UK accepts all 44 recommendations to overhaul how AI medical tools are regulated

On Tuesday, Oct. 6, 2026, the UK government said it will accept in full all 44 recommendations from the National Commission into the Regulation of AI in Healthcare, a doctor-led panel. The medicines regulator, the MHRA, also opened applications for the third phase of AI Airlock, its testing program for AI medical devices.

Most AI medical tools today get checked once, before they're sold, even though the software can drift or get updated after it's in hospitals. The UK is saying that's not enough and it will keep watching these tools for as long as they're in use. That's the right instinct, but right now it's a promise: the draft rules land in December and the full plan not until spring 2027. If it works, the UK could get safe AI tools to patients faster than the US or EU. If it stalls, it's 44 good ideas on paper.

On Tuesday, 6 October 2026, the UK government said it will accept all 44 recommendations of the National Commission into the Regulation of AI in Healthcare. The Commission published those recommendations on 10 September 2026. The written response, from the Medicines and Healthcare products Regulatory Agency, sets out how they will be delivered across the UK. The MHRA is the regulator for medicines and medical devices. The news page says the government confirmed the decision that day. The response itself is titled as published on Tuesday, 6 October 2026, and its foreword says, “We are accepting all 44 of their recommendations.” Those lines are the government’s.

Who wrote the recommendations. The MHRA set up the Commission in September 2025 as an independent advisory body. Independent, here, means it advised the government. It does not itself pass a law. NHS doctors led it. Professor Alastair Denniston chaired it. The government’s notes on the news page say the Commission’s year of evidence gathering involved more than 12,000 people, and call that the largest engagement of its kind on how healthcare technology is regulated. Denniston said the response shows the government listened to the Commission and to “the more than 12,000 patients, clinicians and members of the public who shaped our recommendations.” Those lines are the government’s and Denniston’s, on the same page.

The change the government is accepting. AI medical devices are software tools used in care, and they can be updated after they are already in hospitals. The Commission said the old approach leans too hard on one check before a device goes on the market. The government says these devices should be assessed and watched throughout their working life, so their real performance is known while they are in use. The news page calls that a move away from a single point in time. The response’s foreword says the UK needs regulation that is proportionate, based on risk, and based on that whole working life. Proportionate, in that sentence, means the rules should match how risky the tool is. Those lines are the government’s.

What the MHRA will publish first. It says it will issue draft guidance by December 2026 on how to handle AI medical devices that change as they adapt and improve. Draft guidance is a proposed rulebook, open for comment, not the finished rule. The MHRA also says it will consult next year on how AI devices are qualified and classified. Qualify and classify means deciding whether a piece of software counts as a medical device, and which risk group it sits in. The government says that consultation is the Commission’s first recommendation. Those lines are the government’s.

How a promising tool could reach patients sooner, and what patients are owed. The government says it will explore staged approval routes. A staged route would let a promising AI tool be used in the NHS earlier, under close supervision, while more evidence from real use is gathered. The same announcement says patients should be told when AI is used in their care. The government also says it will strengthen how patients take part in decisions, improve access to redress when care falls short, and share the benefits fairly, including with underserved groups. Redress means a route to raise it, and to seek a remedy, when care is not good enough. Those lines are the government’s.

When the full plan is due. A full implementation roadmap, with timelines for all 44 recommendations, is due by spring 2027. The news page says regular progress updates will follow. The response says this publication is the next stage of the work, not the end of it. Those lines are the government’s.

The testing program that opened the same day. AI Airlock is the MHRA’s regulatory sandbox for AI medical devices. A sandbox, here, is a controlled test: developers, the regulator, and health partners try a real product and learn how the rules should work. The MHRA opened applications for Phase 3 on 6 October 2026. The news page says the program has three more years of government funding, and that Phase 3 focuses on watching devices after they are deployed. A webinar for applicants is on 22 October 2026 at 10:00. The page does not print a time zone next to that clock. The first wave of companies will be picked in November 2026. The application page says people who want to be in the current review wave should apply before 31 October, that this phase is funded until April 2029, and that there is no fee to apply or to take part. Those lines are the MHRA’s.

What the minister and the regulator said. Health Innovation Minister James Frith said “innovation must never come at the expense of patient safety,” and that the government is “accepting in full all 44 recommendations of the Independent Commission.” MHRA chief executive Lawrence Tallon said the Commission, “led by NHS doctors,” has given “a clear and ambitious blueprint” for getting AI products to patients safely, quickly, and in a way that keeps public trust. Those sentences are theirs, on the news page.

How other outlets reported the same announcement. Fierce Biotech, by Ben Adams, said the MHRA will continually monitor AI-based devices, and called that a step up from one initial check. Its page prints Oct. 6, 2026, 8:02 a.m., and does not print a time zone next to that clock. pharmaphorum, by Phil Taylor, said the government will accept all 44 recommendations, including monitoring after a device is deployed, with a full implementation plan by spring 2027 and draft proposals before the end of the year. Those framings are the outlets’. The government’s own line is that it is accepting the recommendations and setting out how they will be delivered. The draft guidance is due in December. The roadmap is due by spring 2027.

The picture is a teal and navy summary card. A large 44 stands for the recommendations the government accepted. Four lines name the commitments beside it: watching AI medical devices through their working life, draft guidance by December 2026, AI Airlock Phase 3 opening, and a full roadmap by spring 2027. It is a summary of the announcement. It is not a photograph of a hospital, a device, or a minister.

In plain terms, the UK said on Tuesday it will take all 44 recommendations from a doctor-led commission on how AI medical tools are regulated. The change it is promising is to keep checking those tools for as long as they are in use, not only once before they are sold. The first draft of the guidance on devices that change is due in December 2026. The plan with dates for every recommendation is due by spring 2027. The same day, the MHRA opened the next round of AI Airlock, the program where companies test those questions with the regulator.