The National Commission on AI in Healthcare: accountability, liability and the future of clinical negligence claims

The publication of the Full National Commission Report Publication by the National Commission into the Regulation of AI in Healthcare represents the UK's most comprehensive review of how artificial intelligence should be regulated across the healthcare sector.

Established by the Medicines and Healthcare products Regulatory Agency (MHRA), the Commission has proposed a new framework designed to ensure AI is deployed in a way that is safe, trusted and capable of supporting innovation.

While much of the report focuses on regulatory reform, one issue that is likely to be of particular interest to healthcare providers, insurers and clinical negligence practitioners is the Commission's concern that clinicians and healthcare organisations may become AI "liability sinks".

The problem of the AI "liability sink"

As AI becomes increasingly integrated into clinical decision-making, questions inevitably arise about who should bear responsibility when things go wrong.

The Commission notes that under the current legal framework, claims are likely to be directed primarily against healthcare professionals and healthcare providers because they owe the clearest duties of care to patients. However, this can create what the report describes as a "liability sink", whereby legal responsibility is concentrated on clinicians and providers even where the underlying cause of an adverse outcome may relate to the design, development or operation of an AI system.

The concern is particularly acute in circumstances where clinicians are expected to exercise independent judgement whilst simultaneously relying upon increasingly sophisticated AI systems. As AI tools become more embedded within healthcare pathways, identifying whether an error arose from the technology, its implementation, organisational governance or clinical decision-making is likely to become increasingly complex.

Is the law fit for purpose?

The Commission stopped short of recommending wholesale reform of clinical negligence law. Instead, it expressly referred to the recent UK Jurisdiction Taskforce statement on AI liability, which we have previously commented on, and noted the view that the common law of England and Wales remains sufficiently flexible to accommodate technological change. It is anticipated that the leading judgments of Bolam and Bolitho will remain central in assessing clinical negligence claims with clinicians being judged by the standards expected of a reasonable and responsible body of opinion. The recent NHS Resolution Guidance highlights that claims concerning the use of AI are likely to centre on areas of challenge such as:

• Did the clinician appropriately rely on the AI, or fail to apply their own judgement?
• Was the AI used in accordance with guidance, training, and regulatory approval?
• Was any known limitations/performance concerns of the AI system disregarded?  

The Commission highlights the need for development of frameworks and in time the relevant tests will be clarified in case law. The report acknowledges ongoing concerns about how responsibility should be allocated between manufacturers, healthcare providers, clinicians and regulators. It also recognises the potential effect of perceived uncertainty, particularly if clinicians fear they will ultimately bear legal responsibility for AI-generated recommendations over which they have limited control. Ultimately, AI has no legal personality, and legal liability will not attach to the system itself.  The determining factor will rest with clinicians applying their professional judgement and establishing whether they used AI in accordance with the standards expected of a reasonable body (with underlying indemnities between developers and organisations to be set out in contract).

The Commission's conclusions are also consistent with the view that existing clinical negligence principles can adapt to AI-assisted healthcare. Whilst questions will inevitably arise concerning the role of AI in clinical decision-making, the report does not advocate the creation of a new liability regime. Rather, it emphasises the importance of clarity regarding responsibilities across the healthcare ecosystem. Organisations permitting the use of AI will face questions as to the adequacy of their governance, risk management, documentation and whether reasonable steps were taken to deploy and oversee AI systems safely.

A shift towards shared responsibility

A central theme of the report is that safe AI deployment cannot be achieved through product regulation alone. The Commission advocates a "system-wide responsibility" model under which manufacturers, healthcare providers, healthcare professionals and regulators each have distinct responsibilities throughout the product lifecycle. It recommends clearer allocation of responsibility at every stage, from development and procurement through to deployment, monitoring and post-market surveillance.

In practical terms, this signals a move away from viewing AI-related incidents solely through the lens of individual clinical decision-making. Instead, scrutiny is likely to extend to procurement decisions, governance arrangements, risk management processes, training programmes and contractual relationships between providers and developers. The NHS Resolution statement accepts it is far more likely that the medical negligence claims will be pursued against the treating NHS organisation than the developers or manufacturers of AI programmes or products.

Governance and organisational oversight

The report repeatedly emphasises that healthcare organisations must be "AI ready" before deploying these technologies. Among the recommendations are an "AI readiness toolbox", national guidance on governance and lifecycle management, provider responsibilities for training staff, and contractual allocation of risk controls between manufacturers and healthcare organisations.

From a litigation perspective, these recommendations are significant. Historically, claims involving medical technology have often focused on the actions of individual clinicians. However, as AI systems become more sophisticated, courts may increasingly be asked to examine whether organisations had appropriate governance structures in place, whether risks were properly understood, whether performance was adequately monitored, and whether appropriate safeguards were implemented. These documents are likely to become part of disclosure in a claim.

Looking ahead

The Commission's recommendations do not create a new cause of action or alter the existing principles governing clinical negligence claims. However, they provide perhaps the clearest indication to date of how policymakers envisage accountability operating in an AI-enabled healthcare system.

The report recognises that AI has the potential to improve patient outcomes and support healthcare professionals, but it also acknowledges that trust depends upon clear accountability when things go wrong. Its warning against clinicians becoming AI "liability sinks" is therefore likely to resonate across the healthcare sector. Likewise, insurers will need to look at indemnities and ensure that there is clarity to policyholders on what is covered.

As AI adoption accelerates across the NHS, the legal debate may shift towards how accountability is shared when AI-assisted care falls below the expected standard. The Commission's warning against healthcare providers becoming AI "liability sinks" highlights the importance of ensuring that responsibility is allocated to those best placed to manage the relevant risks. Clear governance arrangements, effective monitoring and appropriate allocation of responsibilities are likely to become as important as the underlying technology in determining how future claims are investigated, defended and resolved.

Clyde & Co's healthcare group is recognised for its extensive industry knowledge, offering a range of legal services covering public and private sectors as well as inquests, advocacy, professional regulation, product liability and pharmaceuticals/life sciences. Should we be able to assist you, please do contact one of our experts.