The Thesis
The most interesting number in Malaysia’s new banking AI report is not the 92% adoption rate. It is the 25%. That is the share of senior banking leaders who trust AI-generated outputs enough to act on them in key business decisions. In other words, nearly every bank in Malaysia is using AI — and three out of four executives do not trust what it produces. That disconnect is the story.
The Asian Institute of Chartered Bankers (AICB), alongside Ecosystm and the AICB Chief Risk Officers’ Forum, surveyed 87 senior leaders across commercial banks, digital banks, Islamic banks, and development financial institutions. The findings paint a picture of an industry that has moved fast on deployment but slowly on everything else.
The Adoption-Governance Gap
The numbers are striking. 92% of Malaysian banks have begun using AI, mainly for customer onboarding, fraud detection, anti-money laundering, and employee productivity. But higher-risk applications — credit underwriting, pricing, regulatory reporting — remain uncommon. Only 2% of institutions qualified as “Advanced” in AI readiness, while 44% were still at the “Developing” stage.
And here is where it gets uncomfortable: 53% of institutions rely on fragmented or ad hoc AI governance. Only 27% formally classify their AI systems by risk level. Only 30% continuously monitor AI models throughout their lifecycle. The industry is deploying tools it has not built the infrastructure to oversee.
AICB Chief Executive Edward Ling framed it precisely: “The question now is whether institutions have the judgement, ethics, governance and professional capability to use AI responsibly in decisions that affect customers, risk and institutional performance.”
Why This Matters for Islamic Banking
Malaysia is the world’s largest Islamic banking market by assets — and this report surveyed Islamic banks alongside conventional ones. The implications are specific. Islamic finance operates under additional compliance layers: Sharia board oversight, product screening, and ethical mandates that conventional banks do not face. Deploying AI in credit underwriting or product classification without robust governance creates a compounding risk: not just regulatory exposure, but Sharia compliance exposure.
A model that approves a financing structure the Sharia board would reject is not just a governance failure — it is a reputational one. And in a market where consumer trust in Islamic banking products is itself a competitive advantage, the cost of an AI-driven compliance error could be disproportionately high.
79% of institutions reported AI talent gaps or underdeveloped internal capabilities. Only 20% actively encouraged AI-driven decision-making across their workforce. These are not just operational deficits — they are structural constraints on how quickly Islamic banks can move from pilot to production without exposing themselves.
The Headwinds
The talent gap is the most immediate constraint. Malaysia produces strong fintech talent, but the specific intersection of AI model risk management, regulatory compliance, and Sharia governance expertise is vanishingly small. Banks are competing for the same pool — and the AICB report suggests most are losing.
There is also a regulatory timing question. Bank Negara Malaysia (BNM) Governor recently stated that banks must keep humans accountable for AI decisions. But the frameworks for what that accountability looks like in practice — model risk standards, explainability requirements, third-party AI vendor oversight — are still being developed. Banks are building before the rules are final, which means some will have to rebuild.
What to Watch
The report calls for clearer guidance on model risk, explainability, and data governance as banks move from pilot to deployment. The critical question is whether BNM will issue prescriptive standards or principle-based guidance — and whether Islamic banks will be subject to additional AI governance requirements reflecting their Sharia compliance obligations. If the answer is yes, Malaysia could become the first market to define what responsible AI looks like in Islamic banking. If the answer is silence, the 25% trust figure is unlikely to move.