AI in Finance in Aotearoa New Zealand: A Living Whitepaper
Snapshot date: March 9, 2026
Introduction
Artificial intelligence is now firmly on the agenda across New Zealand finance, but adoption is uneven. In banking, insurance, payments, wealth, and financial advice, the dominant pattern is not full autonomous decision-making; it is targeted deployment in fraud detection, customer service, productivity tooling, data analysis, and selected decision-support workflows. Regulators are encouraging innovation, but they are also signalling that AI must sit inside strong governance, cyber resilience, and human oversight frameworks. (fma.govt.nz)
Executive Summary
- AI adoption is real, but still maturing. The FMA’s September 2024 sector research found all 13 participating financial firms either had integrated generative AI or planned to do so soon; by November 2025, RBNZ still described AI use across the financial sector as “exploratory,” but accelerating. (fma.govt.nz)
- The leading use cases are operational, defensive, and customer-facing. Current deployments cluster around fraud detection, risk management, customer communications, cybersecurity, and internal productivity rather than fully automated high-stakes decisions. (fma.govt.nz)
- Open banking is becoming an AI enabler. The Customer and Product Data Act 2025 and the rollout of Payments NZ API standards are creating a better data and payments layer for AI-enabled financial products and services. (apicentre.paymentsnz.co.nz)
- Fraud and scam prevention is the most visible production use case. Banks, card networks, telcos, and platforms are now using shared intelligence and AI-driven detection to block phishing, mule-account activity, and payment fraud at increasing scale. (visa.co.nz)
- Governance is becoming a strategic differentiator. RBNZ, FMA, KPMG, and the Commerce Commission all point to the same pressure points: data quality, explainability, third-party concentration, cyber risk, bias, and the need for reviewable human-controlled decisions. (fma.govt.nz)
- Public trust remains a constraint. New Zealand’s broader AI environment shows rapid uptake but low trust and weak literacy, which matters acutely in finance, where products are high-stakes and heavily dependent on customer confidence. (datacom.com)
The Current State of AI Adoption in New Zealand Finance
Sector-wide position
The clearest official baseline remains the FMA’s Understanding Artificial Intelligence in Financial Services report, published on September 10, 2024. It drew on 13 respondents across asset management, banking, financial advice, and insurance, with responses from six deposit takers, three insurers, and four financial advice or asset management firms. The report found AI adoption was already present across the sample and that every participant either had integrated generative AI or planned to do so soon. (fma.govt.nz)
The FMA research showed current adoption concentrated in fraud detection, risk management, decision-making support, product development, and product management. Looking forward, the most commonly planned areas for new AI use were customer service, risk management, fraud detection, and sales and marketing. All 13 respondents said they wanted to invest in customer service use cases. (fma.govt.nz)
At the same time, the FMA’s findings point to a deliberately conservative operating posture. Respondents reported a cautious, risk-first approach, emphasising controlled deployment, staff training, model validation, pilots, and governance rather than unrestricted rollout. Human oversight was explicitly described as part of current control design. (fma.govt.nz)
RBNZ’s May 2025 and November 2025 Financial Stability Reports reinforce that picture. The May 2025 special topic argued AI could materially improve productivity, model accuracy, risk assessment, fraud detection, and cyber resilience across banking, insurance, and non-bank financial institutions, while also introducing risks tied to model error, bias, privacy, cyber attack, disinformation, and concentration in a small number of providers. By November 2025, RBNZ said AI use across the sector remained exploratory and limited in scope, but that adoption was accelerating. (rbnz.govt.nz)
National context that shapes finance
Finance is adopting AI inside a wider New Zealand AI environment that is moving quickly but has not yet scaled evenly. Datacom’s 2025 State of AI Index found 87% of surveyed New Zealand organisations were using some form of AI, up from 66% in 2024 and 48% in 2023, while only 12% had scaled AI across the whole business. The same report found 88% of AI-using organisations reported positive operational impact. (datacom.com)
Government policy is also now more explicit. MBIE’s New Zealand’s AI Strategy: Investing with confidence, released in July 2025, deliberately emphasises adoption and application rather than frontier model development. In practice, that orientation fits finance, where most value is likely to come from embedding AI into existing workflows, data environments, and customer journeys rather than building foundational models locally. (mbie.govt.nz)
The trust environment is less favourable. KPMG’s 2025 New Zealand AI trust snapshot found 81% of New Zealanders believe AI regulation is required, only 44% believe AI’s benefits outweigh its risks, and only 24% report having undertaken AI-related training or education. For financial services, that combination implies adoption can grow, but customer-facing scale will depend on visible governance and clear consumer protections. (kpmg.com)
Recent News and Market Developments
1) Open banking is becoming the platform layer for smarter financial services
Open banking has moved from concept to implementation. Payments NZ’s API Centre said the Customer and Product Data Act 2025 came into force on March 30, 2025, and that by May 30, 2025 the four largest banks—ANZ, ASB, BNZ, and Westpac NZ—were due to implement version 2.3 of the payment initiation standard, with Kiwibank on track for May 2026. The new standard includes enduring payment consent and decoupled authentication, both important capabilities for more intelligent, embedded, and AI-assisted payment journeys. (apicentre.paymentsnz.co.nz)
By October 2025, Payments NZ reported that more than 100,000 unique customers used open banking services in a single month, completing over 180,000 payments. That is significant because AI in finance is often constrained less by model capability than by secure access to bank-grade data and action rails; open banking improves both. (apicentre.paymentsnz.co.nz)
2) Agentic commerce has arrived in the market narrative
On February 17, 2026, Westpac NZ announced that Mastercard had completed New Zealand’s first authenticated agentic transactions with Westpac. While early-stage, this is an important signal: the sector is starting to experiment not just with AI for analysis or communication, but with AI-mediated commerce and transaction initiation. (westpac.co.nz)
This development matters strategically because it suggests the next phase of AI in finance may move from copilots and analytics into action-taking systems. In New Zealand, that transition is likely to happen first in controlled payment and commerce contexts rather than in fully autonomous lending or wealth decisions. That is an inference from the market pattern and the regulatory emphasis on reviewability and safeguards. (westpac.co.nz)
3) Scam and fraud prevention is the fastest-scaling production use case
Fraud prevention is where AI has the clearest operational evidence today. Visa reported in December 2024 that its AI-based Visa Advanced Authorisation system helped New Zealand financial institutions prevent NZ$273 million in fraud from disrupting New Zealand businesses over a year, and that it assesses more than 500 risk attributes in about a millisecond to score transactions in real time. (visa.co.nz)
The banking sector has also moved toward shared intelligence. On November 17, 2025, the New Zealand Banking Association said retail banks had implemented new fraud intelligence technology to identify mule accounts faster, share intelligence across institutions, freeze funds more quickly, and warn customers before payments to high-risk accounts are made. NZBA said thousands of intelligence-sharing cases had already contributed to the recovery or prevention of millions of dollars in losses. (nzba.org.nz)
On January 15, 2026, ANZ NZ said a pilot with 2degrees had expanded to include Kiwibank, TSB, One NZ, Spark, and Trade Me to detect and disrupt malicious phishing domains within the first hour of a webpage going live. ANZ said the collaboration had already disrupted more than 5,000 domains in two months and reduced ANZ customer card-phishing cases by 39% over the same period. (anz.com.au)
ASB’s February 2026 half-year results point to the same direction of travel. ASB said it was now able to share data between banks related to digital fraud and money-mule activity through the Fraud Reporting Exchange and New Zealand Data Exchange, while continuing to invest in modernised customer journeys. (asb.co.nz)
Research Overview
Official regulator and policy research
The FMA’s 2024 study remains the most directly relevant public research on New Zealand financial-sector AI use. It found the main adoption motivations were improved customer outcomes and operational efficiency (both cited by 9 respondents), followed by fraud detection (6), risk management (5), and improved data analytics (4). The report also found current tools ranged from off-the-shelf systems such as Microsoft Copilot and GitHub to machine-learning and security tools used for fraud detection, pricing, personalisation, and cyber defence. (fma.govt.nz)
The same report identified the top risks as staff training requirements, cybersecurity concerns, data privacy, lack of transparency, regulatory compliance challenges, bias and fairness, and potential discrimination. Most respondents said governance and oversight were integral to an AI risk framework, and more advanced firms were already adapting existing governance structures or creating AI-specific frameworks. (fma.govt.nz)
RBNZ’s May 2025 AI special topic placed these firm-level findings into a systemic context. It highlighted benefits such as more accurate modelling, better threat detection, improved productivity, and more personalised customer offerings, but warned about systemic vulnerabilities from shared models, concentration in a small set of critical AI providers, AI-enabled cyber threats, and the possibility of deepfakes and impersonation increasing financial losses for customers. (rbnz.govt.nz)
RBNZ’s November 2025 Financial Stability Report sharpened the operational message: AI adoption remains early, but third-party risk management, data controls, and human oversight are now central. It explicitly stated that material AI-enabled decisions must be reviewable and subject to effective controls. (rbnz.govt.nz)
Broader New Zealand AI research relevant to finance
Datacom’s 2025 State of AI Index is not finance-specific, but it is highly relevant because it describes the operating environment in which financial institutions hire, procure, and compete. Its finding that adoption is broad but enterprise-scale deployment remains limited mirrors what the finance-specific regulator evidence shows: the sector is moving from experimentation into selective scaling rather than wholesale transformation. That is an inference supported by the overlap between Datacom’s maturity data and RBNZ/FMA sector findings. (datacom.com)
The KPMG and University of Melbourne 2025 trust study is particularly important for finance because trust is a core industry asset. New Zealand ranked among the lowest globally on AI optimism and acceptance, and only 23% of respondents believed current safeguards were sufficient to make AI use safe. In financial services, those sentiment numbers help explain why firms are emphasising explainability, internal controls, and staged rollouts instead of aggressive automation. (kpmg.com)
Research on financial advice and wealth
FPSB’s 2025 global research found two in three financial planning firms are using AI or planning to within 12 months, with the most common uses being client communications (41%), client data collection (33%), and client risk profiling (30%). It also found widespread optimism that AI can improve service quality, reduce cost, and broaden access to advice. (fpsb.org)
New Zealand-specific reporting based on that work indicates local advisers are broadly aligned with the global pattern. Riskinfo NZ, citing Financial Advice NZ and FPSB research, reported in May 2025 that two in three Kiwi advisers were already using AI or planning to do so within the next year, with strong expectations that AI could improve service, advice quality, cost, and access. (riskinfonz.co.nz)
Case Studies
Case Study 1: AI-led fraud defence is moving from bank-by-bank controls to ecosystem coordination
The strongest New Zealand case study is not a single product but a new operating model. NZ banks are now combining institution-level AI and analytics with cross-bank information sharing and cross-sector partnerships. Visa reported card-network-level fraud prevention outcomes; NZBA reported bank-to-bank fraud intelligence sharing; and ANZ reported a live collaboration with telcos and digital platforms that disrupted more than 5,000 phishing domains in two months and cut ANZ card-phishing cases by 39%. (visa.co.nz)
Why it matters
- It shows AI in New Zealand finance is already delivering value in production where use cases are measurable and defensible. (visa.co.nz)
- It also shows that the best AI outcomes in finance may come from shared infrastructure and shared intelligence, not only from internal model development. (nzba.org.nz)
Case Study 2: ASB is using AI both internally and as a customer-enablement tool
On August 27, 2025, ASB launched a programme with the New Zealand Product Accelerator to place AI, data science, and business analytics students into ASB business customers to solve real-world productivity problems. The pilot started with 13 students and was designed to scale toward around 100 placements and internships each semester. In its February 2026 results, ASB said the programme was being scaled to match up to 100 ASB business customers with students. (asb.co.nz)
This is notable because it expands the role of a bank from AI adopter to AI diffuser. ASB is not only using AI to modernise its own services and scam controls; it is positioning AI as a productivity lever for the broader business economy. (asb.co.nz)
Case Study 3: Southern Cross Travel Insurance shows cautious automation in claims
A published transformation case study on Southern Cross Travel Insurance said the insurer’s new claims platform automates customer communications and task allocation, and that 35% of claims can now be safely processed automatically under robust controls. The same case study says SCTI plans to leverage AI further, but cautiously and with customer benefit in mind. Because this is a vendor-published case study, the claims should be read as company-reported outcomes rather than independent benchmarking. (equinox.co.nz)
Even with that caveat, it is a useful marker of where insurance AI is heading in New Zealand: claims automation, triage, workflow orchestration, and customer communication first; broader AI expansion later. (equinox.co.nz)
Case Study 4: Westpac and Mastercard signal the next frontier—AI that can transact
Westpac’s February 2026 announcement of New Zealand’s first authenticated agentic transactions with Mastercard is still early-stage, but it is strategically important. It suggests the market is beginning to test AI that not only informs customer choices but can also participate in authenticated commercial actions. (westpac.co.nz)
The likely implication is that New Zealand finance will see a gradual progression:
- AI for internal productivity and fraud prevention,
- AI for customer communication and decision support,
- AI for consented action initiation in tightly controlled payment and commerce contexts. This sequencing is an inference from the current adoption record and the regulatory focus on control, reviewability, and safe architecture. (fma.govt.nz)
Key Trends
Trend 1: Defensive AI is outpacing offensive AI
In New Zealand finance, AI that protects customers and institutions is scaling faster than AI that replaces judgment. Fraud detection, cyber defence, anomaly detection, and scam prevention have clearer return on investment, lower conduct risk, and stronger internal support than fully automated underwriting, pricing, or advice. (fma.govt.nz)
Trend 2: Customer service is the biggest near-term expansion zone
The FMA found customer service was the top planned future AI application among respondents, and adviser research similarly shows strong use of AI in client communications and data gathering. This suggests the next broad wave of finance AI in New Zealand will likely appear in service operations, onboarding, knowledge retrieval, and workflow support rather than in headline-grabbing autonomous finance products. (fma.govt.nz)
Trend 3: Open banking is turning data access into a competitive variable
As open banking standards mature, firms will have better infrastructure for intelligent payments, personal financial management, and embedded financial journeys. That does not guarantee AI success, but it reduces one of the historical bottlenecks: access to standardised, permissioned data and payment initiation. (apicentre.paymentsnz.co.nz)
Trend 4: Governance is moving from compliance overhead to core capability
FMA respondents reported investments in staff training, AI-specific governance, pilots, and validation. RBNZ has added systemic emphasis on third-party risk, human oversight, and data controls. The sector is converging on a view that AI governance is not a separate policy exercise; it is part of mainstream operational resilience and conduct management. (fma.govt.nz)
Trend 5: Third-party concentration is a material strategic risk
RBNZ has repeatedly highlighted dependence on external providers as a financial stability issue, and the Commerce Commission has separately warned about concentration across cloud, models, and data. For New Zealand finance, which often imports technology rather than builds it, vendor concentration is likely to remain one of the most important structural AI issues over the next 24 months. (rbnz.govt.nz)
Constraints and Risks
Regulatory and conduct risk
New Zealand still does not have finance-specific AI rules, but that does not mean firms are operating in a vacuum. The FMA has framed its 2024 work as research rather than guidance, while RBNZ has made clear that existing obligations around risk management, cyber resilience, governance, and prudential oversight still apply. In effect, firms are expected to manage AI within current regulatory settings while the technology evolves. (fma.govt.nz)
Trust and consumer risk
The trust gap is especially important in finance. KPMG’s 2025 trust data shows low public confidence, and the Commerce Commission has warned that AI can intensify misleading content, deepfakes, manipulative nudging, and consumer imbalance. Financial scams are an obvious example of this risk already playing out. (kpmg.com)
Data quality, explainability, and human review
RBNZ’s November 2025 report states that AI tools perform only as well as the data they are trained on and the prompts they receive, and that material AI-enabled decisions must be reviewable. The FMA research points in the same direction, with respondents focusing on training, oversight, and validation. This means any move toward AI in underwriting, pricing, or advice will be gated by explainability and auditability, not just by technical feasibility. (rbnz.govt.nz)
Overall Assessment
As of March 9, 2026, AI in New Zealand finance is best understood as selectively operational, strategically important, and still governance-constrained. The sector has clearly moved beyond curiosity: regulators have surveyed live use cases, open banking rails are expanding, banks are deploying shared anti-scam technologies, insurers are automating parts of claims, advisers are beginning to embed AI into practice, and early agentic commerce experiments have appeared. (fma.govt.nz)
But the sector has not yet reached broad enterprise-scale transformation. RBNZ still describes most adoption as exploratory, Datacom’s national data shows scale remains difficult, and public trust in AI remains weak. That combination suggests the near-term winners will be firms that treat AI not as a standalone innovation project, but as part of a larger operating model built on secure data, open banking connectivity, cyber resilience, human accountability, and careful customer trust design. (rbnz.govt.nz)
In short: New Zealand finance has entered the implementation era of AI, but it is doing so pragmatically. The strongest momentum is in fraud prevention, service operations, and data-enabled payments; the next frontier is likely to be controlled action-taking systems built on open banking and authenticated consent. The pace of change is increasing, but so is the importance of governance. (visa.co.nz)