NHS Doctors Face Legal Risk as AI Liability Gap Leaves Them Exposed to Negligence Claims

It's Time to Close the AI Liability Gap

As AI advances in healthcare, recommender systems -- which analyse patient data and suggest personalised treatment plans, diagnoses, or medications -- are poised to become increasingly influential in clinical decision-making. While this shift could benefit patients in the future, these systems also raise complex questions about liability -- particularly regarding accountability when a patient comes to harm due to a defective AI system. Would it lie with the clinician? The AI manufacturer? The supplier of the system? All parties? When the Algorithm Gets It Wrong Consider this hypothetical but plausible case: A 69-year-old woman had been taking long-term warfarin for atrial fibrillation; her GP regularly monitored her and provided dosing advice. The patient reported a recent short episode of diarrhoea and vomiting, which had subsequently settled. Recent blood tests had been inputted into an AI system that assisted with warfarin dose titration, and the system inappropriately recommended an increase in the warfarin dose. The GP followed the recommendation and increased the dose. One week later, the patient was admitted to hospital with a gastrointestinal bleed requiring surgery and an ICU stay. She was found to have an international normalised ratio significantly above the recommended range. The patient later brought a claim against the GP, alleging that the increased dose of warfarin was inappropriate and had resulted in the haemorrhage and hospital stay. In this case, the AI-generated recommendation was potentially unsafe and inappropriate. While the GP retains clinical responsibility for prescribing, surely the AI developer or manufacturer should also share liability due to the flawed dosing advice? A Legal Framework That Hasn't Kept Up Currently, this would be unlikely. Under the existing product liability framework, AI systems are not clearly defined as 'products' -- meaning developers, manufacturers, and suppliers of AI tools are likely to be shielded from the liability rules that would usually apply if a defective product caused harm. This loophole has not come about by design -- the Consumer Protection Act 1987 is decades old and was simply never designed with AI in mind, meaning AI systems will likely fall outside of its scope. Nevertheless, under the current framework, it is difficult for patients to bring product liability claims against developers, manufacturers, and suppliers of AI tools, if harm occurs. The default may therefore be to pursue the end users of the AI through a clinical negligence claim -- specifically, the NHS, where taxpayers would ultimately foot the bill, and clinicians in private practice who indemnify or insure themselves against claims. At Medical Protection Society (MPS), we do not feel it is equitable that clinicians are at risk of absorbing all legal responsibility if a patient comes to harm, particularly where the harm arises from a defective AI system. In our new report, Closing the AI Liability Gap, we call on the UK government to address this by developing legislation that would classify AI systems as products subject to strict liability. The EU Has Already Moved -- So Should the UK The law has always struggled to keep up with technological change. But with AI, the pace of change is so rapid that this gap feels less like a step and more like a widening gulf. AI cannot be governed with the tools of the past. The government has made its desire for AI to play a central role in the future of healthcare very clear, backing that ambition with significant investment across the NHS. But this ambition has not yet been matched by an evolution in the legal framework that assigns responsibility when things go wrong. The government is not without inspiration when it comes to developing potentially appropriate legal frameworks for the UK. The new EU Product Liability Directive 2024/2853 modernises the European liability framework to take account of emerging technologies and will come into effect across the EU by December 2026. In the directive, standalone AI systems and AI components embedded in devices, are expressly included within the definition of a 'product'. This ensures that AI systems are subject to a similar liability regime as physical goods and imposes continuing obligations on those who design, update, and distribute AI systems. For EU member states, the directive closes the gap that previously left AI outside the traditional product liability regime. While any new product liability framework for the UK would need to be tailored to the UK context, the government could draw on this directive. The potential benefits from action are clear: fairer outcomes when things go wrong, better protection for the NHS and clinicians when harm arises from defective AI, and the introduction of safer AI tools -- as shared liability creates stronger incentives for developers to prioritise safer design and ongoing testing, rather than externalising risk to users or downstream organisations. Reviews Are Under Way -- But Speed Matters Awareness of the AI liability gap in the UK, and the need to address it, is slowly growing. For example, the Medicines and Healthcare products Regulatory Agency, via a National Commission, is carrying out a review of AI regulation in healthcare, which includes seeking views on how liability should be allocated when AI tools are used. The Law Commission is also undertaking a review of the product liability regime and its application to emerging technologies. We will engage fully with these reviews, but we hope they progress at pace. If the government is serious about the UK being a world leader in AI, it must confront this issue swiftly. Useful Resources At MPS, our aim is to support healthcare professionals in embracing AI while helping them to understand and navigate the associated medicolegal risks. Last year we launched the AI Safer Practice Framework, designed to help healthcare professionals integrate AI safely and responsibly into practice. The framework is made up of two parts: INFORMED and RECORDS. INFORMED guides ethical decision-making in the use of AI, while RECORDS documents AI-assisted decisions for accountability and clinical rationale. The framework has been structured around these acronyms for ease of use. The AI Safer Practice Framework can be accessed here. This article is published as part of an editorial collaboration between Medscape UK and MPS that aims to deliver medicolegal content to help healthcare professionals navigate the many challenges they face in their clinical practice. Dr Sarah Townley, MBChB, is the deputy medical director at MPS, and a board member for the MPS Foundation. Prior to joining MPS in 2009, Townley studied medicine at the University of Leeds before working as a GP. During her time at MPS, Townley has undertaken various roles, including medicolegal consultant and underwriting policy lead. Townley has also gained her diploma in legal medicine, membership of the Faculty of Forensic and Legal Medicine, and CII certificate. She has disclosed no relevant financial relationships. MPS membership provides the right to request access to expert advice and support on clinical negligence claims, complaints, General Medical Council investigations, disciplinaries, inquests, and criminal charges such as gross negligence manslaughter.

Doctors and NHS could be sued for mistakes made by AI tools, report warns

Medical Protection Society calls for law to be overhauled to help medics avoid liability for errors made by technology Doctors and the NHS could be sued for medical negligence over mistakes made by artificial intelligence tools used in diagnosing patients and suggesting their treatment, ministers are being warned. Under the law as it stands, medics and the health service can be held liable for patients being harmed or dying even if it was AI that made the errors that resulted in their suffering. The Medical Protection Society, which represents doctors accused of wrongdoing, says in a report that medics could become the "liability sink" - a target of clinical negligence lawsuits - for mistakes made by AI unless the law is overhauled. The NHS is using AI for more and more purposes, including to analyse scans and X-rays, generate summaries of doctors' conversations with patients, and draft letters to patients. "The law has always struggled to keep up with technological change. But with AI, the pace of change is so rapid that this gap feels less like a step and more like a widening gulf," said Dr Sarah Townley, the MPS's deputy medical director. Giving an example of potential harm from AI errors, the MPS said AI could miss a tumour in a patient's lung when reading an X-ray of their chest. This could result in the patient dying because the false reassurance from the AI would mean no treatment would be given and the cancer could then spread. Similarly, a patient could need surgery and treatment in intensive care for severe bleeding if an AI wrongly recommended increasing their dose of warfarin, a blood thinner used to treat the heart condition atrial fibrillation. In such scenarios there was a real and significant risk that a claim would be brought against a doctor in relation to the use of AI tools, the MPS said. "Under the current product liability framework in the UK, there is a risk that clinical negligence claims could be brought against the clinicians in these cases and that they would be held wholly liable," it warns. The body wants the government to reclassify AI tools and systems as products, so they fall under the scope of the Consumer Protection Act 1987. It believes that would help doctors and the NHS avoid liability for mistakes that the technology has made. Medics in the UK are increasingly worried about doctors being blamed for errors by AI. Public trust in medicine may fall without action to make AI developers and manufacturers liable, they fear. "Innovation and patient safety should move forward together. If AI is advancing at Formula One speed, then legislation, regulation and governance cannot be left sitting in the pit lane," said Dr Ragit Varia, the president-elect of the Society for Acute Medicine. "Clinicians should not find themselves holding a liability hot potato when decisions have been influenced by AI systems developed, supplied and implemented by others without the appropriate structure. We must avoid creating an accountability vacuum where responsibility for harm is unclear." NHS Resolution, which handles negligence claims against hospitals in England, is drafting guidelines on AI liability, the Department of Health and Social Care said. "We welcome the MPS's report and will review its recommendations to ensure patients continue receiving the benefits of AI in healthcare safely and quickly," a DHSC spokesperson said. Ahmed Binesmael, a senior policy analyst at the Health Foundation thinktank, said: "Our research consistently shows that public confidence in AI depends not just on the technology itself, but on the safeguards and oversight that accompany it. As AI adoption grows across the NHS, ensuring clear accountability and robust governance will be essential to maintaining public trust and confidence."