Where an AI policy fits among your existing policies

Standalone, or clauses in what you have?

Both work. A standalone AI policy is easier to point at, easier to review on a short cycle, and easier for staff to find. Clauses distributed through existing policies are less likely to contradict them and less likely to be forgotten.

The practical answer for most schools is a short standalone document that explicitly cross-references the others, rather than restating them. It keeps the thing that changes fastest in one place without creating a second version of rules that already exist.

What it overlaps with

• Acceptable Use — devices, accounts, passwords and what may be installed. The AI policy should name which tools are permitted and leave the rest there.
• Data Protection — lawful basis, minimisation, transparency with parents, DPIAs. The AI policy triggers these; it should not restate them.
• Assessment and academic integrity — what pupils may use and what must be declared. This belongs with the school’s existing assessment rules or it will be missed.
• Child protection — AI touches this where a tool processes safeguarding content or flags concerns. That is DLP territory and the AI policy should defer to it.
• Procurement — what is asked of a vendor before a system with AI features is bought.

The review cycle is the real difference

The argument for keeping it separate is mostly about pace. An AI policy will need revising more often than an admissions policy — the tools change, the Department guidance changes, and a document that is two years old will be describing a landscape that no longer exists.

A separate document can go back to the Board on its own without reopening five other policies. That is a small administrative point and it is the one that decides whether the policy stays accurate.