Governing AI in operations: ethical risks, accountability mechanisms, and responsible digital operations

Artificial intelligence (AI) is becoming increasingly embedded in operational decision-making, including forecasting, scheduling, procurement, pricing, and clinical operations. While these applications can improve efficiency and responsiveness, they also raise important ethical, governance, and regulatory concerns. This study conducts a systematic literature review of 57 peer-reviewed articles to examine how responsible AI is conceptualised and governed in operational contexts. The review finds evidence that responsibility in operational AI cannot be understood solely in technical terms such as accuracy or robustness. Instead, it emerges from the interaction among algorithmic design, organisational governance, stakeholder impacts, and the regulatory environment. The findings reveal function-specific patterns of ethical risk, uneven development of governance mechanisms across the AI lifecycle, and significant gaps in post-deployment accountability. In response, the study develops a risk-by-function taxonomy of ethical harms, provides an accountability-mechanism effectiveness map, and proposes a conceptual framework for responsible digital operations.