Five Fixes to Fight Fraud with Gen AI
Email and messages are the dominant vehicles for bank fraud today
AI can help detect and block fraud, and it can also improve the fraud-fighting experience for both your employees and your customers. It can triage problems so your fraud investigators are not distracted by minutiae, as well as find problems faster than a human. In addition, it can shorten investigation times and improve the customer experience with fast, English language notifications so they know what’s going on and do not need to reach out several times before the fraud investigations complete.
Here are our five fixes:
Blocking fraudulent emails and texts. Scan incoming texts and emails for fraud using both fraud pattern detection (tuned models) as well as customer pattern detection. For example, is this the customer’s first email? Are they texting from a known phone number? Identify in English why the transaction or communication might be fraudulent, such as “larger than normal amount”, “new wire recipient”, “unknown telephone number”, etc. Communicate with both employees as well as customers about what is going on in English.
Employee and customer fraud education. Provide educational material about fraud techniques as well as interactive fraud training for employees. These messages can include basic materials as well as information about the latest scams.
Analyze account behavior. Include cash, deposit, and payment patterns to develop adaptive risk scores and alert thresholds. For high risk accounts and/or suspicious patterns, automatically draft SARs - suspicious activity reports. Detect new beneficiaries or other signs of elder fraud or account takeover. These accounts can be escalated to employees.
Fraud investigation AI agent and copilot. Ask and get answers to questions about everything from the risk of a specific account to trends in fraudulent emails sent to the bank. An AI agent can monitor activities across these areas and independently notify the fraud department when something unusual is happening - an influx of fraudulent emails, a possible account takeover, and more.
Customer fraud recovery journey with AI. Instead of customers calling the bank five times wondering when they will be reimbursed, have AI proactively send email updates to the customer on what the customer needs to do (submit evidence) and where they stand in terms of reimbursement. It is important to keep a human in the loop in these situation.
Of course, if you are working with sensitive data like this, you will need secure, private AI to do it, as well as being able to govern access and maintain compliance with customer protection laws.
This is not a future dream. Banks and other companies are already using AI to fight fraud - and so can you.
AIR-GPT was used in the production and editing of this article.
Copyright © 2025 by Artificial Intelligence Risk, Inc. This article may not be redistributed in whole or part without express written permission from the author.