How AI can help in the fight against fraud
Telecoms fraud is an expensive business, writes Katia Gonzalez, the head of fraud operations at BICS. According to the most recent Global Fraud Loss Survey by the Communications Fraud Control Association, fraud cost the industry a huge US$29.2 billion last year alone. This figure was down by more than 23% on 2015’s total, which is certainly encouraging, but the potential remains for the various types and methods of telecoms fraud to do some substantial financial and reputational damage.
International Revenue Share Fraud, for example, is an especially damaging and hard-to-prevent technique, which has continued to grow in popularity over the past decade and which was responsible for generating more than US$6 billion for fraudsters during 2017. Second to this was Interconnect Bypass, or SIMBox Fraud, a relatively recent technique in which VoIP calls are redirected onto mobile networks, effectively bypassing the interconnect toll charging points, thereby enabling fraudsters to avoid payment of the official call termination fee of an operator or MVNO. This technique alone cost operators more than $4 billion in lost revenue in 2017. And as the number of services offered by operators continues to grow, so too does the opportunity for criminals to use stolen or fabricated identities to access a range of value-added services, from mobile internet to mobile banking. Subscription fraud, the most popular method of fraudulently accessing networks and services, was responsible for a loss of more than US$2 billion last year.
We have previously explored the need for a collaborative, industry-wide approach to tackling fraudulent activity, with operators sharing their knowledge and resources for the benefit of all. Such an approach allows operators to reduce their chances of being hit by fraud schemes and mitigate the amount of fraudulent numbers remaining active around the world.
There is, however, a burgeoning technology that can further inform this intelligence, improving the ability of operators to identify trends and anomalies in network traffic and, in turn, enable them to take a more proactive approach to preventing fraud.
Automation and analytics
One of the hottest topics at this year’s Mobile World Congress, artificial intelligence (AI), is increasingly gaining momentum throughout the telecoms industry, particularly with regard to optimising networks and processes through greater automation, analytics, and virtualisation. Seen as playing a key part in an operator’s digital transformation, it is hoped that AI will enable them to restructure their business models for greater efficiency and revenue.
AI can also be applied to the battle against fraud. Ensuring the security of an operator’s network can be particularly time and resource intensive, for instance, but applying even basic analytical algorithms to automatically determine whether or not it makes basic sense for a packet to be coming from a particular network would immediately free up resources. If the same principle was also applied to analysing voice or SMS traffic, the operator would immediately see an increase in efficiency. As an additional benefit, the security team would then have more time to focus on monitoring network traffic for potentially fraudulent activity.
What’s more, the analytical capability of AI can be used to supplement the intelligence derived from the previously discussed collaborative approach, to more accurately identify trends, and anomalies such as irregular traffic patterns. Indeed, while AI can be used to automatically profile personalities to help operators better understand how to identify and treat customers, it can also be used to learn typically fraudulent behaviours, potentially allowing operators to stop fraud before it even occurs.
Look to the future
Telecoms fraud continues to evolve, employing various different forms and techniques to illegally channel funds from operators and subscribers into their own coffers. Developments in roaming, the global reach of the internet and the advent of new, next-generation networks mean fraud is neither limited to any one county, nor concerned with the size or maturity of an operator. Not only can it erode an operator’s bottom line, it can also lead to increased costs, legal complexities, complications in partner relations, reduced subscriber satisfaction, increased churn and service disruptions.
Fraud must be continuously assessed, and operators must keep an eye to the future for the methods and technologies that will help limit fraud exposure, enabling them to remain agile and stay one step ahead of the fraudsters themselves. Providers of the increasingly popular IPX network interconnection model, for example, already offer new digital services via the platform, including evolved machine learning fraud management, which involves smart data and AI approaches being leveraged to help forecast fraud trends before they occur.
Intelligence is a powerful weapon in the fight against fraud, and as AI technology becomes more refined, so operators should give thought as to how it can be applied to offer them the insight and the edge they need to minimise their losses.