The nature of fraud has changed dramatically since the dawn of the digital age. According to a recent Office for National Statistics (ONS) survey, one in 10 people in the UK have fallen victim to cybercrime. Companies are no different, with so-called business email compromise schemes netting in billions of dollars for criminal gangs. International money transfer company Xoom, for example, was tricked into sending $30.8m of corporate cash to an overseas account.
This is not a new problem, but it would be churlish to blame companies, as the threat is always evolving. In the 2016 Faces of Fraud Survey, sponsored by SAS, just 34% of surveyed security leaders said that they have high confidence in their organization's ability to detect and prevent fraud before it results in serious business impact, 56% of whom cited the high level of sophistication and rapid evolution of today’s schemes.
One of the other reasons many of the respondents gave was a lack of awareness among customers and/or partners and employees, cited by 56% and 52% respectively. In order to correct this, both companies and banks - anyone with a serious interest in preventing fraud on a larger scale - are turning to data analytics to identify potential schemes and prevent them before they can take place.
The speed at which money now moves during transactions is an obstacle that cybercriminals have long overcome, and financial institutions have to prevent fraud at the same speed to protect their customers and assets or be held liable for any theft. Machine learning algorithms can help banks find anomalies in the data indicative of fraud in real time, without disturbing the flow of legitimate transactions that must flow seamlessly.
At the recent Big Data & Analytics for Banking Summit in Melbourne, Steve York, General Manager of Group Compliance, Security & Business Resilience at Bank of Queensland, discussed how the nature of fraud had changed, with criminals knowing that they only have a certain amount of time to take money from accounts of people whose data they have sold, and they are subsequently selling information on using the dark web to other criminals.
Criminals are now working collaboratively as never before, and the approach to using data to stop them must be same. Fiserv is one company leading the way with its new, advanced predictive scoring model specific to Automated Clearing House (ACH) transactions. Designed to deal with fraudulent electronic payments as they happen, Payment Fraud Manager detected more than 90% of fraud in Fiserv model validation tests, while reviewing only 2% of the transactions. The algorithm leveraged industry-level data from hundreds of financial institutions of all sizes across the US, as is now common practice. In the 2016 Faces of Fraud Survey, 68% of insurance respondents and 64% of financial services respondents said that within-industry data would be very valuable or valuable, while 53% of insurance and 48% of financial services reported that outside-industry data would be very valuable or valuable.
Fraud is a growing problem, despite the efforts to prevent it. Criminals are now extremely skilled at getting hold of people’s data, through tricking both individuals and companies who hold that data. For police, it is often almost impossible to track down criminals to track down those who would steal data as they often operate out of countries like North Korea and Romania. Fraud needs to be stopped in real time, and companies, banks, and individuals need to work together and share all the data possible to ensure this happens.