Detecting Fraud Before the Money Moves - How bank-led data sharing is evolving from post-event reporting to proactive fraud prevention

Banks across Asia and other major markets have made meaningful progress in strengthening fraud-response frameworks. Shared mule-account data, coordinated account freezing, improved victim-response processes, dedicated fraud hotlines, and tighter reporting channels have all helped reduce losses and improve recovery.

Central Fraud Registry

Regulators increasingly recognise that fraud and scam activity now moves too quickly for any single institution to tackle alone. Thailand is moving in this direction, with the Bank of Thailand promoting the Central Fraud Registry to support information sharing on money trails and mule accounts among financial institutions, alongside tighter mule-account measures and broader anti-fraud controls.  

That progress is important, but it is still only part of the answer. Fraud today is organised, fast-moving, and increasingly industrialised. Criminal groups reuse devices, identities, phone numbers, SIMs, social engineering scripts, synthetic profiles, mule-account networks, and digital infrastructure at scale. Funds can move across multiple accounts and channels in minutes.

In that environment, reporting fraud after it has already happened remains necessary, but it is no longer sufficient. The strategic shift now underway is towards sharing fraud markers early enough to detect suspicious activity before payment is completed. This is the direction of travel in a growing number of markets, especially where regulators, industry bodies, banks, payment operators, and telecoms providers are trying to move from retrospective response to real-time prevention.

COSMIC

Singapore is one of the clearest regional examples of structured, regulator-backed information sharing. MAS launched COSMIC on 1 April 2024 as a centralised digital platform to facilitate the sharing of customer information among financial institutions to combat money laundering, terrorism financing, and proliferation financing. MAS has also issued requirements and guidance for prescribed financial institutions to establish controls around the sharing of risk information through COSMIC.  

COSMIC is not primarily a retail scam-prevention platform. It is framed around serious financial crime. But it is an important milestone because it shows that regulators are willing to create controlled legal, operational, and governance frameworks for inter-firm intelligence sharing where the risk justifies it. That principle is highly relevant to the next phase of scam and mule-account prevention.

FMLIT

Hong Kong provides one of Asia’s strongest examples of how this model can become operationally effective over time. Through the Fraud and Money Laundering Intelligence Taskforce, the HKMA, banks, and law enforcement have built a public-private partnership that has steadily increased intelligence-led outcomes.

The HKMA reported that intelligence-led suspicious transaction reports filed by member banks increased by 319% in 2022 compared with 2021, contributing to the identification of 6,819 new suspicious accounts and more than HK$120 million in criminal proceeds restrained or confiscated.   By 2024, the HKMA said suspicious transaction reports made by banks following intelligence shared via FMLIT had quadrupled year on year, while criminal proceeds restrained or confiscated increased by 34%.  

Hong Kong has also been expanding bank-to-bank information sharing. FINEST, launched in 2023, was designed to help banks share information to detect and disrupt fraud and mule-account networks more effectively.   In 2025, the HKMA, Hong Kong Police Force, and Hong Kong Association of Banks announced further measures to prevent, detect, and disrupt fraud and associated mule-account networks.  

What makes Hong Kong especially relevant is that the model is evolving beyond static case sharing. The HKMA has repeatedly highlighted the importance of network analytics and more dynamic, customer-behaviour-oriented detection models. In 2023, the HKMA said around 60% of retail banks were deploying network analytics, more than twice as many as three years earlier, and linked this capability to stronger intelligence-led reporting and law-enforcement outcomes.   In 2025, the HKMA also noted that some banks had already shifted away from transaction-focused, rules-based monitoring systems towards more dynamic, customer behaviour-oriented models, while signalling that the industry still needed a coherent framework for wider adoption.  

This is the real next-generation theme: not simply sharing confirmed bad accounts after the event, but using richer data, link analysis, behavioural signals, and earlier indicators to identify suspicious activity before losses escalate.

Downunder

Australia is moving in a similar direction, though through a different model. The Australian Banking Association has positioned the Scam-Safe Accord as a sector-wide framework to disrupt, detect, and respond to scam activity. One of its most important initiatives is Confirmation of Payee, a name-checking service launched in July 2025 to help customers verify whether the receiving account name matches the intended payee before a transfer is made.  

This is a simple but important example of the broader shift from post-event reporting to pre-payment intervention. Rather than waiting for fraud to be reported, the control is designed to reduce the likelihood of funds being sent to the wrong or fraudulent destination in the first place. The ABA has also been explicit that stronger anti-scam outcomes require cooperation across banks, government, and other sectors, not just better controls inside individual institutions.

Europe Focus

The UK is taking this logic further by extending collaboration beyond banking. The Payment Systems Regulator expanded Confirmation of Payee so that, by October 2024, over 99% of Faster Payments and CHAPS transactions were covered by the name-checking service, explicitly positioning it as a vital anti-fraud measure.  

In parallel, UK Finance and the mobile industry have been working on Scam Signal, a model that uses real-time telecoms data to help banks detect signs of authorised push payment fraud attempts.   That matters because many scams begin long before the payment itself: through spoofed calls, social engineering, SIM-related activity, remote access tools, messaging platforms, or other telecom and digital signals. The lesson is clear: the future anti-fraud model will not be banking-only. It will increasingly involve telecom operators, payment-system utilities, digital platforms, and other trusted parts of the ecosystem that hold useful early-warning signals.

Back to South East Asia

Malaysia is also strengthening its framework. Bank Negara Malaysia, PayNet, and financial institutions launched the National Fraud Portal in August 2024 as an integrated platform to strengthen operational anti-fraud capabilities. PayNet describes the platform as a centralised infrastructure that supports faster reporting, tracing, freezing of suspicious transactions, and coordinated intervention across banks, regulators, and enforcement agencies.  

Malaysia’s approach reinforces the same broader lesson visible elsewhere: stronger outcomes depend on centralised coordination, better information quality, faster escalation, and clearer operational workflows across institutions.

For Thailand, the implication is not that current measures are inadequate. On the contrary, the steps already taken are positive and necessary. The Central Fraud Registry, tighter controls around mule accounts, dedicated emergency contact channels, and stronger coordination frameworks all represent meaningful progress. But there is a strong case for accelerating the next phase.

That phase should focus less on reporting known fraud after funds have moved and more on sharing risk indicators that help banks identify suspicious beneficiaries, linked mule patterns, anomalous transaction journeys, and broader ecosystem signals before value is released. The Bank of Thailand’s own direction of travel already supports this, particularly through its push for wider information sharing, more consistent handling of mule accounts, and stronger cross-sector anti-fraud coordination.  

The practical next step for the Thai market would be a focused industry discussion involving the Bank of Thailand, the Thai Bankers’ Association, leading traditional and digital banks, payment-system stakeholders, and selected telecom participants. The objective does not need to be a large, complex new utility from day one. A more practical starting point would be a narrow pilot around pre-payment fraud markers.

That pilot could include shared mule-destination intelligence, beneficiary-risk indicators, linked-account patterns, device or identity reuse markers, and other early-warning signals that could trigger step-up checks, payment delays, beneficiary verification, or intervention before funds leave the sender’s account. A second workstream could examine how telecom and digital-platform signals might be incorporated over time, given how many scams now originate outside the banking perimeter.

The strongest global lesson is that fraud prevention becomes materially more effective when institutions share earlier, act faster, and widen the ecosystem beyond banking alone.

Conclusion

The direction of travel is increasingly clear. The first generation of fraud collaboration focused on reporting suspicious accounts and responding faster after the event. The next generation is about detecting fraud earlier and stopping payments before criminal funds are successfully transferred.

Singapore and Hong Kong show that regulator-backed information sharing can be formalised and scaled. Australia and the UK show how pre-payment controls, beneficiary verification, and broader ecosystem collaboration can push the industry further upstream. Malaysia shows the value of centralised operational coordination. Thailand has already made a strong start.

The opportunity now is to build on that momentum and evolve from a retrospective reporting model to a more proactive, intelligence-led prevention model that reflects how modern scams and mule networks actually operate. The goal should be simple: detect the fraud before the money moves.