With the ongoing pandemic fuelling the growth of ecommerce and the need for digital payments, Asia’s payment sector is projected to exceed US$1 trillion in annual revenue by 2023.
That's according to McKinsey. As consumers increasingly shift towards digital platforms for their daily necessities, financial service institutions (FSIs) are challenged with not only handling increased transaction volumes but also detecting and sieving out fraudulent traffic.
In order to quickly identify bad actors, FSIs need to utilise and access greater intelligence data to better detect and prevent cybercrimes. What this calls for is new ways of structuring and analysing data to enable FSIs to do so at scale and in real-time – without compromising the user experience of legitimate customers.
As organisations move to cloud platforms for better operational and cost efficiencies, up-to-date cybersecurity is paramount. Thales' 2020 Data Threat report revealed that 45% of all data from Asia-Pacific companies is currently stored in the cloud, including 42% that is deemed to be sensitive information.
Yet a whopping 99% of these businesses admit they have some sensitive data in the cloud that isn’t encrypted. In fact, just 52% of sensitive data they have stored in the cloud is secured through encryption, while 42% is protected with tokenization.
Tapping into the cloud for criminal intel
As FSIs in Asia move mission critical workloads to the cloud, there are opportunities to integrate real-time detection engines – powered by machine learning (ML) capabilities – to better use data to quickly detect and block fraudulent transactions.
Banks are also looking beyond their own data, using geolocation and other non-traditional datasets to enhance this capability, allowing them to build a more comprehensive view of their customers and identify any transactional anomalies.
Google’s own search results sometimes offer information sourced from our Knowledge Graph (Google’s source of truth for the internet) that can add additional insights into an institution's data – a database of billions of facts deemed relevant and publicly sourced. External data sources, of course, should only be obtained on user consent.
Knowledge graphs build a database of connections which allows FSI’s to link information using structured and unstructured data to establish facts about an area of interest. They tap ML and graph technologies to give AI models context, integrating information from a wide range of data silos and wrapping this with learning and reasoning.
The Panama Papers, for instance, demonstrates how the ability to analyse 11.5 million documents – or 2.6TB worth of data – and make connections between people, the flow of money, and legal structures, can uncover financial crimes and corruption.
Safeguarding from real-time risks
Data is a critical enabler in threat detection systems, but FSIs need the right infrastructure to crunch the data in real-time and identify risks swiftly.
In a traditional rules-based model, banks need to code in specific patterns that flag potential financial crimes and regularly modify the code to reflect changes in these patterns or rules. FSIs can no longer depend on static rules to detect threats. They have to adopt a dynamic, risk-based model powered by AI/ML that will enable them to proactively seek out changes in criminal activities and move quickly to mitigate risks.
Doing a better job at flagging risks and threats also yields more cost savings since banks can spend less time and money investigating every incident, including false positives. These resources can be redirected toward combating real threats and ensuring customer satisfaction will not be affected.
FSIs such as Australian forex and payments firm, OFX, process thousands of transfers daily and, with transactions climbing, need a smarter monitoring system to keep up with the changing threat landscape. Working with Google Cloud to deploy AI-powered solutions through Quantexa, OFX have been able to slash the time it used to take to crunch the same amount of data by 50%, allowing them to scale as and when its transaction volume requires, all while enhancing its crime prevention capabilities.
To ensure compliance, FSIs need to consider action, including adopting a risk-based detection system that is powered by AI/ ML, and building knowledge graphs that tap consent-based data beyond their own sector.
Ultimately, FSIs need to think about how they can ensure their monitoring systems can scale quickly alongside transaction volume while protecting their users. Without this, we can expect to see more breaches, leaks, and attacks. The pressure is on for FSI leaders to make changes to adapt to the new digital environment, but with tools such as AI-enabled risk detection systems, the only question is why haven’t they already?