How African Universities Are Writing AI and ChatGPT Use Policies for Students

Why most universities do not have a standalone AI policy yet

Generative AI tools spread through African campuses faster than most universities could formalise a written response. Only a small number of institutions, concentrated in South Africa, had published dedicated generative AI policies rather than folding AI use into existing plagiarism and academic misconduct rules. The University of Cape Town and University of Johannesburg are tightening their response to AI-related cheating under updated academic-integrity rules, according to reporting from ITWeb , and UCT is the clearest case of a university with a standalone written AI policy (see UCT News below).

That gap does not mean AI use goes unregulated everywhere else. Most universities still apply their existing academic integrity codes to AI-assisted work, treating undisclosed AI-generated submissions as a form of misconduct alongside plagiarism, even where no AI-specific clause exists yet. If your university has not published a named AI policy, ask your faculty or academic office directly how existing rules on originality and citation apply to AI tools before you submit work you built with ChatGPT or a similar tool.

University of Cape Town's AI in Education Framework

UCT's Online Education Subcommittee began drafting its Framework for Artificial Intelligence in Education in September 2024, and the university's Senate Teaching and Learning Committee endorsed it in June 2025, according to UCT News . The framework sets out three priorities: building AI literacy among staff and students, protecting assessment integrity, and exploring where AI genuinely improves teaching and learning rather than replacing student effort.

UCT also made a practical, debated call within the same rollout: from 1 October 2025, the university stopped using Turnitin's AI Score during grading, stating that AI detection tools are widely considered to be unreliable. That decision reflects a wider concern among academic integrity offices that flagging student work as AI-generated based on a detection score alone produces too many wrong calls to justify using it for discipline. Details on the framework and related guidance are published through UCT's Centre for Innovation in Learning and Teaching .

What UNESCO recommends

UNESCO published its first global Guidance for Generative AI in Education and Research in 2023, building on its 2021 Recommendation on the Ethics of Artificial Intelligence, according to UNESCO . The guidance takes a human centred approach: it calls on education systems to set a minimum age for students to interact with generative AI tools without supervision, protect student data privacy, and train teachers before rolling AI tools into classrooms. It does not mandate a single policy text, so implementation still depends on each country's education ministry and each university's own senate or academic board.

What published policies tend to cover

Where African universities have written AI-specific rules, the common elements are similar to what UCT and its South African peers have set out. Policies generally distinguish between AI use for brainstorming, editing, or study support, which is often permitted, and AI-generated text submitted as original work without disclosure, which is treated as misconduct. Several also require students to name the specific tool used in any assignment where AI played a role, mirroring how academic publishers now require AI disclosure in research papers.

What this means for you as a student

• Check your university's student handbook and academic integrity policy for any AI-specific clause before assuming a blanket ban or blanket allowance.

• Where no AI-specific rule exists, treat undisclosed AI use in graded work as you would undisclosed help from another person, since it carries a similar misconduct risk.

• Do not rely on a low AI-detection score as proof your work is safe. Detection tools produce wrong flags in both directions, part of why UCT dropped Turnitin's AI Score for grading decisions.

• Ask your lecturer directly, in writing, when a policy is unclear for a specific assignment. A short email creates a record of what you were told.