Date: 7 October 2026
By: Sea Jia Wei
Introduction
On 5 October 2026, the Personal Data Protection Department (“JPDP”) issued Public Consultation Paper No. 1/2026 – Artificial Intelligence (AI) and Personal Data Protection Framework (“Consultation Paper”) for public consultation. The consultation is open until 23 October 2026. The proposed Artificial Intelligence (AI) and Personal Data Protection Framework (“Proposed Framework”) seeks to clarify how the Personal Data Protection Act 2010 (“PDPA”) applies where personal data is processed through AI systems.
The Proposed Framework covers AI systems developed in-house as well as those procured from third-party providers, including off-the-shelf, customised and open-source systems, and AI functionality embedded in SaaS, cloud services and APIs. For businesses, however, the more important point is not simply that Malaysia is developing guidance on AI. The Proposed Framework signals that the use of AI involving personal data is increasingly a matter of data governance and accountability, rather than simply an IT or procurement issue.
The practical question for businesses is therefore not only whether an AI system works. Businesses should also ask: What personal data is being used? Why is it being used? Who has access to it? Where does it go? What does the AI provider do with it? And can the business demonstrate that the risks have been properly managed?
AI does not create a “free pass” for using personal data
The Proposed Framework makes clear that the use of AI does not create a separate legal basis for processing personal data. The existing requirements under PDPA continue to apply throughout the AI lifecycle, including planning, development, testing, procurement, deployment, operation, monitoring and decommissioning.
Businesses should therefore continue to consider the purpose of the processing, the applicable legal basis and whether the personal data being processed is necessary, adequate and not excessive for that purpose.
This creates a practical risk for businesses that seek to use existing personal data for new AI-related purposes. The fact that a business already holds personal data does not, by itself, mean that the data can automatically be repurposed for AI. Businesses should consider whether the proposed AI use is consistent with the purpose for which the personal data was originally collected or made available.
For example, a company may have collected customer information for the purpose of providing its services but later wish to upload historical customer records into an AI system to develop customer profiles or improve its services. Before doing so, the business should ask: “Are we allowed to use this data for this particular purpose?”
Businesses should know what happens to personal data when using AI vendors
The Proposed Framework places particular emphasis on due diligence before businesses procure, integrate or use external AI systems. This includes, but is not limited to, considering the purpose, functionality and limitations of the AI system, the categories and sensitivity of the personal data involved, the sources and quality of relevant datasets or pre-trained models where information is reasonably available, whether personal data will be retained, reused or used for model training, fine-tuning or optimisation, whether other processors or service providers are involved, whether personal data is processed or transferred outside Malaysia, the provider’s technical and organisational measures, and any known risks or biases affecting the processing of personal data.
This creates a practical vendor-management risk for businesses. The fact that an AI system is supplied by a third party does not, by itself, address the business’s obligations under the PDPA. Businesses should also determine the respective roles and responsibilities of the business and the AI service provider and, where the provider processes personal data on the business’s behalf, establish appropriate arrangements governing that processing. Importantly, engaging an AI service provider does not transfer or diminish the data controller’s responsibilities under the PDPA.
For example, a business may use an AI-enabled customer service platform to assist employees in responding to customer enquiries. Where the platform processes customer names, contact details, account information or complaint records, the business should consider what happens to that personal data once it enters the AI system. This includes whether the provider retains or reuses the data, whether it is used for model training, whether other service providers have access to it, where the data is processed and what security measures apply.
Businesses should therefore look beyond whether the AI tool performs its intended function. They should understand how personal data flows through the AI system and ensure that appropriate contractual, security and governance measures are in place.
Cross-border processing should not be overlooked
The use of cloud-based and external AI systems may also involve the transfer of personal data outside Malaysia. Where an external AI system involves a cross-border transfer, the Proposed Framework states that the data controller should ensure compliance with section 129 of PDPA. Where appropriate, a Transfer Impact Assessment (“TIA”) should also be conducted, taking into account matters such as where the personal data is stored or processed, who may access it, the provider’s data protection practices, applicable safeguards and any onward transfers.
This creates a potential blind spot for businesses. A business may contract with one AI vendor without realising that personal data is subsequently processed or accessed by other service providers or infrastructure located in different jurisdictions. The Proposed Framework therefore contemplates that businesses should look beyond the immediate vendor and seek sufficient visibility over the wider data-processing chain.
Businesses should therefore not assume that contracting with a local vendor means that personal data will necessarily remain in Malaysia.
AI compliance does not end once the system is deployed
The Proposed Framework adopts a lifecycle approach to AI governance. Businesses are expected to monitor AI systems during operation and periodically review whether the personal data being processed remains adequate, relevant and not excessive. Where an AI system is modified, retrained or updated, the business should consider whether the changes materially affect the processing of personal data or the intended use of the system and, where appropriate, conduct further testing, validation or review.
This creates a practical process risk for businesses. An AI system that was compliant when initially approved may no longer present the same risk after it has been retrained, integrated with new systems or used for a different purpose.
For example, a company may initially approve an AI chatbot to respond to customer enquiries based on a defined dataset. If the chatbot is subsequently retrained using additional customer information and new functions are introduced, the company should consider whether those changes materially affect the processing of personal data before continuing to use the updated system. Similarly, where an external AI provider changes its terms of service, processing activities, subprocessors or data-handling practices, the business should assess whether additional safeguards or further review are necessary. AI compliance should therefore not be treated as a one-off approval exercise completed when the software is first purchased.
Higher-risk AI use cases require closer scrutiny
Not every AI use case presents the same level of risk. The Proposed Framework adopts a risk-based approach and contemplates a Data Protection Impact Assessment (“DPIA”) where appropriate. Where AI involves sensitive personal data or makes or supports decisions that may significantly affect individuals, businesses should consider whether additional risk assessment, testing, human oversight or other safeguards are appropriate.
For businesses, this means that the level of governance and assessment should be proportionate to the nature, scope and risks of the particular AI use case. The level of scrutiny should therefore take into account factors such as the sensitivity of the personal data involved, the complexity of the AI system and whether it makes or supports decisions that may significantly affect the data subject.
What should businesses consider doing now?
Although the Proposed Framework remains subject to consultation and may be amended before finalisation, the Consultation Paper provides a useful indication of the regulatory direction and an opportunity for businesses to review their existing practices.
1. Map current AI use
Identify where AI is already being used across the organisation, including AI functionality embedded within existing SaaS, cloud and business applications. This should not be limited to formally approved AI projects, as employees may already be using AI tools as part of ordinary business activities.
2. Identify personal data flows
For each material AI use case, businesses should determine:
- what personal data is being inputted or otherwise processed;
- why the data is being processed;
- where the data is stored or processed;
- who can access it; and
- whether it is retained, reused or used for training or other secondary purposes.
3. Review AI vendor arrangements
Review relevant contracts, privacy terms and vendor documentation, particularly in relation to data retention, model training, secondary use of data, subprocessors, security measures, deletion and return of data, and cross-border transfers. Where reasonably necessary, businesses may also consider obtaining model cards, system cards or other technical documentation from the AI provider.
4. Assess higher-risk AI use cases
Identify AI systems involving sensitive personal data, significant decision-making or potentially material effects on individuals. Consider whether a DPIA, additional testing, human oversight or other safeguards are appropriate.
5. Keep evidence of compliance
Businesses should consider maintaining records of material AI use cases, including the AI system and its purpose, personal data and data sources involved, applicable legal basis, risk assessments or DPIAs, approvals and material decisions, testing and reviews, material changes, incidents and corrective actions, and oversight of AI service providers.
Concluding remarks
The Proposed Framework signals a potential shift towards treating AI involving personal data as an ongoing data governance, compliance and accountability issue, going beyond making a technology or procurement decision.
For businesses, the key takeaway is not merely to understand the Proposed Framework, but to consider whether their existing AI use, vendor arrangements and internal processes can withstand regulatory scrutiny. Businesses should therefore move beyond general AI awareness and start putting in place practical governance measures, particularly for AI systems that process personal data or present higher risks to individuals.
This article is intended for general information only and does not constitute legal advice.