Steve Hadaway, EMEA General Manager at FICO

Fraudsters will always follow the path of least resistance.

That’s why, looking back to the introduction of EMV chip cards (which protect card payments at the point of sale) in the UK, France, and Australia, we see an immediate shift toward card-not-present (CNP) fraud to circumvent the protection.

Card details are being compromised on a worryingly regular basis in high-profile data breaches;as a result, it’s become easier than ever to buy and use card details online, which facilitates borderless criminal activity. It’s for this reason that CNP now comprises 73% of all fraud committed in Europe. What’s more, fraudsters are increasingly focusing their activities on those financial markets with the least protection against CNP fraud.

Fraud migrates across Europe

Against this challenging backdrop, in 2017 the UK became the first market in many years to achieve a significant reduction in card-not-present fraud -losses fell by 8%, even as online card utilisation continued to rise. This is reflected in FICO’s 2017 European fraud map, which measures and compares fraud levels across 19 countries. CNP fraud still accounts for the lion’s share of card fraud losses in the UK (72%), but the reduction achieved is significant.

Unfortunately,the borderless nature of CNP fraud means that preventing it in the UK does not remove the threat altogether – criminals can find and pursue other targets,just as they did when EMV was introduced.Although UK CNP levels have decreased, losses in Austria, Denmark, Norway, Sweden, Poland, and Russia have all increased. Hungary and Denmark are the biggest victims of CNP migration: Hungary has seen an over 300% increase in CNP fraud losses over the last three years, while losses have doubled in Denmark in the same period. With UK CNP fraud levels expected to fall further in 2018, this trend is set to continue.

How has the UK achieved such an important victory, and what can other markets do to protect themselves from fraudsters looking for new targets?

Tackling a fraud epidemic

For banks looking to build an anti-fraud strategy, here are two tactics they should prioritise:

• Use real-time profiling techniques to better understand customer behaviour

By combining artificial intelligence (AI) with customer profiling and machine learning algorithms,alongside business rule-writing capabilities,banks can make real-time decisions on whether to approve certain payments. This combination is known as real-time dynamic profiling.

It monitors key transaction behaviours and patterns for each profiled entity to ensure uncharacteristic payments or activities can be easily recognised. Profiles are updated constantly and include multiple variables like the customer’s favourite ATMs, countries, currencies, typical amounts stored, time of day and week when the customer typically uses their accounts, as well as other indicators of risk, such as address or phone number changes.

By using these profiling techniques, the system is able to flag anomalous behaviour as it happens. If intervention is necessary, either a fraud analyst or the actual customer is able to review and then confirm or deny the legitimacy of a transaction.

• Train AI anti-fraud models on the largest possible data sets

Introducing new fraud prevention technologies can be a difficult balancing act. Banks need to ensure they prevent a higher quantity of fraud cases without also increasing the false positive rate.Otherwise, they risk irritating customers by denying legitimate transactions. To ensure as high an accuracy level as possible, banks should train their machine learning algorithms on as much data as they can – which means data sharing with competitors can be extremely beneficial.

In order to tackle CNP fraud more effectively, banks need to work together to implement real-time solutions that adapt constantly and make decisions based on dynamic profiling. As countries like the UK continue to record tangible improvements and successes against this type of fraud, those who fail to shore up their defences will see their CNP problem continue to worsen: they must act now if they are to avoid future fraud migration coming their way.

For more insights into fraud trends and technologies, including the five keys to using AI and machine learning for fraud detection, visit our blog at www.fico.com/blogs.