Marqeta has added an AI-powered risk score to its Real-Time Decisioning product, aimed at preventing payment fraud.
Built into the card-issuing company's platform, the score assesses transaction risk at the point of authorisation. Customers can use it alongside existing rules and controls in Real-Time Decisioning to decide whether to approve, decline or review a payment.
The model analyses more than 300 real-time transaction attributes against historical data from Marqeta card programmes. It is designed to identify patterns specific to each customer's use case and adapt to cardholder behaviour and broader market shifts.
The update reflects a broader shift in payments fraud management, as static rules come under pressure from more varied and fast-moving attack patterns. The new score is intended to help customers reduce fraud while cutting false declines, which can block legitimate transactions.
Fraud Pressure
Marqeta cited an estimate that global payment fraud will rise by 153% between 2025 and 2030. In that environment, payment firms and card issuers are placing more emphasis on models that can respond to new behaviour without waiting for manual rule changes.
Its system combines customer-defined authorisation rules with machine learning, allowing businesses to run risk scenarios based on their own cardholder data. Scoring takes place within milliseconds, so decisions can be made within the normal authorisation flow.
Real-Time Decisioning is part of Marqeta's broader RiskControl suite, which also includes Know Your Customer checks, 3D Secure and disputes management. Together, these products extend the company's risk offering around its card-issuing platform.
The launch comes as card and payments providers look for ways to tighten controls without adding friction for users. False declines remain a commercial issue because they can interrupt purchases, hurt customer retention and create support costs, even when fraud controls work as intended.
Marqeta, which focuses on modern card-issuing infrastructure, said it processed nearly USD $400 billion in annual payments volume in 2025 and is certified to operate in more than 40 countries. That scale gives it access to a large volume of transaction data used to train the risk score.
Embedded Scoring
The product is positioned as part of the existing authorisation process rather than as a separate fraud tool. That means the score can feed directly into approval decisions when a card transaction is presented.
Anthony Peculic, Marqeta's interim chief product officer, said the launch is intended to help customers keep pace with faster-evolving fraud threats as they scale their card programmes.
"Today's fraud threats are evolving faster than ever, requiring businesses to keep pace as they scale their card programs," Peculic said.
"By embedding AI-powered controls and advanced machine learning into the authorization process, we enable customers to expand confidently while also strengthening their fraud defense as they scale."