Academic Integrity & Human Oversight

DRAFT — pending review.

Provenance trail: content moves through explicit stages — AI-generated → human-reviewed → moderator-reviewed → approved (or rejected). Each transition is recorded in an append-only trail (tamper-evident by DB trigger).
Human responsibility: AI output is a draft for educator review; publication is a deliberate human approval action, recorded with the approver.
Moderation & external examiners: moderator and external-examiner roles can read the integrity trail for assurance.
Research ethics: for research activities, consent is recorded against a pseudonymous reference; anonymisation status is tracked. No automated destructive processing.
Incident handling: see Incident Response — confirmed personal-data breaches start a tracked 72-hour regulator clock.