The Medical Protection Society (MPS) has called on the government to legislate to classify AI systems as products, warning that a growing mismatch between healthcare AI and the UK's product liability framework risks leaving the NHS and individual clinicians to absorb the full legal cost when defective systems cause patient harm.
In a policy paper published on 9 June 2026, Closing the AI Liability Gap, the MPS argued that because AI systems are not clearly defined as "products" under the Consumer Protection Act 1987, the developers, manufacturers and suppliers of AI tools are likely to sit outside the strict liability rules that would ordinarily apply when a defective product injures someone.
The practical consequence, the society warned, is that patients harmed by a faulty AI system will struggle to bring product liability claims against the companies behind it. Instead, liability is likely to fall to the end user: a clinical negligence claim against the NHS or against clinicians in private practice, who must indemnify or insure themselves.
The organisation says the risk is most acute for AI recommender systems, which analyse patient data and suggest diagnoses and treatment plans, leaving the clinician who acts on a flawed recommendation carrying all of the legal responsibility for the outcome.
Dr Sarah Townley, the society's Deputy Medical Director, said the law has always lagged behind technology, but with AI the gap now feels "less like a step and more like a widening gulf". She described the current framework as inequitable, arguing that the 1987 Act was never designed with AI in mind and that legal responsibility for harm arising from defective systems should be distributed fairly across those who develop, supply and deploy them.
Clear, shared liability would not only protect the NHS and clinicians, Dr Townley argued, but would build trust in AI and give developers a direct incentive to prioritise safety. This, she suggested, should be a pre-condition of the Government's ambition for the NHS to lead the world in AI adoption.
Professor Gozie Offiah, Chair of the MPS Foundation, said AI in healthcare had moved from aspiration to reality, but warned the disconnect between its use and the liability framework risks making clinicians and the NHS the obvious target for negligence claims.
Roger McMillan, head of the healthcare team at law firm Carson McDowell, added that unambiguous legislative classification of AI systems as products would give developers, software markets and end users much-needed clarity about legal risk.
The report has been published ahead of a formal public consultation by the Law Commission, expected in the second half of 2026, as part of its ongoing review of the product liability regime and its application to digital products and emerging technologies including AI.
The MPS has also published an AI Safer Practice Framework (https://www.medicalprotection.org/ai-framework ) to support the safe integration of AI into clinical practice. This week, NHS England extended the deployment of Microsoft’s Copilot AI tool to half a million staff.
