A Modular, Public AI Security Assurance Framework

A practical and principle-aligned system for public sector use, designed from first principles, and released under a public license for review, reuse, and refinement.

Why This Exists

AI systems are entering sensitive and high-trust environments.

This framework was built to help public sector teams adopt AI safely, accountably, and transparently — without needing a PhD, a consultant, or a blank cheque.

It aligns with:

• Aotearoa New Zealand’s legal and cultural context
• Public service values
• Internationally recognised assurance and security principles
• The need for proportionate, transparent, and verifiable systems

It is designed to support:

• 🇳🇿 Alignment with New Zealand’s laws and values
• 📜 Te Tiriti o Waitangi and cultural safety
• 🔍 Proportionate assessment based on real risk
• 🧠 Explainability, oversight, and accountability
• 🧩 Integration with modern AI (including LLMs)

What This Framework Is

This is not a report. It is a system.

• A modular set of templates, controls, policies, and decision pathways