AI in Malaysia · Regulation

Malaysia has no AI Act yet. Here is what already binds your business

The AI Governance Bill has slipped to early 2027, but data protection law, sector regulators and a growing stack of guidelines already apply to how a Malaysian company uses AI. A plain map of what is binding, what is expected and what is coming.

Pikabyte Engineering 7 min read

Early 2027

when the Digital Minister now expects the AI Governance Bill to be tabled in Parliament

On 5 October 2026, Digital Minister Gobind Singh Deo told the Dewan Rakyat that the AI Governance Bill would be tabled early next year instead of this year. On 28 October he described it as still being drafted.

So as of 19 November 2026, Malaysia has no law dedicated to artificial intelligence. That is not the same as AI being unregulated. Several laws already apply, regulators have said how they read them, and a set of guidelines describes what a careful company is expected to do.

This is a map in three bands: binding now, expected, and coming.

Band Instrument Who it touches
Binding Personal Data Protection Act, as amended Anyone processing personal data, vendors included
Binding Bank Negara policy documents Banks, insurers, takaful operators, payment providers
Binding Securities Commission technology risk guidelines Capital market entities
Binding Online Safety Act 2025 Licensed platforms and network providers
Expected National Guidelines on AI Governance and Ethics Developers, deployers and users of AI
Expected Commissioner’s guidelines on impact assessments and automated decisions Anyone using AI on personal data
Expected Banking industry AI frameworks Financial institutions and their suppliers
Expected ASEAN guides on AI governance A regional reference
Coming AI Governance Bill Anyone who develops, provides, uses or operates AI
Coming Court practice direction on AI Lawyers and court officers

Binding now

The Personal Data Protection Act. If your AI feature touches a customer’s or employee’s details, this is the law that applies today. We covered the 2025 amendments, including breach notification and direct duties for vendors, in a separate article.

Bank Negara’s existing rules. In November 2026 the central bank published its feedback statement on AI in the financial sector, drawing on 113 respondents. Two passages settle the question for now. “BNM is not introducing AI-specific regulation at this stage”, it says. And: “Existing requirements, including those on governance, risk management, consumer protection and Shariah compliance, continue to apply to AI-enabled activities.” The statement lists the policy documents it has in mind, among them Risk Management in Technology, Outsourcing, and Fair Treatment of Financial Consumers. If you build for a bank or an insurer, read those as your specification.

The Securities Commission’s guidelines. Its Guidelines on Technology Risk Management took effect on 19 August 2024 for capital market entities. As issued, they set four principles for AI and machine learning: accountability, transparency and explainability, fairness and non-discrimination, and practical accuracy and reliability.

The Online Safety Act 2025. In force since 1 January 2026, with four sets of regulations. It applies to licensed applications, content and network service providers, and platforms with at least eight million Malaysian users have been required to hold a licence since 1 January 2025. It does not impose duties on individual users. Failing its core duties can cost up to RM10 million. Most businesses are not caught directly. It matters here because the minister names it, with the Communications and Multimedia Act 1998 and the Sexual Offences Against Children Act 2017, as the laws that already allow action against AI-generated abuse material, impersonation and non-consensual intimate content. We look at the business side of that in our article on deepfake scams.

Nothing about using AI suspends the laws a company already answers to. Bank Negara’s statement makes the same point about personal data and intellectual property.

Expected

These are not statutes. They are what a regulator, a customer or a judge will measure you against.

The national guidelines. The Ministry of Science, Technology and Innovation published the National Guidelines on AI Governance and Ethics in September 2024. They are voluntary and set seven principles: fairness; reliability, safety and control; privacy and security; inclusiveness; transparency; accountability; and the pursuit of human benefit and happiness. We have argued before that they make a good design brief.

The Commissioner’s three guidelines. On 30 April 2026 the Personal Data Protection Commissioner issued guidelines on data protection impact assessments, data protection by design, and automated decision-making and profiling. According to a law firm summary:

  • An impact assessment is expected where processing is likely to exceed 20,000 people, or more than 10,000 where sensitive data, financial information included, is involved. It is also expected where the risk is high, and the listed factors include innovative technology such as AI.
  • Where automated decisions use sensitive personal data, explicit consent is required.
  • A privacy notice should say that automated decision-making is used, for which kinds of decision, with what consequences, and how far AI is involved.

The same summary describes compliance with the design guideline as voluntary. Ask your adviser how much weight the other two carry.

Industry frameworks in finance. The Chief Risk Officers’ Forum of the Asian Institute of Chartered Bankers issued an AI Governance Framework in 2025, endorsed by the Association of Banks in Malaysia. Bank Negara says a model risk management framework from the same forum is coming soon. These reach suppliers through contracts.

ASEAN. The ASEAN Guide on AI Governance and Ethics dates from 2024. An expanded guide for generative AI, with nine policy recommendations, was introduced at the fifth ASEAN Digital Ministers’ Meeting. It is a reference, not law, and useful if you sell across the region.

Coming

The AI Governance Bill. The Ministry of Digital, through the National AI Office, announced engagement sessions through July 2026 and a public consultation expected to open on 10 July. From the ministry’s statements and the minister’s answers in Parliament:

  • It would be Malaysia’s first horizontal legal framework for AI, complementing existing sector laws.
  • It is risk-based. Obligations depend on the potential harm of a system, not on the industry.
  • Duties would be assigned according to each party’s role and level of control, and would reach those who develop, provide, use or operate AI systems.
  • It encourages incident reporting, safeguards and sandboxes.
  • It “will not directly regulate AI output”. Illegal content stays under existing laws.

The status, precisely: drafted and consulted on, not tabled, not passed, not in force. Tabling is also the beginning. A bill has to pass both houses and be brought into force, and Malaysia’s data protection amendments commenced in stages.

A court practice direction. On 28 October 2026 the Chief Justice said a practice direction on the use of AI in legal proceedings, for lawyers and court officers, would be issued within the next few weeks. Judges already have their own guidelines. Check whether it has been published before you rely on this paragraph. Our view on what law firms should do in the meantime has not changed.

More from Bank Negara. The central bank says it will “progressively provide greater regulatory clarity as AI use cases and risks evolve”, and that what it learns may lead to thematic reviews.

A permanent institution. The National AI Office was launched on 12 December 2024 under the Ministry of Digital. On 28 July 2026 it was formally institutionalised as AI Malaysia Berhad, which is to house the Malaysian AI Safety Institute and develop governance frameworks and ethical standards.

What to do before the Bill arrives

None of this requires waiting.

  1. Keep a register of AI use. Each use, the data involved, the provider, and a named owner. The Bill is expected to assign duties by role, so know whether you are the developer, the provider or the user of each system.
  2. Run the impact assessment test. More than 20,000 people, more than 10,000 with sensitive data, or AI helping to decide something about a person.
  3. Say so. If a system makes or shapes decisions about people, the privacy notice should say it.
  4. Keep a person accountable. Every AI output that reaches a customer or a ledger should have a human who accepted it, and a record.
  5. Know where the data goes. We wrote about where AI requests from Malaysia are actually processed.
  6. Give staff a rule they can remember. A one-page policy is enough to start.

A company that does these six things will not need to rebuild when the Bill is tabled. It will already be doing what the Bill asks.

Sources

This article is general information and is not legal advice.

Written by

Pikabyte Engineering

Pikabyte builds custom business systems, applied AI and secure cloud infrastructure from Malaysia. Talk to us.

Start a conversation

Want this applied to your business?

Tell us what is slowing the business down. You will talk to an engineer, not a salesperson, and get a straight answer on whether software is the right fix.