As criminals adapt to the solutions that banks use to detect fraud across the product spectrum, these require banks to create fresh solutions to counter emerging trends, which spur further strategies of crooks, and so on.
Use of Data Science
The field of Data Science leverages Artificial Intelligence to infer patterns, with the goal of crafting effective strategies against offenders. Such solutions usually take the form of mathematical models, which help ascertain probability of any banking activity being conducted by fraudsters.
In recent times, there has been a steady growth of inexpensive computing power, along with a wider variety of data sources. This has led to a growing sense of confidence within the crime research fraternity of being able to design better solutions. The barrage of data and growing pool of Machine Learning techniques paint a very convincing picture of vigilantes, who are finally gaining the upper hand.
Crime Mutation
However, crooks always strive to slip through counter-crime nets. To illustrate, fraudsters trying out stolen cards will attempt multiple transactions of varying amounts, patiently waiting to sneak past fraud-detection rules.
A careful study of attributes of declined transactions will determine counter-fraud strategies, and they adapt accordingly. Similarly, enhancement of security questions might trigger fresh waves of social engineering, to gather relevant information.
Illusion of Asymmetric Insight
Even though mutations in criminal modus operandi are highly common, data science pushes on to leverage patterns found in historical training sets to build solutions, all the while assuming that such findings will hold true in future. This is a classic example of the Illusion of Asymmetric Insight.
In 2001, researchers at the University of Illinois and Williams College, performed a series of studies looking into how people's perceptions of others, compared to the insights they thought others had of them.
Results showed that we have the tendency to believe that we know other people more than they know us. In addition, we assume our knowledge of other groups to be more than any information members of such groups might have about our groups.
While we presume we know others based on their observable attributes, we think that others cannot truly know us because they do not have access to the inner workings of ourselves, or our groups.
Asymmetric Insight, has significant ramifications. This bias tends to make us overlook analytical prowess and cognitive uniqueness of others. Hence, we do not feel the need to grasp how others might be analyzing our actions.
In the field of fraud detection, this prevents us from acknowledging that counter-fraud initiatives are often analyzed by offenders, in order to help commit fraud. Case in point, vendors selling stolen credit card information on the dark net have been known to use automated systems to detect undercover buys by law enforcement.
Solutions to Crime Mutations
The evolution of fraud trends depends on:
- Transmission strength of counter-fraud signals - Nuances of counter-fraud initiatives have a higher chance of diffusion, than ones taken to detect money laundering, terrorism financing or other financial crimes. This is because each declined fraud transaction passes on the message to fraudsters that their actions are being analyzed. This prompts the need to infer such solutions.
- It is crucial for banks to study transmission possibility across the entire surface of anti-fraud infrastructure to ensure, offenders are never tipped off, especially as unintended consequences of fighting fraud.
- Ease of deduction of inner workings from such signals - Ease of inference can be reduced by using features that are difficult to grasp. Case in point; elements quantifying subtle nuances of online behavior, such as click-journey and style of typing, might differentiate genuine activity from illegitimate. Clearly the odds of criminals being able to trace back alerts to such features, is quite low.
- Ability to adapt, after grasping attributes of such solutions - Fraudsters’ ability to adapt depends on the overarching approach to fighting financial crime. Pursuing a disjointed vision of the crime ecosystem renders it easy for offenders to modify their methods of working. Case in point, replacing counter-fraud solutions working in silo, with strategies that connect the dots between money laundering and (first party) fraudsters, might work better.
Conclusion
Research into fraudsters’ insight requires creation of systems which comb through the landscape of shifting trends in financial crime, studying them as responses to counter-fraud measures. However, all of this is only possible when we accept the fact that algorithms sifting through a barrage of data has unwittingly established a two-way street, connecting the sentinel with the criminal.