Stop AI Cheating in Australian Universities: Unis Warned to Reinstate In-Person Exams

The Escalating Crisis of AI Cheating in Australian Universities

In recent years, Australian higher education has grappled with a profound challenge: the widespread use of generative artificial intelligence (AI) tools, such as ChatGPT, by students to complete assessments. What began as a novel technology has evolved into a normalized practice, where students generate essays, solve problems, and even simulate exam responses without genuine effort. This phenomenon, often termed AI cheating or AI-assisted plagiarism, undermines the core principles of academic integrity—the commitment to honest scholarship and original work that forms the foundation of university education.

The shift accelerated post-COVID-19, with many universities retaining online assessments for convenience and scalability. However, this flexibility has backfired, enabling undetectable cheating on a massive scale. Academics report feeling overwhelmed, as AI-generated content floods submissions, making traditional grading nearly impossible. Students, meanwhile, perceive not using AI as a competitive disadvantage, fostering a culture where 'beating the system' is celebrated rather than condemned.

This crisis threatens the value of Australian degrees, raising questions about graduate preparedness for real-world roles in fields like medicine, engineering, and law. Employers increasingly doubt whether hires possess the critical thinking skills claimed on their resumes, potentially eroding international confidence in Australia's world-class universities such as the University of Sydney, University of Melbourne, and University of New South Wales (UNSW).

Former Chancellors Issue Stark Warning on AI Abuse

Leading the charge against this tide are two prominent figures in Australian higher education: Dr. Alan Finkel, former chancellor of Monash University from 2008 to 2016 and Australia's former Chief Scientist, and Terry Budge, chancellor of Murdoch University from 2006 to 2013. In a compelling op-ed published in The Australian on February 8, 2026, they warn that universities must immediately reinstate in-person supervised exams to halt the 'epidemic' of AI cheating.Read the full article here.

Dr. Finkel argues that online learning's legacy has commoditized education, turning universities into 'degree mills.' He proposes that at least 70% of final grades come from invigilated on-campus assessments, including supervised exams, in-class presentations, oral defenses, and practical labs. 'Universities are facing more criticism for not managing AI abuse,' Finkel states, emphasizing the reputational risk if inaction persists.

Budge echoes this, decrying plummeting lecture attendance—down to 7% at the University of Western Australia (UWA)—as evidence that AI enables absenteeism. 'We cannot allow a situation where online exams are cheatable with AI,' he asserts, calling for campus-based monitoring to restore faith in degrees. Their message has sparked national debate, amplified on platforms like X (formerly Twitter), where educators and students alike discuss the urgency.

Revealing Statistics: How Prevalent Is AI Cheating?

A landmark study, 'AI in Higher Education: Student Perspectives,' surveyed over 8,000 students from the University of Queensland (UQ), Deakin University, Monash University, and UNSW Sydney. The findings are sobering:

• 83% of students use AI weekly or more for studies.
• 44% use generative AI daily.
• 40% admit to unauthorized AI use for cheating.
• 71% believe AI facilitates cheating.
• 91% fear detection and rule violations.
• Only 27% fully trust AI outputs, yet 68% use it to generate text.

Academics estimate even higher rates, with up to 80% of submissions AI-influenced and students claiming 90% 'fullbore cheating' in some courses. TEQSA, Australia's higher education regulator, notes inappropriate use ranging from 10% to over 60% per cohort.TEQSA Report on AI Risks.

Case Study: UNSW's Dramatic Rise in Misconduct Cases

At UNSW, academic misconduct cases surged 219% in 2024 to 530, including 394 minor plagiarism and 136 serious cases involving exams. The Business faculty led, followed by Engineering, with UNSW College (pathway programs) seeing one in 12 students (8.3%) implicated. Penalties ranged from warnings and mark deductions to 35 expulsions.

Students exploit AI by drafting in native languages, then polishing via translation tools—prohibited under UNSW's strict policy. Contract cheating persists at 209 cases, often via illicit tutoring services. This mirrors national trends, with anonymous tips highlighting organized cheating networks.

False Positives: The Double-Edged Sword of AI Detectors

Over a dozen Australian universities deploy AI detection software like Turnitin or GPTZero, but reliability issues abound. The Australian Catholic University (ACU) wrongly flagged nearly 6,000 students in 2024, leading to 90% of misconduct cases being overturned after appeals. False positives cause stress, delays, and financial losses, as students prove innocence.

Experts warn detectors flag human work erroneously (even The Bible), while sophisticated prompts evade them. ABC investigations reveal no national standards, exacerbating inequities for international students. Lecturers like UWA's Dr. Jonathan Albright lament: 'You don’t need to show up if AI is doing your assessments.'

The Case for Reviving In-Person Supervised Exams

In-person exams offer a robust countermeasure, as AI cannot intervene under proctor supervision. Here's how implementation works step-by-step:

1. Schedule invigilated sessions: Allocate campus venues for high-stakes assessments, requiring ID checks and no devices.
2. Diversify formats: Combine written exams with practicals, where hands-on skills reveal true competence.
3. Scale gradually: Start with 30-50% of marks, building to 70% as Finkel recommends.
4. Monitor and adapt: Use CCTV and roaming invigilators to deter collusion.

Benefits include restored integrity, higher attendance, and employer reassurance. Challenges like costs and logistics exist, but former chancellors argue they're essential investments. Universities like Sydney have banned smart devices in exams, proving feasibility.

Complementary Strategies: Oral Assessments and Beyond

Beyond exams, TEQSA endorses multifaceted approaches:

• Oral vivas: 15-minute discussions verify understanding, as at Griffith University.
• Process evidence: Require prompt logs, drafts, and 'show your working' to distinguish AI from original effort.
• Relational learning: Know students via portfolios and interactions to spot inconsistencies.
• AI literacy: Teach ethical use, distinguishing permitted brainstorming from cheating.

UNSW's multi-lane policy exemplifies this, balancing innovation with safeguards.

University Policies and Ongoing Challenges

Australian universities vary: Sydney permits AI except in closed exams; Melbourne handles 'low-risk' breaches leniently. TEQSA urges centralized integrity teams over overburdened lecturers. Challenges include international student volumes (over 50% at some unis), cultural differences in plagiarism views, and resource strains. For lecturers navigating this, resources like higher ed career advice on AcademicJobs.com offer guidance.

Broader Implications: From Degrees to Employability

Unchecked AI cheating devalues degrees, with graduates lacking skills for complex roles. Employers in high-stakes sectors like healthcare question hires: 'Did AI earn this qualification?' Reputational harm could deter international enrollments, vital to Australia's $48 billion sector. Positive note: reformed assessments build AI-fluent professionals, enhancing global competitiveness.

Stakeholders diverge: Students cite pressure and AI ubiquity; lecturers feel pressured to pass suspects; employers demand verifiable skills. Solutions must balance equity, especially for regional or disabled students.

A Roadmap for the AI Era in Australian Higher Education

The path forward demands leadership: sector-wide standards, TEQSA oversight, and vice-chancellor commitment to in-person rigor. Short-term: ramp up supervised assessments. Long-term: redesign curricula for AI integration, emphasizing creativity and ethics. Platforms like AcademicJobs.com support this transition with opportunities in lecturer jobs and professor jobs.

Explore Rate My Professor for insights into teaching quality amid changes, or browse higher ed jobs and university jobs in Australia. For career advice, visit higher ed career advice. By acting decisively, Australian universities can reclaim integrity and prepare graduates for an AI-driven world.

Photo by Datingjungle on Unsplash