Al in Healthcare: are we asking the wrong question?

Looking up at the signage decorating the ceiling of the hall at NHS ConfedExpo last week – you would be forgiven for thinking it was a MedTech conference. The theme continued when the new health secretary took to the stage and promised to “back” NHS leaders who take risks by increasing the use of new technology, in his first major speech in the role. So, was James Murray right when he queried whether currently the NHS is asking itself the wrong question? Should the collective mindset shift to asking not “what if it goes wrong?” but rather “what if we don’t do this and leave things the same?”

The law frequently lags behind technological advances. Think advances in assisted fertilisation, not to mention the regulation around 3D organ printing. Perhaps no surprises then that the case law in clinical negligence stems back to the seminal decision in Bolam in 1957. Furthermore, it has been close to 50 years since the Law Commission’s first report on Liability for Defective Products was published and almost 40 years since the introduction of the product liability regime, in the Consumer Protection Act 1987. The simple truth is that the legal protections of a patient whose care has involved the use of AI, whether it be note taking, diagnoses or clinical decision making, did not have in its sights AI adoption in healthcare.

Medical Protection is putting pressure on the Government to address the “widening gulf” between the pace at which healthcare is adopting AI solutions and the law, highlighting concerns that the buck may stop with the NHS if this gulf is not addressed. In a new report – Closing the AI Liability Gap – Medical Protection lays bare that under the existing product liability framework, clinicians who use AI systems – especially to support diagnoses and treatment plans - are at risk of finding themselves liable in a subsequent civil claim brought by a patient who suffers harm, even if harm arises due to defective AI.

A recent paper exploring public perceptions around the use of AI in healthcare (Kings College London, May 2026) demonstrates the public is split on AI use in NHS clinical decision-making, with around two in five both supporting (37%) and opposing (38%) its use – but strong opposition (15%) is almost double strong support (8%) [2093 UK adults surveyed]. The chief blocker is anxiety about safety and accuracy (39%) when thinking about AI being used for clinical tasks in the NHS, and the public are twice as likely to associate it with a negative feeling compared to a positive one. Interestingly 18-to-24 year-olds are the age group most opposed to clinical use of AI, with half (49%) saying they oppose it – more than those aged 65 and over (36%).

The research interestingly showed the public would hold the individual clinician responsible if a problem was missed by AI (34%) with the Trust or hospital shaping up as the next most likely defendant (24%) and only 20% saying responsibility should be shared and 6% saying it should be the tech company that developed the product.

Just as the Met is urging tech giants to make mobile phones harder to reuse to take the profit out of crime, shared liability would have the double advantage of preserving scarce NHS resources whilst incentivising developers to prioritise safety from the outset. 

The Law Commission is currently reviewing the product liability regime which extends to its application to digital products and emerging technologies such as AI. A formal public consultation is expected in the second half of 2026, with further recommendations to follow.

Improved regulation and the safety blanket of a reimagined product liability regime would unquestionably help increase public confidence in the take up of AI in healthcare and shift the dial away from the focus being concern when AI gets it wrong.

Stay ahead of policy and regulatory change — follow our dedicated NHS Health & Care page.