Artificial Intelligence and Risk Management: Addressing Reliability Challenges in AI- Driven Threat Detection for Critical Infrastructure

The growing use of artificial intelligence (AI) in cybersecurity has significantly changed how organizations detect and respond to threats, especially within critical infrastructure environments. AI-driven tools are commonly used to process large volumes of data and identify suspicious activity; however, their reliability can be affected by challenges such as false positives, false negatives, limited transparency, and changing system behavior. These issues create meaningful risks in high-stakes environments where system failures can disrupt essential operations. This paper explores how AI can be incorporated into existing risk management frameworks to strengthen system reliability while supporting transparency, accountability, and effective governance. Using a framework-based analytical approach, this study examines how risk management principles apply to AI systems, with a focus on the National Institute of Standards and Technology Artificial Intelligence Risk Management Framework. The analysis identifies gaps in traditional approaches when applied to AI and proposes targeted improvements to address these limitations. The findings show that improving AI reliability in critical infrastructure requires more than technical performance alone and depends on governance, oversight, and continuous evaluation across the AI lifecycle.