AI E-commerce Fraud Prevention: A Guide for Australian Sellers

Card-not-present fraud on Australian cards used with domestic merchants fell to a record low in 2024-25, but fraud has not gone away; it has shifted to overseas merchants, scams and abuse of refunds and accounts. Australian Payments Network data shows domestic card-not-present fraud fell 11.1% to $312 million in FY25, while Australians reported $2.18 billion in scam losses in 2025. For online sellers, the best defence combines payment-provider fraud tools, AI-based anomaly detection and tight reconciliation, so unusual orders, refunds and payments are caught quickly.

The Australian fraud picture

Australian Payments Network (AusPayNet) publishes card fraud statistics twice a year. In the 12 months to June 2025 (AusPayNet, 2025):

• Total card fraud fell from $868 million to $854 million, and the overall fraud rate fell from 77.6 to 71.8 cents per $1,000 spent.
• Card-not-present (CNP) fraud on Australian cards used with domestic merchants fell 11.1% to $312 million.
• The domestic CNP fraud rate hit a record low of 75 cents per $1,000.
• The overseas CNP fraud rate was more than ten times higher, at $10.75 per $1,000.

AusPayNet notes that spending online with overseas merchants accounts for only 3% of Australian card spending but 52% of all card fraud on Australian-issued cards. Australian merchants have improved their defences, and fraudsters have moved to where controls are weaker.

Scams are a separate and larger problem. The National Anti-Scam Centre reported $2.18 billion in losses in 2025, up 7.8% on 2024, although still down almost 30% from the 2022 peak. Payment redirection scams, where criminals pose as a supplier or customer and change bank details, cost $166.8 million. The ACCC warned that AI is increasing the sophistication of scam activity.

Fraud types online sellers face

Card-not-present fraud on your own web store: stolen card details used to buy goods, followed by a chargeback that leaves you without the goods or the money.

Refund and return abuse: claims that an item never arrived, returns of a different or used item, or repeated "damaged on arrival" claims.

Account takeover: criminals gaining access to customer accounts, or to your own marketplace and payment accounts through phishing. Phishing was the most reported scam type to Scamwatch in 2025.

Payment redirection: fake emails changing a supplier's bank details, or fake marketplace or carrier messages asking you to "verify" your account.

Marketplace-specific risk: on marketplaces, the platform usually handles payment fraud screening, but sellers still face claims, returns abuse and settlement errors that need to be spotted.

How AI helps

Fraud detection was one of the earliest commercial uses of machine learning, and it remains one of the most useful. AI models score each transaction against patterns such as mismatched billing and delivery addresses, unusual order values, many orders from one device, or a first-time customer ordering high-resale items for express delivery.

The payments industry expects to lean further on AI. A Mastercard survey of 300 senior executives found 90% believe their fraud losses will grow if they do not increase their use of AI fraud prevention, as generative AI helps criminals automate fraud. Among Australian SMEs that use AI, 48% use it for cybersecurity and threat detection, the third most common use.

AI fraud scoring should not be the final word. Automatically cancelling orders can block genuine customers, and if decisions significantly affect individuals using personal information, Privacy Act transparency obligations may apply from December 2026; see AI risks and governance for Australian retailers. The better pattern is AI that flags, and a person who decides on borderline cases.

Practical controls for small and mid-sized sellers

1. Use your payment provider's fraud tools and strong customer authentication such as 3-D Secure on higher-risk orders.
2. Hold, do not ship, flagged orders until checked, especially high-value or express orders to new customers.
3. Use signature or photo proof of delivery for high-value parcels so "item not received" claims can be answered.
4. Verify bank detail changes by phone using a number you already hold, never one from the email.
5. Turn on multi-factor authentication for every marketplace, payment, carrier and email account.
6. Reconcile every payout. Refunds, chargebacks and adjustments you did not expect are often the first sign of fraud or abuse.
7. Track claims by customer and address to spot repeat refund abuse.

Key figures

• Card fraud on Australian cards fell to $854 million in FY25, or 71.8 cents per $1,000 spent (AusPayNet, 2025).
• Domestic card-not-present fraud fell 11.1% to $312 million, a record-low 75 cents per $1,000 (AusPayNet, 2025).
• Online spending with overseas merchants is 3% of card spending but 52% of card fraud (AusPayNet, 2025).
• Australians reported $2.18 billion in scam losses in 2025, up 7.8% (ACCC, 2026).
• 90% of payments executives expect losses to grow without more AI fraud prevention (Mastercard, 2025).

How eComChannels helps

Many fraud and abuse losses show up first in the money, not the order. eComChannels finance auto reconciliation matches every marketplace remittance and settlement to its orders, applies fee rules and flags short-paid or missing payments, so deductions you did not expect stand out and problems surface in days rather than at month-end. That makes it easier to dispute incorrect deductions and spot patterns of refund abuse.

Order management puts orders from every channel in one place, so you can review suspicious orders before dispatch and keep stock accurate if an order is cancelled. Together they reduce losses and the hours spent chasing them. Book a demo or start a 30-day free trial.

FAQ

Do marketplaces protect sellers from fraud?

Marketplaces generally screen payments on orders placed through them, but sellers can still lose money through claims, returns abuse, settlement errors and phishing attacks on seller accounts. Check each marketplace's seller protection terms.

Is card-not-present fraud getting worse in Australia?

For domestic merchants it has fallen to a record-low rate according to AusPayNet's FY25 data. Fraud on Australian cards used with overseas merchants remains much higher.

Should I automatically cancel orders flagged as risky?

Use automatic holds rather than automatic cancellations for borderline cases, and have a person review them. This protects genuine customers and keeps you in control of decisions.