Our school is mid-way through rewriting our AI policy and I'm finding it really hard to land on principles that work across faculties.
The version we drafted last year tried to be one-size-fits-all and frankly didn't survive contact with reality. English teachers wanted permissive use with disclosure; Maths wanted near-total ban during assessment; Art was somewhere else entirely.
A few specific things I'm trying to figure out:
- Where does your policy sit? Is it part of the academic integrity policy, the digital citizenship one, or its own thing?
- How do you handle assessment specifically? Is it per-task, per-subject, or a blanket rule? Do you require students to declare AI use?
- Staff use — is there separate guidance for teachers using AI to plan/mark/give feedback? Or is it all rolled together?
- Review cadence — how often are you re-visiting it? AI is moving so fast that an annual cycle feels too slow.
I'd love to hear from anyone (primary, secondary, or tertiary) about what's actually working. Even better if you can share a public version of your policy I can pinch ideas from.
8 replies
Sign in to join the conversation.
Sign inWe split it: a short whole-school AI Charter (3 principles, one page) and then faculty-specific addenda. The charter sets the tone (transparency, equity, learning-first); the addenda do the heavy lifting for assessment rules. It's working better than the monolith we had before.
this is a reply
For assessment we've gone per-task. Every assessment task now has a small "AI use" badge: Permitted / Permitted with disclosure / Not permitted. Students get a 30-second explainer of what each one means in week 1 and it's referenced on every rubric. Removed a lot of the "is this allowed?" anxiety.
On staff use — we made the call early to write it INTO the policy, not leave it implicit. Staff using AI for planning/marking/feedback has to be disclosed in faculty meeting once a term so we know what's working. It removed the secretive feeling without policing anyone.
Review cadence: we're on quarterly now, but the bigger shift was decoupling principles from practices. Principles get reviewed annually; practices (which prompts, which tools, which workflows) get reviewed every term. The principle layer barely moves; the practice layer moves a lot.
+1 to the charter + addenda split. We tried the monolithic approach too and it got 47 pages long before anyone noticed. Nobody read it. The 1-page charter actually gets stuck on classroom walls.
Genuine question back: how are you handling the equity piece? Some kids have ChatGPT Plus at home, school accounts get the free tier. We're wrestling with whether the school provides paid access or whether the policy explicitly says results can vary by tool and that's OK.
I'm not too sure about this