MAS unveils open-source toolkit for responsible AI-use in finance

It was supported by the collaborative effort of the association of 31 industry players.

The Monetary Authority of Singapore (MAS) has released an open-source toolkit for the responsible use of Artificial Intelligence (AI) in finance.

Developed by a MAS-led consortium of 31 industry players, the Veritas Toolkit version 2.01 assists financial institutions (FIs) in implementing assessment methodologies for the Fairness, Ethics, Accountability, and Transparency (FEAT) principles. 

These principles offer guidance to firms in the financial industry regarding the responsible utilization of AI and data analytics in their offerings.

The toolkit was the first of its kind developed for the financial sector. The first version, Veritas Toolkit 1.0, was released in February 2022. It was focused on Fairness assessment methodology. 

The latest version, Veritas Toolkit 2.0, not only enhances the Fairness assessment methodology but also introduces assessment methodologies for Ethics, Accountability, and Transparency. 

Accenture and Bank of China (BOC) led the development of the toolkit, with pilot testing contributions from BNY Mellon, DBS Bank (DBS), OCBC Bank (OCBC), and United Overseas Bank Limited (UOB).

ALSO READ: MAS suggests digital money standards through whitepaper

The association has also released a white paper that outlines the valuable insights gained from seven financial institutions (BNY Mellon, DBS, HSBC, OCBC, Singapore Life (Singlife), Standard Chartered Bank, and UOB which conducted pilot integrations of the Veritas methodology within their internal governance frameworks. 

Key lessons learned include the significance of: establishing a consistent and robust responsible AI framework across different regions; adopting a risk-based approach to determine the necessary governance for AI use cases; and promoting responsible AI practices and providing training for the emerging generation of AI professionals in the financial sector.

Moving forward, the association’s future endeavours will revolve around training initiatives in responsible AI and promoting wider adoption of the Veritas Methodologies and Toolkit among financial institutions, MAS said. 

Further, MAS has collaborated with AI solution providers like IBM and SAS to integrate the Veritas Toolkit into their AI solutions.