The current direction

Six proposed rights

This is a draft from Mackenzie Conor James Clark and the Lycheetah Framework. It has not been adopted as law or a national standard.

1. Know when AI is used

You are told when an AI system processes your data or shapes a consequential decision about you, and which system was involved.

2. Keep a receipt

You receive a record of the decision, purpose, policy and time. Integrity information supports comparison with a trusted copy; a hash alone does not prove that the event happened.

3. Access, correct and withdraw

You can access your data, request correction and withdraw specified future AI uses under the proposal. Retention duties, exceptions and completed effects must be disclosed.

4. Reach a human

A consequential decision has a named human route for review and correction. This is a proposed service commitment, not a claim that every deployment already provides it.

5. Know where data lives

You are told where data is stored and when it is disclosed overseas, with applicable protections and exceptions made clear.

6. Receive equal treatment

The proposed standard applies equally regardless of ancestry, faith, region or politics. Group-level outcome testing must publish its method, limits and results.

Read the complete proposal →

Evidence and open work

A proposal with specific gaps

The repository includes local policy/receipt and permission-control prototypes. It does not yet provide an end-to-end citizen service, independently validated group-fairness test, universal withdrawal mechanism or data-location attestation.

The earlier WOF v0.1 and implementation sketch are unreconciled research candidates. No certification scheme, independent assessor network, enforcement mandate or public register is established by this work. Proposed numeric thresholds are not validated national pass marks.

Policy and receipt source · Permission and recovery research · Current source guide and research status

Existing context

Proposed rights and current law are distinct

The Privacy Act already provides access/correction protections and conditions for covered overseas disclosures, subject to its scope and exceptions. Those provisions do not automatically enact the full AI-use notice, universal receipt, withdrawal and human-review package proposed here.

IPP 3A requires notice when personal information is collected indirectly. It does not require telling a person that AI was used. The voluntary Algorithm Charter for Aotearoa New Zealand (2020) already commits signatory agencies to a channel for challenging algorithm-informed decisions; rights 4 and 6 would make that enforceable for each person.

New Zealand's Public Service AI Framework provides separate government guidance. This repository is an independent proposal, not a substitute for applicable law, government requirements or genuine rights-holder consent. No agency, iwi or community endorsement is claimed.

Preserved history

Earlier drafts have a clear status

The March cultural-governance direction is superseded as the current NZ policy proposal. Its source documents remain available with historical notices. Their earlier ACTIVE tags and partnership/outreach plans are not current adoption or implementation claims.

Read the earlier cultural-governance drafts