New Zealand's Algorithm Charter, launched in July 2020 by Stats NZ, saw 21 agencies commit to six transparency and fairness principles for algorithm use. An independent review a year later found 27 signatories — over half of government agencies — with near-universal support.
21
Signatory agencies at launch (July 2020)
27
Signatory agencies after one year (December 2021)
more than 50 %
Share of government agencies signed (December 2021)
Details
Maturity
Established
Promoter
Stats NZ (Government Chief Data Officer)
Period
2020–2021 (launch and one-year review)
Keywords
AI governance, data ethics, public administration, digital government
Context
On 28-29 July 2020, New Zealand's Government Statistician at Stats NZ launched the Algorithm Charter for Aotearoa New Zealand, a voluntary cross-agency commitment covering six areas: assessing risk with a probability/impact matrix, being transparent about how algorithms inform decisions, embedding Māori and Treaty of Waitangi perspectives, undergoing regular peer review, engaging affected communities, and providing a channel to challenge algorithm-based decisions.
Activities
Twenty-one agencies signed at launch, including the ministries of Justice, Education and Social Development, Inland Revenue, and Land Information New Zealand. An independent one-year review conducted by actuarial consultancy Taylor Fry, published in December 2021, interviewed signatories, non-signatories and subject-matter experts to assess uptake and impact.
Results
The review found the charter had grown to 27 signatories, more than half of all New Zealand government agencies, with support described as 'almost universal' and no agency reporting that the charter had stifled its use of algorithms. It also found agencies had made only partial progress against some of the six commitments, particularly public-facing transparency and periodic reporting, and recommended better cross-agency information-sharing to close those gaps.
Conclusions
The Charter is voluntary and carries no enforcement mechanism, so each agency decides what to publish and how thoroughly to apply the risk matrix; the documented gains are attitudinal and procedural — signatory growth, self-reported stocktakes and expressed support — rather than measured changes in the algorithmic decisions themselves. The review's findings fed into New Zealand's subsequent Open Government Partnership commitments on algorithm transparency.
Implementation
Indicative cost
Low (< €50k) — No implementation budget is disclosed; costs are primarily each signatory agency's own staff time to apply the risk matrix and the commissioning of the Taylor Fry review.
Time to results
Medium (1–3 years) — Launched July 2020; independent review completed and published December 2021; commitments remain ongoing with no end date.
Staffing & skills
Stats NZ / Government Chief Data Officer (charter owner and secretariat), Signatory agencies (Ministry of Justice, Ministry of Education, Ministry of Social Development, Inland Revenue, Land Information New Zealand and others) self-implementing the six commitments, Taylor Fry (independent reviewer)
Conditions for success
Clear institutional owner (Stats NZ) coordinating a voluntary cross-government commitment
A formally scheduled independent one-year review providing external accountability
Follow-through into a further national commitment (New Zealand's Open Government Partnership algorithm-transparency pledges)
Common failure modes
Voluntary and unenforced — each agency decides what to publish and how thoroughly to apply the risk matrix
Independent review found only partial progress on public-facing transparency and periodic reporting specifically