Stellenbosch University Rejects a Blanket AI Rule — Lecturers Set AI Conditions Module by Module, Tied to Learning Outcomes

Stellenbosch University (South Africa) has a Senate-approved position statement on the ethical use of AI that took effect on 1 January 2025: students and staff remain accountable for the integrity and accuracy of AI-assisted work, and generative-AI tools cannot be credited as authors of scholarly work. In 2025 the university's guidance on AI in assessment, announced by Deputy Vice-Chancellor Liezl van Dyk, rejected a single institution-wide rule.
Instead, decisions sit at module and assessment level. Academics may prohibit, restrict, allow, encourage or require AI depending on the intended learning outcomes, judging whether AI would support or undermine them; a programming task, laboratory report, literature review, design project or professional placement may each need a different approach. Lecturers must explain the educational reason for the AI conditions they set, not merely state them. Students stay accountable for accuracy and originality, must disclose substantive AI contributions (basic editing may not require it), and postgraduate students must discuss AI plans with supervisors before submission. Where an AI platform is mandatory, an alternative must be available. The university warns against relying solely on AI-detection systems when investigating suspected misuse.
Evidence caveats: the approach is a policy design, documented in the university's published statements and trade press coverage; no outcome data (integrity-case numbers, student learning, staff workload) has been published. It leaves a heavy judgement burden on individual lecturers, and there is no mandatory university-wide AI-literacy certification for staff, so consistency depends on local support.

Read the full analysis: https://files.su.ac.za/public/2025-10/Ethical_Use_AI_Eng.pdf