From diagnostic support tools to patient-facing chat, AI is arriving in Thai hospitals quickly. The legal framework has not caught up, but existing laws still apply, and they decide where the risk sits.
Is the tool a medical device?
Software intended to diagnose, monitor or support treatment decisions may be regulated as a medical device. Before deployment, hospitals should confirm the product’s regulatory status with the vendor and ask for evidence of any required approvals.
Data in, data out
AI tools are trained on, and process, patient data. Hospitals need to know what data leaves their systems, where it is stored, whether it is used to train the vendor’s models, and whether transfers outside Thailand comply with the PDPA.
Who is responsible for the decision?
A clinician who relies on an AI recommendation remains responsible for the care provided. Clear policies on when AI output can be used, and how clinicians document their own judgment, protect both patients and practitioners.
AI can support a clinical decision. It cannot take professional responsibility for it.
Contracts with vendors
Procurement is where many of these risks can be managed: warranties on regulatory status, performance commitments, data-protection obligations, audit rights and a sensible allocation of liability if the tool gets it wrong.
This article is for general information only and does not constitute legal advice. For advice on your circumstances, please contact our team.