Leading Back Office Systems for Banks

In-depth Supplier Profiles and User Lists. Subscribe now

In-depth Banking Tech and FinTech Research

Over 300 research reports that's updated quarterly. Subscribe now

RPA in Banking: Strategies and Pitfalls

Insights on RPA Suppliers and Case Studies. Subscribe now

Legacy fraud prevention not keeping up with changing security threats

Frederik Mennes, Director Product Security, OneSpan

Existing fraud prevention systems based on pre-defined rules are not keeping up with the changing nature of security, and financial institutions must turn instead to detection solutions that leverage AI through supervised and unsupervised machine learning, a leading security expert has advised.

Frederik Mennes, Director Product Security, Security Competence Centre with security solution developer OneSpan, said modern risk analytics solutions present the best answer for banks in a world which saw a 78% increase of fraud trials in 2018, according to KMPG. He further noted that British banking customers lost £500m to fraud in the first half of last year while the FCA has said the UK financial services industry is spending over £650 million annually in dedicated staff time to combat fraud, laundering and other financial crimes.

“There are two main drivers for the adoption of risk analytics,” said Mennes. “First is the fraud that banks are experiencing in their digital channels – online and mobile banking. There is also the increasing need for regulatory compliance by financial institutions. In the EU there’s PSD2, mandating the use of transaction risk analysis to make sure they can spot problems in their digital banking channels.”

In terms of approach, said Mennes, it is possible to segment banks into different approaches: “There’s the Tier One banks – HSBC, Citi and so on – that are quite advanced in the adoption of risk analytics,” he explained. “Smaller and medium-sized banks are much less mature. There’s a division too between digital banks, which are aware of the care they need to take of their digital customers, and the incumbents, some of which are lagging slightly.”

It’s important, he said, for banks to use solutions that can address multiple channels at the same time – not one solution for ATMs, one for web-based banking, one for mobile banking. “That’s the only way to spot some fraud,” he concluded. “Banks also need to use machine learning to dynamically analyse fraud. ML and AI will automatically help you identify what constitutes fraud, early and with less manual effort.”

He added that machine learning algorithms can analyse transaction data, only flagging suspicious transactions with higher risk scores; detect complex patterns in milliseconds that are difficult for analysts to identify; reduce false positives, because AI is capable of analysing a much larger set of data points, connections between entities and fraud patterns; etter detect new and emerging fraud.

Related Posts