AI Use Policies at South African Universities: What UCT, Wits and Stellenbosch Require
By Muntasir Minhaz • Published Aug 07, 2026 • Education News & Policy
UCT, Wits, and Stellenbosch all allow AI for coursework support but require disclosure once AI produces substantive content, and none of the three lets AI be listed as an author. Invigilated exams stay untouched because AI cannot reach you in a proctored room.
🎓 UCT's AI in Education Framework took effect in June 2025
📩 Wits requires disclosure and AI literacy across six policy principles
✅ Stellenbosch's Senate position statement took effect January 1, 2025
⏱️ Basic grammar or language help generally needs no disclosure at any of the three
None of the three universities lets AI be credited as an author of your work
Three Universities, Three Documents, One Shared Principle
South Africa's leading universities published their generative AI policies within months of each other, and all three land on the same core rule: use AI to support your learning, disclose it when it shapes your output, and take full responsibility for what you submit.
| University | Policy | Effective | Disclosure rule |
|---|---|---|---|
| UCT | AI in Education Framework | June 2025 | Required for AI-generated submissions |
| Wits | AI Policy (six principles) | Aligned to 2025-2029 strategy | Integrity with disclosure, always |
| Stellenbosch | Ethical Use of AI position statement | January 1, 2025 | Required once AI use is substantive |
University of Cape Town: AI in Education Framework
UCT's AI in Education Framework , published in June 2025, takes a supportive stance rather than a restrictive one, prioritizing ethical integration of AI into learning over blanket bans. You are expected to make sure your submitted work reflects your own effort, even when you used AI tools to prepare it. The framework acknowledges that undeclared AI-generated submissions have already come up as a problem case, while noting that closed, invigilated exams remain largely unaffected because standard exam proctoring keeps AI tools out of the room.
University of the Witwatersrand: AI Policy and Six Principles
Wits University's AI policy , aligned with its 2025-2029 Learning and Teaching Strategic Plan, rests on six principles: integrity with disclosure, AI literacy, adaptive assessment practices, human oversight, risk management, and equity of access. The policy is direct about responsibility: using an AI tool does not remove your accountability for misconduct such as plagiarism, falsification, or improper attribution. Wits reports no formal misconduct incidents tied to AI so far, which it credits to leading with ethical guidance and AI literacy training rather than relying only on after-the-fact detection.
Stellenbosch University: Position Statement on Ethical AI Use
Stellenbosch's Senate-approved position statement on the ethical use of AI took effect on January 1, 2025, and holds both students and staff accountable for the integrity and accuracy of anything they produce with AI assistance. The statement is explicit that generative AI tools cannot be credited as authors of academic or scholarly work. It draws a practical line between limited technical assistance, such as basic language editing or coding support, which may not need disclosure unless a lecturer asks for it, and generative AI that produces substantive material such as text, images, music, or code, which you must always declare. Separately, reporting on Stellenbosch's approach notes the university deliberately avoided a single blanket rule for every assessment type, leaving faculties room to set stricter or looser rules depending on what a specific course tests.
What the Three Policies Have in Common
All three treat undisclosed generative AI use as academic misconduct once it produces substantive content.
None of the three allows AI to be listed as an author or co-author of your work.
All three permit AI for support tasks like language editing, brainstorming, or explaining concepts.
All three leave enforcement detail, such as exact penalties, to faculty or module-level policy rather than a single university-wide penalty table.
Where They Differ
UCT and Stellenbosch both name specific disclosure thresholds tied to how substantive the AI contribution is. Wits frames disclosure as a standing principle rather than a threshold, applying it more broadly across all AI-assisted work. Stellenbosch is the only one of the three to formally reject a single blanket assessment rule, instead pushing disclosure and assessment design decisions down to individual faculties.
What This Means for Your Coursework
Check your specific faculty or module handbook at your university, since all three institutions leave some policy detail at that level rather than setting one campus-wide rule. If a task is a closed, invigilated exam, AI policy changes little for you, since you cannot access AI tools in that setting anyway. If a task is a take-home essay, project, or coding assignment, assume disclosure applies once AI does more than fix your grammar, and ask your lecturer directly if a specific policy is unclear.