When fraud becomes a credit risk problem
New Zealand Credit Risk and Fraud Summit insights series – by Richard Atkinson, Head of Fraud and ID, Experian
Evolving fraud tactics are increasingly challenging the integrity of credit risk decisions, affecting who is being assessed, what evidence is being relied on, and whether the applicant’s apparent capacity and intent are genuine.
We’re seeing AI accelerate familiar fraud methods, making it easier for bad actors to create synthetic identities, manipulate documents and fabricate supporting evidence at scale. For credit-risk leaders, the core consideration is how and when additional checks are integrated into the credit risk decision.
At the New Zealand Credit Risk and Fraud Summit, I joined the panel discussion, “When fraud becomes a credit risk problem.” The discussion centred on what these developments mean for New Zealand lenders: which identity evidence requires stronger testing, which controls may belong at each point in the credit risk journey, and how different signals can inform the outcome of an application.
The impact of different fraud types on credit risk decisions
Fraud affects credit risk in different ways depending on the question it compromises.
In a third-party fraud scenario, the lender may be assessing an application controlled by someone other than the person connected to the identity, credentials or documents being presented.
With first-party misrepresentation, the applicant may be genuine while the evidence gives a misleading view of repayment capacity, financial stability or intent. An overstated salary can influence affordability assessment; a concealed commitment may affect perceived serviceability; and a manipulated bank statement can change the interpretation of cash flow or financial pressure.
Alongside first- and third-party fraud, synthetic fraud is becoming a more prominent risk consideration for lenders. A synthetic identity may combine real identifiers, fabricated attributes and manufactured behavioural signals to create a profile capable of entering ordinary credit workflows. As the profile accumulates surface legitimacy, the credit assessment may need to consider whether the identity and surrounding evidence have developed in a way that is consistent with a genuine applicant.
The practical question is how to identify these risks early enough, distinguish between them clearly enough and respond in a way that protects the credit decision without applying the same approach to every application.
How GenAI raises the bar for identity verification
As generative AI becomes more widely used and its outputs become more sophisticated, plausible-looking application evidence is becoming easier to produce and harder to assess at scale. Documents, images, identity artefacts and supporting material can be created or altered quickly, then submitted through digital application channels.
Once plausible-appearing evidence enters a workflow, it can influence income checks, verification, affordability or review. A generated payslip may support an income claim; an altered bank statement may change the apparent pattern of expenses or financial pressure; and fabricated supporting material can make a marginal application appear more complete, stable or consistent than the underlying facts support.
Experian’s Fraud in the Age of AI reportshows the pressure facing organisations. Sixty-four per cent of businesses report a surge in fraud-related losses over the past year, while 61% of fraud decision-makers identify AI-driven fraud as the greatest future threat. In New Zealand, 58% of fraud decision-makers say existing KYC and identity verification checks are underprepared for GenAI-generated documents.
Document checks now need to examine more than the information shown on the page. Advanced tools can inspect how a file has been built, whether it has been changed and whether its structure is consistent with the document being presented. Resistant AI’sdocument fraud detection capability, for example, is designed to examine document or image structure, composition and context. Those outputs can then be considered within the credit decisioning flow to help determine the next step: accept the evidence, ask for clarification, route the application to review or consider the document alongside other risk signals.
Minimising friction for genuine applicants
Strong fraud defence can lose commercial value when genuine customers are asked to complete additional verification steps that are not clearly targeted to risk. Credit providers need controls that test identity evidence adequately without applying the same heavy journey to every applicant.
Experian’s Fraud in the Age of AI report found that 52% of consumers have abandoned an online sign-up or verification process when it felt too intrusive, while 90% of consumers across Australia and New Zealand are concerned about someone stealing their identity and using it to commit fraud online.
Sequencing gives lenders a practical way to test evidence while keeping additional verification targeted. At the Application stage of the journey, identity and entitlement checks can establish whether the identity document, application details and person completing the application journey are consistent. Behavioural and journey signals can help identify patterns that may point to synthetic identity development, engineered application conduct or concerns about repayment intent. During assessment, income, employment, document and affordability evidence can be tested against the claims made in the application. The final credit decision is then made on a combination of all the factors and data points uncovered throughout the entire application journey.
With that sequence in place, a lower-risk applicant may move with limited interruption, while an application with inconsistent income evidence can receive a targeted step-up. Where document concerns, device anomalies and unusual behaviour appear together, the journey can move to review or decline according to policy and risk appetite.
Key takeaways for New Zealand credit risk and fraud leaders
One point that stood out from the panel was the rapid acceleration
in the use and availability of AI tools by both individuals to facilitate first-party fraud and by professional fraud groups conducting large-scale third-party fraud. Already, the technologies in use have seemingly made human detection of fraud almost impossible, and with the rapid advancement that is occurring in AI capabilities, many lenders will need to consider how advanced detection technologies can support their fraud defence strategies.
Building on this, the discussion highlighted a shift from viewing fraud detection in isolation to understanding how it connects with credit risk decisions. It explored how identity, document and behavioural signals can be embedded into key decisioning moments to inform application outcomes, while keeping the journey workable for genuine customers.
Viewed through that operational lens, two takeaways stand out for New Zealand credit risk and fraud leaders.
First, consider whether existing mechanisms are aligned to current and emerging fraud risks. Identity verification, assessment of document authenticity, device intelligence, behavioural analytics, income consistency checks and fraud detection capability can be assessed against third-party impersonation, synthetic identity activity, first-party misrepresentation and AI-generated evidence. The central question is whether the organisation can test the evidence most likely to influence approval, referral, limit-setting or decline.
Second, consider how the control sequence aligns with the decisioning flow.Fraud and identity signals are most useful when they enter the workflow at points where they can inform the next decisioning step, support explainability and help determine whether an application should proceed, receive further verification, move to specialist review or be declined.
To discuss how Experian supports fraud, identity and credit decisioning across digital onboarding and credit assessment, please get in contact through the form below.