How governments are using AI and analytics for proactive fraud risk management


Artificial intelligence and advanced data analytics can help governments swiftly detect and manage fraud, said private- and public-sector speakers at a recent webinar presented by GovInsider in partnership with SAS.

The webinar “Elevating Public Sector Fraud Management with AI and Advanced Analytics” featured speakers (clockwise from bottom right) Ravin Kaur from the Companies Commission of Malaysia; Robin Ng from the Inland Revenue Authority of Singapore; and Shaun Barry from global data analytics firm SAS. Image: GovInsider

“Fraud in the public sector is an escalating arms race,” declared Shaun Barry, Global Director – Risk Fraud & Compliance at SAS, a global analytics company, at a recent GovInsider webinar titled “Elevating Public Sector Fraud Management with AI and Advanced Analytics” alongside government leaders from the Inland Revenue Authority of Singapore (IRAS) and Companies Commission of Malaysia (SSM). 


Likening the war on fraud to the Cold War’s nuclear arms race, Barry made the case that governments and sophisticated fraudsters are locked in an algorithmic arms race, with mathematics, advanced analytics, and AI the weapons of the day. 


GovInsider highlights the key takeaways from the discussion, in which panelists discussed why agencies are turning to AI and analytics to combat fraud today. 

AI and analytics: From an afterthought to a core competency for governments 


“Today, we’re actively leveraging on AI and advanced analytics to not only detect or manage tax frauds, but it’s actually the foundation of overall compliance risk management,” said Ng Sy Horng Robin, Assistant Commissioner (Data & Digital Ecosystems Division) and Chief Data Officer with IRAS.  


From compliance filing to tax recovery and managing government payout risks, the technologies now cover the “whole gamut” of these functions, he adds. 


Illustrating IRAS’ risk-based approach, Ng shared that the tax authority currently uses AI and advanced analytics to develop risk profiles for taxpayers, and then tailor the interventions – whether it is related to compliance or service – based on the risk profiles. Analytics also helps the agency detect anomalies and trace potentially fraudulent activity. 


Similarly, Ravin Kaur, Senior Advisor with SSM, shared that the regulatory body uses data analytics to detect, prevent, and mitigate potential cases of financial crime. 


This is true for the region at large as well. As Southeast Asian governments increasingly roll out national digital ID programmes, they are increasingly weaving in AI and analytics into integrity checks to detect and deter frauds in these programmes, Barry noted. 

Implementing a clear data and AI strategy 


Even before digitalisation, detecting fraud has always been tedious to begin with, said Ng. With the rapid digitalisation of finance, the manual approach – whether with physical or digital records – is no longer tenable today, amidst increasing transaction volumes, citizen expectation of real-time services, and manpower shortage.   


AI and analytics provide a very efficient way to comb through the data and pick up fraudulent transactions in a matter of seconds, said Ng. Speedy detection is a key motivating factor, especially at the increasing pace of frauds happening in the real world, he added. This is why agencies need a clear data and AI strategy in place, he explained. 


But for such a strategy to be successful, leaders need to get their organisation to adopt a data-first and data-ready culture. 


Kaur explained that it is vital for civil servants to understand the value of machine learning and data analytics, as well as the key risks around the use of such technologies. This is why agencies need to implement a governance framework internally to ensure transparency and accountability when it comes to using AI. 


Similarly, Barry cautioned not to let analytics and AI run its course in decision-making. Rather, there should always be a human in the loop who can be accountable for the decisions made from data gathered from these technologies.  


“Simply allowing an algorithm to make a decision that’s unchallenged or unchecked by a human being is a recipe for disaster in government,” he said. 

Public-private partnerships the way forward 


Kaur added that the public and private sector need to collaborate in terms of governance frameworks around ethical AI, given the rapid pace of AI advancements today.  


This is why financial institutions are working closely with governments to establish AI governance frameworks, she noted. For example, the Monetary Authority of Singapore partnered with the industry to launch a generative AI risk framework for the financial sector last November. 


Aside from partnering on ethical AI, Ng shares the possibility of public-private partnerships extending to co-creating more efficient digital solutions for fraud detection and management. 


“Imagine if audit submissions are built into the software. At the end of the day, when it comes to filing, all I need is one click for submission. Basically, I don’t have to file a separate return, and everything gets submitted to IRAS,” Ng shares an example.  


Additionally, Barry highlighted the challenge of combatting as a common thread binding both public and private sector and noted that organisations can benefit from sharing best practices. 


“There are only so many ways to cheat a system, and there are common patterns you see across governments and even other industries like banking, insurance, telecommunications, and law enforcement.  


“While they are a little bit different, you see the same basic patterns of [fraud] behaviour that someone in the government or private sector has likely experienced. At the end of the day, governments and the private sector can collaborate to facilitate that knowledge-sharing,” Barry said. 


To learn more about how your agency can tackle fraud effectively, protect public services, and ensure taxpayers receive the value they deserve with an enterprise analytics strategy, download SAS' new whitepapers here. You can also watch the webinar discussed above at this link