Blog
Malaysia's AI governance bill is coming: How to prepare your identity security program now
Malaysia's AI Governance Bill has moved past the talking stage. The National AI Office opened public consultation on the Bill on 10 July 2026 and closed it on 31 July. If enacted, the Bill would move the country from voluntary AI guidance to binding law. Organizations that wait for enforcement to begin may be retrofitting compliance under pressure; those that build identity governance for AI agents and machine accounts now can extend controls they already have in place.
The stakes are already real. Bank Negara Malaysia's own AI survey found that 71% of banking institutions and development financial institutions had implemented at least one AI application by the end of 2024, up from roughly 56% a year earlier, for uses like fraud detection, credit scoring, and compliance automation. Insurance operators are moving in the same direction1. Regulation is arriving after adoption, not before it, which means the gap between how much AI Malaysian organizations run and how well they govern it a potential risk to solve for right now.
Why Malaysia's regulatory window is closing fast
Malaysia's approach mirrors a pattern seen elsewhere: voluntary guidance first, binding law second. The EU AI Act followed the same trajectory, and the National AI Office (NAIO), launched on 12 December 2024 following Cabinet approval that August, is now running Malaysia's version.2
Three instruments define the current landscape:
- National AI Governance and Ethics Guidelines (AIGE), voluntary, published September 2024, covering principles including accountability, transparency, and resilience and security.3
- Personal Data Protection Act (PDPA) Amendment Act 2024, mandatory since 1 June 2025, requiring a Data Protection Officer, breach notification, and data portability. The breach notification duty is general (it applies to any personal data breach, AI-driven or not), so an AI system that mishandles personal data falls under it by extension.
- Cybersecurity Act 2024 (CSA), in force since 26 August 2024, requiring national critical information infrastructure entities to conduct risk assessments and audits and report cybersecurity incidents, and requiring licensed cybersecurity service providers to retain service records for at least six years.
The AIGE's voluntary status may not last. NAIO's planned deliverables, an AI Technology Action Plan, an AI Code of Ethics, and an AI Adoption Regulatory Framework, could convert today's ethical principles into enforceable requirements if the AI Governance Bill passes.
Financial services faces the sharpest deadline
BNM's updated Risk Management in Technology (RMiT) policy, effective 28 November 2025, already mandates what many organizations treat as aspirational. Section 10.53 requires an access control policy governing identification, authentication, and authorization for every user of IT assets and data. Section 10.57 requires that user activity in critical systems be logged for audit and investigation, with logs retained for at least three years.
RMiT also names the CISO directly (Section 9.5) as accountable for enforcing these policies, with board-level oversight of technology risk required under Sections 8.1, 8.2, and 8.4. This isn't guidance. It's an explicit, dated mandate for auditable proof of access governance, the kind of evidence identity platforms can help produce.
The blind spot regulation hasn't caught up to yet
Here's what none of these frameworks fully address yet: the explosion of non-human identities behind every AI deployment. Every AI agent, service account, and bot that Malaysian banks and insurers stood up to hit that adoption rate is an identity. Some of them can be invisible to traditional access reviews.
That's a problem, because the PDPA's breach notification duty already applies to AI systems by extension, and RMiT already demands three years of activity logs on critical systems. You can't produce an audit trail for an identity you never registered. Periodic access reviews, run quarterly or annually, leave months-long windows where an ungoverned AI agent can quietly accumulate access no one signed off on.
Malaysia's own AI Governance Bill is expected to take a risk-based approach, similar to the EU AI Act, which means incident reporting and accountability obligations could apply directly to how organizations govern the AI agents making decisions on their behalf. The PDP Commissioner has already issued a Guideline on Automated Decision-Making and Profiling in April 2026, and accountability for the systems doing that deciding is not something regulators are likely to wait on.
Treating AI agents as governed identities, not infrastructure
The organizations best positioned for Malaysia's 2026/27 regulatory shift are the ones that stop treating AI agents as IT infrastructure and start treating them as identities requiring the same lifecycle discipline as employees. That's the model SailPoint® Agentic Fabric™ was built for.
Agentic Fabric extends SailPoint's enterprise-grade identity security platform to give organizations visibility into AI agents, bots, and machine accounts across supported agentic platforms, browsers, endpoints, and SaaS applications. Instead of discovering non-human identities during a breach investigation or an audit, they can be registered, owned, and governed from creation.
How Agentic Fabric can help:
- Discovery & Unified Visibility helps close the visibility blind spot, you can't govern what you can't see, and a unified identity graph maps discovered agents to their human owners, entitlements, and the data they touch.
- Real-time Governance & Audit treats every governed AI agent as a first-class identity with automated provisioning, deprovisioning, and a continuous audit trail, which can assist with goals towards RMiT's Section 10.57 logging requirement and the PDPA's breach notification duty.
- Proactive Protection & Response monitors agent behavior for drift, scores risk dynamically, and can trigger automated remediation, revoking entitlements or disabling agents at quickly rather than during the next scheduled review cycle.
- Data Access Governance maps the full human-to-agent-to-data path, enforcing controls at the data layer to help prevent agents from bypassing existing security to reach sensitive information.
This isn't a bolt-on tool for AI risk. It's an extension of identity governance already trusted by customers representing 53% of the Fortune 5004, applied to the identities multiplying fastest inside Malaysian enterprises today.
Build the evidence trail before it's required
Malaysia's compliance window, mandatory PDPA and CSA obligations today, AIGE principles that may become enforceable, won't stay open indefinitely. Organizations that build continuous, audit-ready governance over AI agents and machine identities now will be better positioned to address the AI Governance Bill's requirements as a natural extension of existing controls, not a scramble.
Want to see where AI agent governance is heading globally and how leading enterprises are getting ahead of it? Download SailPoint's Year of the AI Agent report to benchmark your organization's readiness against what's coming next.
DISCLAIMER: The information contained herein is for informational purposes only, and nothing conveyed herein is intended to constitute any form of legal advice. SailPoint cannot give such advice and recommends that you contact legal counsel regarding applicable legal issues. The development, release, and timing of any features, functionality, capabilities, or services described for SailPoint's products that are not currently available remain at SailPoint's sole discretion on a when, and if, available basis and may not be delivered at all and must not be relied on in making a purchasing or investing decision.
Sources
- Bank Negara Malaysia, AI Survey 2024, as published in the BNM discussion paper “Artificial Intelligence in the Malaysian Financial Sector” (August 2025)
- NAIO was approved by the Malaysian Cabinet on 28 August 2024 and launched on 12 December 2024: NAIO FAQ; MyDIGITAL press release.
- Launched by Malaysia’s Ministry of Science, Technology and Innovation on 20 September 2024 as voluntary guidance; its seven principles include accountability, transparency, and privacy and security (Skrine).
- Fortune 500 rankings. The figure refers to SailPoint’s overall customer base, not Agentic Fabric customers.