- Methodically laying down foundations for AI before presenting Bill to Cabinet in 3Q26 and on to lawmakers in Parliament
- AI Nation 2030 has the ambitious target of wanting Malaysia to be among the top 10 AI nations in the world by 2030
Shamsul Majid’s first year of running Malaysia’s National AI Office (NAIO), an agency under the Digital Ministry, as CEO of a tight 20-person team was less about flexing regulatory muscle, and more about laying down the foundations for AI so the country does not trip over itself later.
[Ed Note: The interview happened before the action taken by Sam’s previous employer, Malaysian Communications and Multimedia Commission (MCMC), on Sunday to temporarily block Grok, an AI bot developed by Elon Musk’s AI company, xAI, which is accused of allowing users to create harmful non-consensual deepfake images of women and children.]
Again, and again, over the course of the hour-long chat with Digital News Asia, Sam as he prefers to be called, returns to a simple sequence: teach first, govern next, legislate last. “Instead of starting off with enforcement, let’s teach people first; teach board directors; teach kids; teach the standards agency, and document all that,” he says, arguing Malaysia should avoid being “so heavy handed” while the ecosystem is still learning what “good” looks like in adopting AI into all facets of life.
That approach is not philosophical but economic. A former CTO of the MCMC who applied for the NAIO CEO job out of a desire to help shape the national conversation, guardrails and adoption of AI, Sam recounts being warned by some EU academics that while the EU AI Act “looks good on paper”, it is proving to be very expensive to enforce with high compliance costs.
Becoming aware about the unintended risk of “imposing on our businesses unnecessary burdens” or saddling government with an enforcement job it cannot realistically execute, he opted to focus on building the foundations.
After all, none of the major AI players have a local legal presence: “if we wanted to enforce any violation of our AI Act, there’s no Open AI Sdn Bhd established in Malaysia. Who do you send the letter to?” he said using a hypothetical example as the AI Act has not been presented to Parliament yet.
He has also proved adept at tapping industry excitement and enthusiasm around the national effort to ensure AI is a force for positive change, to recruit senior executives to volunteer their time to be part of seven different special working groups NAIO set up offering valuable input that has been incorporated into the draft bills of the proposed AI Act.
In his telling, much of 2025’s work – standards discussions, governance layers, civil service enablement, even a kids code of ethics colouring book – was foundations for a single national playbook. With the AI Technology Action Plan 2026-2030 (referred to as AI Nation 2030 in the 13th Malaysia Plan) poised to be the shared roadmap for the country, he declares, “After the work put in last year, 2026 is a year of execution,” adding that AI Nation 2030 has the ambitious target of wanting Malaysia to be among the top 10 AI nations in the world by 2030. Based on a ranking by Stanford University, it currently stands at 25. No pressure.
With its first budget at a paltry US$2.46 million (RM10 million) in 2025, its work has been recognized with its budget doubling this year to US$4.93 million (RM20 million). “That’s still very small compared to what they have to do,” said a senior executive from the private sector who is in one of the seven special working groups.
You won’t hear Sam complaining though. He has built a reputation as a doer in his NAIO role.
That is definitely needed with a lot of work to be done to achieve that top 10 global ranking with NAIO off to a good start in 2025 with a lot done so far, both planned and unplanned. In the following Q&A Sam shares highlights.
Digital News Asia: As 2025 was your first year, what are the barometers you look at to chart your annual progress?
Sam Majid: We’ve been using two external benchmarks: the Stanford AI Index (national comparison) and Oxford Insights’ Government AI Readiness (government comparison). Malaysia was at No 24 and No 26 respectively, based on data from before NAIO existed, so we’ll track how 2025 activities shift that. But, very importantly, we also want a domestic yardstick: over the next five years, we aim to help build the Department of Statistics Malaysia (DOSM) capability to measure Malaysia’s own AI progress, so that by around 2030 we will have our local measurement, and not just rely on rankings from parties based outside Malaysia as their scoring will lack local context.
Digital News Asia: Can you talk about this move to establish a Malaysian AI Standards? Why adopt, with localization, what some first mover nations have established?
Sam: Governance is hard without standards. Procurement teams, especially in government, will ask how to validate the AI systems they intend to purchase, and today Malaysia lacks a local benchmark. That’s why we have engaged with Jabatan Standard Malaysia (Department of Standards Malaysia) to develop Malaysian AI standards, including translating relevant ISO work into Malaysian Standards (MS). This push for Malaysian developed standards links back to the AI Governance Bill we intend to bring to the Cabinet in Q3 2026.
Digital News Asia: Why wait till the 3rd quarter of 2026 to present the AI Governance Bill to the Cabinet and with a further time lag before it is debated and duly approved by lawmakers into law (when the bill becomes as Act) seems to leave a long period before AI regulations come into play that require compliance and following the rules outlined.
Sam: We could do it sooner but we want to have a longer public consultation period. We’re sequencing deliberately: “instead of starting off with enforcement, let’s teach people first.” The Bill covers multiple areas, including safety, testing, sandboxing, and governance, but we’re building layers before any Act is introduced because rushed enforcement tends to fail. The EU AI Act looks good on paper, but enforcement faced pushback when foundational aspects such as establishing standards and governance and, critically, creating awareness and educating stakeholders, were weak. We want the ecosystem ready before any enforcement bites.
Digital News Asia: It looks like the EU experience around AI regulation has been useful in shaping how you are going about this here?
Sam: Yes, because we’ve been listening (to the experience of other countries). I was told by academics from Brussels, when they were in KL, that the EU AI Act is “very expensive to enforce”, and companies struggle with compliance because they “don’t know what to do.” The lesson for Malaysia is to avoid “unnecessary burdens” by building standards and teaching self-governance first, then monitoring, before moving to enforcement.
Digital News Asia: Was there any significant achievement in 2025 that you feel went under the radar and doesn’t seem to be appreciated?
Sam: I would say AI at Work 2.0, that gave 450,000 civil service access to AI tools. It became an important adoption lever that was unexpected for us and was not in our 2025 roadmap. What happened was that when NAIO was launched in Dec 2024, Google Cloud wanted to support the Digital Ministry’s AI agenda and in early Feb 2025 they enabled Gemini and NotebookLM for the 445,000 civil servants using Google Workspace. And because these were AI tools, we were tasked with ensuring adoption and running change management, supporting responsible use, and champion-building, because you can give people tools, but if there’s no training, there’s no adoption. But now you can count these 450,000 civil servants as being AI users.
Google’s supportive move then put pressure on the other global tech players with government business who came forward as well. For example, Red Hat asked how they could support us and we asked them to focus on open-source AI, to show SMEs the adoption path to AI is not only commercial, and this led to the SME adoption report Red Hat produced.
Digital News Asia: SMEs are one of the six sectors you are tasked with to oversee, right? Healthcare, Transportation, Agriculture, Public Services and Education being the others.
Sam: Actually, it is Micro and Small Medium Enterprises (mSME). And we know that one of the challenges mSME face is the cost of technology. Well, in the National AI Action Plan, we want to help create the emergence of a ‘made by Malaysia’ AI in the next five years to reduce the cost of AI subscriptions. This should then allow more SMEs to adopt AI tools and there is opportunity for open-source AI to be part of this.
Digital News Asia: Okay, so with civil servants, are you going to inject a competitive element by introducing a ranking of say which ministry is using AI the best, and acknowledge the most innovative use of AI that generated productivity improvement that can be measured.
Sam: It is possible. There are discussions about a public sector AI competition. We also track two dimensions: day-to-day AI tooling, and AI as an innovation tool that rewires services. We’ve identified between 50–60 government AI use cases, mostly small-scale so far but they’re a start, and the bigger wins come when the process owners agree to redesign entire workflows. This is the harder layer – rewiring end-to-end processes.
Separately, there’s the “AI productivity paradox” work via the Malaysian Productivity Council, which looks at how to avoid spending on AI without getting measurable output. This paradox is on the Prime Minister’s radar as well and we are committed to tackling it to ensure our AI Act benefits the nation and no one is left behind.
Digital News Asia: Because your scope is so wide, which two or three sectors are you going to give priority to for this year, and what measurable targets will you publish?
Sam: Obviously we want to improve beyond the current Stanford ranking position. Priority sectors are healthcare, public services, and transportation and cities, building on the “smart city” momentum towards “AI city” outcomes.
Digital News Asia: Regarding the upcoming AI Act will compliance be voluntary or mandatory?
Sam: Well, mandatory implies legislation, so today, right now, this is guideline-led. The starting point is the Ministry of Science, Technology and Innovation’s (Mosti) Sept 2024 introduced National Guidelines on Artificial Intelligence Governance and Ethics (AIGE), with seven ethical principles (including transparency and human benefit). We’ve also produced some practical governance guides: one aimed at boards of directors for enterprise AI governance, because boards often lack awareness, and a kids’ AI code of ethics colouring book.
Digital News Asia: So where does the AI Incident Reporting database come into the picture? There is hardly any information on this.
Sam: It’s part of the governance approach: a database to capture how AI is misused over time, because incidents evolve quickly, for example, the rise of AI-generated videos used to scam. If we later move to an Act, we should do it “once” and do it right, backed by real quantitative and qualitative evidence, not guesswork. Enforcement is also hard when key AI firms have no local legal entity: “there’s no Open AI Sdn Bhd established in Malaysia… Who do you send the letter to? Rushing risks a “toothless” or “half baked” outcome.
Digital News Asia: Which three workflows for government will you prioritise for 2026?
Sam: Jabatan Digital Negara or JDN (previously MAMPU) has already adapted Mosti’s AIGE into its Public Sector Guidelines for AI (GPAI), setting guardrails. For 2026, the three focus areas are in improving public service delivery (including bringing AI into the next version of the myGov app), optimising internal operations (including trials of agentic AI for meeting minutes, action items, and follow-ups), and AI-assisted, data-driven policymaking, using internal datasets to be more predictive, not reactive.
Digital News Asia: Seems like despite the small team you have, a lot has been done. It’s a good thing your budget has doubled to RM20 million in 2026
Sam: Yes, and we can add around five to 10 people.
Digital News Asia: What is the game plan for 2026?
Sam: NAIO’s scope stays, but we add a key role: secretariat to drive execution of the AI Action Plan (AI Nation 2030). Once launched, someone must “chase after the execution”, debottleneck issues, and report progress via the quarterly MED4IR platform chaired by the Prime Minister which meets monthly. There are four pillars; NAIO sits under one, with a steering committee chaired by Digital Minister Gobind Singh. It’s not a law, but a national aspiration that needs momentum and monitoring.
Digital News Asia: You have talked about data sets related to technology. So what data sets are going to be released?
Sam: It’s less about NAIO “releasing” data, more about guiding data owners. First, a data playbook (completed, pending publication) to help organisations prepare data for AI, and to audit whether datasets are “AI ready”. For public sector execution, NAIO works closely with JDN, since JDN runs government IT projects.
So, again, it’s about self-governing, we can’t force them. Only data owners know what is the risk profile of their data. However, we are giving them a playbook to guide them for AI. So that’s the first part, to allow auditing and labelling of data.
Second, data sharing depends on existing classification rules under the government security office framework (Open, Restricted, Limited, Top Secret). If data is open, it should be shared and can go into www.data.gov.my but owners must assess risk. This also aligns with longer-term plans under the 13th Malaysia Plan, including a Data Commission and a National Data Bank over the next five years.
These are institutions to handle data and a bank for all data sets. If you have a bank for data sets, then the data must be in a form that is ready. So that’s why we have the data playbook that will help data owners organise their data and the Data Commission will take care of all this.
Digital News Asia: Any closing thoughts to share about your hopes for 2026.
Sam: 2026 is a year of execution. With a “single playbook”, Malaysians have a shared roadmap on the economic and societal impact we want from AI. The focus is human-centric: governance, responsible AI, and upskilling, so people are not left behind. If execution is strong, I expect collective gains. Remember that AI Nation 2030 is about “raising the ceiling” (capability), increasing the baseline, and doing it within “good governance”, meaning responsible, safe use.
AI was used to generate the initial draft of this article with the writer and editor responsible for the published version.
Keyword(s) :