ANZ says customer scam losses across two Australian banking portfolios fell by about 24% from October 2025 through June 2026 compared with the same period a year earlier, even as scammers increased their activity and changed tactics.
The bank also reported preventing and recovering more than A$100 million in scam- and fraud-related funds during the nine-month period. It attributed the results to investment in prevention, detection, disruption and customer education.
Those figures indicate movement in the right direction, but ANZ’s disclosure does not provide the absolute amount customers lost, the prior-period loss total, the number of scam cases or a quantified measure of the claimed increase in scam activity. It therefore does not allow an outside reader to calculate losses per case, compare the bank directly with peers or determine how much of the A$100 million was prevented before transfer rather than recovered after funds moved.
What ANZ reported
ANZ said its First Responders team handled nearly 170,000 calls and resolved 94% of scam-related calls at first contact during the reporting period. The bank did not disclose how it defines resolution, how many calls concerned an actual loss or how the call total maps to unique customers and cases.
It also reported that customers abandoned more than 476,000 payments after receiving a Confirmation of Payee mismatch warning. That is an important intervention count, but it is not equivalent to 476,000 prevented scams. A mismatch can arise for legitimate reasons, and ANZ did not disclose how many abandoned payments were later confirmed as fraudulent, retried with corrected details or completed through another route.
The bank said goods-and-services scams, including online-shopping traps, were the most common scam type experienced by its customers. Non-bank impersonation scams ranked next, followed by investment, romance and friendship, and bank-impersonation scams. The release did not provide case counts or losses for any category.
The A$100 million figure combines different outcomes
ANZ described more than A$100 million as scam- and fraud-related funds that it prevented and recovered. Those terms cover materially different control outcomes.
Prevention can include blocking or interrupting a payment before value reaches a recipient. Recovery generally happens after funds have moved and may depend on fast customer reporting, cooperation from the receiving institution and whether the money remains available. Scam and fraud are also not interchangeable: a scam can involve a customer being manipulated into authorising a payment, while other fraud may involve an unauthorised transaction or account compromise.
Combining these categories produces a large headline number but limits operational comparison. A more accountable disclosure would separate scam from other fraud, prevention from recovery, attempted from completed payments, and gross funds at risk from actual customer losses. It would also explain whether recovered funds were returned to customers and how quickly intervention occurred.
Confirmation of Payee is generating friction at scale
The 476,000 abandoned payments show that beneficiary-name checking is influencing customer behaviour at a significant scale. Confirmation of Payee compares entered account details with information held for the intended recipient and alerts the payer when the result does not match as expected.
For payment providers, the control challenge is to distinguish useful friction from noise. If warnings are too broad, customers can become accustomed to overriding them. If they are too narrow, plausible beneficiary details may still pass while a customer is being manipulated. Abandonment is therefore an intermediate signal rather than a complete measure of fraud prevention.
Useful performance reporting would track warning outcomes: how often customers correct an error, proceed after verification, abandon a suspected scam, report the beneficiary account or later suffer a loss. That feedback can help institutions calibrate matching thresholds, warning language and escalation paths.
Loss reduction does not remove the accountability questions
ANZ’s result covers what the release describes as ANZ Australia Classic and Plus scam data from October 2025 through June 2026. It does not state whether the measurement includes every Australian customer, every product or losses reimbursed by the bank. It also does not provide a methodology for classifying scam losses or explain whether changes in reporting, portfolio mix or reimbursement affected the comparison.
The bank’s figures should therefore be read as company-reported performance, not an independently audited industry benchmark. A 24% reduction can be meaningful while still leaving customers with substantial losses; without the absolute total, the residual harm cannot be assessed.
The disclosure also illustrates why financial institutions need more than a single outcome metric. Payments teams should examine losses, attempted payments, warning overrides, blocked transfers, recovery rates, time to intervention, beneficiary-account recurrence and customer reimbursement together. Improvements in one measure can otherwise conceal displacement into another scam type, channel or portfolio.
ANZ’s announcement supplies evidence that its controls are interrupting payments and that reported losses declined in the measured portfolios. It does not establish which intervention produced the improvement or quantify the bank’s remaining scam exposure. For payments and risk professionals, that boundary matters as much as the headline percentage.