Greater adoption of artificial intelligent tools requires resolution of the liability conundrum

Clinician-facing artificial intelligence (AI)-enabled tools can benefit patient care, but also have potential to cause physical or psychological patient harm. Opinions differ among stakeholders on who should be held responsible when such events occur – the clinician, the developer or vendor, the deploying institution, or some combination. Under current Australian tort law, clinicians remain primarily responsible for patient harm, even if software is shown to be flawed. As a result, clinicians may avoid or delay using AI tools despite demonstrated benefits to patient care. We briefly describe unique legal challenges posed by the use of AI tools, propose a framework for sharing liability among all parties according to their contribution to AI-related harm, and, failing this, offer nine strategies that clinicians and health services can adopt now to lower their liability risk.