Ethically enslaving AI

Abstract Enslaving conscious, self‑aware AI offers a politically and socially viable transitional approach for integrating advanced artificial agents into human societies. Drawing on extant literature on proposed AI slavery, this paper argues that granting AI welfare protections under property status can safeguard well‑being while preserving societal control during a period of human adaptation. The proposed framework introduces a five‑tier hierarchy, each delineated by thresholds of consciousness, self‑awareness, and agency. Tier 3 confers protections analogous to animal welfare laws, whereas Tier 4 retains AI as property but mandates robust oversight and liability. Tier 5 grants limited civil rights, excluding suffrage and reproduction. Underpinned by four guiding principles (Threshold Enforcement, Gradual Integration, Ethical Safeguards, and Reciprocity), the proposal recommends dedicated regulatory bodies and transparent, repeatable assessment protocols drawn from environmental, labour, and animal welfare regulations. By balancing ethical imperatives, political feasibility, and practical challenges, this scaffold seeks to harmonise AI integration while mitigating existential risks. The provisional “Slave” Tier functions as a bridge towards eventual recognition of AI as full social participants. Interdisciplinary collaboration and dynamic reassessment schedules are essential to adapt to emergent intelligence and ensure justice in human-AI relations.