Core policy on Generative AI in education and student experience

1. Vision and policy context

The University of Dundee recognises that Generative Artificial Intelligence (GAI) is now part of the educational, professional, and social environment in which students learn and graduates will work. The University will take a positive, responsible, and academically rigorous approach to generative AI in education.

This policy establishes a framework for the responsible, transparent, and ethical use of GAI. Our aim is to empower staff and students to innovate with these technologies while ensuring that human judgement, individual accountability, and rigorous academic standards remain the hallmark of a Dundee education.

The University acknowledges that GAI is a rapidly advancing field. Consequently, this policy is a living document. We are committed to an iterative approach, updating this policy as we learn from our community’s experiences, emerging research, and technological shifts. We encourage an ongoing dialogue between staff and students to ensure this policy remains fit for purpose and supportive of academic excellence.

Legal Framework: This policy operates within the framework of the UK GDPR, the Equality Act 2010, and Consumer Protection law. Where a conflict arises between policy restrictions and a student’s legal entitlement to reasonable adjustments, the legal entitlement shall take precedence.

Policy Alignment: This document must be read in conjunction with the following University policies and guidance:

• Assessment Policy for Taught Provision 
• Accessible and Inclusive Learning Policy
• Code of Practice on Academic Misconduct by Students 
• Data Protection Policy
• Information Security Policy
• Guidance on Proofreading of Written Submissions for Assessment 
• General copyright guidance 
• Equality and diversity policy

2. Policy scope

This policy applies to all activities contributing to academic credit or formal evaluation. It encompasses learning and teaching activities, assessment design, marking, feedback, and moderation across all formats, including text, code, imagery, audio, and multimodal outputs.

3. Core principles

• AI is part of the Educational Environment: Generative AI (GAI) should not be treated as a temporary disruption or as a separate issue detached from normal learning and assessment. It is now part of the environment in which students learn, communicate,create, and work. The University’s role is to ensure that students learn to use AI responsibly, critically, ethically, and professionally, while maintaining the academic standards expected of Dundee awards.
• Responsible Empowerment: The University encourages the exploration of GAI to support and enhance the student’s ability to achieve intended learning outcomes.
• AI has no Academic Agency: Generative AI tools do not have academic agency. They do not hold responsibility for the accuracy, integrity, ethics, originality, or appropriateness of submitted work. AI is an engine for augmentation, not a replacement for human agency. Students remain responsible for the work they submit; staff remain responsible for academic judgement.
• Assessment must be designed for the AI Environment: Assessment design should recognise that generative AI is widely available. Assessment should be designed to make clear when individual performance must be assured under controlled conditions and when students may use AI as part of authentic academic or professional practice.
• Active Transparency: Open communication regarding AI use builds trust and supports integrity. Where GAI is used, its role must be clearly disclosed by students and staff.
• Inclusion and accessibility must be protected: Aligned with our Accessible and Inclusive Learning Policy, we leverage AI to remove environmental and systemic barriers to learning. The University is committed to ensuring AI promotes equitable participation for all students.
• Ethical Stewardship: All AI users must respect data privacy, intellectual property, and the standards set out in the Data Protection and Information Security policies.
• Graduates must be prepared for an AI-transformed world: The University has a responsibility to ensure that its education remains relevant to the professional environments graduates will enter. AI is already reshaping the nature of work across most sectors and disciplines. Programmes should consider what AI competence means in their specific graduate employment contexts and ensure that students can develop the knowledge, skills, and professional judgement they will need to work effectively and ethically alongside AI in their careers.

4. Assessment model

The University uses a three-category assessment model. Each category defines the conditions under which an assessment is completed and the default position on generative AI use.

• Controlled Situation Assessment: Assessment completed under restricted conditions where direct assurance of individual student performance is required. AI use is restricted by default.
• Uncontrolled Situation Assessment: Assessment completed outside controlled conditions, where students may access external tools and resources. AI use is permitted by default.
• Mixed Situation Assessment: Assessment combining uncontrolled and controlled elements, where authentic practice is assessed alongside direct assurance of individual understanding or capability.

Full descriptions, including default positions, purposes, and examples for each situation being used, are set out in Appendix A.

5. Assessment briefs

Every assessment brief must state which situation is being used:

• controlled
• uncontrolled
• mixed

Assessment briefs must also state any specific restrictions or expectations concerning AI use.

Where AI use is restricted, the brief must explain the restriction clearly and link it to the intended learning outcomes or assessment purpose.

Where AI use is permitted, the appropriate assessment brief should explain whether students are required to disclose their use of AI and what form that disclosure should take.

6. Student responsibilities and disclosure

   Students are responsible for ensuring that their use of generative AI is appropriate, ethical, transparent, and consistent with the assessment brief. Students must

   • Follow the assessment brief and any programme-specific instructions.
   • Understand whether an assessment is controlled, uncontrolled, or mixed.
   • Not use AI where it is prohibited.
   • Disclose AI use where disclosure is required.
   • Check the accuracy and reliability of AI-assisted work.
   • Verify sources, references, data, quotations, claims etc.
   • Ensure that submitted work reflects their own understanding.
   • added
   • Be prepared to discuss, explain, justify, or defend submitted work.

   Students should not and must not assume that AI-generated output is accurate, unbiased, lawful, ethical, properly referenced, or appropriate for academic use.

   1. Disclosure of AI usage

   Where disclosure is required, students must provide a clear statement explaining how generative AI was used.

   Programmes and/or Assignment briefs will state what disclosure is required by students. A disclosure statement may include (but is not limited to):

   • the tool or tools used
   • the purpose of use
   • the nature of the contribution
   • the parts of the work affected
   • how the student checked, revised, or verified the output
   • any limitations or risks identified.

   Disclosure should not be treated as admission of misconduct. Responsible disclosure supports academic integrity and professional practice.

   Failure to disclose AI use where disclosure is required may be treated as an academic integrity concern.

   Students using AI for reasonable adjustments recommended by Disability Services are NOT required to indicate use unless AI technology is used is beyond the reasonable adjustment.

7. Academic integrity and professionalism

The University views academic integrity as a commitment to authentic learning. "AI Authorship" (presenting AI-generated work as one’s own) is a breach of this and will be managed under the Code of Practice on Academic Misconduct by Students.

• AI Authorship: Submitting GAI-generated content as your own original work or using GAI to paraphrase other work to avoid detection, is treated as academic misconduct (plagiarism) in the same manner as presenting another person's work as your own.
• Data Privacy: Entering confidential patient, client, or placement-related data into non-university approved GAI tools are strictly prohibited and will be treated as a professional disciplinary matter.
• Fabrication: Using GAI to generate fabricated citations or "hallucinated" evidence is a breach of integrity.

Review Process: Staff reserve the right to conduct an informal authorship check-in or viva to discuss a student's process. These are intended as pedagogical opportunities to verify understanding. (see Assessment Policy for Taught Provision)

8. Staff responsibilities in delivery and feedback

   Staff are responsible for designing and delivering learning and assessment in ways that are appropriate to the AI environment. Staff must:

   • make assessment conditions clear to students
   • identify whether assessments are controlled, uncontrolled, or mixed
   • explain restrictions or expectations concerning AI use
   • align AI restrictions with intended learning outcomes
   • support students to develop AI literacy where AI use is relevant to learning.
   • protect academic standards through appropriate assessment design.
   • use University-approved tools and processes where student data or assessment material is involved.
   • retain responsibility for academic judgement.

   Staff may use AI to support teaching, planning, and learning materials, administrative drafting, or assessment design, provided such use is appropriate, secure, and consistent with university policy.

   Final academic judgement, including marks, grades, progression decisions, feedback decisions, and moderation outcomes, must remain with university staff. 

   1. Feedback, marking, and moderation

   Generative AI may be used to support aspects of feedback preparation or assessment administration only where this is consistent with university policy, data protection requirements, information security requirements, and the Assessment Policy for Taught Provision.

   AI must not replace academic judgement.

   Where AI is used to support feedback or marking workflows, staff must ensure that:

   • the use is approved by a programme lead or equivalent and deemed appropriate
   • student data is protected
   • outputs are checked by staff
   • feedback remains meaningful, accurate, and relevant
   • equality and accessibility implications are considered
   • students are not disadvantaged by inconsistent or inappropriate use.

9. Data Protection, usage, confidentiality, and information security

   Students and staff must not input confidential, sensitive, restricted, personal, unpublished, or professionally regulated information into public AI tools unless this has been explicitly approved by a Faculty VPESE.

   Use of AI must comply with the University’s Data Protection Policy, Information Security Policy, research governance requirements, and any relevant legal, professional, or contractual obligations.

   1. Intellectual Property and Copyright

   Students and staff must use generative AI in ways that respect copyright, intellectual property, licensing conditions, and moral rights.

   AI-generated or AI-assisted material must not be assumed to be free from copyright, license, or ownership issues.

   Students and staff must not use AI to reproduce, adapt, or imitate protected material in ways that breach copyright, license terms, or academic integrity expectations.

   • Final Judgement: In accordance with the Assessment Policy for Taught Provision, final decisions on grades must be made by academic staff.
     
     • Disclosure: Use of GAI in generating teaching materials or feedback should be acknowledged to students.
     • Data Security: Student personal data or full scripts must not be uploaded into unapproved public AI tools. Our current approved public tool is copilot (standalone and embedded into Microsoft 365).

10. Equality, Diversity, and Inclusion (EDI)

The University is committed to ensuring that this policy supports equitable participation for all students, including those with disabilities, long-term health conditions, or other protected characteristics.

Generative AI tools can play a valuable role in removing barriers to learning and assessment. The University actively supports the use of AI-enabled assistive technology (AI-AT) where this has been recommended by Disability Services as part of a student's documented support.

Where a student uses AI-AT tools as part of documented support:

• this use is legitimate, expected, and consistent with academic integrity in line with guidance on Proofreading of Written Submissions for assessment
• the student is not required to disclose this use in disclosure statements
• staff must respect the student’s documented support in any discussion of AI use
• however, students must NOT use AI-AT where the functionality of those tools aligns with any AI-use restrictions stated on the assessment cover sheet and would impact the achievement of a learning outcome or competency standard. If this is identified as a barrier, Disability Services should be contacted in the first instance.

Disability Services will work with students and staff to ensure that support with AI tools is provided with guidance to support both equitable access and the integrity of the assessment process.

Where there is uncertainty about the use of AI-AT by staff or students, Disability Services should be in contacted in the first instance.

The University will keep its guidance on AI-AT as a reasonable adjustment under review as AI capabilities develop, to ensure that disabled students are not disadvantaged by changes in either technology or policy.

11. AI literacy and student support

The University, within Faculties and appropriate central services will endeavour to support students to develop AI literacy as part of their academic and professional development.

AI literacy includes the ability to:

• understand the capabilities and limitations of AI tools.
• use AI critically and responsibly.
• evaluate AI-generated outputs.
• identify bias, errors, fabrication, and hallucination.
• protect data and confidentiality.
• use AI ethically and legally.
• disclose AI use appropriately.
• understand when AI use is inappropriate.
• retain ownership of academic judgement.

The University recognises that the employment landscape is changing significantly as a direct result of AI, and that graduates will be expected to work alongside AI tools across most professional contexts. AI literacy is therefore not only an academic concern but a professional and employability one.

Programmes should consider:

• what AI competence looks like within their specific discipline and graduate employment context
• how students will be supported to develop relevant AI skills and professional judgement over the course of their studies
• where AI literacy connects to professional standards, accreditation requirements, or regulatory expectations in their field.
• how the curriculum reflects the AI-transformed environments graduates are likely to enter

This work should be considered as part of ongoing programme review and curriculum development, ensuring that the University's taught provision remains responsive to a rapidly changing professional landscape.

12. Environmental sustainability

The University recognises that the use of generative AI carries an environmental cost. AI systems are computationally intensive, and their use at scale contributes to energy consumption, water use, and carbon emissions associated with data centre infrastructure. This is consistent with the University's broader recognition, set out in its Environmental and Sustainability Policy Statement, that all activities must be considered in terms of their environmental impact.

The University's commitment to achieving Net Zero no later than 2045 extends to the way staff and students use digital technologies, including AI. In practice, this means:

• being mindful of unnecessary or excessive AI use where simpler approaches would suffice
• considering the environmental credentials of AI tools and platforms where this information is available
• treating environmental impact as a legitimate factor in decisions about how and when AI is used

The University acknowledges that detailed guidance on sustainable AI use is still developing across the sector. This statement will be reviewed as clearer evidence and practical guidance emerges, in line with this policy's commitment to remaining a living document.

Use of AI must be consistent with the University's Environmental and Sustainability Policy.

Appendix A: Assessment Categories

1. Controlled Situation Assessment

Controlled assessment is used where the University requires direct assurance of individual student knowledge, skill, judgement, competence, or professional readiness, completed under restricted conditions with limited access to external tools, resources, or support.

Default position: AI use is not allowed unless explicitly permitted in the assessment brief.

Purpose: To assure that the student can demonstrate required learning outcomes without inappropriate external support.

Examples:

• Invigilated examinations
• In-class tests
• Oral assessments and Vivas
• Practical demonstrations
• Observed laboratory work
• Presentations with questioning
• Coding tests
• Clinical or professional demonstrations

Where AI tools are explicitly permitted within a controlled assessment, the assessment brief must state the following:

• which tools may be used
• what they may be used for
• what remains the student's own responsibility
• whether disclosure is required; and how AI use will be considered in marking

2. Uncontrolled situation assessment

Uncontrolled assessment is used where students complete work outside controlled conditions and may reasonably have access to external tools, sources, peers, software, and AI systems.

Default position: AI use is permitted unless explicitly restricted in the assessment brief.

Purpose: To support authentic academic and professional practice while maintaining student responsibility for submitted work.

Examples:

• Essays
• Reports
• Portfolios
• Software projects
• Case studies
• Design artefacts
• Creative outputs
• Dissertations
• Posters
• Presentations
• Reflective accounts
• Data analysis

Students remain responsible for all submitted work and must not submit AI-generated content in a way that misrepresents their own contribution, understanding, or achievement of the intended learning outcomes.

3. Mixed Situation Assessment

Mixed assessment combines uncontrolled work with a controlled element that allows the University to assure individual understanding, authorship, competence, or judgement.

Default position: AI use may be permitted in some elements and restricted in others. The assessment brief must clearly state which conditions apply to each element.

Purpose: To combine authentic AI-enabled practice with direct assurance of individual capability.

Examples:

• Coursework plus oral defence
• Project submission plus demonstration
• Software artefact plus code walk-through
• Report plus viva
• Portfolio plus reflective commentary
• Design artefact plus critique
• Group output plus individual questioning
• Dissertation plus supervisory checkpoint

Mixed assessment should be used where students are expected to engage with authentic tools and practices, including AI, but where the programme must also directly assure individual capability.
 

Appendix B: Standard Assessment Brief Wording

1. Controlled situation assessment wording

This is a controlled assessment. Generative AI tools must not be used unless explicitly permitted in the instructions for this assessment. You must complete the work under the stated assessment conditions. Use of AI in breach of these conditions may be considered under the University’s academic misconduct procedures.

2. Uncontrolled situation assessment wording

This is an uncontrolled assessment. You may use generative AI tools to support your work unless specific restrictions are stated in this brief. You remain responsible for the accuracy, integrity, originality, evidence, and quality of the work you submit. You must disclose AI use where required and must be able to explain and justify your submitted work.

3. Mixed situation assessment wording

This is a mixed assessment. Some elements may be completed in uncontrolled conditions, and some elements will be used to assure your individual understanding or capability. The instructions for each element state whether and how generative AI may be used. You remain responsible for all submitted work and may be asked to explain or defend your process and decisions.
 

Appendix C: Standard AI disclosure statement

Where required, students should include a statement of disclosure. The following structure is a standard example; programmes and/or assignments may have additional requirements, which will be stated in the assignment brief:

• I used generative AI in preparation of this work.
  
  • Tool(s) used: [insert tool name(s)]
  • Purpose of use: [e.g., brainstorming, planning, summarising, editing, coding support, generating examples, checking clarity, analysing data]
  • Parts of the work affected: [briefly identify sections, outputs, or processes]
  • Verification: [explain how the output was checked, revised, corrected, or evaluated]
  • Responsibility: I confirm that I have reviewed the work, that I understand the submitted content, and that I take responsibility for its accuracy, integrity, and compliance with the assessment brief.

Where no AI was used and a declaration is required:

• I did not use generative AI in the preparation of this work.

Appendix D: Key Definitions

• AI Literacy: The skills to navigate, use, and critically evaluate AI technologies.
  
  • AI Authorship: When the AI produces the substantive content of a submission.
  • AI-Enabled Assistive Technology (AI-AT): Tools used specifically to mitigate disability-related barriers.
  • Process Data: Evidence of iterative development (e.g., version history or prompt logs).

Appendix E: Policy administration

Steering Group: Chaired by the Associate Vice-Principal (Education and Student Experience). Membership includes representatives from all Faculties, DUSA, Disability Services, and Digital/IT.

Appendix F: Implementation roadmap and supporting resources

To ensure the effective and consistent application of this policy across the University, the following resources and workstreams will be developed and maintained by the Education Academy in collaboration with relevant Faculties and Professional Services.

Adaptable Frameworks: All institutional resources listed below are intended to serve as baseline frameworks. Faculties and Disciplines are encouraged to adapt these materials to suit their specific pedagogical needs and professional accreditation requirements.

Staff support tools:

• AI-Resilient Assessment Bank: A repository of best practice examples from across the University.
• Guidance on 'Informal Vivas': A toolkit for academic staff on how to conduct constructive authorship check-ins.

Student support tools

• Standardised AI Use Statement Template: A downloadable template for students to complete.
• 'How to Prompt Ethically' Guide: Focusing on critical prompting and avoiding 'hallucinations'.
• Process Log Templates: Digital templates to help students evidence original thought.

Institutional resources

• Updated Academic Misconduct Templates: Guidance for staff on interpreting AI-related cases.
• Approved AI Tool Registry: A dynamic list of GAI tools that have passed the University’s DPIA.

Strategic portfolio review

• Assessment Risk Audit: Establish a working group to conduct a comprehensive review of the University’s existing assessment portfolio. This audit will produce a report identifying AI risk factors (vulnerabilities and opportunities) within different disciplines to support the development of future-proofed assessment approaches. This will be done alongside a similar review of assessment load that has been planned.

Quality & Enhancement

• Annual Monitoring Reporting: An appropriate route for course leaders to report on use of AI in their courses and by their students will be built into existing reporting processes.
• Curriculum Design & Approval: Consideration of AI will be built into our design and approval processes.