This systematic literature mapping study aim to provide practical insights on the ethics of artificial intelligence (AI) in assessment. It is important to study the divide between what may be ethically permissible and not permissible, especially in fundamental societal institutions like education, when teaching practitioners or researchers apply AI in academic processes such as assessments. This study applied a systematic literature mapping methodology to scour extant research, so as to holistically structure the landscape into explicit topical research clusters. Through topic modelling and network analyses, research mapped key ethical principles to research archetypical domains, and reviewed the influence of these ethical principles in each thematic domain. Results of this study identified five key research archetypical themes, with presence across the system layers of cognitive, information and physical domains of an AI-based assessment pipeline, namely: (i) AI system design and check for assessment purposes; (ii) AI-based assessment construction and rollout; (iii) data stewardship and surveillance; (iv) administration of assessments using AI systems; and (v) AI-facilitated assessment grading and evaluation. Ten AI ethics principles, namely, (i) fairness, (ii) privacy, (iii) explainability, (iv) accountability, (v) accuracy, (vi) inclusivity, (vii) trust, (viii) human centricity, (ix) auditability and (x) cheating, epitomize the key ethics considerations across each of the five research themes; each manifesting varying levels of importance. The findings of this research can provide researchers and practitioners the insights into the application methods of AI in assessments and their intertwined ethical challenges, and in particular, the generalizable key research themes structured across the assessment pipeline, for follow up studies.