Scholarlink

<p>The escalating sophistication of cyber threats has exposed the limitations of traditional AI systems, which remain largely reactive, opaque, and devoid of ethical reasoning. This study introduces a conceptual framework for Agentic Artificial Intelligence, designed to enable autonomous, ethically aligned, and resilient cybersecurity defense. Grounded in agent theory, multi-agent reinforcement learning, machine ethics, and international AI governance principles, the framework integrates four interdependent phases: Design, Simulate, Evaluate, and Govern. An overarching Ethical Governance Layer ensures transparency, auditability, and alignment with regulatory and societal norms. Conceptual validation is achieved through architectural modeling and simulation-based evaluation metrics that assess resilience, explainability, and normative compliance. An illustrative use case in a smart healthcare cybersecurity environment demonstrates the framework’s capacity to make ethically constrained decisions under uncertainty. By unifying autonomy, ethical reasoning, and policy-aware governance, this work offers a foundational pathway for developing trustworthy and accountable AI systems in cybersecurity and other critical domains.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Designing ethically aligned agentic AI systems for cybersecurity: a conceptual framework

  • 期刊论文

  • 发表日期：

    2026年2月18日

查看全文
  • Adabara Ibrahim,
  • Bashir Olaniyi Sadiq,
  • Aliyu Nuhu Shuaibu,
  • Yale Ibrahim Danjuma,
  • Venkateswarlu Maninti,
  • Mutebi Joe

摘要

The escalating sophistication of cyber threats has exposed the limitations of traditional AI systems, which remain largely reactive, opaque, and devoid of ethical reasoning. This study introduces a conceptual framework for Agentic Artificial Intelligence, designed to enable autonomous, ethically aligned, and resilient cybersecurity defense. Grounded in agent theory, multi-agent reinforcement learning, machine ethics, and international AI governance principles, the framework integrates four interdependent phases: Design, Simulate, Evaluate, and Govern. An overarching Ethical Governance Layer ensures transparency, auditability, and alignment with regulatory and societal norms. Conceptual validation is achieved through architectural modeling and simulation-based evaluation metrics that assess resilience, explainability, and normative compliance. An illustrative use case in a smart healthcare cybersecurity environment demonstrates the framework’s capacity to make ethically constrained decisions under uncertainty. By unifying autonomy, ethical reasoning, and policy-aware governance, this work offers a foundational pathway for developing trustworthy and accountable AI systems in cybersecurity and other critical domains.