AI Forum NZ — Generative AI Working Group A living document
Living Whitepaper
Latest

AI in Finance in Aotearoa New Zealand: A Living Whitepaper

Snapshot date: June 10, 2026
Updated from the April 2, 2026 edition

Introduction

AI adoption in New Zealand finance is still best understood as practical, supervised, and selective rather than fully autonomous. The clearest live use cases remain fraud and scam prevention, customer-service augmentation, adviser productivity, workflow automation, and the data-and-payments infrastructure that makes more advanced services possible. Since the last edition on April 2, 2026, the strongest new evidence has come from bank deployments, open-banking rollout milestones, and a sharper Reserve Bank warning that AI can amplify concentration, model, and cyber risk. (fma.govt.nz)

Executive Summary

  • The center of gravity remains defensive and assistive AI. New Zealand banks are most visibly scaling AI in fraud detection, scam controls, customer-service support, and adviser productivity rather than in fully automated consequential decisions. The FMA’s sector baseline and March 2026 advice review, together with current bank announcements, all point in that direction. (fma.govt.nz)
  • What changed after April 2 is operational proof, not a leap to autonomy. Westpac launched an AI-supported contact-centre platform on April 14, 2026; Kiwibank rolled out open banking through all digital channels for personal and business customers on May 28, 2026; and ASB launched a large SME AI-adoption programme on May 13, 2026. (westpac.co.nz)
  • Regulatory risk language has hardened. In its May 6, 2026 Financial Stability Report, RBNZ said AI adoption could amplify risk through dependence on a small number of third-party AI providers, biased or misleading outputs, AI-driven cyber threats, and even mortgage stress if AI materially affects employment in some sectors. (rbnz.govt.nz)
  • Open banking has become more real as an AI enabler. MBIE’s regime remains on track, and Kiwibank has now moved earlier than its December 2026 account-information deadline by enabling payments and data sharing through its app and internet banking in late May. That likely strengthens the permissioned data and payment rails needed for AI-enabled finance products. (mbie.govt.nz)
  • The threat side of AI is rising at the same time as the opportunity side. On April 8, 2026, the FMA warned of a significant increase in fake trading-platform scams using AI-generated deepfakes of public figures and business leaders; it said it identified 110 ads in one 24-hour period and more than 190 linked websites since early March. (fma.govt.nz)
  • The best current description of the market is “hybrid AI finance.” Human oversight remains central, but AI is moving deeper into customer-facing operations, especially advice, contact centres, fraud controls, and payments. That is an inference from the combined regulator and industry evidence. (fma.govt.nz)

What’s New Since April 2, 2026

1) RBNZ has sharpened its warning on AI concentration and cyber risk

The most important new official development since the last edition is the Reserve Bank’s May 6, 2026 Financial Stability Report. RBNZ said AI adoption could amplify financial-sector risk if firms rely on only a small number of third-party AI providers, if models generate biased or fraudulent outputs, or if more capable AI systems materially increase cyber risk. It also linked AI to a possible credit channel, noting that if AI contributes to job losses in some sectors, some borrowers may find mortgage repayments harder. (rbnz.govt.nz)

This matters because it moves the conversation beyond generic “innovation with safeguards.” The May report places AI inside core financial-stability thinking: operational resilience, third-party dependency, cyber capability, and credit quality. (rbnz.govt.nz)

2) Westpac has pushed AI further into frontline customer operations

On April 14, 2026, Westpac NZ said it had begun rolling out Microsoft Dynamics 365 Contact Centre as a Service with built-in AI. The system supports staff during calls by surfacing relevant customer-profile and product information in real time, and Westpac said it expects deployment across all contact centres by August 2026. (westpac.co.nz)

This is important because it shows a live New Zealand example of AI augmenting staff in regulated customer interactions, rather than replacing them. Westpac paired the launch with fresh customer sentiment data: in its April 7 survey, 65% of respondents were comfortable or neutral about banks using AI to support contact-centre staff, and 70% were comfortable or neutral about AI being used to detect fraud and scams. (westpac.co.nz)

3) Kiwibank has turned open banking from timetable into operating reality

On May 28, 2026, Kiwibank said it became the first New Zealand bank to roll out open banking through all its digital channels for both individual and business customers. It said customers could access open-banking-enabled payments and data sharing from that date through internet banking and the Kiwibank app, and that this delivered data-sharing capability about six months ahead of Kiwibank’s December 2026 regulatory deadline. (kiwibank.co.nz)

For AI in finance, the significance is indirect but material: open banking improves access to consented financial data and action rails, which are prerequisites for higher-value AI use cases in payments, cashflow management, personalisation, and embedded finance. That is an inference from the regulatory and bank rollout evidence. (kiwibank.co.nz)

4) The FMA is now dealing with AI as a market-abuse vector, not just an innovation tool

On April 8, 2026, the FMA warned of an increasing number of fake trading-platform scams using fake news articles and AI-generated deepfakes of politicians and business leaders. It said it had identified 110 linked ads in one 24-hour period and more than 190 related fake websites since early March 2026. (fma.govt.nz)

This is a crucial update because it reinforces that AI in New Zealand finance is no longer just about internal productivity or better customer experience. It is also driving a faster, more convincing fraud environment, which helps explain why banks continue to prioritise scam detection and authentication controls. (fma.govt.nz)

5) Banks are starting to act as AI-enablement partners for the wider economy

On May 13, 2026, ASB launched Pathway to Productivity, combining an AI bootcamp co-developed with Xero, an emerging-talent placement programme, and advisory support. ASB said the programme aimed to support more than 4,100 New Zealand businesses in its first year, with an AI bootcamp designed for 4,000 SMEs. (asb.co.nz)

This is not internal bank AI adoption in the narrow sense, but it is strategically relevant. It suggests some New Zealand financial institutions are positioning themselves not only as users of AI, but as distributors of AI capability into the business economy. (asb.co.nz)

Current State of AI Adoption in New Zealand Finance

Sector-wide position

The best broad baseline remains the FMA’s September 10, 2024 research on AI in financial services. It found that all 13 participating firms across banking, insurance, asset management, and financial advice either already used generative AI or planned to adopt it soon, with activity concentrated in fraud detection, risk management, decision support, product development, and customer-service-related opportunities. (fma.govt.nz)

The strongest updated evidence on customer-facing scaling still comes from the FMA’s March 25, 2026 access-to-advice review. It found that 40 Financial Advice Providers had a digital advice facility and that about 165,000 retail clients had received digital advice in the prior year that resulted in acquiring a financial product. The FMA said the predominant use of AI is to support advisers rather than replace them. (fma.govt.nz)

Taken together, the evidence suggests New Zealand finance has moved beyond experimentation, but not into broadly autonomous decisioning. The clearest operating model is still human-led delivery with AI support embedded into workflows, detection systems, and service tools. This is an inference from the combined FMA, RBNZ, bank, and payments evidence. (fma.govt.nz)

Where adoption is currently strongest

  • Fraud, scams, and cyber defence: ASB says it uses advanced AI fraud monitoring to detect unusual behaviour and potential fraudulent activity, while ANZ’s cross-sector phishing-disruption effort reported more than 5,000 malicious domains disrupted and a 39% reduction in ANZ customer card-phishing cases over two months. (asb.co.nz)
  • Customer-service augmentation: Westpac’s April launch shows AI being used inside frontline service operations to improve response quality and speed while keeping humans in the conversation. (westpac.co.nz)
  • Financial advice productivity and hybrid advice models: The FMA’s March review shows digital advice is now material in scale, and vendor-published case evidence from Deloitte says NZHL used generative AI to reduce annual client-review preparation time by 80%. (fma.govt.nz)
  • Payments and data-sharing infrastructure: Kiwibank’s late-May rollout and the broader API Centre implementation framework indicate that open banking is maturing into a real operating layer for future AI-enabled propositions. (kiwibank.co.nz)

Recent News and Market Developments

Advice and wealth

Financial advice remains the most clearly documented scaling channel for AI in New Zealand finance. The FMA says digital tools, hybrid models, and AI-supported advice processes could improve scalability and consistency if implemented with the right design and oversight, and it has signalled further work on AI use in advice. (fma.govt.nz)

A useful practical example comes from NZHL. In a Deloitte New Zealand case study published on January 22, 2026, Deloitte said NZHL used generative AI and Snowflake’s data platform to transform its annual client-review process, cutting preparation time by 80%. Because this is a vendor-published case study, it should be read as directional evidence rather than independent evaluation. (deloitte.com)

Banking and customer operations

Westpac’s April launch is the clearest recent sign that AI is moving deeper into live banking operations. The model is deliberately conservative: AI prepares and surfaces information, while staff remain the accountable customer-facing decision-makers. (westpac.co.nz)

Westpac has also signalled, through changes to customer notices, that it may use automated tools including AI to support risk management, system security, and service improvement, with governance, safeguards, and human oversight. That suggests AI is becoming part of operating architecture, not just isolated pilots. (westpac.co.nz)

Fraud, scams, and trust

The fraud story has become even more central since April. The FMA’s April warning on deepfake investment scams, Kiwibank’s public warning about scams using the image of CEO Steve Jurkovich, and Westpac’s criticism of platform inaction after the FMA alert all show that AI-enabled fraud is becoming a major market conduct problem. (fma.govt.nz)

Banks are responding with layered controls. ASB highlights advanced AI fraud monitoring, payment-interruption prompts, and Caller Check verification; Kiwibank highlights call verification, multi-factor authentication, and scam support; and ANZ continues to emphasise ecosystem intelligence-sharing to disrupt phishing campaigns earlier. (asb.co.nz)

Payments and open banking

The open-banking stack is becoming more concrete. MBIE’s regulations designate ANZ, ASB, BNZ, and Westpac from December 1, 2025, with Kiwibank designated for payments from June 1, 2026 and account information from December 1, 2026. (mbie.govt.nz)

The API Centre says four banks are live with key open-banking standards covering over 80% of customer accounts, and that once all five major banks are live the framework will cover more than 90% of consumer accounts. It has also previously reported more than 100,000 unique customers using open-banking services and more than 180,000 payments in October 2025. (apicentre.paymentsnz.co.nz)

That does not itself equal AI adoption, but it materially improves the infrastructure for AI-enabled payments, account aggregation, consented data analysis, and workflow automation. This is an inference from the payments and open-banking evidence. (apicentre.paymentsnz.co.nz)

Research and Policy Overview

Core public evidence base

The current public evidence stack for AI in New Zealand finance is still anchored by four sources:

  • FMA, September 2024: sector baseline on AI in financial services. (fma.govt.nz)
  • FMA, March 25, 2026: strongest current evidence on digital advice, hybrid advice, and AI-supported adviser workflows. (fma.govt.nz)
  • RBNZ, May 6, 2026: latest official statement linking AI to concentration, cyber, model, and credit risks in financial stability. (rbnz.govt.nz)
  • MBIE/API Centre/Kiwibank, 2025–2026: open-banking rules, implementation milestones, and bank rollout progress that create the data-sharing and payment-initiation layer for future AI services. (mbie.govt.nz)

What the research now says

The picture is increasingly coherent:

  • AI is real in production, but mostly in bounded use cases. That is clear from fraud systems, advice workflows, and service-assist tools. (asb.co.nz)
  • Human accountability remains the dominant design principle. FMA’s advice findings and Westpac’s service model both point to AI as augmentation rather than replacement. (fma.govt.nz)
  • Infrastructure matters as much as models. Open banking, authentication, consent, and payment standards are becoming critical enablers of the next wave of finance AI. (apicentre.paymentsnz.co.nz)
  • Risk management is becoming more explicit and systemic. RBNZ now treats AI not simply as a firm-level innovation issue, but as part of third-party dependency, cyber resilience, and credit-risk thinking. (rbnz.govt.nz)

Case Studies

Case Study 1: Westpac NZ’s AI-supported contact centre

What happened: On April 14, 2026, Westpac NZ announced rollout of Microsoft’s Dynamics 365 Contact Centre as a Service with built-in AI. The system supports staff in real time with relevant customer and product information, and Westpac expects full deployment across contact centres by August 2026. (westpac.co.nz)

Why it matters:

  • It is one of the clearest public examples of AI entering a live, regulated customer-service workflow in New Zealand banking. (westpac.co.nz)
  • The design keeps humans in the loop, which aligns with the broader New Zealand pattern of cautious, supervised AI adoption. (westpac.co.nz)

Case Study 2: Kiwibank’s all-channel open-banking rollout

What happened: On May 28, 2026, Kiwibank said it became the first bank in New Zealand to make open-banking-enabled payments and data sharing available through both internet banking and its app for personal and business customers. (kiwibank.co.nz)

Why it matters:

  • It moves open banking from policy timeline to customer-usable infrastructure. (kiwibank.co.nz)
  • For AI, it expands the consented data and payment rails needed for more intelligent financial experiences. That is an inference, but a strong one. (kiwibank.co.nz)

Case Study 3: ANZ’s anti-phishing intelligence coalition

What happened: ANZ’s January 2026 update said its collaboration with telcos, platforms, and other firms had disrupted more than 5,000 phishing domains in two months and helped reduce ANZ customer card-phishing cases by 39% over the same period. (anz.com.au)

Why it matters:

  • It remains the strongest public operational proof point for AI-adjacent defensive systems in New Zealand finance. (anz.com.au)
  • It also shows that the most effective finance AI deployments are often ecosystem deployments, not just internal bank tools. (anz.com.au)

Case Study 4: NZHL’s adviser workflow automation

What happened: In a case study published by Deloitte New Zealand on January 22, 2026, Deloitte said NZHL used generative AI and a governed data foundation to improve its annual client-review process, reducing preparation time by 80%. (deloitte.com)

Why it matters:

  • It illustrates the most plausible next step for AI in New Zealand advice: not adviser replacement, but major compression of admin and preparation time. (deloitte.com)
  • It supports the FMA’s broader finding that AI in advice is predominantly an adviser-capacity multiplier. (fma.govt.nz)

Trend 1: Defensive AI still leads the market

Fraud detection, scam interruption, behavioural monitoring, phishing disruption, and identity verification remain the strongest proven use cases. This is consistent across ANZ, ASB, Kiwibank, Westpac, and FMA scam warnings. (anz.com.au)

Trend 2: Customer-facing AI is expanding, but only in supervised forms

The recent Westpac rollout and the FMA’s advice findings both point to a human-plus-AI operating model. New Zealand finance appears more willing to scale AI when accountability remains clearly with staff or licensed advisers. (westpac.co.nz)

Trend 3: Open banking is becoming the enabling layer for the next wave

With Kiwibank’s rollout, API Centre progress, and MBIE’s regulatory timetable, open banking is increasingly an operating substrate rather than a future concept. That likely matters more for the next phase of finance AI than any single model announcement. This is an inference from the infrastructure evidence. (kiwibank.co.nz)

Trend 4: Trust is conditional, not absent

Recent Westpac survey results show many New Zealanders are comfortable or neutral about AI being used to help staff or prevent fraud, but the framing matters: support, speed, and safety perform better than replacement or opaque autonomy. (westpac.co.nz)

Trend 5: Governance is becoming product architecture

RBNZ’s May 2026 language, Payments NZ’s resilience framing, and bank terms updates all point the same way: identity, authority, accountability, privacy, security, and third-party controls are becoming built-in design requirements for AI-enabled finance products. (rbnz.govt.nz)

Constraints and Risks

Conduct and consumer risk

The FMA’s March advice work supports innovation, but only with oversight and consumer-outcome discipline. Meanwhile, the FMA’s April deepfake warning shows how quickly AI can be used to create misleading financial promotions and fraudulent investment journeys. (fma.govt.nz)

Third-party and concentration risk

RBNZ has now made this explicit: dependence on a small number of third-party AI providers can create new operational and model risks. In a small market like New Zealand, where many core technology layers are externally sourced, that risk is likely to stay prominent. (rbnz.govt.nz)

Cyber amplification risk

RBNZ said increasingly capable AI systems could materially amplify cyber risks from malicious actors, and Payments NZ has also been discussing the need for clearer identity and accountability as AI-enabled systems increasingly act on behalf of users. (rbnz.govt.nz)

Open banking increases opportunity, but only if customers trust how data is accessed, shared, and governed. The API Centre continues to emphasise security, consent, and operational standards, which suggests the market understands that AI value in finance depends on trusted data governance. (apicentre.paymentsnz.co.nz)

Overall Assessment

As of June 10, 2026, AI in New Zealand finance is best described as operationally established in narrow domains, increasingly visible in customer-facing support, and still tightly constrained by trust, governance, and infrastructure discipline. The market has advanced since the April 2 edition, but the change is mostly in proof points: Westpac has moved AI into frontline service operations, Kiwibank has accelerated the open-banking layer, ASB is pushing AI enablement into the SME economy, and RBNZ has made the risk side of AI more explicit. (westpac.co.nz)

The dominant New Zealand model remains AI as copilot, detector, summariser, recommender, and workflow engine. The strongest near-term winners are likely to be firms that combine AI with secure data access, high-confidence authentication, human accountability, and strong third-party controls, rather than those pursuing autonomy for its own sake. That conclusion is an inference from the current regulator, bank, and payments evidence base. (fma.govt.nz)

← Back to all articles View as Markdown