What we expect
Every risk on our risks page has a practical answer. These are the five things we think New Zealanders should be able to expect from AI.
Someone is accountable when AI harms you.
Voice-cloning scams and AI companion chatbots are already affecting people. If a cloned voice scams your family, or a chatbot harms someone vulnerable, there should be a clear place to turn.
AI-generated content is clearly labelled.
Unlabelled AI-generated content is flooding social media, and it’s getting harder to tell what’s genuine. We’re starting with political ads, so voters know what they’re looking at before the 7 November election.
If AI helps decide your job, insurance or benefits, you’re told and can challenge it.
AI systems that can’t explain their decisions are already used in areas like hiring, insurance and benefits. Nobody should be stuck with a decision that no one can explain.
New Zealanders can see where public data goes, and who answers when critical systems rely on AI.
New Zealand’s cultural and public sector data can be collected by overseas AI companies without New Zealanders having a say. Our data and infrastructure shouldn’t depend on systems nobody here can check.
The most powerful AI isn’t released until it’s shown to be safe.
AI is advancing faster than anyone can check it, and new failures keep turning up. Until these systems can be shown to be safe, they shouldn’t be rolled out widely, and New Zealand should help push for international rules that make that the standard. If safety can’t be shown, development should slow or stop until it can.
Talk to your MP.
Tell them what you expect. It takes two minutes, and we’ll draft the email for you.