Ecommerce fraud is accelerating as criminals leverage AI and automation to execute familiar schemes faster, at greater volume, and with less manual effort. A Q4 2025 Merchant Risk Council survey of 1,278 fraud prevention respondents across 37 countries found that the percentage of fraudulent ecommerce orders increased from 3.0% to 3.5% year-over-year. For a typical online merchant processing 2,500 orders per month at a $50 average order value, that 0.5-point rise translates from $3,750 to $4,375 in monthly fraud losses.
The survey identified five dominant fraud categories. Refund abuse ranked first at 41% of merchants, with fraudsters falsely claiming non-delivery or damage—now easier to fake with AI-generated photographs and shipping documents. Payment fraud affected 38%, exploiting near-instant bank transfers that are difficult to reverse, with AI-generated emails, voice messages, and videos making impersonation schemes more convincing. Credential theft and account takeover hit 37% of respondents, where criminals impersonate trusted entities to steal login details; fraudulent purchases then appear legitimate because they come from existing customer accounts. Card testing affected 33%, as criminals use stolen card numbers on small orders to verify which ones work before larger fraud or resale. Identity theft impacted 31%, with criminals using stolen personal data or creating synthetic identities—combinations of real and fabricated information—to bypass account-based defenses.
Commerce practitioners face rising fraud prevention costs and must prioritize detection and mitigation across multiple vectors: reviewing refund claim patterns and delivery records, implementing bot protection and transaction velocity controls, monitoring for unusual payment declines and small-transaction spikes, and deploying identity verification tools to catch synthetic shopper profiles. The acceleration of these risks in 2027 will depend on how quickly merchants adopt AI-powered fraud detection to match the sophistication of attacker automation.