CDU Generative Artificial Intelligence Policy
CDU recognises generative AI as an emerging technology with social, environmental and academic implications. The University commits to educating all stakeholders about responsible and informed gen AI use, while exploring its potential for innovation across education, research and operations. This policy sets the principles for the use of generative AI at CDU. It applies to all employees and students, and frames responsible adoption while maintaining excellence and integrity.
Purpose and Scope
This policy establishes principles for the use of generative artificial intelligence (gen AI) at the University. It applies to all employees and students, and governs gen AI use across all university activities including teaching, learning, assessment, research and professional operations.
Core Principles
CDU's approach to generative AI is guided by seven principles that ensure responsible adoption across the University.
- Exploration and learning. CDU encourages innovation and ensures all stakeholders understand ethical, effective gen AI use, building AI literacy across all areas of operation.
- Guidance and training. The University will provide discipline-specific guidance, resources and training to support staff and students.
- Equity. CDU will actively mitigate risks arising from inequitable access to gen AI tools, ensuring fair participation for all.
- Transparency. Gen AI systems must be transparent; decision-making involving AI should be understandable and explainable.
- Stakeholder notification. Users must be informed about how, when and where gen AI operates within university systems and processes.
- Security. CDU will implement robust measures for monitoring and safeguarding all gen AI systems used within the university environment.
- Bias mitigation. The University will address algorithmic bias and work to prevent unintended discriminatory outcomes.
Platforms and Usage
The University does not maintain a static "approved" or "endorsed" platform list. Gen AI platforms and tools evolve rapidly, and CDU's approach reflects that reality.
Platform adoption
New platform adoption requires Design Authority approval when it involves institutional implementation, system integration, or the processing of University data. This ensures appropriate governance oversight for enterprise-level AI deployments.
Platform restrictions
Restrictions may apply when platforms present unacceptable risks related to security, privacy, data sovereignty, compliance, ethics, or Information Security Policy violations. Staff should consult IT Services if uncertain about a platform's suitability.
Integrity
Known risks
Gen AI platforms may contain incorrect, out-of-date and biased data. They can hallucinate and reference non-existent sources and facts, and outputs can appear authoritative despite being inaccurate. Users must verify all information to avoid submitting false content. Content generated by AI may also infringe copyright, so all users must comply with CDU's Copyright and Intellectual Property policies when using AI-generated material.
Student use
Academic teachers and HDR supervisors provide guidance covering which learning activities permit gen AI use, the extent of permitted usage, referencing requirements, and associated risks. Students and employees must be transparent about, and disclose, gen AI use.
Research use
Researchers must observe expected standards of integrity under the Responsible Conduct of Research Policy and associated procedures when using gen AI tools in their work.
Data, Privacy and Security
Gen AI platforms may incorporate user prompts into responses to other users. This creates significant data protection obligations for all CDU staff and students.
Personal identifiers (names, addresses, contact details); sensitive dates and identification numbers; identifiable photographs, video or audio; employment, educational or health information; location data or linked opinions; assessment or examination content; and confidential University information.
Research data protection
Researchers must protect unpublished work by obtaining vendor assurance that data will not be used to train future models. This is critical for maintaining research integrity and intellectual property protection.
Social and Environmental Justice
Equity of access
CDU acknowledges the subscription costs and equity concerns associated with gen AI platforms. If course fees apply for gen AI platform access, this must be specified prior to student enrolment so people can make informed decisions.
Environmental impact
The University recognises that gen AI involves significant greenhouse gas emissions and the use of water and other natural resources, and encourages judicious use to minimise environmental impact.
Teaching and Learning
Gen AI challenges traditional assessment reliability. CDU commits to agile assessment reviews that keep processes current and fit for purpose given AI capabilities. The University's strategies are to ensure understanding of gen AI implications and Academic Integrity Policy compliance, to develop competent and ethical gen AI users among staff and students, and to assure learning outcomes through robust assessment design.
Assessment and marking
Gen AI use in assessment marking or feedback must comply with the data, privacy and security requirements in this policy. VET competency judgements require qualified assessors, and gen AI platforms cannot fulfil that role.
The Academic Integrity Policy
The Generative AI Policy does not stand alone. When a question involves student work, the Academic Integrity Policy is usually the instrument that answers it. The current version took effect on 31 October 2025, approved by Academic Board, and it takes a deliberately educative approach: the University's stated priority is preventing academic dishonesty through education, with investigation and discipline as the response when prevention fails.
Three parts of the policy do most of the work in AI questions.
What staff must do
Staff who design and assess student work must provide clear guidance on the permitted use of technologies such as generative AI, support students to critically evaluate AI output, stay aware of emerging academic technologies and their implications for assessment, and clearly explain what constitutes contract cheating, plagiarism, collusion and other forms of cheating within their unit. The obligation to be clear about what is permitted sits with the staff member, not the student; a student cannot follow a boundary that was never drawn.
What students must do
Students must declare and acknowledge use of generative artificial intelligence when creating academic content. They are responsible for the academic integrity of everything they submit, which includes keeping appropriate records: sources, drafts, back-up copies, and records of their contributions to group work. The policy also recognises the evolving role of generative AI in academic work: where a task requires it, students demonstrate ethical, transparent and purposeful use, and their work may be evaluated on the quality and appropriateness of that use.
What happens when authenticity is questioned
Where concerns arise, the policy allows the University to ask a student for evidence that the work is their own: research notes, annotated bibliographies, drafts, or version histories. Students may also be asked to present, explain or defend their ideas verbally, face-to-face or online. Allegations against students are then managed under the Student Academic Misconduct Procedure, with the policy requiring investigations that are fair, confidential and timely, and decisions that are evidence-based and consistent with procedural fairness.
Scenario discussion: test it against the policy
Policies read differently when a real situation is on the table. Each scenario below is drawn from situations that recur in Australian universities and TAFEs. Working in small groups, take one scenario, find the parts of the two policies that bear on it, and decide what the policies actually permit, require or rule out. The point is not to reach a comfortable answer; several of these scenarios expose places where the policies give less guidance than people assume, and noticing that gap is the exercise.
The assessment rewrite
A lecturer is rewriting a tired assessment task and wants to calibrate the new version against real student performance. They propose uploading a folder of previous student submissions to a gen AI tool and asking it to analyse how well the old task elicited evidence of the learning outcomes. The submissions are marked, the unit is finished, and no student names appear in the file names.
What the Data, Privacy and Security section prohibits inputting; whether marked student work is "assessment content" or "educational information" even with the names stripped; who owns the intellectual property in a student submission; and whether any form of this idea could be done within policy.
The multilingual student and the detector score
A detection tool reports that a final-year student's essay is 87 per cent likely to be AI-generated. The student speaks English as an additional language, has a clean record across three years of study, and denies using AI. The unit coordinator believes the score and wants to proceed straight to a misconduct allegation on the strength of the report.
What the Academic Integrity Policy actually permits when authenticity is questioned; what "evidence-based" and "procedural fairness" require before an allegation proceeds; the equity and bias-mitigation principles in the Generative AI Policy; and what the student can be asked to provide or do, as distinct from what the score alone can establish.
The retrospective demand for drafts
After marking has begun, a lecturer becomes suspicious of several polished submissions and emails those students asking each to produce their drafts and version history within 48 hours. Nothing in the unit outline or assessment brief told students to keep drafts, and one student wrote their essay in a single sitting in a library computer lab, in a session that saved nothing.
The policy's record-keeping responsibility and where students are supposed to learn about it; whether a request for evidence is fair when the requirement was never signalled; what the absence of drafts can and cannot prove; and what the staff obligation to give clear guidance implies for next semester's unit outline.
The accessibility collision
A student with an approved access plan uses an AI-based writing-support tool as part of their reasonable adjustment. The unit's assessment brief states plainly that any use of generative AI is prohibited. A tutor notices the tool's characteristic phrasing in a submission and raises it with the unit coordinator, who says the rule is the rule and every student must be treated the same.
The equity principle in the Generative AI Policy; what "treating every student the same" means when one student has an approved adjustment; whether a blanket prohibition was ever compatible with the student's plan; and who needed to talk to whom before the assessment brief was published.
Marking with Copilot
A time-poor lecturer with 140 submissions to mark wants to paste each essay into Copilot and ask it to draft feedback against the rubric, which they will then review and edit before releasing. A colleague in the TAFE faculty likes the idea and wants to do the same for a batch of VET assessment tasks awaiting competency judgements.
The Assessment and Marking clause of the Generative AI Policy and the data, privacy and security requirements it invokes; whether a student essay is the student's personal information and intellectual property; the difference between the higher education case and the VET case, where the policy is explicit that competency judgements require qualified assessors; and what disclosure students are owed about how their work is being processed.