Authorised push payment scams, account takeover and synthetic identities are not edge cases. They are the result of product decisions made long before a loss occurs.

The authorised-payment problem

Traditional card fraud involves an unauthorised transaction. Authorised push payment fraud is different: the customer approves the transfer, often after being manipulated by someone impersonating a bank, a police officer, an investment platform or a romantic partner.

Controls designed for unauthorised transactions — chip, tokenisation, risk-based authentication — do not solve this. The customer is present, the device may be genuine, and the payment instruction is valid. The failure sits in the journey that led to the approval.

Illustrative schematic accompanying the analysis
Original schematic prepared for this paper. Illustrative only; not a live data feed or a map of any specific market.

Friction has to be intelligent

Adding warnings to every payment trains customers to dismiss them. The useful approach is targeted friction: a pause, a clear explanation and a route to a human when the combination of payee, amount, channel and behaviour suggests elevated risk.

Confirmation-of-payee services reduce misdirection but do not prove that the payee is legitimate. A scammer can own the account. Matching the name to the account number prevents a particular error; it does not establish intent.

Liability shapes behaviour

Who bears the loss determines who invests in prevention. Rules that place all responsibility on the customer encourage banks to add disclosures rather than controls. Rules that place all responsibility on the bank can encourage over-blocking and financial exclusion.

The more effective frameworks apportion responsibility based on what each party could reasonably have done: scheme rules, receiving-bank controls, sending-bank warnings and customer behaviour. They also require fast information sharing so that a reported scam account can be acted on before the next victim pays.

Information sharing is a control

Fraud is a network problem. A single institution sees only part of the pattern. Sharing indicators of compromise, mule-account behaviour and scam typologies can improve detection across the system, provided it is done within privacy and competition rules.

Supervisors and regulators have a role in making that sharing lawful and predictable. Without a clear legal basis, institutions default to silence and the network stays fragmented.

Design for the moment before payment

The highest-leverage point is the moment before the customer approves. Good design makes the payee, the amount and the consequence legible. It slows down the unusual case without punishing the routine one. It gives the customer a way back.

Fraud prevention is not only a security function. It is a product discipline, and it belongs in the design of every payment journey.

About this paper This is an explanatory working paper, not investment, legal or regulatory advice. It contains no invented statistics or quotations. Sources and scope are set out in the methodology. Corrections: editorial@bankingwebinar.com.