Short answer
AI for nonprofits in New Zealand is most useful for the writing and assembly work that eats small teams’ time: pulling grant evidence into funder templates, drafting donor communications, and answering volunteer enquiries from approved information. People still decide what outcomes are claimed and what is sent. Charities Act reporting duties, the Privacy Act and the Unsolicited Electronic Messages Act all apply to AI-assisted work.
- Grant evidence assembly is a strong first project if every figure links to a source record.
- Large charities must have audited financial statements; AI-drafted reports carry the same duties.
- Researching donors from public sources is indirect collection, with notice duties since 1 May 2026.
- MBIE’s voluntary Responsible AI Guidance explicitly covers non-profits.
Where AI saves time in a charity
In most New Zealand charities the scarce resource is staff time, and a large share of it goes on writing: funder reports, applications, appeal letters, thank-yous, board papers and replies to people who want to help. AI is good at first drafts and at assembling information that already exists. It is bad at knowing what is true about your programmes, which is why the useful projects tie every sentence back to your own records.
Most AI adoption surveys cover businesses, not charities, so treat them as context. The Deloitte Access Economics study for 2degrees, published in April 2026 from 259 firm responses collected in January and February 2026, found 82% currently using AI, 13% planning to and 5% with no plans. Its definition counts AI built into software people already own, and most respondents used built-in or publicly available tools rather than anything purpose-built. For a charity that usually means the AI features in Microsoft 365, Google Workspace or a donor CRM are the starting point, and MBIE’s Responsible AI Guidance is written to include non-profits.
Nonprofit workflow map
| Workflow | What AI does | What a person still decides | Data involved | Difficulty |
|---|---|---|---|---|
| Grant evidence assembly | Pulls attendance, outputs, case summaries and finance figures into the funder’s template, linking each figure to its record | Which outcomes are claimed and how they are described | Programme records, attendance data, financials, consented stories | Medium: every number must trace |
| Donor communication drafting | Drafts thank-yous, appeals and updates in your voice, personalised from the donor record | What is sent and to whom, and anything sensitive such as memorial gifts | Donor CRM, gift history, programme stories | Low: a person sends |
| Volunteer enquiry handling | Answers common questions from approved information, collects availability and routes applications | Who is accepted, and any safety checks | Enquiry emails, forms, role descriptions | Low: approved answers only |
| Annual return narrative | Drafts the narrative parts of the performance report from board minutes and programme data | Final figures, statements and sign-off | Minutes, programme reports, financial statements | Medium: reporting standards apply |
| Funding application drafting | Drafts sections from past applications, the funder’s criteria and current data | Whether to apply and what is promised | Past applications, criteria, budgets | Medium: promises must be deliverable |
| Board paper summaries | Summarises long reports into decisions needed and actions | The decisions themselves | Board papers, management reports | Low: internal audience |
| Safety-check tracking | Tracks which volunteer checks are complete or expiring and drafts reminders | Whether someone may work with children or vulnerable people | Check records, role requirements | High: safeguarding stakes |
| Plain-language and te reo drafts | Produces plain-language versions and first-pass translations for a fluent reviewer | Final wording and cultural fit | Resources, forms, notices | Medium: translation quality varies |
Worked guides cover the first three: assembling grant evidence, drafting donor communications and handling volunteer enquiries.
Charity, privacy and email rules that apply
- Charities Act 2005. Registered charities file an annual return with financial statements (s 41). Under section 42C, statements must be audited where total operating expenditure was $1.1 million or more in each of the two preceding accounting periods, and audited or reviewed at $550,000 or more. Knowingly filing statements that fail the applicable reporting standard is an offence with a fine of up to $50,000 (s 42B). An AI-drafted narrative carries exactly the same duties.
- Incorporated societies. The window to re-register under the Incorporated Societies Act 2022 ran from 5 October 2023 to 5 April 2026, and the Companies Office says that transition period has ended; removed societies can apply to be restored. Confirm your status before automating any governance paperwork.
- Fundraising email. Section 9 of the Unsolicited Electronic Messages Act 2007 prohibits sending unsolicited commercial electronic messages with a New Zealand link, and an unsubscribe takes effect five working days after use. Check whether your appeals and promotions fall within it.
- Privacy and donor research. Building a picture of a prospective donor from public sources is indirect collection. Since 1 May 2026, IPP 3A requires notice as soon as reasonably practicable unless an exception applies.
- Māori data. For kaupapa Māori organisations and services working with whānau, Te Mana Raraunga’s principles favour keeping Māori data in Aotearoa and grounding every use in free, prior and informed consent.
This is general information; your auditor, Charities Services or a lawyer can advise on specifics.
The tools charities already use
Most charities run Xero for accounts, a donor or member database such as Infoodle, Salesforce Nonprofit Cloud, Raisely or Donorbox, and Microsoft 365 or Google Workspace for everything else. Start with the AI already inside those tools before buying anything new, and check where each one processes data. When you need the pieces joined up, for example programme data flowing into a funder template, that is a small integration job rather than a new platform. Our notes on Xero, Salesforce and Google Workspace cover the connection points.
A first project: grant evidence assembly
Funder reporting is repetitive, deadline-driven and checkable, which makes it a good first project for a small team.
- Pick one report. Choose a funder report due next quarter with a template you have filled before.
- Map questions to records. For each template question, write down where the evidence lives: attendance spreadsheet, case system, Xero, survey results.
- Draft with links. The system fills the template and attaches the source of every number and quotation. Anything without a source is marked for a person.
- Review in two passes. The programme lead checks the story; the finance lead checks the figures.
- Compare. Record hours spent against last time, and any funder queries.
Illustrative example: a 25-person social service in Christchurch files 14 funder reports a year, each taking a manager about 16 hours to gather data and write, so 224 hours. If assembly and drafting drop each report to about 7 hours of review and editing, the total is 98 hours, a saving of roughly 126 hours, about three working weeks of a manager’s year. The measure that matters as much as hours is that 100% of claims trace to a record. Our AI document processing work starts with this kind of mapping, and the lesson on creating a source-checked content draft shows the method.
Where AI goes wrong for nonprofits
Workflows we map in this sector
- Grant evidence assemblyPull numbers, outcomes and consented stories from your records into funder-ready drafts, with every claim sourced.
- Volunteer enquiry handlingReply to would-be volunteers in minutes, match them to open roles and book inductions; coordinators decide placement.
- Donor communication draftingPersonalised thank-yous and impact updates drafted from your CRM, with a fundraiser approving every batch.
Sources
- Productivity Propelled: impact of AI adoption, Deloitte Access Economics for 2degrees, 28 April 2026
- Responsible AI Guidance for Businesses, MBIE, July 2025
- Charities Act 2005, New Zealand Legislation, version as at 6 June 2026
- Incorporated Societies Register, Companies Office, accessed 19 October 2026
- Unsolicited Electronic Messages Act 2007, New Zealand Legislation, version as at 28 October 2021
- IPP 3A guidance, Office of the Privacy Commissioner, accessed 19 October 2026
- Principles of Māori Data Sovereignty, Te Mana Raraunga, October 2018
Written by the Levy Street team in Wellington. Last updated 19 October 2026.