MAS Sets New Accountability Standards for AI in Finance

Overview of New AI Guidelines

The Monetary Authority of Singapore (MAS) has introduced important guidelines aimed at ensuring that financial institutions remain accountable for the use of artificial intelligence (AI) tools, including chatbots, even when these technologies are sourced from third-party developers. These guidelines were announced on October 7, 2023, and address the potential risks such as data breaches and misinformation that can arise from AI applications in the financial sector.

Phased Implementation

Set to be rolled out in phases starting from October 2027, these guidelines come at a time when there are growing concerns about the regulation and safety of advanced AI technologies that possess autonomous functionalities. MAS stated that it's crucial for financial institutions to maintain accountability for AI utilized in their services, irrespective of whether the technology is developed externally.

Key Responsibilities and Risk Management

Institutions are expected to carry out thorough assessments before deploying third-party AI tools. This includes evaluating model transparency and explainability, as well as preparing contingency plans for any unexpected behaviors encountered. In addition, greater emphasis is placed on human oversight, necessitating the creation of an inventory that tracks various AI use cases and ensures safeguards are applied proportionally based on assessed risks.

In November 2025, MAS conducted a public consultation seeking input regarding AI risk management, leading to clarifications in the updated guidelines. Notably, it emphasized that basic governance structures could adequately manage AI risks for lower-stakes applications, such as using AI for tasks like drafting emails or creating marketing content.

Accountability Measures

Boards and senior management will have the duty to continuously review AI usage policies and ensure there are internal processes to manage any incidents stemming from AI utilization. Importantly, existing governance frameworks can be leveraged, allowing institutions to avoid the necessity of establishing entirely new AI committees unless highlighted by their specific operational needs.

Furthermore, MAS mandates institutions to maintain an inventory detailing all AI systems and models in use, including their approved utilization methods, training data sources, and responsible personnel. Assessing risks will require consideration of potential repercussions from AI failures and the level of autonomy provided to the AI systems.

Implementation Timeline

The rollout of the guidelines will occur in two significant phases. By October 7, 2027, institutions must establish foundational governance systems and ensure they have inventories of their AI tools and the capability for robust risk assessments. Subsequently, by October 8, 2028, they will need to implement rigorous testing and monitoring processes, emphasizing the importance of maintaining significant human oversight while securing skilled personnel and the requisite technology infrastructure.

Looking Ahead

MAS plans to further engage the financial sector in 2027 to discuss more guidance regarding the use of autonomous AI systems. AI holds considerable promise for the financial industry, offering enhancements in customer service, risk management, and product innovation. However, realizing these benefits sustainably necessitates a firm understanding and careful management of the risks that sophisticated AI implementations may introduce.

According to MAS Deputy Managing Director Ho Hern Shin, with clearer regulations surrounding AI usage in finance, institutions can foster innovation confidently while preserving customer trust and maintaining the integrity of Singapore's financial landscape.