AI in Hong Kong and Singapore

Having explored Europe we now look at Asia where artificial intelligence (AI) has been all the rage for some time in Hong Kong and Singapore, and the demand for hardware to power this technology is massive.

From a financial services perspective, the Hong Kong Monetary Authority (HKMA) and the Monetary Authority of Singapore (MAS) have both published AI principles. MAS published the so-called "FEAT Principles" (Fairness, Ethics, Accountability, Transparency) in November 2018 (updated in February 2019) and these were developed in partnership with Singaporean and international financial institutions and AI experts, and describe aspirational ethical properties that an AI system should have. The HKMA published high level principles on AI and consumer protection principles in respect of the use of big data analytics and AI shortly thereafter in November 2019 and banks are expected to take these principles into account when designing and adopting their AI and big data analytics.

The HKMA is currently seeing banks use AI applications in their anti-money laundering (AML) work and notes that these applications have the potential to transform many of the data intensive processes currently deployed for financial crime risk management. The HKMA published a report in April 2024 which include key observations and good practices identified in a thematic review of transaction monitoring systems. In the report, the HKMA set out its regulatory expectations on the use of AI in AML settings which adds to its thematic guidance on banks' transaction monitoring systems.

In response to AI's impact on the workforce, the HKMA updated its Supervisory Policy Manual in May 2024 in relation to manpower and training. The HKMA highlighted that the banking industry must proactively plan ahead for manpower development to enhance the professional knowledge and skill-sets of employees to enable them to complement and co-exist with technology in the age of AI.

The HKMA has a number of on-going initiatives regarding AI. Perhaps the best known is the adoption in June 2021 of the "Fintech 2025" which aims to encourage the financial sector to adopt technology comprehensively by 2025. A Fintech Promotion Roadmap was unveiled by the HKMA in the summer of 2023 which outlined key initiatives including those relating to AI. More recently, in April, the HKMA launched the FiNETech series, gathering around 100 banks, securities and insurance companies as well as technology firms to explore next-level collaboration in the areas of Wealthtech, Insurtech, Greentech, AI, and Distributed Ledger Technology.

In the securities and futures markets, the Hong Kong Securities and Futures Commission (SFC) is speeding up its adoption of AI for monitoring social media platforms and flagging potential risk content thereby allowing it to intervene at an early stage. Operational efficiency is another key area for the SFC. It continues to look at how AI-assisted automation can make a difference. In 2019, the SFC released Guidelines on Online Distribution and Advisory Platforms, a chapter of which addresses robo-advice, a tool where AI can be deployed. Earlier in 2016, the HKMA updated the Guidelines for the Regulation of Automated Trading Services, which is relevant when AI is used in trading or trading controls.

The Privacy Commissioner for Personal Data (PCPD) has also released non-sector specific guidance on the use of AI. The Guidance on the Ethical Development and Use of AI was issued in 2021 so that organizations (such as banks) can understand and comply with the relevant requirements of privacy legislation when they develop and use AI systems. The PCPD has signaled that it will publish the "Artificial Intelligence: Model Personal Data Framework" in Q2 2024, which will include a series of recommendations on governance and mitigation of privacy risks for organizations that handle personal data when procuring, implementing and using AI systems. Additionally, the Government Chief Information Officer published the Ethical Artificial Intelligence Framework in 2022 (and updated multiple times in 2023). The Framework was originally developed for internal adoption within the Government regarding the applications of AI and big data analytics but was revised for general reference by organizations when adopting AI and big data analytics in their projects.

For Singapore, despite the lack of specific legislation governing the use of AI, several sector-agnostic initiatives and tools have been implemented which could nevertheless impact AI developments within the financial services sector. This includes:

1. The National Artificial Intelligence Strategy 2.0 (NAIS) (first launched in 2019, updated in 2023), which outlines Singapore's ambition and commitment to build a trusted and responsible AI ecosystem, driving innovation and growth through AI. Singapore aims to accelerate the development and deployment of AI solutions in certain key domains, including financial services.

2. The draft Model AI Governance Framework for Generative AI (the Draft). Announced by Singapore's Infocomm Media Development Authority (IMDA), the Draft provides 9 key recommendations to harmonize the management of generative AI risk whilst maintaining an environment for technological innovation. The Draft is currently undergoing a public consultation since January 2024.

3. Model AI Governance Framework (the Model Framework). The Model Framework was launched in 2019 and then updated a year later. Lauded as a world first, the Model Framework provides detailed and readily implementable guidance to private sector organizations, addressing key ethical and governance issues when deploying AI solutions.

4. Advisory Guidelines on Use of Personal Data in AI Recommendation and Decision Systems (Personal Data Guidelines). Issued in March 2024, the Personal Data Guidelines provide organizations with direction on the use of personal data to develop and train AI recommendation and decision systems (AI Systems), as well as the applicability of Singapore's Personal Data and Protection Act 2012 when personal data is collected as input from data subjects for use in AI Systems.

Singapore has also implemented specific tools and testing frameworks aligned with the FEAT Principles to facilitate the adoption of AI within Singapore's financial services industry. For instance, the MAS has been encouraging banks to use AI and data analytics to improve money laundering risk management. When supervising the use of AI models for AML, MAS has emphasized the explainability of model outputs and effectiveness over conventional methods as key measures of success. It has required banks to put in place adequate model governance and oversight to achieve these outcomes.

In June 2023, MAS issued an open-source toolkit to enable the responsible use of AI in the financial industry. The Veritas Toolkit version 2.0, developed by an MAS-led consortium of 31 industry players, is designed to help financial institutions carry out the assessment methodologies for the FEAT Principles. The IMDA has also released AI Verify in May 2022, an open-source AI governance testing framework and software toolkit aimed at assisting organizations in the evaluation of AI systems against internationally recognized AI ethics principles. This foreshadows the launch of the AI Verify Foundation in June 2023, which aims to support the development and use of AI Verify.

MAS is also working on Project MindForge which seeks to develop a risk framework for the use of Generative AI (Gen AI) for the financial sector. It aims to develop a clear and concise framework on the responsible use of Gen AI in the financial industry, and to catalyze Gen AI-powered innovation to solve common industry wide challenges and enhance risk management. Phase 1 of Project MindForge involved the development of a Gen AI framework to enable financial institutions to use Gen AI in a responsible manner. It identified risk dimensions in certain areas including accountability and governance, transparency and explainability and legal and regulatory. A platform-agnostic Gen AI reference architecture has also been developed, providing a list of the building blocks and components that organizations can use to create robust enterprise-level Gen AI technology capabilities. In the next phase, the MindForge consortium will expand its scope to involve financial institutions from the insurance and asset management industries.

As regards asset management, the MAS is mindful that Gen AI offers the potential to improve operational efficiency – for example, in the communication of returns and performance attribution – and mitigate risks such as cyber-attacks. It is also being increasingly used in fund managers' toolkits for asset allocation and the selection of securities. However, MAS also feels that it is imperative that asset managers be mindful of the risks Gen AI carries when used in complex investment and compliance tasks. For example, AI hallucinations can result in inaccurate and misleading outcomes which could lead to poor investment decision-making. MAS is currently studying this further with industry stakeholders under Project MindForge.

In the field of financing, an MAS-led consortium also announced in November 2023 the completion of an AI technical platform known as NovA! (via a minimum viable product) that banks can tap onto to address key challenges when issuing sustainability linked loans in the real estate industry.