Securing academic references with referencing source information
DOI:
https://doi.org/10.65106/apubs.2025.2753Keywords:
academic integrity, assessment security, Generative AI, referencingAbstract
The release of Generative Artificial Intelligence (GenAI) tools forced [our institution] to rethink our assessment design by incorporating a process-oriented component that exploited GenAI weaknesses to limit its use and secure the integrity of student submissions (Shishavan, 2024).
In our capstone subject, we require students to engage with academic literature. Using the right prompt, GenAI tools can easily summarise literature. In the early stage of GenAI availability, we frequently identified cases of hallucinated references in student submissions. Once students were aware of this issue, we began to see frequent references to older academic textbooks used in assessments with suspected GenAI misuse. However, even when referred to academic integrity investigations, proving any misconduct was a time-consuming exercise and often inconclusive, partly due to the difficulty in accessing the cited textbooks.
With the aim of ensuring students are engaging critically with the academic literature at a level required in post-graduate studies and to disincentivise the unethical use of GenAI tools, we developed and trialled a process-oriented engagement called a “Reference Source Information document” (RSID). The RSID requires students to record their research and use of literature by providing the exact quotes, with page numbers, that formed the basis of their citations/ paraphrased content in their assessment. In addition, we interlinked the RSID within the marking rubrics. Submission of an incomplete or falsified RSID would result in a failing grade, without the need for an academic integrity investigation (unless egregious misuse was apparent).
The rationale for the RSID was the following:
- Ensure students are engaging with the literature critically and methodically, rather than providing GenAI-generated summaries, and therefore create an incentive for the ethical use of Gen AI.
- Students who use GenAI ethically would only see a small amount of extra work, if any, in completing the RSID as its process reflects good research practice expected from a post-graduate student.
- Current popular GenAI tools struggle to provide direct and accurate quotes, implying that without engaging with any literature, producing an RSID is not possible (Jazwinska & Chandrasekar, 2025; Peters & Chin-Yee, 2025).
- When students have substantially relied on GenAI to produce a section of an assessment, reverse engineering the RSID requires them to engage with the literature to find a direct quote that matches the claim made by GenAI.
- The requirement for students to perform a verification process underlines the ethical best-practice use and norms of GenAI in a professional context (Victorian Legal Services Commission, 2024).
We found that requiring students to provide the RSID was an effective way to secure academic references. The requirement greatly reduced the number of hallucinated references and textbooks.
While there were some student comments on the extra workload, these comments were in the minority. We found a key way to avoid this type of concern was for academics to make it clear why the document was important: to ensure cognitive engagement and learning from research, credibility of the submission, uphold academic integrity and ensure that subject and course learning outcomes are met.
Downloads
Published
Issue
Section
Categories
License
Copyright (c) 2025 Uwe Kaufmann, Edward Willson, Liz Heathcote

This work is licensed under a Creative Commons Attribution 4.0 International License.