Visa Deploys AI to Modernize Payments
Visa has launched "Intelligent Authorization," a new AI-powered system designed to help banks optimize payment approvals and reduce fraud. The move signals a major push by financial incumbents to embed AI at the core of their infrastructure. This sets a new technical baseline for fintech startups competing in the payments space.
Visa's use of AI in payments is not new; the company first deployed AI-based technology for managing risk and fraud back in 1993. This decades-long investment in AI infrastructure now helps prevent an estimated $40 billion in fraud annually on Visa's network. The "Intelligent Authorization" system is part of a broader suite of services called VisaNet +AI. This suite includes "Smarter Stand-In Processing" (Smarter STIP), which uses deep learning to approve or decline transactions on behalf of banks during outages, mirroring their decisions with up to 95% accuracy. The scale of the problem AI is tasked with solving is immense. Global losses from online payment fraud are projected to hit $362 billion by 2028. Visa itself processed over 233 billion transactions in 2024, and its AI models can evaluate up to 182 risk attributes in a millisecond to generate a fraud score. This new system is designed to streamline a complex legacy infrastructure for acquirers—the financial institutions that process payments for merchants. It provides a single API connection to handle transactions across major card networks, aiming for a 96.3% average approval rate and 99.999% uptime. The competitive landscape is heating up as both established players and startups leverage AI. Mastercard has its own AI-driven fraud risk solutions, while fintechs like Adyen use machine learning to optimize payment routing, and PayPal deploys AI to analyze millions of transactions for suspicious patterns in real time. Beyond just authorization, Visa is positioning its AI platform for future forms of commerce. Its "Visa Intelligent Commerce" initiative is designed to allow AI agents to securely make purchases on behalf of consumers, using tokenized credentials instead of traditional card numbers and allowing users to set specific spending limits.
