[rev_slider_vc alias=”finance_application_desktop”]
[dt-section-title title=”AI And Fraud Detection”]

Join The Discussion | Ai4 Finance 2019

Within Financial Services, fraud detection refers to a system for identifying suspicious behavior aimed mostly at detecting fraudulent credit card purchases and mobile transactions, identity and social misrepresentation, as well as money laundering and internal fraud. 

The hard but true fact is that fraud is a multi-billion dollar business and growing. Surveys show that over half of all large corporations have experienced some type of fraud and for the financial services industry that figure is much higher.

In 2017, American financial institutions filed ~5 million “suspicious activity” reports, identifying transactions with potential links to fraud. That figure represents an increase of 2,000% in five years¹.  25% of all malware attacks hit banks and other financial services organizations. On top of that, the number of compromised credit cards has increased year over year at alarming rates – as high as 200%+. Credential leaks and malicious apps have also recently seen 100% year over year increases.

While the digitization of financial services opens new possibilities for crime, it also allows for technologically advanced solutions such as artificial intelligence and machine learning technology.

[dt-space height=”30″]

Here are some examples of how artificial intelligence and machine learning technology can applied to fraud detection:

1) Avoiding false positives and lost customers: Most fraud detection gateways rely on a set of rules to classify a transaction as fraudulent or not. This binary classification system causes a high rate of false positives, or safe transactions being marked as fraudulent. If an honest customer is trying to purchase a new toaster and keeps getting marked as fraud and can’t complete the purchase, both the immediate transaction becomes less likely and the customer’s lifetime value goes down. In place of a rules-based, binary classifier, a machine learning model can provide a more nuanced review in microseconds. By taking into account a vast amount of customers purchase behavior data, the machine learning model can build a more complete view of what a fraudulent or safe transaction looks like. The traditional system of human-written rules can’t encompass as complete of a view of customer purchase behavior as AI, giving AI the upper hand when it comes to fraud detection.

At Ai4 Finance 2019, Mastercard’s SVP of Cyber & Intelligence Solutions will highlight the topic of AI + Fraud Detection in a talk entitled AI’s Role in Authenticating Identity with Speed, Accuracy and Security.

[dt-space height=”15″]

2) Avoiding the cost of human reviewers: Visa’s 2016 Annual Fraud Detection Benchmark Report shows that 83% of North American businesses conduct manual reviews, and on average they review 29% of orders manually. 46% of the survey respondents said their human review staff makes up the biggest cost of their fraud management budget. Point being, human review is common and costly. With AI, financial firms have been able to cut the number of fraud alerts that require human review in half. As AI continues to approve over time, the level of human intervention required will continue to decrease, saving fraud management programs some serious coin while improving their effectiveness.

[dt-space height=”30″]

Join the AI + Fraud Detection discussion at Ai4 Finance!

[rev_slider_vc alias=”aboutsection”]