issueisitpossibletodetectinreal-timeifsomeoneis.docVIP

Issue: Is it possible to detect in ‘real-time’ if someone is fraudulently using a legitimate person’s credentials to obtain service? Real-time could be 5 seconds to 5 days depending on the eCommerce product. This detection is very important with the addition of the Internet as a new customer eCommerce channel. Background: Roughly 4 billion U.S. dollars were lost to fraud in 1999 in the credit card channel alone. Research and data have determined that organized crime is the main threat to financial loss. Organized crime can easily obtain large numbers of identification and authentication credentials of legitimate people. Current technology tracks patterns of behavior. This pattern matching mechanism checks if the current purchase/payment pattern matches previous patterns. This creates the scenario where one is always checking what the losses were 30 to 60 days ago. Pattern matching requires a huge database. Many fraudulent activities fall below the threshold of pattern matching detection. A recent fraud occurred where $19.95 was charged on a large subset of 900,000 credit cards in some cases multiple times over different billing cycles. Only 200 card holders came forward to complain. Financial institutions are starting to use the Internet as an additional channel to provide product to its customer base. This globalization of the footprint for doing business injects new risks. Data shows that the customers outside the footprint attempt to defraud the system. In the telco arena the technology (pattern matching) today takes about 10+ days to detect someone stealing service. This is probably one reason that the wireless community is supporting the U.S. Government requirement to have a GPS location system in cell phones by 2001. One such GPS solution can locate people in canyons or elevators very effectively. There is currently, in New York City, a gang of 200+ foreign nationals selling cell phones with a warranty that it can b