The Dominance of Misconduct Narratives in AI Policies

A recent critical analysis by James Cook University’s Centre for the Advancement of Digital and Scholarship in Education (CADSI) examined policies from all Australian universities, revealing a striking pattern. Using Carol Bacchi’s ‘What’s the problem represented to be?’ (WPR) framework, the study dissected how generative artificial intelligence (GenAI) is constructed as primarily a threat to academic integrity rather than an opportunity for innovation. Institutions frame AI through the lens of plagiarism, cheating, and unauthorised assistance, assuming a rigid divide between human originality and AI-generated content. This risk-averse approach overshadows discussions on how AI could enhance learning, foster collaboration, or redefine knowledge creation in a digital era.

The analysis highlights ‘silences’ in policies: minimal guidance on ethical, transparent AI integration for pedagogy. Instead, compliance and detection dominate, potentially stifling meaningful educational uses. This misalignment risks portraying universities as enforcers rather than enablers of 21st-century skills.

Key Findings from Policy Review Across Australian Institutions

The CADSI project reviewed publicly available documents from Universities Australia members, identifying academic integrity as the overwhelming narrative. AI is positioned as eroding student authorship and effort, with policies emphasising penalties over proactive teaching strategies. While international parallels exist, Australian policies lag in balancing threats with benefits, limiting forward-thinking reforms.

TEQSA’s 2024 report echoes this, noting GenAI as a catalyst for rethinking integrity processes. All Australian higher education providers submitted action plans by July 2024, but implementation challenges persist amid rapid tech evolution. Estimates show 10-60%+ student AI use, but inappropriate application remains the focal concern, prompting calls for systemic assessment overhauls.

Case Study: Australian Catholic University’s AI Detector Controversy

The Australian Catholic University (ACU) exemplifies policy pitfalls. In 2024, nearly 6,000 misconduct referrals—90% AI-related—stemmed from Turnitin’s detector. About 25% were dismissed, including all solely detector-based cases, after revealing false positives. Students endured months-long probes, ‘results withheld’ notations impacting job prospects, and burdens to prove innocence via notes and histories.

ACU scrapped the tool in March 2025, introducing AI ethics modules. Deputy VC Tania Broadley called the figure ‘overstated’ but confirmed half of breaches involved undisclosed AI. This highlights unreliable detectors, staff AI illiteracy, and policy ambiguities, mirroring wider sector struggles.

Statistics Highlighting the Scale of AI-Related Misconduct Concerns

AI misuse surges: ACU’s 6,000 cases; UNSW reported 100 exam incidents in 2024; nearly 80% of Aussie students used AI by 2025, creating an ‘illusion of competence’ masking shallow learning. Over a dozen unis deploy detectors, but errors erode trust. Confirmed UK cases hit 7,000 in 2023-24, signaling Australia’s trajectory without reform.

• 90% ACU referrals AI-linked (2024).
• 25% dismissed post-investigation.
• 80% student AI adoption (2025 survey).
• Detector abandonment trend (e.g., 36 unis in 33 months per one analysis).

These figures underscore urgency but also critique over-reliance on punitive measures.

University-Specific Policies: Risk Management Prevails

University of Sydney’s ‘two-lane’ system permits AI in open assessments with disclosure, shifting to process verification. UNSW’s multi-lane approach clarifies use levels. Melbourne mandates declarations, treating non-disclosure as misconduct. Yet, CADSI notes these still prioritise compliance over pedagogy, with few exploring AI for personalised learning or skill-building.

Australian Framework for AI in Higher Education advocates ethical deployment, but uptake varies. Macquarie leads with safe AI initiatives emphasising integrity and privacy.

Pedagogical Opportunities: Underutilised Potential

Beyond risks, AI offers tutoring, brainstorming, and accessibility aids. Experts like USYD’s Danny Liu advocate teaching ethical use over bans: “Academics are teachers, not police.” TEQSA recommends showing ‘working’ (prompts, integration), oral assessments, and student partnerships.

Deakin interviews reveal teachers redesigning for authentic tasks, but policy lags hinder innovation. Balanced frames could equip graduates for AI-driven jobs.

Stakeholder Perspectives: Students, Staff, and Experts

Students report confusion; staff face literacy gaps. Jason Lodge (UQ) warns against ‘wait-and-see’, urging awareness and centralised processes. Unis like Sydney foster relational learning, knowing students individually to spot inconsistencies.

• Students: Burdened by probes, unclear rules.
• Staff: Overwhelmed, low AI skills.
• Experts: Shift to verifying learning, not policing.

Implications for Australian Higher Education

Misconduct focus risks digital divides, equity issues (e.g., disabilities), and stifled innovation. It undermines trust, as ACU shows, and fails future-proofing grads. Broader: Credential devaluation if AI erodes skills.

Recommendations for Balanced AI Policies

CADSI urges forward-looking policies supporting ethical uses. TEQSA: Transparent frameworks, process documentation, sector collaboration. Actionable steps:

• Centralise misconduct handling.
• AI literacy training.
• Authentic assessments (vivas, portfolios).
• Student co-design.

Learn from pioneers like Macquarie’s safe AI trailblazing.

Photo by International Student Navigator Australia on Unsplash

Future Outlook: Towards Pedagogical Integration

As AI evolves, unis must evolve policies. 2026 trends: Detector scepticism, assessment redesigns, national frameworks. Potential: AI literacy as graduate attribute, boosting employability. Challenge: Equity, ethics amid tech pace.

Australia can lead by reframing AI as ally, not adversary.