Navigating the dark side of AI in service ecosystems: an ethical leadership framework for risk mitigation

Abstract

The rapid integration of artificial intelligence (AI) into service ecosystems is transforming value cocreation while generating significant ethical risks that threaten customer trust, organisational legitimacy, and social sustainability. Building on Kaartemo and Helkkula's (2025) framework of human-AI resource relations, this paper develops the Ethical AI Risk Mitigation (EAIRM) model to examine how different configurations of human-AI collaboration create distinct ethical challenges across fairness, autonomy, transparency, and accountability dimensions. Drawing on a structured literature synthesis, we identify four leadership approaches, compliance-oriented, values-based, stakeholder-engaged, and anticipatory, that systematically mitigate ethical risks while enabling service innovation. This paper provides service managers and scholars with actionable insights for responsible AI adoption. Through integrative theory building, the model contributes to service research and practice by: (1) revealing how identical ethical risks operate through different causal mechanisms depending on human-AI resource configuration (e.g., accountability failures in composite relations arise from emergent hybrid agency opacity, whereas in hermeneutic relations they arise from AI reasoning opacity); (2) specifying multi-actor governance structures for service ecosystems where no single actor controls ethical outcomes; (3) theorizing leadership mechanisms and organisational mediators that convert ethical principles into operational practices; and (4) generating testable propositions with boundary conditions, moderators, and feedback dynamics. This framework advances service ecosystem theory by demonstrating that resource relations have ethical risk implications requiring polycentric governance, not just value creation potential.