Short answer: three rules, applied consistently. Staff review every piece of AI-generated content before it reaches pupils; pupils don't use general AI tools directly; and no pupil personal data enters any AI system. A primary school that holds those three lines has covered the overwhelming majority of the real risk.
Why review-before-use matters most day to day
AI produces confident, plausible, occasionally wrong or subtly inappropriate content. The safeguard isn't complicated — a member of staff reads it first, the same professional filter applied to any resource from any source. Build that habit into your staff guidance and it becomes automatic.
The structural layer behind the daily rules
Behind the daily habits sit the structures: an AI policy everyone knows, the DSL involved in AI decisions from the start, filtering and monitoring standards applied to anything pupil-facing, and a risk register reviewed termly. None of it is new machinery — it's your existing safeguarding culture, extended to cover one more technology. Our full safeguarding and AI guide walks through each piece.