Card issuer fraud costs rank near top for 42% of issuers

Card issuer fraud costs are among the largest platform-related expenses for a sizeable share of U.S. issuers, according to a PYMNTS Intelligence report produced with Visa Digital Processing Solutions. The study found that 42% of issuers rank fraud and disputes as their largest or second-largest platform operating cost, excluding employee expenses.

The findings come from The Issuer Risk Playbook, which surveyed 500 executives in head-of-payment roles at U.S. bank and non-bank card issuers. The report points to growing pressure on fraud systems as payment decisions increasingly have to be made in milliseconds while issuers try to separate criminal activity from legitimate customer spending.

Issuers also reported more concern about the tools they use to make those decisions. The share citing insufficient fraud systems and high false declines as a processor pain point rose to 21% from 14% a year earlier, according to the report. Cybersecurity and network concerns increased to 36% from 30% over the same period.

What is agentic commerce in card payments?

Agentic commerce refers to transactions in which artificial intelligence software may recommend a product, choose a payment method or carry out a purchase on a consumer’s behalf. For card issuers, the issue is whether a transaction initiated by software fits the authority a customer gave, while still needing approval or rejection at authorization speed.

That adds a new layer to a long-running fraud-control task. Issuers already compare a transaction with known cardholder behavior, merchant patterns and risk signals. With AI-mediated purchasing, they may also need enough information to judge whether an automated action is within the customer’s delegated instructions.

How are issuers using AI for fraud detection?

Real-time fraud detection has become common across issuers, according to PYMNTS Intelligence. Roughly six in 10 issuers across customer lifetime value tiers are deploying or improving AI-powered card fraud detection and prevention, suggesting that the capability is becoming a standard part of issuing operations.

The report found sharper differences by customer lifetime value, or CLTV, a measure used to group issuers by the long-term value of their customer relationships. Sixty-eight percent of high-CLTV issuers said stronger security and fraud prevention are necessary for agentic commerce. That compared with 47% of medium-CLTV issuers and 43% of low-CLTV issuers.

High-CLTV issuers are also putting near-term AI investment behind real-time fraud detection and operational automation. PYMNTS Intelligence said 41% of high-CLTV issuers identified each of those areas as an implementation priority over the next 12 months.

Real-time detection matters because card authorization is the point at which an issuer can approve or block a payment. Better decisioning can reduce suspicious transactions while limiting false declines, which can create dispute costs and damage the customer relationship when legitimate purchases are rejected.

The report frames fraud controls as a requirement whether AI agents become common payment initiators quickly or develop more slowly. If agentic commerce gains traction, issuers will need systems that can evaluate delegated purchases during authorization. If it does not, issuers still face the existing problem of more sophisticated fraud alongside the cost of rejecting good transactions.

This story draws on original reporting from PYMNTS.