AI, Scams and Mule Accounts: Five Questions Bank Directors Should Ask About Fraud

Meet the Innovators: A Q&A with Ravi Loganathan, Banking and Policy lead at Sardine and president of Sonar

From AI-generated scams to mule accounts and real-time payments, the fraud landscape is changing quickly. To help community bankers separate hype from reality, we caught up with Ravi Loganathan, Banking and Policy lead at Sardine and president of Sonar. Sardine, an ICBA ThinkTECH Accelerator alum, helps financial institutions combat fraud through real-time intelligence and collaboration. We discussed the trends shaping fraud prevention and what banks can do to strengthen their defenses

To kick things off, tell us a bit about yourself, your role at Sardine, and the company’s mission. What problem was Sardine founded to solve?

I lead Banking and Policy at Sardine and serve as President of Sonar, our collaborative fraud intelligence network. Before Sardine, I spent much of my career in banking, including consumer banking, risk and operations at Bank of America. I was also part of the early team at Early Warning Services that launched Zelle.

Sardine was founded around a simple problem. Financial services have become faster and more digital, but many of the controls used to identify fraud and financial crime were built for a slower world. Money now moves in seconds across multiple rails and institutions. Sardine helps financial institutions understand risk in real time using identity, device, behavioral and transaction intelligence. Increasingly, that also means helping institutions collaborate because you cannot fight network fraud with information from a single institution.

What are the most pressing fraud trends affecting community banks right now, and how are fraudsters using AI and other emerging technologies to evolve their tactics? What should banks be paying closest attention to?

The biggest shift with AI is not simply that it makes fraud more sophisticated. It makes fraud cheaper and easier to scale.

AI can dramatically reduce the cost of creating convincing phishing messages, impersonating customers or employees, fabricating documents and identities, and running social engineering campaigns at scale. At the same time, faster payments have compressed the time banks have to identify fraud and intervene.

Community banks should pay particular attention to scams, account takeover, synthetic and manipulated identities, and mule accounts receiving and rapidly moving stolen funds. Increasingly, the important question is not just, “Is my customer legitimate?” Banks also need to ask, “Who is on the other side of this transaction?”

What are the most common gaps or mistakes you see in how community banks approach fraud today? In particular, many banks still view AML and fraud as separate functions. Why is that distinction becoming less relevant, and what opportunities are missed when those teams operate in silos?

One of the biggest gaps I see is organizational rather than technological. Fraud and AML frequently see different parts of the same criminal activity.

A scam may begin as a fraud problem at the sending institution, become a mule account problem at the receiving institution, and ultimately surface as suspicious money movement for the AML team. When those systems, data and investigators operate independently, each sees only part of what is happening.

There is a real opportunity in connecting those signals. Fraud teams have valuable real time behavioral and transactional intelligence. AML teams often have deeper information about counterparties, networks and patterns over time. Bringing those perspectives together can improve both prevention and investigation.

What fraud related questions should bank directors be asking management?

I would start with five questions.

1. What are our actual fraud losses, including customer reimbursement, operational expenses, and losses that may be classified elsewhere?

2. Which fraud types and payment rails are driving those losses, and how is that changing?

3. How much fraud are we stopping before money leaves the bank versus investigating after the fact?

4. Are our fraud, AML, and payments teams actually sharing data and intelligence?

5. What can we see about the receiving account and institution before we authorize an irrevocable payment?

Directors do not need to become fraud technologists. They should understand whether the bank’s controls are keeping pace with the speed and interconnectedness of the payment system.

If a community bank could prioritize just one area of its fraud strategy this coming year, what should it focus on and why? Looking ahead, where do you see AI having the greatest impact on fraud and risk management, and where do you think its potential may be overstated?

I would prioritize real time intelligence at the point of decision, particularly for irrevocable payments.

The industry has spent decades getting better at understanding the customer initiating a transaction. We now need comparable intelligence about the counterparty receiving the money. Once a real time payment, wire or other irrevocable transaction leaves the bank, investigation is a poor substitute for prevention.

I think AI will have its greatest impact in connecting signals that humans and traditional rules struggle to connect. That includes identity, behavioral, device, transactional and network data. It can also help investigators understand complex relationships much faster.

Where I think AI is overstated is the idea that it can replace sound risk fundamentals. AI does not fix fragmented data, weak governance or disconnected fraud and AML operations. The institutions that benefit most will be the ones that get the underlying data, controls and accountability right.

Visit sardine.ai for more information about how Sardine unifies fraud prevention, AML compliance, and real-time transaction monitoring in one platform - trusted by leading banks, merchants, and fintechs.