Onyinye Agatha Obioha-Val

This study evaluated the adversarial robustness of machine learning-based fraud detection systems by comparing classifier vulnerability profiles and assessing adversarial training as a mitigation strategy. Using the IEEE-CIS Fraud Detection dataset, comprising 590,540 transactions with a fraud incidence of 3.5%, four classifiers—logistic regression, random forest, gradient boosting, and a feed-forward neural network—were trained under identical preprocessing and class-weighting conditions and then subjected to Fast Gradient Sign Method and Projected Gradient Descent attacks at a perturbation budget of 0.02. Adversarial examples were constructed directly using closed-form and backpropagated gradients for the differentiable classifiers and using a logistic regression surrogate for the non-differentiable ensembles, before adversarial training was applied as a post-attack mitigation stage. Logistic regression proved the most adversarially vulnerable architecture, sustaining a 31.12-percentage-point recall loss under Projected Gradient Descent, while adversarial training subsequently restored its recall from 0.39 to 0.999 at an accuracy cost of 0.10 percentage points. Random forest and gradient boosting were not degraded by the surrogate-based attack, indicating that comparative robustness claims for tree-based ensembles require attack methods suited to their non-differentiable structure rather than transfer-based evaluation alone. Within the scope of this single-dataset evaluation, the findings support the adoption of adversarial training for gradient-based fraud detection models and suggest that robustness claims should be accompanied by disclosure of the attack methodology used to establish them.

This study examines how autonomous artificial intelligence agents inherit and exercise digital permissions within enterprise and cloud computing environments and evaluates the security implications of applying traditional access control frameworks to agent-driven systems. A quantitative analytical design was employed using three publicly available datasets: the Google Cloud Identity and Access Management policy dataset, the MITRE ATT&CK Enterprise dataset, and the AWS IAM Access Advisor dataset. Network centrality analysis, K-means clustering, and permission utilization ratio modeling were applied to examine permission inheritance structures, identify dominant cybersecurity risks, and assess the efficiency of RBAC and ABAC authorization models. The analysis indicates that autonomous agents possess substantially higher average permissions (63.7) than human users (11.8), while agent identities exhibit the highest network centrality (0.66). Additionally, credential propagation attacks produced the highest impact with an average data exposure of 68 GB. The findings support the need for agent-specific identity governance, dynamic permission scoping, and enhanced credential monitoring mechanisms.

This research presents a behavioral biometrics-powered continuous authentication framework designed for zero-trust remote work environments in healthcare. The study integrated keystroke dynamics, mouse movement patterns, and contextual risk factors into a multimodal system enabling seamless, real-time identity verification. Using a fused dataset of 3,600 samples from 24 users derived from public keystroke and mouse-dynamics repositories and augmented with healthcare-specific synthetic data where 79 behavioral features were engineered and normalized. Six machine learning and deep learning models were trained, including Random Forest, XGBoost, Support Vector Machine (SVM), LSTM, CNN-LSTM, and CNN, with Random Forest and XGBoost achieving the best performance at 98.25% accuracy, 0% Equal Error Rate (EER), and an AUC-ROC of 1.0000. The framework operated frictionlessly, with inference times below 2.6 ms, ensuring zero disruption to clinical workflows. Dynamic trust scoring enabled adaptive access control, while attack simulations across six threat scenarios yielded a 90.3% detection rate, including 100% for insider threats and zero-effort impersonation. Full compliance with HIPAA standards was validated through continuous monitoring, audit logging, and real-time threat response. The system outperformed traditional authentication methods in accuracy, usability, and security resilience. Despite strong results, limitations include constrained user diversity and simulated environments. The framework advances zero-trust principles by providing passive, high-precision authentication suitable for distributed healthcare systems. Future work should focus on longitudinal field deployment and adaptive modeling to address behavioral drift and emerging threats.

This study investigates the effectiveness of collaborative digital solutions in addressing cybersecurity governance gaps within the educational sector, focusing on resource constraints, awareness deficits, and regulatory compliance challenges. Data from the National Institute of Standards and Technology Cybersecurity Framework Usage Dataset, the Global Threat Intelligence Sharing Alliance Data Repository, and the World Values Survey were utilized. Quantitative methodologies, including descriptive analysis, Difference-in-Differences, and logistic regression, were employed to analyze gaps, evaluate solution effectiveness, and explore barriers. Findings revealed significant governance gaps, with resource constraints showing a mean frequency of 140.3, the highest among categories. Collaborative solutions demonstrated a 49.3% reduction in breach incidents post-intervention. Logistic regression identified awareness as a major barrier with an odds ratio of 2.46. Recommendations include prioritizing cybersecurity awareness programs, enhancing access to collaborative solutions, standardizing data-sharing protocols, and investing in capacity-building initiatives to fortify institutional resilience.

This study investigates the regulatory and ethical dimensions of Open Source Intelligence (OSINT) and deepfake technologies, analyzing their impact on public trust, privacy, and societal stability. Using data from the Global Dataset of Events, Location, and Tone (GDELT), sentiment analysis and time-series regression identified a significant decline in public sentiment (β = -0.23, p = 0.01) and societal stability due to deepfake incidents and OSINT misuse. The Deepfake Detection Challenge Dataset (DFDC) was analyzed using machine learning models, with neural networks achieving the highest accuracy (92%) and precision (91%). Regulatory frameworks were evaluated using the OECD database, where enforcement capacity demonstrated the strongest impact on reducing misuse cases (β = -0.42, p = 0.002). Recommendations include the establishment of globally coordinated regulatory frameworks, public awareness campaigns, investment in advanced detection systems, and ethical integration of AI into OSINT practices.

This study investigates the integration of Artificial Intelligence (AI) and Open Source Intelligence (OSINT) to enhance predictive threat modeling in cybersecurity, addressing the growing complexity and frequency of cyber threats. Integrating AI and OSINT offers transformative potential by enabling organizations to transition from reactive to proactive security measures, a critical need in the evolving digital landscape. Leveraging data from the Twitter Academic API, Common Crawl Dataset, and MITRE ATT&CK Framework, the analysis employed descriptive statistical analysis, logistic regression, and multivariate regression methodologies. Results indicate high data completeness (90.41%) and relevance (81.44%) in OSINT datasets, supporting their suitability for AI model training. Logistic regression demonstrated strong predictive capabilities, achieving 94.98% accuracy, 88.69% precision, and an AUC score of 0.91. However, risks such as data bias (-0.36 coefficient) and adversarial manipulation (-0.33 coefficient) significantly impact predictive performance. The ethical implications of this integration, including concerns about privacy, data fairness, and the potential for misuse, are highlighted as critical considerations for broader adoption. Recommendations include robust preprocessing protocols, advanced adversarial defenses, ethical guidelines, and continuous AI innovation to address these challenges. These findings underscore the potential of AI-OSINT integration while emphasizing the need for ethical and technical safeguards to enhance cybersecurity effectiveness.

This study investigates the transformative role of artificial intelligence (AI) in state-sponsored cyber espionage, focusing on its dual use in offensive and defensive operations. Using data from the MITRE ATT&CK Framework, FireEye APT Groups Database, UNSW-NB15 Intrusion Detection Dataset, and the Cyber Conflict Tracker by CFR, this research applied network graph analysis, multi-criteria decision analysis (MCDA), ensemble classification models, and Difference-in-Differences (DiD) analysis. Results revealed that AI-driven offensive techniques, phishing (degree centrality 0.85), and adaptive malware (betweenness centrality 0.81) significantly enhance operational precision and scalability. Defensively, ensemble classification models achieved up to 95.8% accuracy, highlighting AI's efficacy in intrusion detection. AI regulatory frameworks reduced misattribution rates by 20% and escalation incidents by 10%, demonstrating their critical role in mitigating geopolitical risks. The findings impress AI's transformative potential in advancing cyber operations and shaping international policy and governance. By addressing challenges such as attribution, escalation risks, and ethical dilemmas, this study highlights the necessity for stronger global cooperation and regulatory frameworks to navigate the dual-use nature of AI, providing actionable insights for policymakers, cybersecurity professionals, and researchers, emphasizing the urgency of aligning technological advancements with strategies for enhancing global cybersecurity resilience.

This study explores the dual implications of Artificial Intelligence (AI)-driven Open Source Intelligence (OSINT) in enhancing cyber defense capabilities. Using publicly available datasets, including IBM X-Force breach metrics, MITRE ATT&CK adversarial tactics, GDPR privacy violations, AI-driven phishing incidents, and case-specific data from the Colonial Pipeline ransomware attack and Russia-Ukraine conflict, the research employs multivariate regression, logistic regression, and K-Means clustering. The findings indicate that AI investments improve detection time (-0.68), accuracy (+2.09), and resolution rates (+1.55) with statistical significance (p < 0.001). However, risks associated with algorithmic opacity, weak regulatory frameworks, and reactive AI systems pose ethical and operational challenges. Clustering reveals variability in AI applications, with optimized systems achieving 95.2% detection rates and 5.5-hour response times. Recommendations include investing in scalable tools, strengthening regulations, fostering public-private collaborations, and enhancing reactive AI oversight. The results highlight AI’s transformative potential in cyber defense while emphasizing the need for ethical and regulatory alignment. Future directions include testing these models in diverse operational environments to validate effectiveness and exploring hybrid AI approaches to balance proactive and reactive capabilities, ensuring robust and adaptive defense mechanisms.

This study investigates the role of Artificial Intelligence (AI) in enhancing cybersecurity for U.S. public schools, with the primary objective of evaluating AI's effectiveness in reducing cyber threats and safeguarding student privacy. Specifically, the study assesses AI-driven security systems such as threat detection and anomaly detection algorithms, which help schools monitor network traffic and identify potential breaches in real-time. Using logistic regression on data from the K-12 Cybersecurity Resource Center, findings reveal that schools implementing AI solutions are 75% less likely to experience cyber breaches (p < 0.001), highlighting AI's protective impact. Furthermore, a comparative analysis of FERPA and COPPA compliance reports highlights a substantial reduction in privacy violations among AI-using schools, with an average of 0.57 violations per school, compared to 1.50 in schools without AI. A K-means cluster analysis identified budget constraints (65.75%) and IT staff shortages (55.25%) as primary barriers to AI adoption. To address these obstacles, the study recommends phased technology upgrades and increased funding for workforce training as critical strategies to facilitate AI integration and enhance cybersecurity across educational institutions. These strategic interventions are essential for optimizing the effectiveness of AI-driven security systems, making it feasible for resource-constrained schools to adopt and maintain advanced cybersecurity measures. The study’s findings contribute to the growing body of knowledge on educational cybersecurity and provide actionable insights for policymakers and administrators seeking to strengthen data protection and privacy in school environments.

This study investigates the integration of machine learning (ML) algorithms with smart sensor technologies across manufacturing, energy, and healthcare sectors, focusing on their impact on real-time industrial monitoring, predictive maintenance, and operational efficiency. By utilizing data from the UCI Machine Learning Repository and Kaggle, this research measures the effectiveness of ML-enabled sensors in reducing machine downtime and enhancing fault detection. Time series analysis and regression modeling reveal that sensor integration leads to a significant 5.5% improvement in machine uptime, raising average uptime from 91.5% to 97%, thus validating the role of predictive maintenance. Cost-benefit analysis further highlights that the energy sector achieves the highest financial returns, with a 33.3% ROI and a positive Net Present Value (NPV) over five years, demonstrating substantial cost savings relative to initial investment. Findings underscore the importance of sensor infrastructure compatibility, emphasizing the need for adaptable frameworks such as edge computing and digital twin technology to ensure efficient integration with legacy systems. Recommendations include industry-wide adoption strategies that leverage these technologies to optimize predictive maintenance and maximize sector-specific financial returns.