Privacy-preserving deployment measures for an ITS under GDPR and EU AI Act¶

Strategy · All strategies
Evidence · no claims cited

Description¶

To operate the ITS in a German university regulatory environment, the authors implemented layered privacy measures: strict separation of IP address logs from the ITS, pseudonymous usernames, no e-mail storage beyond registration, voluntary (non-mandatory) use, explicit fine-grained consent for keystroke-level research data recording, teacher exclusion from servers until after anonymization and grade finalization, and self-hosting of an LLama-70b model instead of commercial LLM interfaces to prepare for the EU AI Act.

Design Implications¶

Context¶

Requirements¶

• Explicit consent acquisition for research data recording
• Self-hosted or partner-hosted open-weight language models rather than commercial LLM interfaces
• Manual anonymization of data and finalized grades before teacher access to servers

Constraints¶

• Strict data privacy meant mandatory task submissions could not be implemented, so use had to remain entirely voluntary
• The precise implications of the EU AI Act for AI in education are not yet fully clear
• Fine-grained consent acquisition is described as a research topic in itself that is far from resolved

Target Learners¶

• University students in EU/German higher education contexts

Target Learning Goals¶

• Programming skill practice with legally and ethically compliant data handling

Related Strategies¶

• LA researchers and practitioners should adapt to changing regulations across jurisdictions, use regulatory sandboxes, and carefully consider data nature and privacy risks before cross-border data transfer
• Four-step anonymization balancing disclosure risk and data utility
• Secure enclaves for privacy-protecting researcher access to student-level data

Examples¶

-

Key Sources¶

• Alina Deriyeva, Jesper Dannath, and Benjamin Paaßen. (2025). SCRIPT: Implementing an Intelligent Tutoring System for Programming in a German University Context. Artificial Intelligence in Education (Springer). https://doi.org/10.1007/978-3-031-99261-2_16