IBM to Apply Machine Learning to Combat Fraud by Acquiring IRIS Analytics

Security Artifacts – The Hunt for Forensic Residue

As part of an effort to combat billions of dollars in fraudulent transactions, IBM today announced it has acquired IRIS Analytics, a provider of fraud detection software that makes use of machine-learning algorithms to make it possible to detect fraud in milliseconds.

Bob Griffin, general manager for the IBM Safer Planet initiative, says IBM is looking to more broadly apply artificial intelligence to combat fraud in a way that is simpler for IT organizations to consume. Instead of requiring months of upfront consulting work to set up, Griffin says the IRIS Analytics software is designed to increase the amount of fraud transactions detected by as much as 40 percent right out of the box. The machine-learning algorithms over time become more fine-tuned to the point where just about all fraudulent transactions are identified before being processed, regardless of the online or offline channel used.

That’s critical, says Griffin, because the goal is to not identify fraudulent transactions after they have already taken place, but rather prevent them from occurring in the first place.

Just as importantly, Griffin says IRIS Analytics has developed a “white-box” approach that enables organizations to dynamically adjust transaction rules. Instead of disrupting the customer experience, Griffin says that approach makes it simpler for organizations to determine whether a transaction should be processed in the moment versus the issuer of a credit card automatically putting a hold on a specific credit card until they can determine the validity of a transaction hours, sometimes even days, later.

In general, as machine-learning algorithms advance and the amount of compute horsepower required to process transactions and analytics becomes more broadly available, the volume of transaction fraud being perpetrated should decline substantially in the years ahead. Of course, the criminal enterprises that have turned fraud into a multi-billion industry will no doubt be making IT investments of their own. But at the very least, most of the low-level fraudulent transactions that plague everything from local retail outlets to health care facilities should be little more than a bad business memory.