For card issuers, artificial intelligence is moving closer to the transaction itself.
Eighty-nine percent of issuers plan to adopt or expand at least one AI fraud capability, according to the PYMNTS Intelligence report “Investing Forward, Building the Edge,” done with Visa DPS. Half identify fraud detection and scoring as the single most important AI capability to add over the next year, while 63% plan real-time fraud detection specifically.
Fraud decisions happen continuously, against live transactions, and often have to be made quickly enough to approve a legitimate purchase without creating unnecessary friction for the cardholder.
Issuers are putting AI into that environment at the same time that they are shortening their technology investment cycles. The average time until an issuer’s next platform upgrade has fallen from 7.8 years to 2.4 years, with 7 in 10 issuers now planning an upgrade within three years.
There’s a connection between the infrastructure being replaced and the AI capabilities issuers expect that infrastructure to support, according to the data.
Fraud already occupies a large part of the current development pipeline. Real-time transaction alerts are in progress at 40% of issuers, scam detection at 38%, dynamic CVV at 38%, tokenization at 33% and card fraud detection at 31%.
AI adds another layer to those controls by putting more of the analysis behind a fraud decision into the technology stack. Transaction risk scoring, real-time detection and related tools operate on information generated as transactions occur, making the issuer’s ability to process and act on data part of the fraud-control architecture.
We’d love to be your preferred source for news.
Please add us to your preferred sources list so our news, data and interviews show up in your feed. Thanks!
The scale of planned adoption also distinguishes fraud from other AI use cases. When issuers were forced to select their most important AI capability for the coming 12 months, fraud detection and scoring captured 50% of responses. Operational process automation, the next-highest choice, reached 14% among high-CLTV issuers and lower shares among the other groups.
Fraud also crosses the performance differences that otherwise separate issuer investment strategies. Between 49% and 54% of issuers in every customer lifetime value tier select fraud detection and scoring as their top AI priority.
AI Moves Closer to Transaction Decisioning
Other artificial intelligence investments are developing alongside fraud, but they address different parts of the card business.
Transaction categorization is planned for adoption or expansion by 62% of issuers, operational automation by 48% and portfolio risk analytics by 35%. On the customer side, 53% identify customer-service automation, 50% marketing and segmentation, and 48% personalized rewards among their AI growth priorities.
Fraud carries a different operational requirement because the output can affect an authorization while the transaction is underway. That puts the technology alongside systems responsible for transaction processing and risk controls, rather than limiting it to customer service, marketing or back-office automation.
Real-time fraud detection is planned by 70% of prime and mainstream issuers, 63% of affluent-serving issuers and 51% of subprime and inclusion-focused issuers. Transaction categorization, meanwhile, is clustered at 61% to 62% across all three groups.
With 89% planning some form of AI fraud investment, fraud prevention is becoming an early test of whether AI can operate reliably inside one of the card issuer’s most time-sensitive functions.
Add as Preferred Source
