AI Governance in APAC: What Senior Leaders Need to Know

APAC is never a single market, and the same rules apply around AI regulation. Three of its most commercially significant economies, Singapore, Japan, and Malaysia, are leading by example in their own ways, each taking a distinct approach to governing AI. For senior TA leaders operating across these markets, understanding the differences is a precondition for responsible deployment, along with hiring the right capability to manage it.

In this article, we cover what AI governance entails, along with what these Asian economies are doing to set the standard in a relatively untested realm of legislation.
 

What is AI Governance?

AI governance refers to the policies, controls, and accountability structures that organisations put in place to ensure AI systems are developed and used responsibly. In practice, this means defining who is accountable for AI decisions, what limits are placed on how AI systems operate, how risks are identified and managed before and after deployment, and how organisations remain answerable to regulators, employees, and the people affected by AI-driven outcomes.

As governments across APAC move from voluntary guidance toward formal legislation, AI governance is shifting from an internal best-practice question to an external compliance requirement with direct legal and operational consequence.

Agentic AI Governance Framework in Singapore

Singapore's Infocomm Media Development Authority (IMDA) published the Model AI Governance Framework for Agentic AI at the World Economic Forum in January 2026. Unlike general AI guidelines, it addresses agentic AI specifically: systems capable of planning across multiple steps and taking actions on behalf of users with a significant degree of autonomy.

The framework is structured around four dimensions:

1.        Assessing and bounding risks: reviewing appropriate use cases and placing explicit limits on what an agent can access or do before deployment

2.      Enabling useful human accountability: requiring organisations to define meaningful checkpoints at which human approval is required, not simply logged

3.      Implementing technical controls and processes: structuring breakpoints across the agent lifecycle, including pre-deployment testing for task execution accuracy and policy adherence, and real-time monitoring once live

4.      Enabling end-user responsibility: informing through transparency and training, including ensuring that users understand what an agent can do, who to escalate issues to, and where human skills risk eroding as agents take over functions

The framework was updated in June 2026 with additional case studies and best practices from Singaporean and international companies from Ant International to OCBC, and continues to set the reference standard for responsible agentic AI governance. These dimensions also set a precedent for the roles required for such a transformation, along with the remits they should hold.

Innovation and Rights Tensions in Japan's AI Governance Model

Japan enacted its "Act on the Promotion of Research, Development and Utilisation of Artificial Intelligence-Related Technologies," commonly referred to as the AI Promotion Act, in May 2025. The approach is deliberately light-touch and innovation-oriented, prioritising voluntary compliance and flexible governance rather than the binding, risk-classified obligations of the EU AI Act.

One year on, this position is under sustained pressure. GR Japan's May 2026 industry analysis reports that creators, performers, publishers, and rights holders have mounted an organised campaign for stronger protections, particularly around copyright, voice rights, and AI-generated deepfakes. In October 2025, the Content Overseas Distribution Association submitted a formal request to OpenAI demanding it cease using member companies' content without permission, a move that contributed to the suspension of OpenAI's Sora 2 service in Japan in April 2026.

The government has responded through measures including the Principle Code for Protection of Intellectual Property and Transparency toward Appropriate Use of Generative AI, a comply-or-explain model rather than binding law, and the LDP's AI White Paper 2.0, which argues the challenge is not a binary choice between innovation and rights protection but the construction of a governance model that serves both.

Japan's trajectory is one of careful recalibration, and organisations with significant creative, content, or IP-intensive operations in the market should monitor developments closely.

Malaysia's AI Governance Bill: Moving Toward Mandatory Compliance

Malaysia's National AI Office (NAIO) released a public consultation paper for a proposed AI Governance Bill on 10 July 2026, with Digital Minister Gobind Singh Deo having confirmed earlier in the year that the Bill was near completion and targeted for Cabinet presentation in June 2026. The Bill is drafted as risk-based legislation covering the full AI lifecycle, with provisions addressing AI-related harm, incident reporting, and ethical principles, and including enforcement mechanisms for negligence and harm.

This marks a meaningful shift for Malaysia. Where previous AI governance relied on voluntary guidelines, the Bill moves toward enforceable obligations for organisations operating AI systems in the country. For multinational enterprises, the compliance implications extend to third-party service providers and outsourced technology functions, which are now also directly exposed under Malaysia's updated personal data protection framework, a development that sits alongside, and interacts with, the incoming AI Bill.

What This Means for Leaders

The regulatory picture across these three markets has a direct bearing on workforce strategy. Deploying AI within a structured governance environment, whether in Singapore, Japan, or Malaysia, calls for talent with traditionally separate capabilities: procurement and implementation specialists on the bleeding edge of today’s AI development, AI ethics and governance professionals with regulatory technology experience, and HR support and legal functions capable of applying these frameworks to operational decisions.

For many organisations active across APAC, the demand is already a burgeoning skills gap. The pace at which these regulatory requirements are being introduced means that workforce planning for AI governance roles needs to be happening in real-time, proactively ahead of the compliance deadlines rather than in response to them. The organisations that will manage AI regulation well are those that treat governance capability as a hiring priority today, not a remediation exercise tomorrow.

If you are building out AI governance capability across one or more APAC markets, or currently grappling with specialist hiring pressure stemming from such a deployment, speak with one of Robert Walters’ outsourcing specialists to discuss how a bespoke talent model could quickly adapt to your organisation’s short- and medium-term hiring demands.