Selected work

Representative Papers

Evaluating Large Language Models for Symbolic Security Protocol Analysis

Jul 22, 2026

This study presents the first systematic evaluation of the feasibility of using large language models (LLMs) as substitutes for formal verification tools—such as ProVerif and OFMC—in symbolic security protocol analysis. The authors assess GPT and DeepSeek in both chat and reasoning modes on 130 obfuscated AnB/AnBx protocols, comparing their outputs against formal verification results across 388 security properties. In reasoning mode, LLMs achieve a peak F1 score of 95.7% on confidentiality tasks, yet exhibit less than 50% detection accuracy for authentication attacks. These findings highlight the potential of LLMs in confidentiality analysis while underscoring their current limitations in handling authentication properties. Although LLMs cannot yet replace formal methods, they show promise as effective pre-screening tools in protocol analysis workflows.

0 citationsRead paper

Operational AI Deployment Assurance: Governance-State Orchestration Under Threshold-Sensitive Deployment Conditions -- A Governance Framework for High-Stakes AI Systems

May 26, 2026

This work addresses the limitations of existing AI governance frameworks, which rely on static metrics and post-hoc audits and thus lack the capacity for dynamic, real-time assessment of deployment readiness in high-risk systems—particularly regarding fairness discrepancies, threshold sensitivity, and remediation progress. To bridge this gap, the paper proposes the Operational AI Deployment Assurance (OADA) framework, which uniquely models governance uncertainty as an operational challenge within the deployment pipeline. OADA introduces mechanisms such as deployment assurance scores, readiness categorization, threshold stability zones, and governance escalation states to enable closed-loop, dynamic governance from evaluation to deployment. By integrating the Fairness Discrepancy Index (FDI) and FairRisk-FDI with threshold sensitivity analysis and repair-aware assurance evolution, OADA successfully identifies models deemed “compliant” by conventional metrics yet operationally unstable, offering a scalable deployment assurance paradigm for high-stakes domains like medical AI.

0 citationsRead paper

Routing Cybersecurity Awareness Training by FFM Personality Trait: A Quasi-Experimental Evaluation

May 23, 2026

This study addresses the limited efficacy of traditional cybersecurity awareness training, which often overlooks individual differences. To bridge this gap, the authors propose and implement TailoredSec—the first mobile-based adaptive training system grounded in the Five-Factor Model (FFM) of personality. Leveraging the BFI-10 scale to assess users’ personality traits, the system dynamically routes individuals to one of four tailored training modules. The work provides the first end-to-end validation of a personality-driven content delivery mechanism in cybersecurity education. Results demonstrate that participants in the experimental group achieved significantly higher post-test scores than those in the control group (mean: 35.88 vs. 30.75), with a pass rate of 100% compared to 77.5% and a medium effect size (Cohen’s d = 0.62), confirming that personalized training substantially enhances security awareness outcomes.

0 citationsRead paper

GenAI-FDIA: Physics-Informed Generative Models for False Data Injection Attacks

May 15, 2026

This study addresses the limitations in false data injection attack (FDIA) detection research—namely, the scarcity of realistic data and the lack of physical consistency in handcrafted attacks—by proposing GenAI-FDIA, a framework that systematically evaluates the capacity of 20 generative models, including Wasserstein GANs, MMD-VAEs, normalizing flows, diffusion models, and their hybrid architectures, to produce physically consistent FDIAs on IEEE benchmark power grids. The work introduces a data-driven bad data detection (BDD) threshold calibration mechanism and makes three core contributions: identifying a novel failure mode wherein affine physical projection in normalized space nullifies attacks, proposing a training-free inference-time coordinator to restore attack stealthiness, and diagnosing and mitigating covariance collapse in hybrid architectures. Experiments demonstrate BDD evasion rates ≥86.6% on the 14-bus system; the coordinator boosts evasion from <2% to 100% on the 30-bus system; and a 50-epoch warm-up schedule markedly improves covariance alignment (κ rising from −0.076 to 0.785).

0 citationsRead paper

Strategic commitments shape collective cybersecurity under AI inequality

May 10, 2026

This study addresses the systemic security risks arising from AI capability disparities, which hinder resource-constrained defenders from deploying effective protections. To mitigate this issue, the authors propose an evolutionary game-theoretic model that integrates commitment-based defense strategies with targeted subsidy mechanisms—a novel combination designed to incentivize the widespread adoption of strong defensive behaviors. Leveraging finite-population evolutionary dynamics, social learning rules, and multi-parameter simulations, both theoretical analysis and empirical results demonstrate that the proposed mechanism reliably converges to a high-security equilibrium in heterogeneous AI environments. This approach significantly increases the adoption rate of robust defenses, substantially reduces attack success rates, and simultaneously enhances defender welfare while effectively curbing attacker payoffs.

0 citationsRead paper

Recent publications

Latest Papers

Evaluating Large Language Models for Symbolic Security Protocol Analysis

Jul 22, 2026

This study presents the first systematic evaluation of the feasibility of using large language models (LLMs) as substitutes for formal verification tools—such as ProVerif and OFMC—in symbolic security protocol analysis. The authors assess GPT and DeepSeek in both chat and reasoning modes on 130 obfuscated AnB/AnBx protocols, comparing their outputs against formal verification results across 388 security properties. In reasoning mode, LLMs achieve a peak F1 score of 95.7% on confidentiality tasks, yet exhibit less than 50% detection accuracy for authentication attacks. These findings highlight the potential of LLMs in confidentiality analysis while underscoring their current limitations in handling authentication properties. Although LLMs cannot yet replace formal methods, they show promise as effective pre-screening tools in protocol analysis workflows.

0 citationsRead paper

Operational AI Deployment Assurance: Governance-State Orchestration Under Threshold-Sensitive Deployment Conditions -- A Governance Framework for High-Stakes AI Systems

May 26, 2026

This work addresses the limitations of existing AI governance frameworks, which rely on static metrics and post-hoc audits and thus lack the capacity for dynamic, real-time assessment of deployment readiness in high-risk systems—particularly regarding fairness discrepancies, threshold sensitivity, and remediation progress. To bridge this gap, the paper proposes the Operational AI Deployment Assurance (OADA) framework, which uniquely models governance uncertainty as an operational challenge within the deployment pipeline. OADA introduces mechanisms such as deployment assurance scores, readiness categorization, threshold stability zones, and governance escalation states to enable closed-loop, dynamic governance from evaluation to deployment. By integrating the Fairness Discrepancy Index (FDI) and FairRisk-FDI with threshold sensitivity analysis and repair-aware assurance evolution, OADA successfully identifies models deemed “compliant” by conventional metrics yet operationally unstable, offering a scalable deployment assurance paradigm for high-stakes domains like medical AI.

0 citationsRead paper

Routing Cybersecurity Awareness Training by FFM Personality Trait: A Quasi-Experimental Evaluation

May 23, 2026

This study addresses the limited efficacy of traditional cybersecurity awareness training, which often overlooks individual differences. To bridge this gap, the authors propose and implement TailoredSec—the first mobile-based adaptive training system grounded in the Five-Factor Model (FFM) of personality. Leveraging the BFI-10 scale to assess users’ personality traits, the system dynamically routes individuals to one of four tailored training modules. The work provides the first end-to-end validation of a personality-driven content delivery mechanism in cybersecurity education. Results demonstrate that participants in the experimental group achieved significantly higher post-test scores than those in the control group (mean: 35.88 vs. 30.75), with a pass rate of 100% compared to 77.5% and a medium effect size (Cohen’s d = 0.62), confirming that personalized training substantially enhances security awareness outcomes.

0 citationsRead paper

GenAI-FDIA: Physics-Informed Generative Models for False Data Injection Attacks

May 15, 2026

This study addresses the limitations in false data injection attack (FDIA) detection research—namely, the scarcity of realistic data and the lack of physical consistency in handcrafted attacks—by proposing GenAI-FDIA, a framework that systematically evaluates the capacity of 20 generative models, including Wasserstein GANs, MMD-VAEs, normalizing flows, diffusion models, and their hybrid architectures, to produce physically consistent FDIAs on IEEE benchmark power grids. The work introduces a data-driven bad data detection (BDD) threshold calibration mechanism and makes three core contributions: identifying a novel failure mode wherein affine physical projection in normalized space nullifies attacks, proposing a training-free inference-time coordinator to restore attack stealthiness, and diagnosing and mitigating covariance collapse in hybrid architectures. Experiments demonstrate BDD evasion rates ≥86.6% on the 14-bus system; the coordinator boosts evasion from <2% to 100% on the 30-bus system; and a 50-epoch warm-up schedule markedly improves covariance alignment (κ rising from −0.076 to 0.785).

0 citationsRead paper

Strategic commitments shape collective cybersecurity under AI inequality

May 10, 2026

This study addresses the systemic security risks arising from AI capability disparities, which hinder resource-constrained defenders from deploying effective protections. To mitigate this issue, the authors propose an evolutionary game-theoretic model that integrates commitment-based defense strategies with targeted subsidy mechanisms—a novel combination designed to incentivize the widespread adoption of strong defensive behaviors. Leveraging finite-population evolutionary dynamics, social learning rules, and multi-parameter simulations, both theoretical analysis and empirical results demonstrate that the proposed mechanism reliably converges to a high-security equilibrium in heterogeneous AI environments. This approach significantly increases the adoption rate of robust defenses, substantially reduces attack success rates, and simultaneously enhances defender welfare while effectively curbing attacker payoffs.

0 citationsRead paper