Ecommerce fraud may be growing thanks to AI and automation.
Criminals who have long stolen credit cards, taken over customer accounts, and filed fraudulent refund claims may now pursue those schemes faster, at greater volume, and with relatively less effort.
Merchants may face more fraud attempts in 2027, even if the methods remain familiar.
5 Fraud Risks
A Q4 2025 survey from the Merchant Risk Council found that, on average, the percentage of fraudulent ecommerce orders increased from 3.0% to 3.5% from the previous year.
Imagine an online shop processing 2,500 orders per month at a $50 average order value. A 3.0% fraud rate equates to 75 bad orders, or $3,750. A 3.5% fraud rate means 87 orders worth $4,375.
Yet fraud prevention is not free and requires merchants to identify the top threats to their business.
Here are five likely culprits.
Refund abuse
Refund and return policy abuse was the most commonly reported fraud category among ecommerce businesses in the aforementioned MRC survey. Overall, 41% of merchants surveyed reported refund abuse.
A fraudster places an order, then falsely claims it never arrived or requests a refund for supposedly damaged goods.
Generative AI has made it easier to fake claims, with convincing photographs of damaged products and shipping documents.
To reduce exposure, sellers can:
- Review repeated claims,
- Keep delivery records,
- Verify suspicious damage evidence.
Payment fraud
Real-time payment fraud affected 38% of merchants in the MRC survey and ranked second among ecommerce businesses.
Real-time payments are near-instant transfers between bank accounts, typically settling within seconds. Unlike conventional credit card transactions, the transfers are difficult or impossible to reverse.
Fraud involving real-time payments is either “authorized,” where someone is duped into sending the money, or “unauthorized,” where a criminal hacks into a bank account and initiates the transfer.
AI-generated emails, voice messages, and videos can make impersonation schemes relatively more convincing and easier to scale.
Credential theft
More than one in three (37%) MRC survey respondents experienced phishing and related credential theft fraud.
For this tactic, criminals pretend to be a trusted entity, such as a bank, a carrier such as FedEx, or a colleague. The goal is to get someone to reveal credentials or payment details or approve a fraudulent transaction.
Credential theft and account takeover fraud can be especially troublesome for ecommerce businesses because a fraudulent purchase appears to come from a legitimate customer with an order history.
Once again, AI and automation could make this form of attack somewhat easier to scale in 2027.
Card testing
Some criminals use stolen credit card numbers to place small orders on ecommerce sites to reveal which cards work before making larger purchases or even selling the verified credentials.
Merchants can become unwitting testing vehicles for stolen payment information. The MRC survey found that 33% of sellers experienced card testing, unchanged from the prior year.
Automated card testing can generate many more fraudulent attempts, disrupt checkout systems, and draw scrutiny from payment providers.
Merchants can limit exposure with bot protection and transaction velocity controls while monitoring unusual increases in declined payments or repeated small transactions.
Identity theft
Identity theft involves criminals using stolen personal and financial information to impersonate legitimate shoppers and place fraudulent orders.
Unlike credential theft, which provides access to existing accounts, identity theft allows criminals to impersonate shoppers without necessarily compromising their accounts.
The MRC survey found that 31% of merchants experienced identity theft.
AI could make the problem worse in 2027 by helping criminals create synthetic identities, combining stolen personal information with fabricated names, addresses, photographs, and other details to appear legitimate.



