Clinicians need to ensure they do not lose the relevant clinical skills and knowledgeAround the world, governments and regulators have been struggling to keep pace with the rapid adoption of artificial intelligence (AI) across all sectors. Healthcare is no exception, and as the use of AI expands within clinical decision-making, diagnostics, and health system management, there is a growing need for clear guidance to ensure the technologies are used safely and ethically.
The Malaysian Medical Council (MMC) is one of the few regulators that have already taken the step to publish the MMC Guideline on the ethical use of artificial intelligence (AI) in medical practice—over 10 pages of explicit guidance, released in January 2025. [https://tinyurl.com/477djpva] There is an emphasis on the avoidance of harm, and the guidance speaks directly to the issue of transparency and explainability—the need to understand the AI technology so that its output can be understood, bias reduced, and decisions explained to patients. At the heart of the guidance appears to be the core foundations of patient autonomy, patient safety and confidentiality, and the need for these to continue to be upheld.
Within the publication, the MMC states the need for guidelines specific to AI. They warn of the risks of over-reliance on AI recommendations, losing nuance and individuality of patient care, the requirement for extensive use of patient data and confidentiality as well as the dangers of biases. However, the MMC does not advise avoiding AI; on the contrary, it encourages clinicians to embrace AI but to do so safely, advising them to equip themselves with the skills and knowledge in its application, using clinical judgment to critically evaluate its recommendations and being able to communicate with patients about AI to facilitate informed consent.
The MMC makes clear in this guidance that clinicians’ professional obligations remain the same, when it comes to good clinical practice, consent and confidentiality, as per the MMC Code of Professional Conduct 2019. [https://tinyurl.com/4mm3nxs2] Whether using AI technology or not, clinicians are still bound by the other ethical guidelines published by the MMC.
Reasonable confidence
The MMC advises that clinicians should only use AI which meet the standards of ‘reasonable confidence’, the criteria of which are as follows:
· Trustworthiness – adherence to clinical guidelines and best practice
· Appropriateness – align with clinical evidence for its use in the clinical settings
· Transparency – and explainability of its decision making
· Validated – and conforms to regulatory standards
· Clarity – on liability
Consent and confidentiality
As is the case with incorporating telemedicine into one’s practice, the same standards of consent and confidentiality will apply when using AI, but with additional considerations and obligations.
The MMC advises that before a consultation using AI, clinicians should advise the patient of how the AI is used in their care, any known risks, and the basis for AI-generated output. It recommends informing patients of safeguards that are in place to prevent errors/data breaches, that the patient has a right to know of adverse outcomes resulting from the AI and any conflicts of interest the clinician may have. If the clinician’s view differs from the AI recommendations, then the decisions and rationale behind them should be communicated to the patient, considerations taken for the patient’s individual circumstances and wishes, and a way forward agreed.
Clinicians must also be mindful of the usual patient confidentiality requirements, ensuring that patient data is not shared on AI systems without consent and that appropriate encryption and controls to prevent data breaches are in place.
AI as a tool, not a clinician
The MMC makes the key point that ‘registered medical practitioners remain responsible for their clinical decisions’ and that ‘AI should complement, not replace clinical judgment’ (para 9). Clinicians need to ensure they do not lose the relevant clinical skills and knowledge that may be facilitated by AI, as they need to be able to take over if the technology fails, to critically evaluate AI outputs and be confident in overruling or challenging the output, as clinically required.
Useful AI resources
Medical Protection has developed the AI Safer Practice Framework to help clinicians to integrate AI safely and responsibly into their practice. [https://www.medicalprotection.org/ai-framework] The Framework is made up of two parts: INFORMED and RECORDS; it has been structured around these acronyms to ensure it is practical and memorable.