Article Preview
Human-Centred Artificial Intelligence in Child Protection Social Work: An African and Kenyan Framework for Safeguarding Children of Incarcerated Mothers
Children affected by parental incarceration experience a distinctive combination of family separation, stigma, disrupted caregiving, poverty, educational interruption, and fragmented access to protection services. These risks are often intensified when a mother is arrested or imprisoned and no reliable mechanism immediately identifies the children left in the community, establishes safe care, or coordinates referrals across correctional services, children’s officers, schools, health facilities and civil-society organisations. This article develops a human-centred artificial intelligence (AI) framework for child protection social work, using children of incarcerated mothers in Kenya as a critical case through which to examine opportunities and limits for African child welfare systems. It draws on a practitioner-grounded conference presentation from Justice Nest, relevant Kenyan legislation and case-management guidance, African child-rights instruments, and international scholarship on AI-supported child welfare decision-making. The paper uses a conceptual, narrative policy-analysis approach rather than reporting a systematic review or an evaluation of a deployed algorithm. It proposes four connected functions: early case flagging at prison intake; referral matching across authorised services; carefully bounded indicators of unmet needs; and mandatory caseworker review and action. The analysis argues that AI may improve the timeliness, consistency and traceability of case coordination, but it can also reproduce institutional bias, expose highly sensitive information, misclassify family circumstances and shift accountability away from practitioners. A rights-based approach must therefore place the best interests of the child, privacy, participation, non-discrimination, explainability and human decision-making at the centre of design. For Kenya, implementation should build on, not bypass, the Children Act 2022, the Data Protection Act 2019, existing child-protection case-management procedures and justice-sector commitments concerning children affected by incarceration. The article concludes with a staged research and implementation agenda, including co-design, dataprotection impact assessment, independent audit, community consultation and evaluation of child-centred outcomes. AI should function as accountable infrastructure for social work, not as a substitute for relational practice or adequately resourced child-protection services. Keywords: Artificial Intelligence; Child Protection; Social Work; Children of Incarcerated Mmothers; Data Protection; Human-Centred AI
