FOR HIGHER EDUCATION
Why Restricting AI on Campus Is Not the Answer — and What Works Instead
Founder, AI User Safety Initiative
Faced with AI and academic integrity, many institutions reach for the most intuitive tool: restriction. Ban it, detect it, penalise it. The logic feels sound — protect the integrity of assessment by keeping AI out. In practice, restriction-based approaches achieve close to the opposite of what they intend.
What prohibition actually produces
A ban does not remove AI from campus. It removes the institution's visibility into how students use it. Use moves underground, students optimise to avoid detection rather than to develop understanding, and the relationship between staff and students becomes adversarial — a posture fundamentally at odds with the purpose of education.
Detection tools make this worse, not better. They create an adversarial dynamic that is fundamentally at odds with education's core function.
Detection also fails on its own terms: it is unreliable, it produces false accusations that damage trust, and it trains students to game a system rather than reason about appropriate use. The institution spends its energy policing instead of teaching.
What works instead: a literacy-centered approach
The alternative is not permissiveness — it is governance through literacy. Institutions that handle AI well replace blanket rules with structures that teach judgment:
- Context-specific guidance — clarity on where and how AI is appropriate across different assessment types, rather than one rule for everything.
- Curriculum-integrated AI literacy — instruction that builds critical evaluation and ethical reasoning, embedded in courses.
- Redesigned assessment — tasks built for an AI-integrated environment, where demonstrating judgment matters more than producing text a machine could.
- Professional development for staff — so academics can make sound, consistent decisions rather than improvising under pressure.
This approach is harder than a ban. It is also the only one that prepares students for a world where AI is everywhere — and that keeps the institution in a teaching relationship with them rather than a policing one.