<p>Recent years have witnessed transformative advancements in generative artificial intelligence, with profound implications for higher education. As universities increasingly integrate AI into teaching, learning, and research, there is an urgent need to craft policies that confront ethical challenges and promote its responsible use. This paper takes a first step towards developing a foundation for policy development by evaluating the current landscape of generative artificial intelligence policies across South African universities. Specifically, it examines how South African universities are beginning to govern the ethical use of generative AI in teaching and learning, by analysing the extent to which institutional guidelines embed core ethical principles. Drawing on four international AI-in-education frameworks, five ethical principles: transparency, accountability and responsibility, privacy, human agency and oversight, and inclusiveness and diversity were synthesised and used as a deductive coding frame for a qualitative document analysis of publicly available generative AI guidelines issued by South African universities, collected during February 2025. Only 46% of universities had institution-wide generative AI guidelines at the time of analysis, and ethical principles were unevenly operationalised. Transparency and student-centred accountability were strongly foregrounded through detailed expectations for disclosure and responsible use, whereas privacy, institutional data governance, staff accountability, and inclusiveness were addressed more briefly and often framed as advisory rather than enforceable obligations. The paper contributes to evolving discourse on higher education practice in the current era. The findings would enable actionable implications, providing a policy guideline for higher education, and the paper provides a national baseline of policy maturity regarding AI in education and a call for harmonised, privacy-by-design guidelines.</p>