Singapore's financial regulator is moving to transform how the country's banking sector combats scams, announcing a landmark study that will attempt to build a shared, artificial intelligence-powered detection infrastructure spanning the entire industry. The Monetary Authority of Singapore (MAS) confirmed it will launch the study in early 2027 to design what it is calling an industry-wide Scam Detection Utility — an ambition that signals a fundamental shift from institution-by-institution fraud defences toward a collective, data-driven shield for Singaporean consumers and businesses alike.
The announcement was made by Deputy Prime Minister and MAS Chairman Gan Kim Yong at the Singapore Fintech Festival on 18 November, one of Asia's most closely watched gatherings for financial technology leaders, regulators, and investors. The timing was deliberate: the Festival provides a high-visibility platform, and the choice to unveil a study of this magnitude there underscores just how central the fight against financial scams has become to Singapore's broader fintech and banking agenda.
The 2027 study does not emerge from a standing start. It draws directly on an active trial already underway in which five banks are securely pooling their transaction data to test whether AI models can detect scam patterns more effectively when trained on cross-institutional information rather than any single bank's siloed dataset. That pilot represents a significant operational and legal undertaking in its own right: convincing competing financial institutions to share sensitive transaction data — even under controlled, secure conditions — requires robust governance frameworks, legal agreements on data handling, and a shared belief that collective defence outweighs competitive considerations. The fact that five banks have already reached that threshold is itself a material development.
The logic underpinning the utility concept is well-established in cybersecurity theory but has historically been difficult to execute in banking. Scammers rarely confine their activity to a single institution; a fraud campaign typically touches multiple banks as victims transfer funds across different accounts and payment rails. A bank analysing only its own transaction flows therefore sees an incomplete picture of the threat, potentially missing the cross-institution behavioural signatures that would otherwise trigger an alert. By pooling data, AI models gain exposure to the full arc of a scam event, theoretically improving both detection speed and accuracy while reducing false positives that frustrate legitimate customers.
MAS has long positioned Singapore as a jurisdiction where regulatory rigour and innovation coexist, and the Scam Detection Utility concept fits squarely within that philosophy. The regulator has in recent years tightened the shared responsibility framework between banks and consumers in scam cases, raising the stakes for financial institutions to invest meaningfully in preventive infrastructure. An industry-wide utility — if the 2027 study concludes it is technically and legally viable — would represent the most structurally ambitious step yet in that direction, shifting the defensive perimeter from individual compliance departments to a centralised, continuously learning intelligence layer accessible to all participating institutions.
There are, of course, non-trivial challenges ahead. Designing a utility of this kind demands agreement on data standards, liability allocation when the system fails to flag a scam, governance of the AI models themselves, and mechanisms to ensure that smaller financial institutions can participate on equitable terms alongside the larger incumbent banks already involved in the pilot. Privacy considerations, particularly under Singapore's Personal Data Protection Act, will require careful navigation. The 2027 study will presumably need to address each of these dimensions before any operational deployment can be contemplated.
Regional context matters here as well. Scam losses across Southeast Asia have reached systemic proportions in recent years, with organised criminal networks operating cross-border infrastructure that individual national regulators struggle to disrupt unilaterally. Singapore, as the region's leading international financial centre, faces particular exposure: its deep integration with global payment networks makes it both a target and a potential model for how advanced economies can respond. If MAS succeeds in operationalising a cross-bank AI detection utility, the blueprint could carry significant weight for peer regulators across the Association of Southeast Asian Nations (ASEAN) bloc and beyond.
What This Means for the Industry
For Singapore's banking sector, the trajectory is clear: the era of purely institution-level fraud defences is giving way to a collaborative, infrastructure-driven model where data sharing under regulatory oversight becomes the competitive baseline rather than a differentiator. The five banks currently participating in the trial are helping shape the norms and architecture that will likely define industry obligations for years to come. When the MAS study formally kicks off in early 2027, its findings will not merely be an academic exercise — they will almost certainly form the foundation of regulatory expectations, and potentially mandates, for every licensed bank operating in Singapore. For institutions not yet engaged in the conversation, the window to influence its design is narrowing.
Written by the editorial team — independent journalism powered by Codego Press.