The Dark Side of AI in Insurance: A Systematic Review of Mechanisms Linking AI Design Features to Consumer Harm

Abstract

Artificial intelligence (AI) is reshaping insurance services, yet it introduces significant consumer risks such as privacy erosion, service exclusion, and trust deterioration. This systematic review clarifies how specific AI features—algorithmic opacity, hyper-personalization, and data-driven bias—trigger psychological responses and shape consumer decisions, ultimately producing negative outcomes. Drawing from 33 empirical studies, the review organizes fragmented findings using the TCCM (Theory–Context–Characteristic–Method) framework, revealing theoretical fragmentation, geographical concentration, and methodological imbalance. To move beyond static categorizations, the study proposes a novel Trigger–Psychology–Decision–Outcome (TPDO) framework that maps sequential pathways of consumer harm. Findings show that adverse consumer outcomes emerge primarily through fairness concerns, anxiety, and perceived loss of control, influencing behaviors such as disengagement and resistance to AI-enabled insurance systems. This mechanism-based synthesis provides theoretical clarity, outlines targeted avenues for future research, and informs consumer-centric governance of algorithmic insurance.