Qurexa Logo
Qurexa

Trust. Care. Deliver.

Back to Blog & Insights
Technology

All 44 of them: the government has accepted a doctor-led plan for regulating AI in the NHS

Qurexa Editorial Team6 October 20267 min read 0 0
All 44 of them: the government has accepted a doctor-led plan for regulating AI in the NHS

What happened

On Tuesday 6 October 2026, the government published its response to the National Commission into the Regulation of AI in Healthcare. It accepted all 44 of the commission's recommendations. The commission was chaired by Professor Alastair Denniston, an NHS doctor. Its deputy chair was Professor Henrietta Hughes OBE, who is also the Patient Safety Commissioner for England. It spent a year gathering evidence, and more than 12,000 patients, clinicians and members of the public took part. The headline change is a shift in how AI medical devices are checked. At the moment, a device is largely assessed once, at the point it is approved. The new approach is described as lifecycle-based: a device is monitored continuously, with real-world data, for as long as it is in use. Several specific commitments were set out. The Medicines and Healthcare products Regulatory Agency (MHRA) will publish draft guidance by December 2026 on how changes to an AI device should be managed. New staged authorisation routes will let some tools reach NHS patients earlier, under supervision. There are commitments on involving patients in decisions and on improving routes to redress when care falls short. A full implementation roadmap is promised by spring 2027. Applications also opened for Phase 3 of the MHRA's AI Airlock, a regulatory sandbox where developers test tools under the regulator's eye. This phase focuses on post-market surveillance. The first cohort will be chosen in November 2026, with an applicant webinar on 22 October. Health officials from Wales, Scotland and Northern Ireland endorsed the framework.

Why this matters

AI is already in NHS care. It is reading chest X-rays and mammograms, flagging possible strokes on brain scans, sorting referral lists and transcribing consultations. Most patients have no idea when it has been involved. The problem the commission was set up to solve is that software does not behave like a hip replacement or a blood pressure tablet. A hip replacement is the same object in five years' time. A machine learning model is not. It can be retrained. The population it sees can shift. Its accuracy can quietly drift, and nothing on the outside looks any different. Approving that kind of product once, at the start, and then leaving it alone is a mismatch between the rules and the thing being regulated. Continuous monitoring is the obvious answer, and it is harder than it sounds, because somebody has to collect the data, somebody has to look at it, and somebody has to act when the numbers move. The second thing that matters here is redress. If an AI tool contributes to a missed diagnosis, patients have historically found it very difficult to work out who was responsible: the hospital, the clinician who accepted the output, or the company that built the model. Putting redress in the recommendations, and having the Patient Safety Commissioner as deputy chair, suggests that question was taken seriously rather than left to lawyers later.

What the evidence actually says

This is a government response to recommendations. It is not, yet, a change in the rules you could point to. That distinction matters. What exists today is agreement in principle on 44 points, a date of December 2026 for draft MHRA guidance on managing device changes, and a date of spring 2027 for a full implementation roadmap. Draft guidance is not final guidance, and a roadmap is not a regulation. There is a genuine tension running through the package, and it is worth naming rather than glossing over. The Health Innovation Minister, James Frith MP, said that innovation must never come at the expense of patient safety. The MHRA's Chief Executive, Lawrence Tallon, said the aim is to make the UK the best place to get AI products to patients safely and quickly. Those two statements are both reasonable, and they pull in slightly different directions. Staged authorisation, which lets tools reach patients earlier under supervision, is precisely where that tension sits. The AI Airlock gives some reassurance here, because it is a sandbox rather than a shortcut: developers test under the regulator's supervision. Phase 3's focus on post-market surveillance is the right focus, because post-market surveillance is the part of medical device regulation that has historically been weakest, for AI and for everything else. One honest caveat on the numbers. More than 12,000 people taking part in evidence gathering is a large exercise, and it is not the same as 12,000 people endorsing the conclusions. Participation and agreement are different things. Also worth noting: Lord David Willetts, who chairs the Regulatory Innovation Office, and Kanishka Narayan MP, the AI Minister, were both named in the announcement, which tells you this is being treated as industrial strategy as well as health policy.

Practical advice

For patients, there is not much to do today, but there are things you are entitled to ask. You can ask whether software was involved in reading your scan or test. Staff may not always know the detail, but the question is legitimate and increasingly common. You can ask who reviewed the result. In NHS practice, AI tools in imaging are generally used to assist a human reader rather than replace one. If an AI flag is the only thing standing between you and a decision, that is worth querying. If a consultation is being recorded or transcribed by software, you can ask what happens to the recording and how long it is kept. That is a data protection question and you are allowed to ask it. If you believe you have been harmed by care, including care involving software, the NHS complaints process is the starting point, and the Parliamentary and Health Service Ombudsman can look at complaints that are not resolved locally. One of the commission's recommendations is specifically about making redress clearer, so this is an area to expect change in. For developers and NHS organisations: AI Airlock Phase 3 applications are open now, the applicant webinar is on 22 October 2026, and the first cohort is selected in November. If post-market surveillance is your weak point, that is the cohort's focus.

What to know

On 6 October 2026 the government accepted all 44 recommendations of the National Commission into the Regulation of AI in Healthcare, chaired by Professor Alastair Denniston with Professor Henrietta Hughes as deputy chair. The central change is from one-off approval to lifecycle regulation: AI medical devices monitored continuously with real-world data. The MHRA will publish draft guidance on managing device changes by December 2026, and a full implementation roadmap is due by spring 2027. New staged authorisation routes will let some tools reach patients earlier under supervision. Applications are open for AI Airlock Phase 3, focused on post-market surveillance, with the first cohort chosen in November 2026. This is agreement in principle, not yet a rule change. The commitments to patient involvement and clearer redress are among the more significant parts for patients. Sources: MHRA / GOV.UK, 'Government backs recommendations of NHS doctors-led AI Commission', 6 October 2026, https://www.gov.uk/government/news/government-backs-recommendations-of-nhs-doctors-led-ai-commission ; MHRA / GOV.UK, 'Government Response to the National Commission's Recommendations on the Regulation of AI in Healthcare', 5 October 2026, https://www.gov.uk/government/publications/government-response-to-the-national-commissions-recommendations-on-the-regulation-of-ai-in-healthcare ; MHRA / GOV.UK, 'AI Airlock Phase 3 application', 5 October 2026, https://www.gov.uk/government/publications/ai-airlock-phase-3-application This article is for general information and does not replace advice from a doctor, pharmacist or other qualified healthcare professional.

#artificial intelligence#mhra#medical devices#nhs#regulation#patient safety

Related articles