Fintech 2026

SINGAPORE Trends and Developments Contributed by: Adrian Ang and Benjamin Samynathan, Allen & Gledhill LLP

Singapore’s vibrant fintech ecosystem has led to a rich and varied scope of business models over the years. In 2015 and 2016, peer-to-peer lending and equity crowdfunding platforms were active in the market. In 2017, more robo-advisers began to set up shop in Singapore, and there was global interest in initial coin offerings and cryptocurrency exchanges. From 2018 onwards, there was a flurry of activity in the fintech space ranging from e-wallets, remittance businesses and digital banks, to cryptocurrency funds, security token exchanges, and cryptocurrency broker-dealing and market making. While the years leading up to 2024 were defined by regulatory sandboxes and proofs-of-concept, the 2025–2026 period has witnessed the crystallisation of certain long-term strategic initiatives. This was accompanied by a shift from experimental pilots to commercial-grade infrastructure. The Monetary Authority of Singapore (MAS) has orchestrated this evolution through a calibrated regulatory approach that balances the fostering of innovation with the pres - ervation of financial stability. This piece will focus on how three themes may define the Singapore fintech market in 2026: (i) the use of agentic AI; (ii) institution - al-grade infrastructure; and (iii) shared responsibility in consumer protection. Agentic AI While 2024 was dominated by the adoption of Gen - erative AI for content creation and summarisation, 2026 would be defined by the rise of Agentic AI. Unlike passive models that respond to prompts, or rule-based workflows such as Robotic Process Auto - mation, Agentic AI systems possess the capacity for autonomous decision-making, multi-step planning, and direct execution of tasks within enterprise work - flows. This shift may also require financial institutions to move from “human-in-the-loop” type workflows to “human-on-the-loop” or even “human-out-of-the- loop” configurations for specific low-risk processes. In the context of Singapore’s financial services, Agen - tic AI refers to systems that can (i) perceive and ana - lyse; (ii) reason and plan; and (iii) act autonomously. To consider a hypothetical scenario relating to portfolio management and AML/CFT, under stage (i), an agen - tic AI may ingest real-time market data, transaction

histories and regulatory updates without explicit user input. Under stage (ii), the Agentic AI may formulate strategies to achieve high-level goals (eg, “optimise this portfolio for tax efficiency” or “investigate this suspicious transaction network”). Finally, under stage (iii), the Agentic AI may execute trades, send notifica - tions, freeze accounts or generate code, often inter - acting with other software agents or APIs. In June 2023, MAS launched Project MindForge to examine the risks and opportunities of AI for financial services. The first phase resulted in the release of a whitepaper on the Emerging Risks and Opportuni - ties of Generative AI for Banks in November 2023. The second phase was broadened to encompass the insurance and asset management sectors as well as the emerging field of Agentic AI, and resulted in the release of the “ AI Risk Management Executive Hand - book ” in November 2025. This handbook acts as a high-level companion to the MAS’ proposed Guide - lines on AI Risk Management (see below) and aims to translate regulatory expectations into actionable industry practices. It addresses 17 key considera - tions for AI governance, grouped into four main pil - lars: Scope and Oversight, AI Risk Management, AI Lifecycle Management, and Enablers. • Under Scope and Oversight, the handbook empha - sises the necessity of a clearly defined governance operating model where accountability is estab - lished at the level of the Board and Senior Manage - ment. • The AI Risk Management pillar focuses on integrat - ing AI-specific risks (such as model hallucination or bias) into an institution’s broader enterprise risk framework, ensuring that policies and third-party vendor assessments are robust enough to handle new technologies. A central theme of the hand - book is proportionality, encouraging institutions to adopt governance measures that are scaled according to the severity of the risk and the nature of the specific AI use case. • The AI Lifecycle Management section details the need for rigorous controls throughout an AI model’s life (from development and testing to deployment and monitoring), ensuring that “human-in-the-loop” oversight and explainability are maintained, espe - cially for high-impact applications.

748 CHAMBERS.COM