With scammers in overdrive, fuelled by the twin accelerants of a global downturn and cheap AI, telecos need to act. Carl Eastwood takes a look at what sorts of fraud they are up against and how they can use the self-same AI to tackle the problem
In times of economic difficulty, scammers thrive. They exploit people’s increased financial vulnerability, anxieties and fears. During the Covid-19 pandemic, UK telecoms regulator research showed almost 45 million people received at least one scam text or call over a three-month period. This is more than two-thirds of the entire population of the UK.
The same picture was true across Europe. According to a report by Hiya, French residents received 12.4 billion spam and fraud calls in 2021, with Spain and Germany receiving 8.9 billion and 5.9 billion calls from scammers respectively.
More recently, scammers have been quick to exploit the cost-of-living crisis which continues to put pressure on millions across Europe. Scammers will lure people in with messages most likely to resonate with them – help or advice relating to energy prices is a good example, as well as claiming customers are entitled to rebates and refunds before directing them towards malicious websites.
Understandably people are concerned, with Sky News reporting that in the context of the cost-of-living crisis, 42 per cent of adults in the UK are more worried about scams than they were before the COVID-19 pandemic.
To help reassure and protect customers, communication service providers (CSPs) are already taking a leading role in preventing scammers abusing the network. There has been positive action in recent months – with more developments on the horizon.
The progress so far
At the end of last year, major UK CSPs signed a voluntary charter to coordinate efforts with the government and other industries to reduce the impact of scams – including the scam calls and SMS phishing directly affecting telecommunications customers.
Having acknowledged the rise in suspicious calls and texts, UK regulator Ofcom’s new rules and guidance on spoof calls will officially come into force in May, requiring all telephone networks involved in transmitting calls to identify and block spoof calls where technically feasible. Although UK CSPs are already making significant progress in reducing scams originating from their network there is more work to be done.
And in the EU, the European Electronic Communications Code (EECC) has been in place for just over two years, requiring CSPs to implement measures to protect consumers from scams, including establishing a system for reporting and blocking fraudulent numbers.
These steps are crucial to enhancing protection for customers, however, it is important CSPs do not lessen their efforts after complying with regulations – for optimum and timely protection, more layers of defence are needed.
The challenges to overcome
CSPs are fighting the battle on two fronts. They are charged with preventing phone and SMS-based scams that originate from their network that lead to identities or monies being stolen. However, the scams can also cause financial harm to the CSP if the stolen identities are used to originate new accounts or take over existing ones, resulting in Subscription Fraud.
As with during the pandemic and the cost-of-living crisis, scammers will seize any opportunity to adapt their efforts, often making it difficult for providers to pre-empt and prevent new scams.
It’s not just the content of scams that’s evolving but the technology which fraudsters use too. In turn, the industry can expect to see more sophisticated scams and cases of fraud arise. For one, the rise of generative AI chatbots has meant poor English and grammar may no longer be the first indicator of a malware text or email.
To keep one step ahead and make the most of technological developments themselves, CSPs need to invest in fraud detection systems that keep up – in real time. Those who fail to take meaningful action to protect customers will likely see a negative impact on their reputation, revenue, and customer trust.
How AI can help
Artificial intelligence (AI) enabled prevention and detection measures have had a significant impact on deterring fraudsters – helping to spot patterns and suspicious behaviour.
The existing legacy fraud systems CSPs use to detect fraud are often business rules based, with high false positives, and do not update instantaneously which can be critical in a world where a customer can walk in and obtain devices within minutes from a retail store. By contrast, combining automated real-time decisioning with AI and machine learning can prevent fraud while reducing friction for genuine customers.
The analytics lifecycle can also be automated so that when the performance of a machine learning model falls below a defined threshold it can trigger the automated retraining and deployment of that model, so that any attempt by the fraudsters to stay one step ahead is mitigated.
Finally, AI can augment the human-based investigation process to provide more insight and guidance in what to look for. This could be through use of natural language processing to help explain to an investigator why something has raised an alert, or by presenting more visual-based insight like social network/link analysis to find links between customers/fraudsters.
Many providers now offer more services across different channels, including those related to calls, internet services and TV – yet while upselling is good for business, the potential for subscription fraud also increases.
Subscription fraud happens when a fraudster uses a stolen, synthetic or even their own identity to acquire devices and services – with no intention to pay. Within more mature markets where pay monthly mobile services are dominant, it is the prevalent fraud type. However, it is also the fraud type best tackled by application of machine learning due to the variety of data collected during the application process.
One of SAS’ CSP customers has been able to reduce subscription and dealer fraud by over 80% after replacing business rules with machine learning. It has also reduced investigations by nearly 75% after automating the decisioning process.
Getting the right data at the right time is key to making the right decision in time to prevent the fraud loss. Using data intelligently to make accurate and automated decisions requires CSPs to pursue the use of AI, otherwise they risk being the fraudsters’ CSP of choice when determining who to target first.
Getting the balance right
Safeguarding customers from scams and preventing fraud is all about enhancing the customer experience – the last thing companies will want to do is alienate their customers in the process through false positives in their fraud detection.
AI technology can help eliminate false positives, avoiding customer frustration should they be unable to access their services or if their accounts are blocked. The CSPs succeeding here will increase customer loyalty – striking the right balance between caution and protection.
Faced with the increasing scale and sophistication of scammers, only the providers who have built strategies, and invested in fraud detection capabilities, will be able to tackle it.
Carl Eastwood is Fraud & Financial Crimes Domain Lead at SAS, a global business providing AI and cloud analytics services to communication service providers