The Monetary Authority of Singapore (MAS) is testing AI models that use cross-bank and public-private data to identify suspicious accounts and transactions.
The regulator is working with law enforcement and the banking industry on the initiative.
“The aim is to detect sooner, intervene faster and reduce losses,”
MAS Managing Director Chia Der Jiun said at the Global Fintech Fest in India.
The regulator expects to have findings from the project by the end of 2026.
MAS Expands AI Access and Oversight
Financial institutions in Singapore already use AI for fraud detection, credit underwriting, risk management, regulatory compliance, customer service and document processing. Many of these applications now operate at scale.
However, Chia noted that AI still falls short in areas requiring complex interpretation, strategic judgement and human interaction.
While many companies use AI, fewer have recorded significant productivity gains.
Chia expects this to improve as employees receive more training and organisations redesign their processes and products.
MAS also wants to prevent AI from becoming a competitive tool available mainly to the largest financial institutions. It launched Pathfin.ai to help smaller firms find and deploy validated AI solutions.
The platform and programme now have more than 300 participants, with a growing number of successful matches.
“Innovation must be founded on trust and stability if it is to scale,”
Chia said.
MAS and the financial industry published a generative AI risk framework in 2023, followed by two AI Risk Management Handbooks in 2025.
The regulator has also released proposed Guidelines on AI Risk Management for public consultation.
These set out its expectations for governance, risk management and controls throughout the AI lifecycle.
The guidelines establish what firms should do, while the handbooks provide practical examples of how to meet those expectations.
MAS has also turned its attention to autonomous AI agents. It published the Safeguards for Agentic Finance at Runtime white paper with the financial industry in July 2026.
Known as SAFR, the framework covers controls applied while an agent is operating.
These include verifying its identity and authority, checking proposed actions before execution and maintaining an audit trail.
AI Raises Cybersecurity Risks
Chia warned that more capable AI models can discover vulnerabilities, support cyberattacks and create convincing forms of deception, including deepfakes.
He cited a sixfold increase in high-severity Common Vulnerabilities and Exposures to 2,200 in 2026 compared with the average over the previous three years.
He also referred to CrowdStrike data showing an 89% rise in AI-enabled cyberattacks.
However, successful breaches have not increased at the same rate. Chia attributed this partly to safeguards built into AI models and the continued effectiveness of layered cyber defences.
These measures include strong authentication, rapid patching, network segmentation, access controls and incident response.
Firms can also use AI to scan code continuously, detect vulnerabilities, monitor threats and respond to incidents faster.
AI is advancing more quickly than the other emerging technologies discussed by Chia.
Tokenisation may take several more years to scale, while quantum computing could remain five to ten years away.
However, he said preparations for quantum resilience should begin now.
Singapore and India Build Financial Links
Chia also highlighted the growing financial partnership between Singapore and India.
The PayNow-UPI linkage launched in 2023 supports faster and cheaper cross-border retail payments and remittances.
Transaction volumes have more than doubled each year since its introduction, with MAS expecting faster growth in 2026.
Both countries are also founding members of Nexus, an initiative to connect national instant payment systems through a common framework.
This would allow each payment network to reach multiple markets without building separate bilateral links.
Chia pointed to a partnership between Singapore-based Pints AI and a large Indian insurer.
The companies used AI to automate checks and prepare information for underwriters, while human experts retained responsibility for decisions.
They are now developing a white paper to provide other firms in the sector with a practical reference based on the project.
Chia expects AI to reduce friction across cash management, payments and investments.
As agents make and execute decisions more quickly, regulators will need to assess the effects on competition and financial stability.
Featured image: Edited by Fintech News Singapore, based on image by perig76 via Magnific
