HUMAN - AI COLLABORATION IN SECURE SYSTEM DEVELOPMENT: DECISION PROCESSES AND ORGANIZATIONAL IMPLICATIONS

Abstract

Artificial intelligence is increasingly embedded in secure software development, yet its impact on security outcomes depends less on technical capability than on the socio-technical integration of its outputs. This paper examines human - AI collaboration, focusing on the critical interface between automated decision-support tools and team-level decision-making. Adopting an interdisciplinary perspective, the study analyzes how AI-generated insights inform threat modeling, risk assessment, and security architecture design. The paper proposes the Human–AI Decision Framework (HADF) as an analytical framework for organizing the interaction between AI outputs, human expertise, team deliberation, and organizational context. It identifies key organizational moderators - including trust, explainability, and coordination - that shape whether AI enhances security or introduces new vulnerabilities through misalignment risks. In doing so, the study bridges a gap between technical AI capability and the social processes through which security decisions are actually formed. By positioning AI as a supporting actor within a dynamic organizational system, the study provides a grounded foundation for future empirical research on AI-driven security governance. The framework also offers a structured basis for operationalizing human - AI collaboration in security-critical environments, enabling more consistent evaluation and refinement of decision practices.