From our Bloggers

A fight against financial crime with cloud and AI as shield and sword

A fight against financial crime with cloud and AI as shield and sword
By Karan Kapoor

For profitable financial sector it is important to have smooth, omni-channel consumer experiences. Yet the new world also carries with it a multi-layered challenge: to defend the company and its clients from unremitting attempts at financial fraud.

Regulators now expect banks to handle fraud risks for real-time transactions and operations and the emphasis is on to more accurately detect, control and reduce risk. The concern is that many of today's rules-based fraud detection systems yield over 90% false positives, allowing thousands of researchers to expend hours manually checking warnings. Strengthening the model to reduce the number of false positives will therefore raise the probability that significant criminal conduct would be missed.

The secret to overcoming this problem is cloud and artificial intelligence (AI) technologies. According to McKinsey & Co, banks investing in data processing, process integration and advanced predictive modelling using deep learning and AI will lower enforcement error rates from more than 30 per cent to less than 5 per cent and minimise false-positive warnings to less than 50 per cent.

This potential is recognised by financial companies. "They understand that technology will play a role in helping them fulfil both the needs of their customers and daunting regulatory criteria in the fast-changing, tech-driven environment banks that now live in," writes Bill Borden, Microsoft's executive vice president of global financial services in an industry blog. “The industry has a tremendous opportunity to accelerate the use of cloud and AI as a catalyst to improve competitiveness, drive growth, elevate customer experiences and keep ahead of changing regulations and cybersecurity threats.”

In Hong Kong, for example, HSBC is developing new cloud platforms to allow its data to be properly structured and analysed, and to make more key consumer experiences and apps accessible in a mobile-first environment while maintaining bank-grade protection. Based on Microsoft Azure, the bank's PayMe for Business software helps companies to receive payments quickly and use based-in insights to boost revenue and operations. Adopting an architecture based on micro-services with Azure Kubernetes System and Azure Database for MySQL allowed HSBC to develop independent services with greater protection and better up-time apps. A versatile machine learning analytics framework, based on Azure Databricks, makes it easy to easily recognise user-commercial-transaction interactions and trends and provide customised customer service while rapidly identifying suspicious behaviour.

One main problem for financial institutions is that fraudsters are increasingly evolving their tactics and testing out new strategies, reacting more than the company will to regulatory changes. Yet clever technologies – such as the Microsoft Azure-based BioCatch behavioural biometrics system – are delaying this cat and mouse activity. These tools use AI to research how a person communicates with the company's website or smartphone device, from how it handles and contacts the smartphone to how it corrects typos. When comparing the experience of today with the precedent, they will determine whether it is anomalous.

Microsoft's own fraud detection technology leverages AI to allow accurate, near-real-time online fraud detection on a scale – whether the data is on-site, in the digital cloud, or at the accurate edge of mobile phones. Financial institutions can use Azure Machine Learning to adapt models for fraud detection to their own online banking systems and events, identify and train them to suit the digital technologies of the bank. We can detect potentially suspicious practices and analyse real-time interactions, events, and transactions by reviewing client log data, transactional records, and user data from internet app-based banking and other digital platforms.

Microsoft Dynamics 365 Security Defence also helps banks and e-commerce retailers to reduce security costs and raise approval rates for consumer payment purchases while providing a streamlined shopping experience. This uses AI technologies that constantly studies and adapts to emerging fraudulent trends, and its Transaction Acceptance Booster functionality often fixes declines which are unlawful.

Microsoft's dedication to reliability, protection, privacy, accountability, and openness is behind much of this operation, supported by services such as the Azure Trust Center and the Azure Security Centre.

“Because of the hyper-connectivity, we know that regulatory compliance is and must be a top priority, which is why Microsoft continues to work with financial services regulators and customers to ensure our cloud services help customers meet their strict regulatory requirements,” says Borden. “This longstanding engagement with regulators around cloud computing is one of the reasons more than 90 per cent of global systemically important financial institutions depend on Microsoft cloud services.”