New guide lays out how Singapore boards should oversee AI

SINGAPORE – Company directors now have a guide to help them oversee the use of artificial intelligence, including how to assess AI investments, manage risks and ensure the technology produces tangible results for their organisations.

Launched at the inaugural SID Digital and Tech Forum 2026 on July 31, the guide reflects growing expectations for boards to take a more active role in AI governance, rather than leaving such issues to management or IT teams.

The launch at the tech governance event comes as the Cyber Security Agency of Singapore strengthened board oversight requirements for owners of critical information infrastructure, amid growing risks from AI-enabled cyber threats.

The 116-page guide is developed by the Singapore Institute of Directors (SID) in collaboration with American tech companies OpenAI and Microsoft, and the Infocomm Media Development Authority.

In her speech at the event, Minister for Digital Development and Information Josephine Teo said companies should tap workers’ existing AI capabilities, as some employees are already using AI in more sophisticated ways than their organisations.

“The question is how do you harness their energy, knowledge and their abilities to create an environment where they can bring their individual-level productivity improvements into the workplace,” said Teo.

As companies tap these capabilities, they also need institutional guardrails. These include documenting AI-built tools, establishing governance over their use, and ensuring safety requirements are met.

SID vice-chair Max Loh said in a speech that as AI reshapes industries and redefines companies’ competitive advantage, boards are facing decisions that would have been “unimaginable” a few years ago.

“For boards, AI is also reshaping the questions directors need to ask about... governance, ethics, cyber resilience, talent and long-term value creation,” said Loh, adding that the guide is designed to offer practical guidance.

Among the key considerations in the guide is how companies can build AI-ready boards. It cautions against placing AI oversight in the hands of a small number of individuals.

Instead, it recommends that all directors have a basic understanding of AI terminology, its business implications and governance considerations. A number of directors should have enough working knowledge to challenge management on AI strategy, deployment, data governance and risk trade-offs.

At least one director should have deep expertise to support oversight of more complex technical, regulatory or operational issues.

The guide also said that boards should look beyond activity or adoption figures when evaluating whether AI investments are paying off.

It identifies metrics such as the number of AI pilots launched, licences purchased, training classes conducted, and volume of prompts generated as potentially weak or misleading when used in isolation, as high usage may reflect experimentation without producing business value.

Instead, directors should look for evidence that AI is improving actual business outcomes, such as workflow improvements, better customer experience and operational impact.

Stronger metrics include the percentage of workflows augmented by AI, average time saved per workflow, and percentage of pilots converted to production.

The guide also recommends that boards set clear boundaries around where AI can make decisions, including “hard red lines” where automated decision-making is prohibited and human oversight is required for decisions affecting customers, employees, safety or privacy.

The level of oversight should depend on the risks involved.

For instance, the guide says AI used in workforce decisions such as hiring, promotion or performance assessment should be subject to human oversight, bias testing and data-quality reviews.

There should also be clear avenues for affected individuals to challenge or seek a review of AI-assisted decisions.