A Lifecycle-Integrated Moral Architecture for Responsible AI Design: The Case of Fitrah as Moral Orientation

Responsible AI research has advanced principle-based frameworks emphasizing fairness, accountability, and transparency. Yet persistent implementation gaps suggest that ethical commitments are frequently articulated at the level of governance rather than embedded within system architecture. To fill this gap, this study develops a lifecycle-integrated moral architecture for responsible AI design through a PRISMA-guided systematic review and grounded qualitative synthesis of 56 studies. Drawing on Fitrah, an Islamic moral concept describing an innate orientation toward purposiveness, accountability, and balance, the analysis identifies three structural discontinuities. These are moral ontology without architectural anchoring, construct-principle translation without design-stage encoding, and governance oversight without lifecycle reflexivity. In response, the paper proposes a three-layer architecture integrating moral ontology anchoring, construct-level technical encoding, and governance-driven recalibration across AI development stages. By embedding moral orientation upstream in design and sustaining alignment through lifecycle reflexivity, the study advances responsible AI from principle articulation toward structurally integrated moral architecture.