A Layered Defense Against AI-Driven Fraud

AI has made fraud faster, more polished, and easier to scale. It is also giving banks and credit unions new ways to detect altered documents, stop spoofed calls, monitor transactions, and prioritize suspicious activity. 

In a reported feature for the June 2026 ProSight Executive Report, Jason Bartolacci, director of the ProSight Fraud Alert Network, examines both sides of that contest, which is also a key theme in the just-launched ProSight State of Fraud Prevention Survey. Using Logix Federal Credit Union as the primary operational case study, he shows how one institution is responding to AI-enabled account takeovers, synthetic identities, voice cloning, and other rapidly evolving threats. 

Several practical lessons emerge from Logix’s experience: 

Prepare for attacks aimed directly at customers. The fraud landscape has shifted as criminals increasingly target customers rather than trying to break into accounts through bank contact centers. AI-generated phishing, cloned voices, and personalized messages can make those approaches appear legitimate while allowing criminals to launch them rapidly and at scale. Logix sees account takeover attempts through fake calls and texts hundreds of times a day. 

Verify documents before fraud becomes a loss. The credit union uses an AI-powered document authenticity tool to review bank statements, proof of residency, pay stubs, W-2 forms, and other materials submitted with membership and loan applications. It can identify alterations, shell businesses, and AI-created employment documents. Matt Overin, manager of risk management at Logix, says, “We’ve saved literally millions of dollars each year in loan fraud alone.” 

Close the communication gaps criminals exploit. Fraudsters often schedule attacks for after hours, when customers cannot easily reach a financial institution representative. They can also spoof calls so they appear to come from a bank or credit union. Logix uses anti-spoofing technology to verify calls that appear to come from one of its phone numbers. If a call did not actually originate from a Logix extension, the system terminates it. 

Use AI to direct human attention. Agentic AI could increase the speed and volume of account takeover attempts. Logix also plans to use its technology defensively, with an investigations tool generating lower-level alerts so employees can concentrate on higher-risk cases. 

Keep training and intelligence sharing in the mix. Logix uses online learning tools, employee training resources, and customer education to address tactics such as voice cloning and synthetic identity fraud. Bartolacci also highlights the Fraud Alert Network, an authenticated platform where fraud professionals can exchange intelligence, insights, and best practices. 

The takeaway: Logix’s experience illustrates why AI tools are only one part of the response. Effective fraud prevention also depends on customer and employee education, attention to operational gaps, and trusted information sharing across institutions. 

Financial-services leaders at banks and credit unions are encouraged to participate in ProSight’s inaugural State of Fraud Prevention Survey. The confidential survey takes about 10–15 minutes, and participants will receive exclusive access to key findings before the public release.