Voice Clones and Bank Fraud: New Risks for CFIs

When a Los Angeles man answered his phone, he heard his son anxiously describing a car accident, his arrest, and the urgent need for $25K to post bail or face jail. The man withdrew the money and passed it on to a man who said he was the son’s attorney. He wasn’t. There was no accident, the voice on the phone was a clone, and the money was gone. Voice fraud is on the rise, and so is the quest for defenses by financial institutions. Thanks to artificial intelligence (AI), voice clones are increasingly realistic and being used by scammers against individuals and businesses. In a recent report from Banking Dive and Modulate, 84% of financial services and retail organizations surveyed said their organizations had experienced moderately to highly sophisticated voice attacks in the past year. About 87% were from fraudsters impersonating customers, and 84% were impersonating employees. Why AI Voice Fraud Is a Growing RiskMore than half of voice fraud victims reported an average loss per successful incident ranging from $5K to $25K. However, the unseen cost is how much time and energy companies spend trying to combat this expanding and persistent threat. A shocking 8 of 10 respondents said they spent at least 51 hours each year investigating voice fraud cases.What Most Financial Institutions Are Doing TodayFinancial institutions (FIs) are, of course, fighting back. Nearly every FI uses at least one voice fraud detection/prevention method, and most are using three or more as the attacks have grown more sophisticated with the rise of AI. Here are four of the main ones: 

Costs of Fighting BackFIs that fall prey to voice fraud attacks can wind up losing customer confidence and trust, which can be very costly. However, there is also a downside to all the extra voice verification security: annoying legitimate customers. The more steps customers must take for authentication, the longer it takes and the more frustrated they become. FIs now regularly field customer complaints about all the extra steps to access their accounts, with 44% of those surveyed citing customer complaints regarding a more robust verification process as a major customer experience issue.How FIs Can Balance Security with Customer ExperienceTo reduce the impact on customers, FIs are seeking ways of speeding up the verification process and also reducing the incidence of false positive results that are particularly stressful for legitimate customers.One place to start is with a risk analysis that better defines the areas of greatest vulnerability. That includes an audit of all the ways voice-based transactions come into play. For a typical community financial institution, that might include mapping every inbound voice channel — branch direct lines, call center queues, lender desk numbers, and even ad hoc callbacks from staff mobiles — to see where authentication relies heavily on manual judgment.Where employees interact with customers, FIs can enhance training in voice fraud detection, including how to watch for emotional cues and known fraud tactics. Supervisor training can be bolstered as well. Training exercises can help improve employee and supervisor abilities and reaction times. For example, you might build a simple “pause and verify” script for emotionally charged calls (i.e., urgent bail money, wire requests, elder care emergencies) that require a second factor or supervisor review before any transaction is processed.FIs also need to use the latest and most powerful AI-assisted fraud detection and prevention methods. The goal is not just to quickly identify a suspicious voice but also to be able to tell when the suspicion is unwarranted — a false positive. For example, a call from abroad might pass muster if it is from a known customer who travels a lot and is calling from a known mobile number. But a call from an unknown device might arouse more suspicion. Some FIs are also tailoring their verification rules so that low-risk requests (balance checks, card replacement) face lighter friction, while high-risk actions (wire setup, limit increases) always trigger stronger voice and device checks.As AI-assisted voice fraud grows more sophisticated, community financial institutions should pair stronger, AI-driven detection with smarter triage and training — tightening controls where risk is highest and easing friction where customers simply need quick, routine help.