MAS to trial issuance of tokenised bills, releases guidelines for AI in financial sector

[SINGAPORE] The Monetary Authority of Singapore (MAS) will be trialling the issuance of tokenised bills to primary dealers, which will then be settled with central bank digital currencies (CBDCs).

More details on the tokenised MAS bills will be available next year, said the authority’s managing director Chia Der Jiun at the Singapore FinTech Festival on Thursday (Nov 13).

This follows the successful completion of a live trial for settlement of interbank overnight lending transactions using Singapore dollar wholesale CBDC. The trial involved the three local banks – DBS, OCBC and UOB.

Rachel Chew, DBS’ group chief operating officer and head of digital currencies, global transaction services, said that this pilot paves the way for broader adoption and a wider range of use cases.

She added that by leveraging wholesale CBDCs as a common settlement asset, Singapore has developed a viable blueprint for the next generation of “always-on” financial market infrastructures.

Chia said that tokenised assets have “lifted off the ground” and have yet to “escape velocity”.

Optimists in the industry believe that the financial industry is headed for a future where most financial assets will be tokenised, he added.

When an asset is tokenised, it means that it is converted to a digital asset that can be traded on the blockchain. Tokenised assets, a relatively new financial development, are expected to speed up transactions and enhance the liquidity of assets.

Increasing fragmentation

The industry is risking increasing fragmentation in the tokenisation scene, where each operator will operate in their own private network, Chia said. “We could see a fragmented landscape of sub-scale walled gardens or even a small number of monopolies posing concentration risk,” he added.

SEE ALSO

Companies, consumers must have basic understanding of AI to reap potential benefits: DPM Gan

To avoid these “subpar outcomes”, Chia believes the industry needs to develop and adopt a model of cooperation where participants collaborate to build a marketplace for asset-backed tokens while competing to bring products, clients and liquidity to the market.

“Standardisation and interoperability will mitigate liquidity fragmentation,” he said.

He noted that MAS has been working with a consortium of global policymakers and major financial institutions to develop standards for tokenisation under Project Guardian.

Project Guardian is a collaboration between MAS and financial institutions that seeks to enhance liquidity and efficiency of financial markets through asset tokenisation.

Chia also noted that other than tokenised assets and CBDCs, MAS has also looked into stablecoins as a digital asset.

“There has been a lot of attention on stablecoins,” he said.

A stablecoin is a type of cryptocurrency that is pegged to a fiat currency, such as the US dollar, and backed by reserve assets. Popular stablecoins include USDT and USDC, which are both pegged to the greenback.

Chia noted that stablecoins would be able to work across many different applications and use cases. “While agility is a strength, stability needs to be reinforced,” he said.

He acknowledged that stablecoins have a “patchy record” of keeping their pegs. This can erode confidence and trigger runs on other stablecoins.

“Regulated stablecoins, while nascent, offer the prospect of value stability. Sound and robust regulation of stablecoins will be critical to underpin their stability,” he said.

Chia noted that MAS has recognised this, and has finalised the features of the stablecoin regulatory frameworks.

MAS will be preparing draft legislation for the framework, he said.

AI at the forefront

Apart from tokenisation, Chia described artificial intelligence (AI) as one of the two “transformative themes” for the financial sector. “We are seeing the momentum of AI adoption and experimentation build up across our financial sector at a foundational level,” he said.

He noted that more than 30 financial institutions have their own AI innovation centres based in Singapore.

Financial institutions are applying AI through different means such as summarisation, transcription and translation.

The aim for financial institutions is to adopt AI productively and safely, and for the workforce to adopt AI safely, he said.

To address this issue, MAS on Thursday published a set of guidelines on AI risk management to guide financial institutions on the responsible use of AI in the financial sector.

MAS said that the proposed guidelines will apply to all financial institutions. These may be applied in a proportionate manner, in line with the size and nature of financial institutions’ activities, use of AI, and their risk profiles.

Under the guidelines, financial institutions are expected to ensure that their board and senior management play a key role in the governance and oversight of AI risk management.

This would include the establishment and implementation of frameworks, structures, policies and processes for AI risk management and fostering the appropriate risk culture for the use of AI.

Financial institutions are also expected to “establish clear identification processes for AI usage across firms, to maintain accurate and up-to-date AI inventories and implement risk materiality assessments that factor impact, complexity and reliance dimensions”, MAS said.

Lastly, financial institutions should plan for and implement robust controls in key areas, including data management, fairness and human oversight.

“Such controls should be applied based on their relevance and be proportionate to the assessed risk materiality of AI usage,” MAS said.

It added that financial institutions should also ensure that their capabilities are adequate for their use of AI.

Another initiative announced by Chia was BuildFin, which will bring together technology providers and research institutes to work with financial institutions on complex problems of common interest.

“The aim is to create shared resources and solutions that benefit the ecosystem,” he said.

MAS and financial institution partners have identified their first common problem statement, which is that Singlish presents “a level of complexity” that existing large language models are not fully ready to handle.

Chia revealed that the Agency for Science, Technology and Research will partner financial institutions to develop a voice-to-text AI model. “By working together, they can pool their data to develop a better model and serve customers better,” he said.