5
Singapore
The Singapore financial industry has adopted AI in various forms and for a number of years — from
internal risk management purposes (e.g., onboarding of clients, assessing credit risk and fraud
prevention), to client-facing tools (e.g., chatbots) and AI-driven solutions and products (e.g., robo-
advisers), among others. This trend is expected to continue and potentially at a more rapid pace.
Regulatory approach
As the financial regulator in Singapore, the Monetary Authority of Singapore's (MAS) approach toward
AI can be summarized as a "pro-innovation, risk-sensitive approach." An example of MAS' "pro-
innovation" approach can be seen in the robo-advisory space. In response to the proliferation of robo-
advisory solutions in the last five years, MAS has adjusted its regulations to address specific risks
arising from algorithm-based technologies forming the core of the robo-advisory offering. At the same
time, MAS took care to ensure that regulations did not stifle the introduction of such an alternative
product and service offering to a wider group of investors. In this regard, MAS issued the Guidelines
on Provision of Digital Advisory Services in 2018, which set out its views on how robo-advisors could
comply with existing regulatory standards. At the same time, MAS introduced several exemptions,
such as the exemption from the need to hold a fund management license for portfolio rebalancing and
the exemption from the need to hold a license for dealing in capital markets products if trade orders
are passed to another appropriately licensed entity, to reduce the regulatory burden on robo-advisors.
An example of MAS' "risk-sensitive" approach can be seen in its reaction toward the advent of digital,
non-face-to-face client onboarding by financial institutions. MAS issued circulars, first in 2018 and
then subsequently in 2022, to guide the industry on the measures that financial institutions are
expected to take to address anti-money laundering and counter-terrorism financing risks arising from
non-face-to-face onboarding. These measures include the expectation to place appropriate controls
during the videoconferencing process to verify the identity of the customer and the authenticity of
identification documents sighted via videoconferencing (including, where appropriate, conducting a
sum digit test) and the need to commission an independent assessment to certify the effectiveness of
new technology solutions in managing impersonation and fraud risks.
AI principles and governance
What perhaps has the widest scope of application, albeit in a nonbinding form, are the Principles on
the use of AI and data analytics (AIDA) issued by the MAS in 2018. The principles aim to provide
firms with a set of foundational principles to consider when using AIDA in decision-making in providing
financial products and services. In summary, MAS encourages firms to consider the principles of
fairness, ethics, accountability and transparency (FEAT) when assessing existing or developing new
internal frameworks to govern the use of AIDA.
In February 2022, MAS released the Veritas Toolkit version 1.0, which focused on the assessment
methodology for fairness. In June of the same year, MAS released an information paper, where it
shared observations and good practices noted during its AI thematic review of financial institutions.
More recently, in June 2023, MAS issued Veritas Toolkit version 2.0, an open-source toolkit designed
specifically for the financial industry to help financial institutions carry out assessment methodologies
for the FEAT principles, covering beyond the principles of fairness as seen in version 1.0.
For financial institutions, it should be borne in mind that AI regulations are cross-sectoral and will