This report examines how structural gender inequalities may become embedded in AI systems through the processes used to generate, collect, select, classify, label and reuse data within the African AI ecosystem. Drawing on examples from credit scoring, content moderation, digital identity, and public service delivery, it identifies pathways through which incomplete or inadequately contextualised data may reproduce unequal outcomes for women. The report further evaluates relevant continental governance frameworks and identifies key considerations to strengthen dataset representativeness, gender-responsive impact assessments, participatory data curation practices, and sector-specific safeguards and oversight throughout the AI lifecycle.