New Zealand's Use of Anthropic's Claude for Government Work Raises Governance Concerns
Key Takeaways
- ▸New Zealand's Department of Prime Minister and Cabinet is using Anthropic's Claude-based tool (Paerata) to draft government honours citations and process sensitive personal health and political data
- ▸Anthropic is a signatory to New Zealand's Christchurch Call and is one of three major U.S. AI companies with formal relationships to the New Zealand government
- ▸The government obtained a special exemption to use AI with personal data despite its own policies restricting such use, with public discovery only occurring through Official Information Act requests
Summary
New Zealand's government has begun using Anthropic's Claude AI through a tool called Paerata to assist with administrative functions, including drafting New Year's Honours citations and processing sensitive personal data. The government obtained a special exemption to use the AI system despite its own policies restricting AI use with personal information. This development occurs as Anthropic, one of the world's leading AI companies, faces increased scrutiny from the U.S. Trump administration over national security concerns and refuses Pentagon demands for mass surveillance and autonomous weapons capabilities.
The use of Anthropic's technology in New Zealand's government infrastructure highlights a significant gap between the pace of AI deployment and the development of appropriate governance frameworks. New Zealand's National AI Strategy, released in July 2025, prioritizes economic opportunity over regulatory safeguards and relies on a light-touch approach with no AI-specific legislation and only voluntary guidance for the public sector. Despite identifying AI as a national security threat in its 2023 National Security Strategy, the government has failed to implement binding oversight mechanisms, instead relying on non-mandatory toolkits and templates that agencies are encouraged but not required to use.
- New Zealand's governance response to AI remains inadequate, with no binding AI-specific legislation, voluntary guidance frameworks, and continued internal discussions about whether to disclose AI use in government
Editorial Opinion
New Zealand's approach to AI governance reveals a troubling disconnect between the pace of technological deployment and the development of democratic oversight. While the government uses AI systems built by private American companies facing national security scrutiny, its own regulatory framework remains toothless—relying on voluntary guidelines rather than binding policy. The fact that officials are still discussing whether to disclose AI use in government documents three years after identifying AI as a national security threat suggests a governance crisis that could have far-reaching implications for public trust and national security.
