Credit Card Fraud

Credit Card Fraud refers generally to any fraudulent transaction using a credit card as a source of funds. The fraudulent transaction may be committed to obtain goods or services or to illegally obtain funds from an account. Credit card fraud may occur simultaneously with identity theft, but can also occur when a legitimate consumer makes a purchase with no intention of paying for the goods or services, sometimes referred to as chargeback fraud or friendly fraud. Credit card fraud is related to debit card fraud, differing primarily in the form of payment. Another form of credit card fraud is new application fraud, in which a perpetrator applies for a credit card in a victim's name, then uses the card to purchase goods and services illegally. A victim’s credit card information can be acquired in a number of ways, by being purchased on the deep/dark web, by using skimmers at retail points of sale or ATMs, or through corporate data breaches.. The true cost of credit card fraud for merchants is more than just the cost of lost merchandise — it also includes lost profits, bank fees and chargeback costs.


CVV (Card verification value)

Machine learning (ML) refers to the development of computer algorithms and statistical models to perform predictions and specific tasks without explicit instructions, rather using inferences and patterns instead. Machine learning is a subset of artificial intelligence and generally falls into two main categories: 1) supervised learning, in which the outcomes are known and labelled in training data sets and 2) unsupervised learning, in which no outcome is known and the goal is to have items self-organized into clusters based on common characteristics or features. Supervised learning uses techniques like neural networks, bayesian models, regression models, statistical models, or a combination thereof. Unsupervised learning uses techniques like k-means clustering and is often used for anomaly detection. Some computer systems have the ability to “learn” or make progressive improvements on a task based on algorithms and subsequent outcomes. As an example, machine learning in fraud prevention allows algorithms to make immediate decisions on new transaction decisions, but over time to "learn' from the outcomes of the purchases and from that new data, self-correct to make increasingly accurate predictions going forward. The fastest and most reliable path towards the learning component relies on analysts’ insights, assisted by machine-learned predictions, to make well-informed decisions.