Artificial intelligence and disability in the workplace: A systematic review of ethical issues

Background The use of artificial intelligence (AI) in workplace disability contexts is expanding, raising important ethical concerns. Existing frameworks, such as the OECD Framework for the Classification of AI Systems, provide structure for understanding AI applications but lack an explicit focus on ethics in relation to disability. Objective This systematic review examines ethical issues arising from the use of AI for employees with disabilities and aims to interpret the OECD framework through an explicit ethics lens. Methods We conducted a systematic review of 23 peer-reviewed studies identified through PRISMA-guided searches. Using the OECD AI framework as a foundation, we developed an extended multidimensional framework consisting of five interconnected domains: (1) Context, (2) People, (3) Data & Input, (4) AI Model, and (5) Task & Output and analyzed through the lens of ethics. Results Four recurring ethical concerns were identified: (1) automation bias, (2) data privacy and informed consent, (3) discrimination and inequitable access to opportunities, and (4) AI governance gaps. Ethical risks were found to be context-dependent, disproportionately affecting disabled workers, and exacerbated by biased or incomplete data and opaque AI models. Conclusions This review provides a structured synthesis of current evidence and highlights the need for responsible and inclusive AI practices. We propose key directions for research and practice, including algorithmic audits, inclusive testing, transparent data practices, interdisciplinary oversight, and worker-centered governance. Integrating ethics throughout a multidimensional framework offers a roadmap for advancing equitable AI in workplace disability contexts.