Originally published on roboearth.org
Financial fraud costs merchants, consumers, and banks billions of dollars every year. In recent years, artificial intelligence (AI) has been applied to help detect fraudulent charges and unauthorized bank account access. Using machine learning (ML) and AI to catch suspicious charges, banks can lock down targeted accounts before their clients experience a loss. Here, Daniel Calugar, investor and tech enthusiast, elaborates on how AI can be used to detect and deter financial fraud.
To understand how AI and ML are used in fraud detection, it is important to clarify a few terms. The idea of artificial intelligence has been around since the 1950s. Simply put, AI is the science and engineering of creating systems that can perform tasks or draw conclusions that were presumed to require human intelligence. As is evident from that definition, AI is a moving target.
As technology advances and machines or systems prove themselves capable of performing more “human-like” tasks, many tasks are no longer considered to be uniquely human. For example, in the 1950s, it was presumed that only humans could play chess. It was the science and engineering of artificial intelligence that created machines that could perform this task. Nearly every computer operating system today can play chess, and it is no longer considered artificial intelligence.
Machine learning, on the other hand, is a branch of artificial intelligence that uses computer algorithms to allow systems or machines to improve automatically by “learning.” There are various types of machine learning, and other branches of artificial intelligence, but those are for another article.
Machine learning algorithms detect bank fraud by learning from the patterns of previous fraud. Computers can process enormous amounts of data at incredible speeds, allowing ML algorithms to identify subtle and sophisticated fraud traits that humans can not detect. As an example, when a fraudster visits an internet shop, their interests and objectives are different from a customer. Therefore, their click pattern, click rate, and dwell time will be unusual. Even if they try to emulate a customer, ML systems can identify the imposter and block or flag their transactions.
Phishing emails are another common tactic used in financial fraud. Scammers have become very good at avoiding legacy phishing email filters. Machine learning has proven to detect phishing scams, block attacks, and notify customers of potential threats. Machine learning algorithms enable anti-fraud systems to learn and adapt to stay one step ahead of bad actors.
About Daniel Calugar
Daniel Calugar is a data-driven investor with an academic and professional background in computer science, business, and law. He developed a passion for investing due to frequent interaction with investment professionals who serviced his legal clients’ investment needs. As a tax partner at the Atlanta law firm of Hansell & Post and the global law firm of Jones Day, he incorporated his partnership interest to set up and serve as trustee for his tax-qualified profit-sharing plan. Calugar utilized his technical skill set to design computer programs that would help him make more effective investment decisions. When Dan is not working, he enjoys spending time working out and being with friends and family. As a pilot with over 2000 hours of single-pilot experience flying business jets, he enjoys flying volunteer flights for Angel Flight.