Banking Frontiers

AI - tools for perpetrati­on and countering frauds


While AI is today touching virtually every aspect of life, what is of great concern is its misuse, especially in perpetrati­ng frauds. It is very widely used in: creating fake identities to evade security procedures phishing and related crimes creating fake voice identities and perpetrati­ng frauds deepfake videos for character assassinat­ion

Fraudsters appear to be one or more steps ahead of regulators in making use of new technologi­es and cause immense losses.

One of the latest applicatio­ns of AI in fraud activities is the creation of synthetic identities - fraudsters use real and fake data and generate fictitious identities. Such identities are used to defraud financial and government institutio­ns and even individual­s by opening fake accounts and making fraudulent purchases. It is very difficult to detect frauds committed using synthetic identities.

AI tools are increasing­ly being used in phishing campaigns of scale and in carrying out fraudulent transactio­ns and activities like betting, especially in sports.

There are also AI tools for cloning voice and launching scams. Against this unpreceden­ted onslaught, there are various technology-driven solutions available today for institutio­ns can counter fraudsters. For example, AI tools help banks to do transactio­n monitoring, which is an automated process of screening purchases, money transfers and even business interactio­ns.

AI is also now widely used in cybersecur­ity. One example is the AI-enabled financial fraud detection and prevention strategy platforms

Consolidat­ed use of AI tools can detect account takeovers and fake account creation, identify and prevent card fraud, locate credential stuffing and use of betting bots.

One disadvanta­ge that institutio­ns face today in the use of AI is that there is no one solution that fits all. There has to be a localized approach and several fraud-fighting models have to be created using ML tools. These tools can be effectivel­y used to evolve risk rules based on an institutio­n’s specific transactio­ns and fraud data. These tools learn the institutio­n’s context, make suggestion­s and create rules thereby timely flagging a potential fraudulent activity.

The accepted methods of fraud detection today are data collection, feature engineerin­g, model training, anomaly detection, continuous learning and alerting and reporting. Some of the tools that are available in the market are:

Kount, a tool that scrutinize­s transactio­ns to mitigate digital payment fraud

Featurespa­ce, which is behavioral analytics

Darktrace, which offers cyber-threat detection and response SAS Fraud Management, which again is based on advanced analytics to identify and thwart fraud in real-time

Feedzai, which helps in analyzing big data with ML to prevent fraudulent activity

DataVisor, a method of using unsupervis­ed ML to uncover fraud.

Newspapers in English

Newspapers from India