By Dr. Jun-E Tan, Senior Research Associate, Khazanah Research Institute
The regulation of Malaysia’s AI ecosystem is a multifaceted challenge that requires careful navigation. The roundtable discussions revealed a spectrum of perspectives on how best to approach AI governance, particularly in defining the roles of various stakeholders and the nature of regulatory frameworks.
Stakeholders prefer soft standards and sectoral regulation
At the roundtable, a broad definition of “regulation” was provided, attributed to Nathalie Smuha: “a means to intentionally influence and/or constrain the behavior of actors, be they individuals, groups, or legal entities such as companies.” Participants engaged in discussions on whether the focus should be on building standards or establishing laws as primary vehicles for AI governance. A virtual poll conducted during the roundtable indicated a strong preference for standards, 81% (25 votes out of 31 responses) favoring this softer approach.
Furthermore, participants were asked whether they preferred a horizontal or generalised regulatory approach, similar to the EU AI Act, or a vertical or specialised approach that tailors regulations to specific sectors (e.g., finance, healthcare). A majority of over 75% (23 votes out of 30 responses) favored the vertical approach, underscoring the need for flexibility to allow existing regulators and ministries to adapt regulations to their specific contexts.
Coordination is Key to Streamlining Initiatives on Adoption and Governance
Government initiatives related to AI are often run by different agencies, resulting in a siloed and uncoordinated approach. This lack of coordination leads to overlapping efforts and inefficiencies in utilising limited resources.
In response to KRI’s suggestion for a central AI agency to facilitate national coordination and communication, there was broad support among participants. Discussions primarily focused on implementation details, including governance structure and the necessity for strong leadership and a clear mandate. It was emphasised that the agency should have well-defined objectives and functions. There was also considerable debate over whether it should oversee regulations.
The proposed agency can fine-tune and implement the national strategy on AI, building upon the National AI Roadmap (2021-2025), which is set to expire next year. Other non-regulatory functions that it could take on include fostering connections between experts across different sectors and locations, as well as gathering and disseminating information about AI’s impacts to the public.
More AI Readiness Will Help in Risk Management and Meaningful Adoption
Given that participants represented different stakeholder interests, there was a spectrum of preferences on how to balance free innovation and regulation for safety. The conversation on AI risks oscillated between fears of being left behind and perceived risks of unsafe AI, unintended consequences, or AI being weaponised by malicious actors.
However, there was some consensus around the lack of readiness by Malaysian institutions and the general public, which exacerbates AI risks and reduces the country’s ability to adopt AI effectively. Boosting readiness in terms of governance structures, state and industry capacity, and public AI literacy will be good interim measures as Malaysia figures out its risk appetite and governance approach.
Policy Recommendations
From the key takeaways of the roundtable and preliminary findings of the AIIG study, three policy recommendations for the Malaysian context are proposed:
- Establish and Empower a Central National AI Agency: As of the time of writing, the Ministry of Digital has announced that a National AI Office has been approved by the Cabinet and will begin operations in November 2024. The established AI Office can assume some of the functions proposed above. Clear communication about its governance structure and mandate is essential for building public trust in the office and its operations.
- Develop a Flexible Regulatory Framework: Malaysia requires a regulatory framework that adapts to the evolving AI landscape while being grounded in local context. An agile regulatory framework should be accompanied by initiatives to improve readiness for regulations and compliance. Efforts ensuring that AI adoption is safe and trustworthy should also acknowledge the cross-border nature of emerging technologies; therefore, the country needs to participate actively in global governance and rules-setting of AI.
- Prepare Citizens and Businesses for Future AI Challenges: As the lack of readiness will exacerbate AI risks and hamper successful adoption, policies need to be in place to prepare Malaysian citizens, consumers, and businesses for future challenges and opportunities that AI can bring. These may take the form of AI literacy campaigns for the public and capacity-building for businesses, focusing on effective adoption and awareness of potential risks and best practices for responsible use.
Conclusion
In keeping with global trends of AI governance, conversations in Malaysia have moved beyond ethical principles and are now focused on regulatory frameworks and practical actions to maximise the benefits and minimise the risks of AI adoption.
Different actors, state or non-state, are gearing up efforts to prepare the ground. A National AI Office can coordinate these initiatives and connect stakeholders to optimise resources spent. A concerted focus on boosting readiness and building an agile regulatory framework will ensure that Malaysia’s AI ecosystem is robust and inclusive, serving the needs of its communities and industries.
About the writer
Dr. Jun-E Tan is a Senior Research Associate at Khazanah Research Institute. Her current research interests include digital rights and AI governance in the contexts of Southeast Asia and Malaysia. She has also worked in the areas of environmental and climate policy, social and solidarity economy, as well as sustainable development in general.
Jun-E holds a PhD in Communication from Nanyang Technological University in Singapore, a Master’s degree in Public Policy (University of Malaya), and a Bachelor’s degree in Information Systems Engineering (Universiti Tunku Abdul Rahman).
The views and recommendations expressed in this article are solely of the author/s and do not necessarily reflect the views and position of the Tech for Good Institute.
