhanced lightweight LLMs as adaptable and resource-efficient solutions for next-generation IoT intrusion detection.

Fog and mobile edge computing (MEC) will play a key role in the upcoming fifth generation (5G) mobile networks to support decentralized applications, data analytics and management into the network itself by using a highly distributed compute model. Furthermore, increasing attention is paid to providing user-centric cybersecurity solutions, which particularly require collecting, processing and analyzing significantly large amount of data traffic and huge number of network connections in 5G networks. In this regard, this paper proposes a MEC-oriented solution in 5G mobile networks to detect network anomalies in real-time and in autonomic way. Our proposal uses deep learning techniques to analyze network flows and to detect network anomalies. Moreover, it uses policies in order to provide an efficient and dynamic management system of the computing resources used in the anomaly detection process. The paper presents relevant aspects of the deployment of the proposal and experimental results to show its performance.

Python is one of the most popular programming languages; as such, projects written in Python involve an increasing number of diverse security vulnerabilities. However, existing state-of-the-art analysis tools for Python only support a few vulnerability types. Hence, there is a need to detect a large variety of vulnerabilities in Python projects. In this paper, we propose the SAGA approach for detecting and locating vulnerabilities in Python source code in a versatile way. SAGA includes a source code parser that extracts control- and data-flow information and represents it as a symbolic control-flow graph, as well as a domain-specific language that defines static aspects of the source code and their evolution during graph traversals. We have leveraged this language to define a library of static aspects for integrity, confidentiality, and other security-related properties. We have evaluated SAGA on a dataset of 108 vulnerabilities affecting 82 PyPi packages and covering 37 vulnerability types, obtaining 100% sensitivity and 99.15% specificity, with only one false positive, while outperforming four common security analysis tools. This analysis was performed in less than 31 seconds, i.e., between 2.5 and 512.1 times faster than the baseline tools.

Traditional defenses against Deep Leakage (DL) attacks in Federated Learning (FL) primarily focus on obfuscation, introducing noise, transformations or encryption to degrade an attacker's ability to reconstruct private data. While effective to some extent, these methods often still leak high-level information such as class distributions or feature representations, and are frequently broken by increasingly powerful denoising attacks. We propose a fundamentally different perspective on FL defense: framing it as a spoofing problem.We introduce SpooFL (Figure 1), a spoofing-based defense that deceives attackers into believing they have recovered the true training data, while actually providing convincing but entirely synthetic samples from an unrelated task. Unlike prior synthetic-data defenses that share classes or distributions with the private data and thus still leak semantic information, SpooFL uses a state-of-the-art generative model trained on an external dataset with no class overlap. As a result, attackers are misled into recovering plausible yet completely irrelevant samples, preventing meaningful data leakage while preserving FL training integrity. We implement the first example of such a spoofing defense, and evaluate our method against state-of-the-art DL defenses and demonstrate that it successfully misdirects attackers without compromising model performance significantly.

While the literature features a number of proposals to defend against transaction manipulation attacks, existing proposals are still not integrated within large blockchains, such as Bitcoin, Ethereum, and Cardano. Instead, the user community opted to rely on more practical but ad-hoc solutions (such as Mempool.space) that aim at detecting censorship and transaction displacement attacks by auditing discrepancies in the mempools of so-called observers. In this paper, we precisely analyze, for the first time, the interplay between mempool auditing and the ability to detect censorship and transaction displacement attacks by malicious miners in Bitcoin and Ethereum. Our analysis shows that mempool auditing can result in mis-accusations against miners with a probability larger than 25% in some settings. On a positive note, however, we show that mempool auditing schemes can successfully audit the execution of any two transactions (with an overwhelming probability of 99.9%) if they are consistently received by all observers and sent at least 30 seconds apart from each other. As a direct consequence, our findings show, for the first time, that batch-order fair-ordering schemes can offer only strong fairness guarantees for a limited subset of transactions in real-world deployments.

Verifiable delegation in digital identity systems remains unresolved across centralized, federated, and self-sovereign identity (SSI) environments, particularly where both human users and autonomous AI agents must exercise and transfer authority without exposing primary credentials or private keys. We introduce a unified framework that enables bounded, auditable, and least-privilege delegation across heterogeneous identity ecosystems. The framework includes four key elements: Delegation Grants (DGs), first-class authorization artefacts that encode revocable transfers of authority with enforced scope reduction; a Canonical Verification Context (CVC) that normalizes verification requests into a single structured representation independent of protocols or credential formats; a layered reference architecture that separates trust anchoring, credential and proof validation, policy evaluation, and protocol mediation via a Trust Gateway; and an explicit treatment of blockchain anchoring as an optional integrity layer rather than a structural dependency. Together, these elements advance interoperable delegation and auditability and provide a foundation for future standardization, implementation, and integration of autonomous agents into trusted digital identity infrastructures.

The emergence of quantum computing poses a fundamental threat to current public key cryptographic systems. This threat is necessitating a transition to quantum resistant cryptographic alternatives in all the applications. In this work, we present the implementation of a practical hybrid end-to-end encryption system that combines classical and post-quantum cryptographic primitives to achieve both security and efficiency. Our system employs CRYSTALS-Kyber, a NIST-standardized lattice-based key encapsulation mechanism, for quantum-safe key exchange, coupled with AES-256-GCM for efficient authenticated symmetric encryption and SHA-256 for deterministic key derivation. The architecture follows a zero-trust model where a relay server facilitates communication without accessing plaintext messages or cryptographic keys. All encryption and decryption operations occur exclusively at client endpoints. The system demonstrates that NIST standardized post-quantum cryptography can be effectively integrated into practical messaging systems with acceptable performance characteristics, offering protection against both classical and quantum adversaries. As our focus is on implementation rather than on novelty, we also provide an open-source implementation to facilitate reproducibility and further research in post quantum secure communication systems.

Recent provably secure linguistic steganography (PSLS) methods rely on mainstream autoregressive language models (ARMs) to address historically challenging tasks, that is, to disguise covert communication as “innocuous” natural language communication. However, due to the characteristic of sequential generation of ARMs, the stegotext generated by ARM-based PSLS methods will produce serious error propagation once it changes, making existing methods unavailable under an active tampering attack. To address this, we propose a robust, provably secure linguistic steganography with diffusion language models (DLMs). Unlike ARMs, DLMs can generate text in a partially parallel manner, allowing us to find robust positions for steganographic embedding that can be combined with error-correcting codes. Furthermore, we introduce error correction strategies, including pseudo-random error correction and neighborhood search correction, during steganographic extraction. Theoretical proof and experimental results demonstrate that our method is secure and robust. It can resist token ambiguity in stegotext segmentation and, to some extent, withstand token-level attacks of insertion, deletion, and substitution.

Federated Rank Learning (FRL) is a promising Federated Learning (FL) paradigm designed to be resilient against model poisoning attacks due to its discrete, ranking-based update mechanism. Unlike traditional FL methods that rely on model updates, FRL leverages discrete rankings as a communication parameter between clients and the server. This approach significantly reduces communication costs and limits an adversary's ability to scale or optimize malicious updates in the continuous space, thereby enhancing its robustness. This makes FRL particularly appealing for applications where system security and data privacy are crucial, such as web-based auction and bidding platforms. While FRL substantially reduces the attack surface, we demonstrate that it remains vulnerable to a new class of local model poisoning attack, i.e., fine-grained control attacks. We introduce the Edge Control Attack (ECA), the first fine-grained control attack tailored to ranking-based FL frameworks. Unlike conventional denial-of-service (DoS) attacks that cause conspicuous disruptions, ECA enables an adversary to precisely degrade a competitor's accuracy to any target level while maintaining a normal-looking convergence trajectory, thereby avoiding detection. ECA operates in two stages: (i) identifying and manipulating Ascending and Descending Edges to align the global model with the target model, and (ii) widening the selection boundary gap to stabilize the global model at the target accuracy. Extensive experiments across seven benchmark datasets and nine Byzantine-robust aggregation rules (AGRs) show that ECA achieves fine-grained accuracy control with an average error of only 0.224%, outperforming the baseline by up to 17x. Our findings highlight the need for stronger defenses against advanced poisoning attacks. Our code is available at: https://github.com/Chenzh0205/ECA

Agentic Large Language Models (LLMs) are models able to reason, plan, and execute tools over unstructured data. These abilities enable transformative applications in domains spanning personal assistants, financial, and legal domains, etc. While these systems can substantially improve productivity and service quality, effective agency typically requires access to sensitive personal or organizational information. However, this access introduces critical inference-time privacy risks, specifically regarding contextually appropriate information disclosure. While recent studies highlight the inability of agentic LLMs to consistently adhere to privacy norms, existing defenses often rely on auxiliary LLM-based monitors. However, these defenses are expensive and offer limited protection against attacks that are robust to semantic censorship. To contrast this background, this paper proposes a notion of privacy filters based on activation probing. We show that these filters are both computationally efficient and effective for both single-turn and multi-turn conversational settings. Furthermore, this work provides the first systematic investigation into probing model internals across a conversation trajectory, moving beyond static, single-prompt analysis to capture the evolving state of privacy-sensitive interactions, illustrating their applicability across contexts, attack styles, multiple base models, and challenging settings such as mosaic attacks, while providing nuanced insights into the nature of the discovered concepts and modular construction of probes.

Cybersecurity superintelligence – artificial intelligence exceeding the best human capability in both speed and strategic reasoning – represents the next frontier in security. This paper documents the emergence of such capability through three major contributions that have pioneered the field of AI Security. First, PentestGPT (2023) established LLM-guided penetration testing, achieving 228.6% improvement over baseline models through an architecture that externalizes security expertise into natural language guidance. Second, Cybersecurity AI (CAI, 2025) demonstrated automated expert-level performance, operating 3,600x faster than humans while reducing costs 156-fold, validated through #1 rankings at international competitions including the $50,000 Neurogrid CTF prize. Third, Generative Cut-the-Rope (G-CTR, 2026) introduces a neurosymbolic architecture embedding game-theoretic reasoning into LLM-based agents: symbolic equilibrium computation augments neural inference, doubling success rates while reducing behavioral variance 5.2x and achieving 2:1 advantage over non-strategic AI in Attack & Defense scenarios. Together, these advances establish a clear progression from AI-guided humans to human-guided game-theoretic cybersecurity superintelligence.

With the spread of generative AI in recent years, attacks known as Whaling have become a serious threat. Whaling is a form of social engineering that targets important high-authority individuals within organizations and uses sophisticated fraudulent emails. In the context of Japanese universities, faculty members frequently hold positions that combine research leadership with authority within institutional workflows. This structural characteristic leads to the wide public disclosure of high-value information such as publications, grants, and detailed researcher profiles. Such extensive information exposure enables the construction of highly precise target profiles using generative AI. This raises concerns that Whaling attacks based on high-precision profiling by generative AI will become prevalent. In this study, we propose a Whaling countermeasure framework for university faculty members that constructs personalized defense profiles and uses large language model (LLM)-based agents. We design agents that (i) build vulnerability profiles for each target from publicly available information on faculty members, (ii) identify potential risk scenarios relevant to Whaling defense based on those profiles, (iii) construct defense profiles corresponding to the vulnerabilities and anticipated risks, and (iv) analyze Whaling emails using the defense profiles. Furthermore, we conduct a preliminary risk-assessment experiment. The results indicate that the proposed method can produce judgments accompanied by explanations of response policies that are consistent with the work context of faculty members who are Whaling targets. The findings also highlight practical challenges and considerations for future operational deployment and systematic evaluation.

Cloud environments face frequent DDoS threats due to centralized resources and broad attack surfaces. Modern cloud-native DDoS attacks further evolve rapidly and often blend multi-vector strategies, creating an operational dilemma: defenders need wire-speed monitoring while also requiring explainable, auditable attribution for response. Existing rule-based and supervised-learning approaches typically output black-box scores or labels, provide limited evidence chains, and generalize poorly to unseen attack variants; meanwhile, high-quality labeled data is often difficult to obtain in cloud settings. We present Holmes (DDoS Detective), an LLM-based DDoS detection agent that reframes the model as a virtual SRE investigator rather than an end-to-end classifier. Holmes couples a funnel-like hierarchical workflow (counters/sFlow for continuous sensing and triage; PCAP evidence collection triggered only on anomaly windows) with an Evidence Pack abstraction that converts binary packets into compact, reproducible, high-signal structured evidence. On top of this evidence interface, Holmes enforces a structure-first investigation protocol and strict JSON/quotation constraints to produce machine-consumable reports with auditable evidence anchors. We evaluate Holmes on CICDDoS2019 reflection/amplification attacks and script-triggered flooding scenarios. Results show that Holmes produces attribution decisions grounded in salient evidence anchors across diverse attack families, and when errors occur, its audit logs make the failure source easy to localize, demonstrating the practicality of an LLM agent for cost-controlled and traceable DDoS investigation in cloud operations.

We study the optimal design of additive mechanisms for vector-valued queries under $\epsilon$-differential privacy (DP). Given only the sensitivity of a query and a norm-monotone cost function measuring utility loss, we ask which noise distribution minimizes expected cost among all additive $\epsilon$-DP mechanisms. Using convex rearrangement theory, we show that this infinite-dimensional optimization problem admits a reduction to a one-dimensional compact and convex family of radially symmetric distributions whose extreme points are the staircase distributions. As a consequence, we prove that for any dimension, any norm, and any norm-monotone cost function, there exists an $\epsilon$-DP staircase mechanism that is optimal among all additive mechanisms. This result resolves a conjecture of Geng, Kairouz, Oh, and Viswanath, and provides a geometric explanation for the emergence of staircase mechanisms as extremal solutions in differential privacy.

Infrastructure as Code (IaC) enables automated provisioning of large-scale cloud and on-premise environments, reducing the need for repetitive manual setup. However, this automation is a double-edged sword: a single misconfiguration in IaC scripts can propagate widely, leading to severe system downtime and security risks. Prior studies have shown that IaC scripts often contain security smells–bad coding patterns that may introduce vulnerabilities–and have proposed static analyzers based on symbolic rules to detect them. Yet, our preliminary analysis reveals that rule-based detection alone tends to over-approximate, producing excessive false positives and increasing the burden of manual inspection. In this paper, we present IntelliSA, an intelligent static analyzer for IaC security smell detection that integrates symbolic rules with neural inference. IntelliSA applies symbolic rules to over-approximate potential smells for broad coverage, then employs neural inference to filter false positives. While an LLM can effectively perform this filtering, reliance on LLM APIs introduces high cost and latency, raises data governance concerns, and limits reproducibility and offline deployment. To address the challenges, we adopt a knowledge distillation approach: an LLM teacher generates pseudo-labels to train a compact student model–over 500x smaller–that learns from the teacher's knowledge and efficiently classifies false positives. We evaluate IntelliSA against two static analyzers and three LLM baselines (Claude-4, Grok-4, and GPT-5) using a human-labeled dataset including 241 security smells across 11,814 lines of real-world IaC code. Experimental results show that IntelliSA achieves the highest F1 score (83%), outperforming baselines by 7-42%. Moreover, IntelliSA demonstrates the best cost-effectiveness, detecting 60% of security smells while inspecting less than 2% of the codebase.

Robust access control is a cornerstone of secure software, systems, and networks. An access control mechanism is as effective as the policy it enforces. However, authoring effective policies that satisfy desired properties such as the principle of least privilege is a challenging task even for experienced administrators, as evidenced by many real instances of policy misconfiguration. In this paper, we set out to address this pain point by proposing Restricter, which automatically tightens each (permit) policy rule of a policy with respect to an access log, which captures some already exercised access requests and their corresponding access decisions (i.e., allow or deny). Restricter achieves policy tightening by reducing the number of access requests permitted by a policy rule without sacrificing the functionality of the underlying system it is regulating. We implement Restricter for Amazon's Cedar policy language and demonstrate its effectiveness through two realistic case studies.