Malaysia set to table AI bill, focus shifts to board accountability

Malaysia set to table AI bill, focus shifts to board accountability
AI accountability in Malaysia

As Malaysia prepares to table an artificial intelligence governance bill in early 2027, attention is increasingly turning to how institutions assign responsibility when AI systems influence material decisions. The debate goes beyond legal compliance alone, as the core issue is whether boards, management and organisations are truly prepared to oversee the risks arising from the use of this technology.

Highlights

  • The Malaysian government is expected to table a risk-based Artificial Intelligence Governance Bill in the Dewan Rakyat in early 2027.
  • Company directors and boards remain responsible for AI risks, with an emphasis on oversight, accountability and due diligence under the Companies Act 2016 and Section 17A of the MACC Act.
  • The Securities Commission Malaysia and Bank Negara Malaysia stress that AI oversight and corporate governance disclosures must be a key focus for boards and cannot be delegated to algorithms.

This article was translated from the original. Read the original version by our correspondent here.

AI framework and the question of board responsibility

As reported by The Edge Malaysia, the government is preparing an Artificial Intelligence Governance Bill, which Digital Minister Gobind Singh Deo said is expected to be tabled in the Dewan Rakyat in early 2027. The bill aims to take a risk-based approach across sectors and complement existing laws, while a July consultation paper proposed national governance principles, centralised oversight and obligations proportionate to the level of risk.

AI accountability was also a recurring theme at the inaugural Financial Lines Forum in Kuala Lumpur. In his keynote address, Tan Sri Abdul Rashid Hussain said governance failures typically do not arise from a lack of rules, but from areas the rules do not reach, linking the issue to whether boards see risks but fail to act.

The consultation paper clearly states that responsibility for AI must remain with identifiable human beings, not with the AI system itself. For companies using AI, this puts the boardroom under scrutiny, particularly when the technology already influences credit approvals, transaction flagging, and information provided to customers and patients.

Under the Companies Act 2016, directors have duties of care, skill and diligence. The article argues that these duties do not make directors personally liable for every system error, but do make their oversight of material risks a matter that merits scrutiny.

Section 216 of the Companies Act, although it concerns the delegation of authority to individuals rather than systems, provides an important analogy for the need to assess the authority delegated, the basis for that delegation and the safeguards put in place. In the context of AI, the board's questions remain the same: what authority has been given to the system, on what grounds, and subject to what controls?

Governance implications for companies and regulators

The article outlines several examples showing how AI risks can arise from outside or within an organisation. In early 2024, an employee at Arup's Hong Kong office was deceived during a pre-recorded video conference that used deepfakes to impersonate senior management and approve a transfer of approximately US$25 million, while that same year a Canadian tribunal found Air Canada liable for misleading bereavement fare information provided by its website chatbot.

Both cases reinforce the argument that organisations cannot shift responsibility to technology. The article also cites a warning arising from Section 17A of the MACC Act, which places the burden on directors and officers to demonstrate that they exercised due diligence, showing that excuses such as "I didn't know" or "the algorithm did it" are not sufficient on their own.

Abdul Rashid proposed four key questions for boards: where is AI being used, who is responsible, how does the organisation know if something goes wrong, and can the system's decisions be explained to regulators, customers or courts? According to the article, boards need to demand testable answers from management rather than simply accept general assurances.

Malaysian regulators have also begun emphasising the same issues. In its July consultation, the Securities Commission Malaysia proposed clearer expectations for board oversight of technology and disclosures by listed companies on their use and governance of AI, while the governor of Bank Negara Malaysia said in a 8 July address to the banking industry that AI "sits firmly on the board agenda" and that "responsibility cannot be delegated to an algorithm".

The article goes on to list several practical considerations, including appointing identifiable senior officers to oversee high-risk uses, ensuring providers disclose material limitations, making human oversight genuinely effective, continuously assessing directors' fitness and propriety, and understanding the limits of D&O insurance coverage. It also stresses the need to provide meaningful human review pathways for individuals affected by AI decisions, including the ability to obtain useful explanations, correct inaccurate information and have the decision reconsidered by someone with the authority to change it.

Ultimately, the article's main argument is that legislation only establishes a framework, while integrity, courage and the habit of questioning assumptions must be embedded in an institution's culture. With the bill expected to be tabled early next year, the message is that boards need not wait for new legislation to begin demonstrating that they understand, challenge and oversee the responsible use of AI.

As a reminder, we previously reported on the write-down of Zetrix AI bonds to zero, which put pressure on AmanahRaya's fund NAV.

This material may contain third-party opinions, none of the data and information on this webpage constitutes investment advice according to our Disclaimer. While we adhere to strict Editorial Integrity, this post may contain references to products from our partners.