UK accepts all 44 AI-in-healthcare recommendations. Most of the detail is due by Spring 2027

The UK government said on 6 October 2026 that it will accept all 44 recommendations of the National Commission into the Regulation of AI in Healthcare, the independent body the MHRA set up in September 2025 and which published its report on 10 September. The MHRA opened applications for the third phase of AI Airlock, its sandbox for AI medical devices, the same day.

What the government actually signed up to

We counted the response paper's action plan line by line. It lists 44 recommendations, and each one carries the single label "(Accept)". None is marked as partial or accepted in principle. The ministerial foreword, signed by Health Innovation Minister James Frith, says: "We are accepting all 44 of their recommendations."

Accepting is not the same as delivering, and the paper says as much: "this publication does not mark the end of the process." Many of the commitments are worded as the MHRA or the Department of Health and Social Care "will consider" or "will explore" options.

The dates that are set

Two of the 44 entries name a specific deliverable with a date. The other 42 say "Further details to be shared" by, or in, Spring 2027.

• Changes to AI devices (Recommendation 6): the MHRA "intends to issue draft guidance by December 2026" on predetermined change control plans, which would let manufacturers "establish boundaries to define the scope of allowable change".
• What counts as a medical device, and its risk class (Recommendation 1): the action plan says the MHRA "intends to launch the consultation by Spring 2027". The press release puts it more loosely, saying the MHRA will "begin next year by consulting".
• Roadmap: an implementation plan and roadmap is due "by Spring 2027". The paper promises "annual progress updates thereafter"; the press release says "regular progress updates".

AI Airlock Phase 3

The MHRA's application page says Phase 3 focuses on "Post-Market Surveillance (PMS) and oversight across the lifecycle", meaning how AI devices are watched once they are in use. The details:

• A webinar for applicants is on 22 October 2026 at 10:00, and the first wave of innovators will be selected in November 2026, according to the press release.
• Applicants who want to be in the current review wave should apply before 31 October. The window then stays open, with a further shortlisting period in Spring 2027.
• The phase is "funded until April 2029". Individual projects should generally run for 6 to 12 months.
• There is no fee. The MHRA's page adds: "Currently, candidates are expected to fund their own studies".

The Department of Health and Social Care had already allocated £1.2 million a year for 2026 to 2029 to the programme. In July we reported on the Airlock's Phase 2 findings, where the MHRA's report said that, in the TORTUS ambient-scribe case study, "without active guardrails, out-of-scope performance was detected in approximately 39% of real-world notes, falling to 20% with guardrails active".

What changes for NHS staff and patients, and when

On our reading, nothing changes on a ward yet: the response is a set of commitments, with most detail due by Spring 2027. The planned changes that would reach staff and patients include:

• Being told when AI is used. The health department will consider how to "clarify transparency expectations where AI is used in patient care, what choices patients have". It will also consider "when opt-out arrangements may be feasible and appropriate".
• Public information. The MHRA says it will work to publish "a regularly updated list of authorised AI-enabled medical devices in the UK", and will work to introduce a public database where people can search reports of adverse incidents.
• Who is responsible when it goes wrong. The health department, regulators and NHS Resolution will form a working group to make responsibility clear across the lifecycle. A separate group, which includes the Patient Safety Commissioner, will look at redress for patients. NHS England and NHS Resolution will also "continue to progress guidance on indemnity". On our reading, this bears on the gap we described in September: a June Medical Protection report warned that clinicians risk absorbing the legal responsibility when an AI system errs.

Reaction

Apart from the Health Research Authority's, the quotes below come from the government's own press release.

Professor Alastair Denniston, who chaired the Commission, said the response "shows that government has listened". Professor Henrietta Hughes, the Commission's deputy chair and Patient Safety Commissioner for England, said she "particularly" welcomed the commitments on monitoring and on "clear routes to redress when things go wrong".

Health Innovation Minister James Frith said "innovation must never come at the expense of patient safety". AI Minister Kanishka Narayan said: "We're working to make promising new tools available to people across the country sooner."

Steve Lee, executive director of regulation at the ABHI, said "getting implementation right will be important" and that "Building on, and aligning with, international best practice will be key." Scotland's Health Secretary, Angela Constance, said Scotland would "in due course" set out its own response. The Health Research Authority's chief executive, Matt Westmore, said the HRA would keep working with the MHRA so that "regulatory and research governance arrangements remain clear, proportionate and connected".

Why it matters

On our reading, the direction is now set: the government says it wants AI devices monitored across their working life, not just assessed once. The timetable is mostly still to come. The two dates to watch are the December 2026 draft guidance on device changes and the Spring 2027 roadmap. Until then, the paper says what has been accepted more than how it will be done.