South African universities are moving fast to manage artificial intelligence on campus without turning learning into a policing exercise. The focus is on protecting academic integrity while still helping students build practical skills for a world where AI tools are everywhere.

AI Forces a Rethink on Integrity

With tools like ChatGPT able to generate essays or solve quizzes in seconds, universities are being pushed to draw clearer lines on what counts as acceptable support and what crosses into dishonesty. Instead of blanket bans, many institutions are leaning towards a human-centered approach that emphasizes responsible use and transparency.

NWU Formalizes its AI Policy

North-West University (NWU) says it has become the first South African institution to move from interim AI guidelines to an official AI Framework Policy. Professor Anné Verhoef, director of the NWU Artificial Intelligence Hub, says the policy is built on three pillars: human-centered, ethical and responsible AI use across teaching, learning, research and management.

NWU allows certain AI editing tools, but students must declare their use. Verhoef distinguishes between editing support and ghostwriting. If AI produces the full work and it is presented as the student’s own, that is unethical. Even when AI is used for editing, the student remains responsible for the final submission.

Training Students and Staff for “AI Literacy”

NWU has introduced a free online course for students, AI for Academic and Career Success. It aims to help students understand how AI learns from data and how that can embed cultural bias. For lecturers, NWU developed an AI course focused on assessment design and on deciding when AI should be prohibited or incorporated. Staff also have access to additional international AI training through the Digital Education Council.

Stellenbosch and UKZN Lean Into Trust Over Surveillance

Stellenbosch University says it does not rely on a single fixed AI policy. Instead, it uses an ethical position statement guided by authenticity, fairness, accountability and transparency. The university has discontinued an AI text detection tool linked to plagiarism checking because of concerns about reliability and its usefulness in high-stakes assessments. Postgraduate students also use an AI declaration process designed to support ongoing discussions with supervisors.

At the University of KwaZulu-Natal (UKZN), academics say their approach promotes education rather than surveillance. Their principles include “innovation, not intimidation” and “trust, not control”, with guidelines aimed at helping students use AI responsibly without undermining learning.