Mounting Concerns Over AI Cheating in South African Universities

The rapid proliferation of artificial intelligence (AI) tools like ChatGPT has ignited a firestorm in South African higher education, where universities are grappling with an unprecedented wave of academic misconduct. Generative AI—software capable of producing human-like text, code, and analyses—has made cheating easier than ever, allowing students to submit entire assignments generated by machines as their own work. This phenomenon, often termed AI cheating or AI plagiarism, threatens the very foundation of academic integrity, raising profound questions about the value of degrees from institutions like the University of South Africa (UNISA), University of Cape Town (UCT), University of the Witwatersrand (Wits), and Stellenbosch University (SU). 70 71 In February 2026, UNISA issued stark warnings of a 'student integrity crisis,' highlighting marked increases in AI-assisted submissions that have overwhelmed disciplinary processes. 48 As South Africa's largest distance-learning provider enrolls over 400,000 students, the stakes could not be higher for a sector already strained by funding shortages, inequality, and post-pandemic recovery.

This crisis is not isolated; it's part of a global shift where AI blurs the line between legitimate assistance and outright deception. In South Africa, cultural and linguistic diversity—many students are English language learners—complicates detection, exacerbating inequities in public universities. Yet, amid the alarm, institutions are pivoting toward innovative solutions, from redesigned assessments to AI literacy programs, positioning South African higher education as a potential leader in ethical AI integration.

🔍 UNISA's Urgent Warnings Signal a National Crisis

At the epicenter is UNISA, where Vice-Chancellor Professor Puleng LenkaBula confirmed a surge in AI misuse during exams and assignments. Hundreds of students have been flagged recently, contributing to plagiarism spikes and massive disciplinary backlogs—echoing a 2024 incident where over 5,000 cases were investigated. 71 UNISA's zero-tolerance policy treats unauthorized AI generation as misconduct, punishable by zero marks, suspensions, or expulsion. Legitimate uses, like brainstorming or grammar checks, require declaration, but entire AI-produced essays qualify as plagiarism.

"AI use blurs the line between help and cheating," notes AI engineer Akhil Boddu, underscoring how tools embedded in student workflows enable undetectable submissions. 70 UNISA employs Turnitin's AI detector, but up to 30% false positives—often hitting non-native speakers—have led to appeals and stress. This has prompted short course suspensions and VC scrutiny amid R1.1m bonus controversies.

Other Universities Grapple with Detection Dilemmas

Across South Africa, the pattern repeats. At the University of Pretoria (UP), Associate Professor Charne Lavery found 70-80% of 800 second-year English essays bore AI hallmarks: flawless grammar, uniform structure, synthetic tone—yet Turnitin cleared them. 72 Stellenbosch University's postgraduate survey revealed 65% knew peers cheating with large language models (LLMs). Computer science departments at UCT report foundational skill erosion as students outsource thinking to AI.

UCT made headlines in July 2025 by scrapping Turnitin's AI Score from October, citing unreliability and fairness risks—especially for diverse learners. Instead, UCT's AI in Education Framework promotes ethical use via literacy training and process-based assessments like oral vivas. 69 Wits outlines six ethical principles, SU mandates declarations, and UKZN favors education over punishment. North-West University (NWU) pioneered SA's first official AI policy in January 2026, emphasizing governance.

Challenges: Why Detecting AI Cheating Proves Elusive

AI detectors falter against evolving models; paraphrasing tools evade them, while false positives disproportionately affect black and ESL students, fueling equity debates. Large enrollments hinder manual checks, and limited licenses strain budgets. Pandemic-era online learning normalized paraphrasers, fostering skill gaps in research and critical thinking.

Cultural context matters: In inequality-plagued SA, AI access varies—privileged students thrive, others lag. Faculty workloads balloon with reviews, eroding morale. Universities SA deems AI a 'serious risk' to integrity, urging national adaptation. 61

Real-World Cases Highlight the Human Cost

• UNISA: Hundreds caught submitting AI exams; innocent flagged endure viva defenses, rewrites.
• UP: 70-80% AI-tainted essays undetected.
• UJ History: Covid bots invented 'Stronghold Bunny University' citations.
• Stellenbosch: 65% peers admit knowing cheaters.
• General: IT pros delegate sensitive tasks to ChatGPT, risking professions like medicine/law.

These cases underscore eroded trust: Employers question graduate competencies amid youth unemployment crisis.

Impacts: Eroding Trust and Future Employability

AI cheating undermines degree credibility, vital in SA's job-scarce economy. Skill deficits emerge—students bypass learning, entering fields unprepared. Faculty burnout rises; inequality widens as public unis lag private peers. Globally, degrees lose luster; locally, social mobility suffers.

For higher ed jobs, integrity lapses deter ethical employers. Check professor ethics via Rate My Professor.

Current Policies: From Zero-Tolerance to Ethical Integration

UNISA: Zero-tolerance, anonymous reporting (0800 07 5278). UNISA Integrity Policy

NWU: Comprehensive AI policy. USAf compiles institutional guidelines. Shift: Declarations, transparency principles.

Solutions: Redesigning Assessments and Building AI Literacy

• AI-Resilient Assessments: Oral defenses, in-class writing, vivas, portfolios tracking process (drafts, peer reviews).
• Literacy Programs: Mandatory modules at UCT/NWU; ethical use training.
• Hybrid Tools: Proctoring + human oversight; collaborative platforms logging AI.
• Incentives: Reward originality, reasoning.
• National Push: DHET guidelines, tech partnerships for free access.

Experts like Carla Lever advocate adaptation: 'Detection impossible; rethink pedagogy.' 72

Read career advice on AI skills.

Expert Perspectives: Adaptation Over Alarm

Prof Jonathan Jansen: Detectors imprecise. Aidan Bailey (UCT): Skill gaps catastrophic. Stephen Sparks (UJ): Forces better teaching. Universities SA: Proactive policies essential.

Photo by Jolame Chirwa on Unsplash

Future Outlook: Opportunity in the AI Era

By 2026, SA unis lead ethical AI via NWU/UCT models. Invest in infrastructure, training; national framework imminent. Proactive adaptation preserves integrity, equips graduates for AI-driven world—boosting university jobs in tech-savvy roles.

Engage via higher ed career advice and recruitment for ethical opportunities.