Identity theft is a hot topic these days. There is a lot of talk about steps you can take to help prevent it - myFICO has a great set of suggestions for instance - and one can also be very alert to it having happened. All that said, how can technology help PREVENT it?
Well early detection and effective resolution involves combining several distinct things:
- A broad data set to help identify people accurately
- Sophisticated analytics to leverage this data to make predictions of risk
- The ability to manage rules defined by privacy and other regulations
- The additional ability quickly change or add rules to respond to new fraudster approaches
- An efficient human review capability
Let's take each of these. Data is a double-edged sword these days - there is more data available to verify someone's identity but all this data is also available to be hacked or abused, increasing risk. Data of different types, drawn from a wide variety of sources, can be used to cross-reference facts that have been stolen, for instance, with data that has not and so find out that someone applying for credit, say, does not know things they should know. For instance multiple lines of business within a bank or other products and services used by the same person. Clearly than get expensive so you also want rules-based technology that let's you just enough data as you need it.
In addition this data powers the second item - analytics. Neural networks and other sophisticated analytic models can examine the complex interrelationships between data to predict how likely something is to be abnormal. These models can use relationships that are complex and subtle making it doubly hard for fraudsters.
Not only do business rules add value in the data aspect, they are key to making sure that regulations and policies are enforced on every transaction. In addition they allow for new kinds of fraud to be identified before there is data to support them. For instance, an FBI alert about a particular kind of identity theft gang can be turned into a set of rules and implemented in minutes or hours rather than the days or weeks that traditional coding approaches take and without having to acquire loss data to drive analytics.
Lastly all of this enhances but does not replace the human element - that will always be key.
To successfully combat identity fraud all of this has to work in real-time, making the integration and sophistication of these technologies key.
For more on this, search for my colleague Ted Crooks who is widely quoted on this topic and has some good articles such as this one Sharper tools for hunting down ID fraudsters. Here's another article on Fair Isaac and its focus on fraud detection.
In the interest of full disclosure, Falcon ID is one product aimed at this problem sold by Fair Isaac.
