From Vulnerability to Security: Managing GenAI Risks in Assessment Design
- sasha97518
- Jun 5
- 2 min read
In 2024, my multi-institutional team undertook a large assessment stress testing exercise against various Generative Artificial Intelligence (GenAI) tools. As an output, we were able to identify and quantify risks and generate a list of short- and long-term security options as well as integration opportunities.
While the full study provides detailed evidence on the process, the key outcome is Table 10, the GenAI Assessment Security & Opportunity Matrix. A sample of the matrix is shown below. It is a perfect guide for any teaching staff that is unsure of the risks associated with any assessment types, and what changes or opportunities are available.

Please take note that at the time of writing, the study is one year old. Due to the rapid pace at which the technology is improving, some of the short-term security options identified are already obsolete. This highlights the longevity of working with a short-term option.
Within the supplementary materials of the full study, we also provide instructions on carrying out a risk assessment. This is an opportunity to best understand the risks and mitigation opportunities. An extract of those procedures:

A key takeaway from the study is that all assessments are impacted by GenAI in some way. It is important to identify the risks and apply mitigating factors. However, do not make rash decisions and make changes to assessments without any thought. Assessment validity (ensuring we accurately measure the intended learning outcomes) is just as important as academic integrity. Another important study from my team highlights the validity struggles in aligning the correct assessment to the required learning objective. It is also a great read for anyone interested in assessing laboratory-based competencies.
Sasha Nikolic
05 June 2025



Comments