Payments processor EastNets is to apply machine learning technology to detect fraudulent payments transactions passing over the Swift network.
The new AI-based tool, en.SWPG continuously adapts its predictive models based on historical transactions. These patterns form the basis upon which future transactions are evaluated and classified.
The technology, which will be released to all client globally next month, constantly optimises the accuracy of its models, says the vendor, sorting out genuine transactions from suspicious ones, markedly reducing the number of false positives and reliance on the human factor.
Hazem Mulhim, CEO, EastNets says: “Artificial intelligence is building on old, static, rule-based systems in compliance and risk management solutions by automating customer onboarding, watch list filtering and fraud detection. en.SafeWatch PaymentGuard tackles the rising risk and magnitude of payments fraud in the financial industry by classifying transactions according to past patterns in order to detect anomalous or suspicious transactions."
The EastNets Bureau connects 260 banks to the Swift messaging network, including some of the biggest financial institutions in the Middle East. It hit the headlines in April when the ShadowBrokers hacking crew dumped a cache of old hacking tools on the Internet alongside claims that the NSA had used the highly-classified technology to infiltrate a Swift Service Bureau run by EastNets.
EastNets at the time said the published documents lacked credibility and the claims made by the hackers were "totally false and unfounded".
The EastNets tech upgrade follows the publication at the weekend of Swift's 2016 annual report which showed a 31% decline in profit to EUR47 million, following additional investments in security. In the wake of multiple hacks at Swift member banks, the interbank co-operative has tripled the size of its security team over the past three years and plans to add additional staff. Executive management goals and incentives have also been aligned with security targets.