SINGAPORE Trends and Developments Contributed by: Adrian Ang, Alexander Fong and Benjamin Samynathan, Allen & Gledhill LLP
A further information paper on AI model risk management was published in December 2024. This information paper focused on AI oversight and governance, risk management systems and processes (including materiality assessments), and the standards and processes for develop - ment and deployment of AI. In addition to MAS-led efforts, Project Mind - forge also released an industry-led whitepaper on GenAI risks and opportunities for banks. While these papers focus more on the practices of banks, MAS has stated that it is in the midst of considering supervisory guidance for all financial institutions in 2025, not just banks, by building upon focus areas of the papers. We expect that such measures will be particu - larly relevant to entities rolling out “agentic” AI models in connection with “algorithmic trading” or “robo-advisory” products. Algorithmic trading Algorithmic trading today already exists in the form of sophisticated codes or “trading bots” which are able to implement and execute trad - ing strategies within pre-defined parameters. The revolution brought about by AI will invariably change existing algorithmic trading workflows. Already, there is talk of “agentic AI” essentially playing the role of a tireless, incorporeal trader, which is able to iterate trading strategies and improve upon its own trading systems by learn - ing from market data in real-time. All of this, while being immune to the emotions and needs of an actual human trader. While a true “agentic AI” algorithmic trader does not yet exist, given the exponential growth of AI, it will not be surprising that artificial general
intelligence will reach these capabilities in the foreseeable future. Once it does, it will be crucial for financial institutions to have implemented the measures proposed by MAS, to the extent appli - cable, to forerun the risks that are attendant in the adoption of such “agentic AI” . Robo-advisory services Similarly, robo-advisory services as they exist today run on sophisticated algorithms and are able to provide investment strategies or portfolio recommendations, considering a specific inves - tor’s financial position, investment goals and risk appetites. However, the incorporation of AI into these workflows could drastically streamline such robo-advisory operating models and dramati - cally increase their ability to deliver for cus - tomers. That being said, increased AI usage, particularly in the context of handling sensitive customer information (for the purposes of tai - loring bespoke investment strategies) will give rise to increased data protection risks and other vulnerabilities. Accordingly, as these models improve, we would expect more service providers to take advan - tage of these business cases, with a resurgence in robo-advisory services now powered by AI. However, providers of these services will need to ensure that they assess the exposure to the risks created by such AI usage, including the risks posed to their consumers in relation to data leakages and accuracy. Single-Currency Stablecoins – The Proposed Stablecoin Regulatory Framework As the overview to this Fintech 2025 Guide notes, stablecoins generally achieve value sta - bility by one of two means: (i) through an algo - rithm which buys and sells stablecoins in reserve
776 CHAMBERS.COM
Powered by FlippingBook