AI and Professional Liability in Health and Care — Opportunities for Action

These workshops convened patients and the public, medical defence organisations, insurers, law firms, professional bodies, regulators, accredited registers, independent experts and others. The report describes the findings from each workshop, identifies gaps in professional guidance and regulation, and makes recommendations for how collaboration between regulators and others can address those gaps.

AI tools are already being used by health and care professionals across the NHS and social care system. In many settings they are being applied robustly, thoughtfully and to real benefit. For example, AI systems supporting imaging and diagnostics are helping patients to be diagnosed more swiftly, and are met with broad support from clinicians who recognise their potential to improve clinical efficiency. But in other cases, AI is being used without adequate training, organisational support or oversight, and health and care professionals report uncertainty about their liability when things go wrong.

Professionals are particularly concerned that they may be held accountable for the performance of a tool that they cannot personally verify. Manufacturers of AI tools may secure contracts that minimise liability on themselves, placing the burden largely on the deploying organisation and the individual professionals using the tool. Regulatory guidance has not kept pace, indemnity arrangements contain gaps, and patients have no reliable way of knowing when AI has been used in their care. The risk that clinicians become “liability sinks” was a prevailing concern across the workshops.

Five themes emerged consistently: a regulatory guidance vacuum for professionals, a legal framework that has failed to keep pace with AI, governance and procurement practices that lack consistent application, an AI education and literacy gap that is already surfacing clinical risk, and equity concerns.

To address these challenges, there was strong consensus across all workshops that cross-regulatory collaboration is both possible and necessary. Workshop attendees emphasised the need for high-level principles for regulators to be coordinated by the PSA, and cited this as the most impactful near-term opportunity for action. Longer term ambitions include model contract clauses for AI procurement, AI transparency registers for each provider, and a national AI literacy training programme for health and care professionals.

Following the workshops, the PSA has committed to coordinating professional regulators and accredited registers to issue a joint statement of intent and to develop a high-level principles framework. CERSI-AI will also work with partners to produce model contract clauses to support providers with the procurement of AI systems. These commitments are timed to align with the recommendations of the National Commission into the Regulation of AI in Healthcare, which will be published on 10 September 2026.

The report is available to read in full here on the CERSI-AI website.