New Zealand's Ministry of Education Publishes National Generative-AI Guidance for Schools

New Zealand's Ministry of Education published 'Generative AI: guidance and resources for education professionals' on 25 November 2024, and has kept it under active revision since — the version reviewed here was last updated 22 May 2026. The guidance takes a principles-based approach centred on teacher responsibility and professional judgement, human oversight, and appropriate use cases, stating that the teacher-student relationship must remain central and that AI-assisted marking should support, not replace, teachers' professional judgement.

Schools are directed to publish their own AI policies covering purpose, scope, guidelines for internal and external use, risk mitigation, professional-development needs, data-privacy protocols and a regular review cycle — placing the operational burden of governance on individual school boards. On data protection, the guidance is concrete: schools are told not to enter copyright, confidential, personal or sensitive information into AI models, citing Office of the Privacy Commissioner guidance and warning that most free AI tools reuse submitted prompts for further model training. Unusually for an official guidance document, it explicitly names a cultural-equity limitation of the technology itself: AI models are trained overwhelmingly on 'dominant cultures and languages', and are 'likely to be weak on mātauranga and te reo Māori, as well as Pacific languages and Polynesian cultures'. Generative AI is prohibited in NCEA external assessments, and any internal-assessment use must still meet achievement-standard conditions requiring that submitted evidence represents the student's own, independently demonstrated work.

This is a policy and governance document, not a learning intervention, so it carries no learning-outcome data of its own. A separate NZCER national teacher/student survey (tracked as a distinct practice) provides classroom-adoption context, but the Ministry has not published system-wide compliance-monitoring data showing how consistently the roughly 2,500 state schools have implemented the recommended board-level AI policies.