University AI Policies Create "Fragmented Justice Landscape"

Published: 3rd August 2025

A concerning pattern is emerging across UK higher education: identical AI use cases are being celebrated at some universities whilst resulting in misconduct proceedings at others. This inconsistency is creating what legal experts describe as a "fragmented justice landscape" that undermines student confidence and may disproportionately affect international learners.

The Problem of Inconsistent Enforcement

The issue became prominent following several high-profile cases where AI detection software triggered investigations without human oversight. In one documented case from Manchester, a student faced academic misconduct allegations after their essay was flagged solely by Turnitin's AI writing detection tool—despite no concerns being raised by the human marker who assessed the work.

The student, who was not a native English speaker, was not given access to the detection report or the basis for suspicion, nor could they test the "evidence" against them. This case highlights a fundamental problem: AI detection tools are being used as definitive proof rather than preliminary screening devices.

Scale of the Challenge

Education Week reports that of more than 200 million written assignments reviewed by Turnitin's AI detection tool in 2024, one in ten flagged some AI use. However, the accuracy of these detections remains questionable, particularly for non-native English speakers whose writing patterns may differ from training data used by detection algorithms.

The variation in institutional responses is stark. A student using ChatGPT for idea generation might be celebrated for innovative thinking at one university but face disciplinary proceedings at another. This inconsistency is particularly problematic given the international nature of UK higher education.

Impact on International Students

Legal experts note that international students face additional vulnerabilities in this environment. They may be less familiar with UK academic norms, struggle with unclear policy language, and lack access to adequate support when facing allegations. The reliance on algorithmic detection potentially creates discriminatory outcomes, as AI tools may be more likely to flag writing that deviates from standard patterns.

Some universities have begun to recognise these issues. The University of Queensland, for example, disabled Turnitin's AI writing indicator functionality for all assessments from Semester 2, 2025, acknowledging concerns about accuracy and fairness.

Institutional Responses Vary Wildly

The policy landscape remains deeply fragmented:

• Complete prohibition: Some institutions treat any AI use as automatic misconduct

• Conditional permission: Others allow AI for specific tasks under supervision

• Declaration requirements: Many require students to declare AI use but provide unclear guidance on what constitutes use

• Laissez-faire approaches: A few institutions have minimal restrictions

This variation occurs even within institutions, with different departments applying different standards to similar AI use cases.

Recommendations for Researchers and Students

For Students:

• Document your research and writing process using version control (Google Docs history, Word track changes)

• Seek clarification on AI policies before beginning assessments

• Understand your rights in misconduct proceedings and access available support

For Supervisors and Faculty:

• Provide clear, specific examples of acceptable and unacceptable AI use

• Ensure policies are consistently communicated across all modules

• Advocate for evidence-based rather than algorithmic-only misconduct procedures

For Institutions:

• Develop transparent, consistent policies that acknowledge AI's legitimate research applications

• Establish minimum standards for evidence disclosure in AI-related misconduct cases

• Provide adequate support for students facing AI-related allegations

  
  

The Need for Regulatory Clarity

Legal experts suggest that a national regulatory response—similar to frameworks used for data protection—may be necessary to ensure fairness and consistency. Without sector-wide oversight, the current system risks creating disciplinary outcomes driven by software rather than evidence.

The fundamental challenge is balancing legitimate concerns about academic integrity with the reality that AI tools are becoming integral to research workflows. Universities that fail to address this balance risk both undermining student confidence and missing opportunities to enhance learning through appropriate technology use.

As AI capabilities continue to expand, the need for clear, consistent, and fair policies becomes increasingly urgent. The current fragmented approach serves neither academic integrity nor educational innovation effectively.

Sources:

AI in universities: the fine line between help and misconduct - 5SAH Barristers

Academic integrity and student conduct - University of Queensland