AI pitched to fight mobile‑banking fraud
Recent coverage highlights AI as a practical tool to detect and flag suspicious mobile‑banking patterns, enabling faster intervention against fraud. AI models can surface anomalies in transaction behaviour and speed up responses, framing fraud detection as an easier‑to‑defend use case for enterprise AI programmes. That positions fraud control as a measurable application where AI investment can be justified on loss reduction. (thedailystar.net)
Artificial intelligence is being pitched as one of banking’s clearest near-term uses: spotting suspicious mobile-money behavior fast enough to stop fraud before cash leaves an account. (thedailystar.net) The Daily Star reported on April 12 that one proposed system watches how a customer types, taps and swipes inside a mobile-banking app, then compares that behavior with the customer’s usual pattern. A mismatch can trigger a warning or extra verification before a transfer goes through. (thedailystar.net) That is a shift from older fraud controls built around fixed rules, such as blocking a payment above a set amount or from a new device. Mastercard said banks are now using artificial intelligence to analyze transaction data in real time and make authorization decisions based on unusual patterns rather than only preset thresholds. (mastercard.com) Banks and regulators have been treating fraud detection as a practical artificial-intelligence use case for months. Federal Reserve Vice Chair for Supervision Michael Barr said on April 4, 2025, that traditional forms of artificial intelligence were already “essential” in areas including fraud detection, even as banks tested newer generative systems. (federalreserve.gov) The pitch is straightforward: fraud losses can be counted, false alarms can be measured, and response times can be tracked. Mastercard’s 2025 fraud-prevention report said 42% of card issuers and 26% of acquirers saved more than $5 million in fraud attempts over the prior two years from artificial-intelligence tools used in triage, pattern recognition and real-time detection. (mastercard.com) The pressure is rising because criminals are also using artificial intelligence. A February 2025 Federal Reserve research paper said payment fraud remained high and warned that generative artificial intelligence was giving fraudsters new capabilities, while Barr said in a separate April 17, 2025 speech that deepfake voice and image tools could help criminals impersonate bank clients and relatives. (federalreserve.gov; federalreserve.gov) The catch is that better detection depends on better data. Mastercard said artificial-intelligence systems need high-quality data feeds to improve risk decisions, and Visa says the same tools must separate genuinely suspicious transactions from legitimate ones so banks do not wrongly block customers trying to pay bills or move money. (mastercard.com; corporate.visa.com) Regulators are also signaling that banks cannot treat these systems as black boxes. In her November 22, 2024 speech, Federal Reserve Governor Michelle Bowman pointed to machine-learning fraud tools as an established use of artificial intelligence in finance, while stressing that institutions still need governance, risk controls and oversight. (federalreserve.gov) So the current argument for artificial intelligence in mobile banking is not that it will replace banking staff. It is that a model that notices a strange swipe pattern, a new device, or an odd transfer sequence may give a bank enough time to pause a transaction, ask another question and keep the money where it is. (thedailystar.net; corporate.visa.com)
