What Do Other Bankers Do?
Wire transfers are getting considerable attention these days, not only because of their importance in money laundering investigations, but also because of greater risk due to the quicker availability of funds.
One bank is addressing this problem with the use of a program added to their computer to determine which wires are good and which are bad or fraudulent.
Wells Fargo Bank is planning to use a type of artificial intelligence that incorporates the knowledge and guidelines of experts.
The experts? They are the employees now working in their wire transfer area, who each have an average of 15 years experience with the bank. When they retire, their expertise will retire with them.
Making matters more complicated, more and more customers make requests for wire transfers through their computer systems, rather than through telephone calls, making it easier to carry out a fraudulent transaction.
After getting the input from the 'experts' in the wire transfer area, the computer program developers built a system that knows the typical wire transfer for each customer, based on past experience with that customer.
Most requests for wire transfers are repeat business, they learned. If, for instance, the wires went repeatedly to a business account in the distant institution, and suddenly went to a private account number, an alert would be sounded.
The computer would alert wire transfer personnel that something may be wrong, and give the reason it believes the request may be suspicious.
Early tests ran the program while the wire room employees continued to apply their own surveillance procedures. The system did not find suspicious activity that the bankers missed, but they learned that by using the system they could accomplish their checking procedures in less time. According to a survey by American Banker, 25% of the largest banks in the United States plan to develop expert systems like the one at Wells Fargo in the near future.
Copyright © 1990 Bankers' Hotline. Originally appeared in Bankers' Hotline, Vol. 1, No. 11, 11/90
First published on 11/01/1990