Institutional safeguards for safe, transparent AI use in assessment¶

Strategy · All strategies
Evidence · no claims cited

Description¶

The paper assigns institutions minimum responsibilities for enabling AI-mediated assessment: completing a privacy/data-protection impact assessment (PIA/DPIA) for any AI tool used in assessment with clear data-flow maps and student-facing summaries; vetting and approving tools against privacy, bias, accessibility and auditability criteria; and standardising prompt/process transparency conventions such as prompt-log formats and version control. These conditions let students evidence how they worked without disclosing personal data.

Design Implications¶

Context¶

Requirements¶

• Institutional settings must enable safe and transparent use so students and staff can be explicit about AI-mediated work without privacy or equity penalties

Constraints¶

• Free tiers and institutional AI offerings vary in availability, capability and accessibility

Target Learners¶

• higher education students and staff

Target Learning Goals¶

• responsible participation in digital professional environments
• data privacy awareness

Related Strategies¶

• Establish comprehensive institutional AI governance frameworks including task forces, roadmaps, and transparent policies
• Establish AI-specific evaluation and accountability systems using independent audits, role-based dashboards, and privacy and equity safeguards
• Address core ethical challenges—data privacy, algorithmic fairness, and transparency—when deploying AI-driven personalized learning
• Require process transparency artefacts: process logs, prompt records, reflective commentaries and oral defences
• Institutional governance should establish practical conditions supporting AI literacy, learner agency, and critical evaluation rather than relying on prohibition and detection
• Deploy soft barriers transparently: institutional approval, policy-level disclosure, and accessibility testing

Examples¶

-

Key Sources¶

• Kickbusch, S., Ashford-Rowe, K., Kemp, A., Boreland, J., & Huijser, H. (2025). Beyond Detection: Redesigning Authentic Assessment in an AI-Mediated World. Education Sciences, 15(11), 1537. https://doi.org/10.3390/educsci15111537