AI in Healthcare in Malaysia: Courses and Rules for Doctors, Nurses and Hospitals (2026)
AI in healthcare in Malaysia, 2026: what doctors, nurses and hospitals can use it for, what the MMC guideline and patient data rules require, plus courses and CPD.
AI is already usable in Malaysian healthcare for documentation drafts, literature work, patient education and teaching, but the Malaysian Medical Council holds the doctor fully accountable for every outcome. Doctors should seek appropriate training first (MMC guideline clause 3(4)), keep identifiable patient data out of AI systems without consent, and verify every output before it affects care.
Part of the guide: AI in healthcare and medicine

On this page
- What does the MMC say about AI in medical practice?
- Is there Malaysian law on AI in healthcare?
- How can AI help doctors?
- What are the real risks?
- Which AI tool should a doctor choose?
- What does AI mean for nurses?
- How is AI used in medical education?
- What should hospitals put in place before a pilot?
- Where do emergency physicians see AI fitting?
- How do you train a clinical team in AI?
- Do AI courses give CPD points?
- Can an HRD Corp levy pay for healthcare AI training?
- Who regulates what when AI enters a Malaysian hospital?
- What words do clinicians need to know?
- What are the common mistakes in the first year?
- What does a 30-day starter plan look like?
- What should you do next?
- Sources
- Frequently asked questions
Healthcare in Malaysia is moving to AI from two directions at once. Clinicians already use chat assistants on their own phones, and hospital managers are being asked what their governance is. This guide collects what we can verify: what the regulator says, what the evidence supports, what must stay out of AI tools, and how to train a team. It links to a separate article for each question so you can go straight to yours.
Facts and clause numbers checked on 22 October 2026 against the primary sources listed at the end. Where a point comes from a news report or a review rather than the regulator, we say so.
What does the MMC say about AI in medical practice?
The Malaysian Medical Council (MMC) allows doctors to use AI and holds them fully accountable when they do. Its Guideline on the Ethical Use of Artificial Intelligence (AI) in Medical Practice, Version 1/2025, applies to registered medical practitioners. Three duties matter most in daily work:
- Training first. Clause 3(4): practitioners "should seek appropriate training before using AI tools and/or technology in their clinical practice".
- No patient data without consent. Clause 3(6)(b): "Patient data must not be shared on any AI system without the patient's consent."
- Human oversight. Clause 3(13): practitioners "must validate AI recommendations and intervene as necessary", and clause 3(5)(c) keeps them "responsible and accountable for the patient's clinical outcomes".
The guideline says it will be reviewed annually until 2026 and every two years after that, so check the MMC website for a newer version before you rely on a clause number. The long version, clause by clause, is in The MMC guideline on AI in medical practice, explained. The practical question most doctors ask first is answered in Can doctors use ChatGPT?.

Is there Malaysian law on AI in healthcare?
Not as a dedicated AI law. A peer-reviewed review of AI in Malaysian healthcare notes that AI is not named in any Malaysian legislation and that existing laws apply indirectly. The national position is the voluntary National Guidelines on AI Governance and Ethics, published by MOSTI in September 2024, which set out principles for users, policymakers and developers but are not binding. The Malaysia AI regulation overview 2026 covers the rules outside healthcare.
What does bind healthcare staff today:
| Source | Applies to | What it adds for AI |
|---|---|---|
| MMC AI guideline (Version 1/2025) | Registered medical practitioners | Training, consent, confidentiality, validation, accountability |
| Personal Data Protection Act 2010 (as amended 2024) | Commercial processing of personal data | Health data is sensitive personal data; breach notification; cross-border transfer rules |
| Hospital or ministry policy | Staff of that institution | Approved tools, approved data flows, incident reporting |
| MOSTI AI governance guidelines (2024) | Voluntary, all sectors | Principles such as fairness, reliability and accountability |
Patient data rules deserve their own article: Patient data and AI in Malaysia: what is allowed.
How can AI help doctors?
AI helps doctors most where the work is language: drafting letters, summarising papers, preparing teaching, and structuring notes that the doctor then corrects. It helps least, and carries the most risk, where the task is a judgement about a particular patient.
The best-documented gain is documentation. Researchers at The Permanente Medical Group in the United States reported that 3,442 physicians used an ambient AI scribe across 303,266 encounters in a ten-week period (Tierney et al., NEJM Catalyst, 2024), and a later account of the first year estimated about 15,800 hours of documentation time saved across roughly 2.5 million encounters. These figures come from the health system's own survey-based evaluation, not a controlled trial, and the authors stress continuing attention to accuracy.
A realistic list of uses is in How can AI help doctors?. The documentation question has its own article: Can ChatGPT write clinical notes?

What are the real risks?
Fluent output hides errors. Three results show why the guideline insists on validation:
- A widely deployed sepsis prediction model was tested on 38,455 hospitalisations at the University of Michigan. It had a sensitivity of 33% and a positive predictive value of 12% (Wong et al., JAMA Internal Medicine, 2021). A tool can be in use at many hospitals and still perform poorly in yours.
- Four commercial chatbots were asked medical questions that probe race-based misconceptions. All four produced examples of race-based medicine, and the answers were not always consistent when the question was repeated (Omiye et al., npj Digital Medicine, 2023).
- ChatGPT scored between 52.4% and 75.0% on public USMLE sample questions in early 2023, around the pass line of roughly 60% (Kung et al., PLOS Digital Health, 2023). Exam performance on multiple-choice items says little about safety with a real, messy patient.
Clauses 3(10) and 3(11) add a quieter risk: over-reliance and skills degradation. For teaching hospitals this is the issue that decides how trainees are allowed to use AI. See AI in emergency medicine for a setting where speed makes the temptation strongest.
Which AI tool should a doctor choose?
The guideline names no products. It asks for "reasonable confidence" in a tool, built on five criteria: trustworthiness, appropriateness, transparency, conformity to standards and clarity on liability. A general chat assistant can pass that test for drafting a teaching handout and fail it for anything involving an identifiable patient.
Between the two assistants most doctors ask about, the decision turns on data terms, not on which one writes better. We lay out the comparison in Claude vs ChatGPT for doctors.
What does AI mean for nurses?
AI changes nursing tasks and does not remove the nurse. Handover drafts, patient education sheets, roster questions and study notes are all places where a tool can save time. Bedside assessment, escalation and the relationship with the patient are not tasks a language model performs.
Two articles cover this in depth: Can AI replace nurses? and How AI helps nurses day to day. Our AI course for nurses is built around those uses. A wider list of working uses is in Examples of AI in healthcare.
How is AI used in medical education?
Teaching is where AI is safest to start, because no patient is in the loop. Educators use it to draft cases, build simulation scenarios, prepare question banks for expert review and give students feedback at scale. A systematic review of large-language-model virtual patients for history-taking (Li and Lutfi, JMIR Medical Informatics, 2026) found that the systems mostly simulated internal medicine and mental health cases, with few covering multimorbidity or rare conditions. That is a useful warning for anyone who wants a full curriculum from a chatbot.
Students are already using the tools. A survey of 185 undergraduates at a public medical school in Sarawak found that 40.0% reported using AI tools, with ChatGPT the most used. The sample was convenience-based, so treat it as an indication. The full picture is in How can AI be used in medical education?, with a student view in AI for medical students in Malaysia and a design view in Simulation-based training with AI.
What should hospitals put in place before a pilot?
Start with rules, then tools. A hospital that writes the rules first can say yes quickly to a good use and no to a bad one. A minimum set:
- A list of approved tools and the tasks each is approved for.
- A data rule stating that no identifiable patient data goes into a public AI tool, and which institutional systems are approved for patient data.
- A verification rule naming who checks AI output before it is used in care or teaching.
- A disclosure rule for when a patient is told that AI is part of their care.
- An incident route, because clause 3(14) requires practitioners to report untoward complications from AI use to the tool's developers and the relevant regulators.
- A training record that shows who has been trained, in line with clause 3(4).
Pilots that go wrong in other sectors usually skip step 1 or 3; the same pattern is described in Why AI pilots fail in production. For a hospital-level plan, see Integrating AI into clinical training.
Where do emergency physicians see AI fitting?
Emergency departments combine time pressure, incomplete information and heavy documentation, which makes them both the most promising and the riskiest place for AI. The two physicians who lead our medical track are emergency physicians: Prof. Dato' Dr. Rishya Manikam (Universiti Malaya and UM Specialist Centre) and Datin Dr. Sri Latha Balakrishnan (Hospital Putrajaya). Our own view, which is an opinion and not a finding, is that documentation, teaching and simulation are the places to start, and live triage or diagnosis support needs a validated, locally tested system and a hospital governance decision. The details are in AI in emergency medicine.
How do you train a clinical team in AI?
Train in short, practical blocks on the team's own de-identified material, and keep the rules visible. Our courses run as a half-day briefing, a one-day course or a two-day educators' programme, for groups of up to 25. Content follows the MMC clauses: how the systems fail, documentation and correspondence, literature, talking to patients about AI, consent and data, and supervising trainees.
| Format | Length | Group | Suits |
|---|---|---|---|
| Department briefing | 4 hours | Up to 40 | CME slots, grand rounds, society meetings |
| Course for doctors or nurses | 1 day | Up to 25 | Departments and wards |
| Educators' programme | 2 days | Up to 25 | Faculties, simulation centres, training programmes |
Half-day sessions start from MYR 4,500 per group, as stated on our healthcare pages. What a course for doctors covers is explained in AI course for doctors in Malaysia. The syllabus for educators is on the AI in medical education course page. These are education programmes about using AI. They are not clinical training and they validate no product for clinical use.
Do AI courses give CPD points?
Not by themselves. Doctors need at least 20 CPD points a year to renew the Annual Practising Certificate, and points are awarded for events run through an approved CPD provider. A company is not in that position. A hospital, university or society can apply for points for its own event and use our programme as the content. We do not advertise points. The rules are in CPD points for AI courses in Malaysia.
Can an HRD Corp levy pay for healthcare AI training?
For private hospitals, clinics and healthcare companies that are registered HRD Corp employers, a course from an HRD Corp Registered Training Provider can be HRD Corp Claimable, subject to HRD Corp approval, levy balance and the cost caps. Government hospitals and public universities are outside the levy system. The rules, including the 14-day rule from Employer's Circular 2/2026, are in HRD Corp Claimable AI training in Malaysia: the complete guide, and the wording is explained in What does "HRD Corp claimable" mean?.
Who regulates what when AI enters a Malaysian hospital?
Several bodies touch the topic, and none of them covers all of it. Knowing which one answers which question saves weeks.
| Question | Where the answer sits |
|---|---|
| What may a doctor do with an AI tool? | Malaysian Medical Council (MMC guideline, Version 1/2025) |
| Is a specific AI product a regulated medical device? | The review we cite notes that AI tools are usually treated as medical devices; ask the vendor for the registration and your biosafety or procurement unit |
| What may be done with patient data? | PDPA 2010 for commercial processing; Ministry of Health rules and hospital policy in public hospitals |
| What is good practice for AI in general? | MOSTI National Guidelines on AI Governance and Ethics (voluntary) |
| What is allowed in my hospital? | Your hospital's policy, which should be written down |
The same review lists gaps that matter in practice: no central ethics review body for AI, ethics committees without AI expertise, and no explicit AI liability law. Plan on the basis that your institution has to fill those gaps with its own policy.
What words do clinicians need to know?
| Term | Plain meaning |
|---|---|
| Large language model (LLM) | The system behind tools such as ChatGPT and Claude. It predicts text; it does not look up verified facts |
| Hallucination | A fluent but invented statement, such as a reference that does not exist |
| Ambient scribe | A tool that listens to a consultation and drafts a note |
| De-identification | Removing or altering details so a person cannot be recognised |
| Validation | Testing a tool on data like yours to see how well it works |
| Prompt | The instruction you give the tool |
What are the common mistakes in the first year?
- Buying a tool before writing a rule. The tool arrives, the policy never does.
- Letting personal accounts carry patient data. It starts with one convenient shortcut.
- Skipping the checking routine. Fluent output feels checked when it is not.
- Training only enthusiasts. The people who most need the rules are the sceptics and the seniors who sign off.
- Measuring attendance, not outcomes. Track hours saved, errors caught and incidents reported.
What does a 30-day starter plan look like?
| Week | Action | Owner |
|---|---|---|
| 1 | List current AI use (ask people honestly); pick one low-risk use | Department head |
| 2 | Write the data, verification and disclosure rules on one page | Clinical lead with data protection contact |
| 3 | Train the group on its own de-identified material | Trainer |
| 4 | Run the pilot, log errors and time saved, decide to extend or stop | Department head |
What should you do next?
Pick one group (a department, a ward team or a faculty) and one use (documentation drafts, patient education, case writing). Write the data and verification rules first, then train that group on its own de-identified material. If you want a second opinion on the plan, book a free 30-minute call and we will tell you whether a course, a policy or neither fits. You can also read about how winsym.ai handles HRD Corp or see all courses.
Sources
Links opened and checked in November 2026 (JAMA could not be opened directly, so that one was checked in October 2026 only).
- Malaysian Medical Council, Guideline on the Ethical Use of Artificial Intelligence (AI) in Medical Practice, Version 1/2025, mmc.gov.my, clauses 3(2) to 3(15), read 22 October 2026.
- MOSTI, National Guidelines on AI Governance and Ethics (September 2024), mastic.mosti.gov.my, voluntary guidelines; official page not read in full, content taken from secondary summaries, 22 October 2026.
- "Navigating Artificial Intelligence in Malaysian Healthcare: Research Developments, Ethical Dilemmas, and Governance Strategies", PMC12304322, read 22 October 2026.
- Personal Data Protection (Amendment) Act 2024, commencement dates summarised by DFDL, dfdl.com, 22 October 2026.
- Tierney et al., "Ambient Artificial Intelligence Scribes to Alleviate the Burden of Clinical Documentation", NEJM Catalyst 2024, via divisionofresearch.kaiserpermanente.org, 22 October 2026.
- Wong et al., "External Validation of a Widely Implemented Proprietary Sepsis Prediction Model in Hospitalized Patients", JAMA Internal Medicine 2021, jamanetwork.com (figures checked against the JAMA page), 22 October 2026.
- Omiye et al., "Large language models propagate race-based medicine", npj Digital Medicine 2023, PMC10589311, 22 October 2026.
- Kung et al., "Performance of ChatGPT on USMLE", PLOS Digital Health 2023, doi.org/10.1371/journal.pdig.0000198, 22 October 2026.
- Li and Lutfi, LLM-based virtual patient systems for history-taking, JMIR Medical Informatics 2026, PMC12811743, 22 October 2026.
- Sarawak undergraduate survey on AI tools in medical education, ir.unimas.my/id/eprint/47157, 22 October 2026.
- MMC CPD Guidelines, mmc.gov.my, and MMA CPD for providers, mma.org.my, 22 October 2026.
General information, not medical or legal advice.
In this guide
- AI Course for Doctors in Malaysia: What to Look For (2026)
- AI for Medical Students in Malaysia: How to Use It Without Losing Skills
- AI in Emergency Medicine: Where It Helps, Where It Fails and What to Ask First
- Can AI Replace Nurses? What Nurses in Malaysia Should Learn
- Can ChatGPT Write Clinical Notes? What Is Safe in Malaysia
- Claude vs ChatGPT for Doctors: How to Compare Them in Malaysia
- CPD Points for AI Courses in Malaysia: Who Can Award Them
- How AI Helps Nurses Day to Day: Practical Uses and Firm Limits
- How Can AI Be Used in Medical Education? Seven Uses and Their Limits
- How Can AI Help Doctors? What Works, What Does Not and What the MMC Expects
- Patient Data and AI in Malaysia: What Is Allowed
- The MMC Guideline on AI in Medical Practice, Explained for Doctors
- Simulation-Based Clinical Training with AI: How to Design It Safely
- Integrating AI Into Clinical Training — What Hospitals Should Build First
Frequently asked questions
Is it legal for Malaysian doctors to use AI?
Yes. The Malaysian Medical Council's Guideline on the Ethical Use of AI in Medical Practice (Version 1/2025) allows registered practitioners to use AI tools. It adds duties on training, consent, confidentiality and validation, and it says the practitioner stays responsible for the clinical outcome.
Do nurses need training before using AI at work?
The MMC guideline binds registered medical practitioners, not nurses. We did not find a standalone AI guideline from the Nursing Board Malaysia as of 22 October 2026. Hospitals normally apply the same principles of consent, confidentiality and verification to every clinical role and set their own policy.
Can AI courses for doctors give CPD points?
Only if an approved CPD provider, such as a hospital, university or professional society, applies for points for its own event. A course from a company like winsym.ai does not carry CPD points by itself, and we do not advertise points.
Does the PDPA cover patient data in AI tools?
Health information is sensitive personal data under the Personal Data Protection Act 2010, which generally needs explicit consent for processing. The Act covers personal data in commercial transactions, and section 3(1) says it does not apply to the Federal Government and State Governments, so public hospitals follow Ministry of Health rules instead. Ask your hospital's legal or data protection lead.
Who teaches AI to doctors and nurses in Malaysia?
Several universities and societies run sessions. winsym.ai runs in-house courses for hospitals and faculties led by two emergency physicians, Prof. Dato' Dr. Rishya Manikam and Datin Dr. Sri Latha Balakrishnan. The courses teach the use of AI tools in practice and education. They are not clinical training.