Proposed AI Governance Bill a vital step, but more clarity needed on liability arrangements

Medical Protection Society (MPS) has welcomed Malaysia’s proposed AI Governance Bill as an important step towards developing a comprehensive legal framework for safe AI use in healthcare.

In its response to the National AI Office (NAIO) consultation on the proposed AI Governance Bill, MPS said a clear governance framework is essential in encouraging innovation while maintaining public confidence, and it supports the intention to draw from established international approaches to AI governance.

The world’s leading protection organisation for healthcare professionals - which supports over 4,000 healthcare professionals in Malaysia - has however said the proposed Bill should provide greater clarity in some areas, including on liability arrangements should AI contribute to patient harm.

Dr Sarah Townley, Deputy Medical Director at MPS, said:“We support the proposed Bill and commend the Malaysia Government and the NAIO for taking the steps to develop a comprehensive legal framework for AI systems throughout their entire lifecycle.

“AI has significant potential to improve healthcare by supporting clinical decision-making, improving administrative efficiency and enhancing patient outcomes. A clear and proportionate governance framework is therefore essential to encourage innovation while maintaining public confidence and protecting patient safety.

“While the proposed Bill correctly places emphasis on accountability and governance, it does not cover how legal liability should operate where AI contributes to public harm. This is particularly important in healthcare.

“AI systems increasingly influence clinical reasoning through decision support, documentation tools, triage systems and diagnostic assistance. Where harm results from defects within an AI system, responsibility should not automatically solely rest with the clinician who relied upon or interacted with the technology. Rather, liability should be shared appropriately between all parties who may have contributed to the harm.

“A governance framework that allocates responsibility according to the degree of control over an AI system should be supported by a liability framework that reflects the respective responsibilities of developers, deployers and users. Failure to provide this clarity risks creating uncertainty for healthcare professionals and could hinder the adoption of AI.

“This proposed Bill presents a key opportunity to recognise that liability arrangements should be appropriately attributed across the AI supply chain where defects in AI systems contribute to patient harm.

“We would welcome the opportunity to discuss any aspect of our submission to this consultation with the NAIO, and offer our expertise as the proposed Bill progresses.”