Speaking about identity fraud and security concerns in the telecom market, Rohit Maheshwari, Head – Strategy & Products, Subex had a conversation with Voice&Data. In this discussion, he elaborates the effects of identity fraud and how handset fraud solution can be one of the ways to prevent them.
Has there been an increase in identity fraud in the telecom market?
Sophistication and impact of identity fraud in the telecom market have certainly increased and there are multiple reasons for that.
• Handset prices climbing northwards: You would be surprised to learn that over the last decade or so, telco revenue mix, especially in developed markets, has dramatically shifted from usage-based revenues to recurring monthly fees which are primarily related to handset costs. Handset prices have continued to move northwards and have breached the $1000 mark. The increase in prices makes it particularly lucrative for fraudsters to use identity fraud as an attack vector to steal handsets.
• Continued instances of data breaches: Data breaches at Facebook, Yahoo, Quora, Sony, and so on have fueled a very lucrative underground marketplace for buying and selling of stolen identities.
• Digitalization of sales channels: To buy a telco service, one no longer needs to walk into a store. Fraudsters can now order hundreds of handsets, all from the anonymity of web/mobile, without having to be present at a physical store using stolen IDs and Credit Cards. Contrast this with having to physically walk into stores each time one needs to buy a telco service.
How differently would handset fraud affect both subscribers and telcos?
Handset fraud is typically committed using stolen identities. This results in telcos suffering significant monetary losses which are made up of the direct cost of the handset, cost of sale and cost of provisioning. On the subscriber front, the sufferer is usually the rightful owner whose identity has been stolen to perpetuate handset fraud. The subscriber faces intrusive scrutiny, the risk of reputational damage, lowering of credit risk and denial of access to services.
How is Subex’s handset fraud management solution better than other management solutions?
Traditional approaches to detecting and combating handset fraud focused on the limited transaction data available at the time of customer on-boarding. However, we believe that to be able to successfully combat handset fraud, a lot more can be done and should be done. The key is in being able to connect the dots from across different sources of data and applying advanced AI driven analytics to detect handset fraud. We combine sale transaction data, customer demographic data, syndicated data such as real-time credit bureau data, national ID data, facial recognition, video captcha, geolocation, IP location, social network analytics, dealer/ retailer performance data, incentives data for dealers/ internal staff. This ability to connect dots across data sources and help identify all in real-time sets Subex’s fraud management solution apart from the competition.
How does profiling of subscribers work in the handset fraud management solution to recognize the fraudsters? Is SubexSecure the profiling system that Subex speaks of?
Just to clarify, SubexSecure is aimed only at IoT and Digital security. It leverages a one-of-its-kind honeypot network that combines physical devices and device emulations to generate IoT / ICS signatures. However, it is not aimed at identity or handset fraud.
In the handset fraud management space, Handset fraud solution (also called as Subscription Fraud solution) is a component of Subex’s ROC Fraud Management solution and is available to telcos both as a standalone solution or as part of ROC Fraud Management Solution. As mentioned above, a successful strategy is to be able to combine several different sources of data and then let Artificial Intelligence (AI) take over. This involves building a Fraud Management Solution which can integrate easily with myriad sources including third-party syndicated data, support feature extraction and engineering, application of AI models and consumption of output all in real-time to avoid unnecessary friction in the customer onboarding experience. Having worked on this problem at several customer sites, we have developed deep insights on relevant data sources and feature selection which we fine tune for local requirements.