UK Accepts All 44 AI in Healthcare Recommendations: Draft Guidance on Changes to AI-Enabled Devices Due December 2026, AI Airlock Phase 3 Intake Closes 31 October

UK Accepts All 44 AI Healthcare Recommendations: Draft AI Device Change Guidance Due in December, AI Airlock Phase 3 Round Closes 31 October

Published 27 October 2026 09:31 — Registration and Market Access

Two actions on 6 October 2026

On 6 October 2026, the UK Government published its formal response to the report of the National Commission on the Regulation of AI in Healthcare, confirming that it accepts all 44 recommendations put forward by the Commission. In the same announcement, MHRA opened the application channel for phase 3 of the AI Airlock and aligned the phase 3 testing theme directly with the Commission's core conclusions.

These two actions happened together; they are not separate events. The Commission's report was published on 10 September 2026 under the chairmanship of Professor Alastair Denniston. The Government issued its full acceptance in less than a month and opened the implementation tool — the sandbox — first.

Three dates that can go into a project plan

The first is December 2026. MHRA has committed to publishing draft guidance before that date on how to manage changes to AI-enabled medical devices as they adapt and improve over time. For products with continuously learning models or iterative model versions, this draft is more important than any statement of principle.

The second is 2027. MHRA will hold a public consultation next year on qualifying and classifying AI devices — the Commission's first recommendation.

The third is spring 2027, when the complete implementation roadmap covering the 44 recommendations will be published, setting out timelines and responsible parties for each recommendation. To date, the announcement only refers to a Programme Board having been established and lists ten priority areas.

The announcement also refers to exploring a phased authorisation pathway that would allow promising AI tools earlier access to NHS use, subject to continued collection of real-world evidence. This remains an exploratory direction and does not yet constitute an assessment pathway that can be applied to.

What AI Airlock phase 3 is looking for

The AI Airlock is MHRA's regulatory sandbox for AI medical devices. It has completed its pilot and second phase and has secured three additional years of government funding. According to the phase 3 application page, the current funding covers the period up to April 2029.

Phase 3 has a single overarching priority regulatory challenge: post-market surveillance and total lifecycle management of AI devices. The page breaks this down into three specific questions.

• How can deployment-context factors be captured proactively in the design, including local infrastructure, local workflow integration, data availability and quality, and user interaction?
• How can performance or safety signals be detected and acted upon in environments with varying degrees of local oversight — the examples given are remote, direct-to-consumer and more autonomous systems?
• How can information on performance, safety, updates and rollback be communicated to regulators, institutions, professionals and patients at the right time?

On timing, the application window is open year-round, but projects must be submitted by 31 October 2026 to be considered in the current review round. Projects that miss this round are generally deferred to the next round, with the next screening in spring 2027. Individual project testing plans typically run for 6 to 12 months.

Relationship to the 2030 Strategy and UK-US cooperation

The MHRA 2030 Strategy, published on 14 October 2026, embeds the same agenda at the organisational level: establishing an MHRA Centre for AI and Digital Health to provide clearer regulatory expectations for AI-driven digital devices; building a modern, adaptive framework for AI as a medical device; improving device data and lifecycle traceability; and expanding reliance and mutual recognition pathways with trusted regulators, with the US FDA specifically named.

On 19 October 2026, Jessica Paulsen, MHRA's Deputy Director responsible for AI and software, elaborated at the AdvaMed conference: the MHRA-FDA Pioneer Programme announced in June establishes reciprocal liaison posts within both agencies, and MHRA is recruiting a Regional Deputy Director for US Partnerships in Washington to serve as liaison officer embedded at FDA. The same source makes clear that both agencies continue to make regulatory decisions independently.

Scope and boundaries of the source text

All content in this article is drawn from public UK Government and MHRA pages and does not involve the assessment outcome of any specific product. Three categories of information should be distinguished: confirmed items (all 44 recommendations accepted, AI Airlock phase 3 open for applications, 31 October 2026 as the current round deadline, funding covering to April 2029); items with a clear time commitment but no published text yet (the December 2026 draft change management guidance, the 2027 consultation on qualification and classification, the spring 2027 implementation roadmap); and items still exploratory (the phased authorisation pathway and the specific conditions for earlier NHS access). The third category should not be written into a registration plan as a current requirement.

Sources and applicability

Sources: GOV.UK / MHRA, 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; GOV.UK / MHRA, AI Airlock phase 3 call for applications, 6 October 2026, https://www.gov.uk/government/publications/ai-airlock-phase-3-application; GOV.UK / MHRA, MHRA 2030 Strategy to modernise patient-centred regulation for the era of AI and personalised medicines, 14 October 2026, https://www.gov.uk/government/news/mhra-2030-strategy-to-modernise-patient-centred-regulation-for-the-era-of-ai-and-personalised-medicines; GOV.UK / MHRA, UK regulator brings new AI and MedTech agenda to AdvaMed, 19 October 2026, https://www.gov.uk/government/news/uk-regulator-brings-new-ai-and-medtech-agenda-to-advamed.

Verification scope and access date: the four GOV.UK pages above, together with the publication-date labelling on MHRA's news list on GOV.UK, were accessed and checked on 27 October 2026. All dates, deadlines, funding periods and theme descriptions in this article are taken from the wording of those pages; no media reports were cited.

Applicability note: this article is an interpretation of public regulatory information and does not constitute a conclusion or commitment regarding any specific product, company or registration application. The draft guidance, consultation document and implementation roadmap have not yet been published, and actual requirements will be governed by subsequent official texts; the relevant information may change as official updates are issued.

Information sources

• GOV.UK / Medicines and Healthcare products Regulatory Agency (MHRA) — Government backs recommendations of NHS doctors-led AI Commission — 6 October 2026
• GOV.UK / MHRA — AI Airlock phase 3 call for applications — 6 October 2026
• GOV.UK / MHRA — MHRA 2030 Strategy to modernise patient-centred regulation for the era of AI and personalised medicines — 14 October 2026
• GOV.UK / MHRA — UK regulator brings new AI and MedTech agenda to AdvaMed — 19 October 2026

AIMEILI Regulatory Interpretation

The UK is moving from principles to operational instruments on AI-enabled medical devices, and the sequencing matters for regulatory planning. Full acceptance of the 44 recommendations is a policy commitment; the AI Airlock is the mechanism through which MHRA will generate the evidence base for future requirements; and the December 2026 draft on managing device changes is the first document likely to translate the policy into assessable expectations.

For manufacturers, the most consequential signal is the explicit focus on change. Traditional device regulation assumes a frozen, validated configuration, with significant changes handled through new submissions or notifications. Adaptive and continuously learning AI devices challenge that model, and MHRA has committed to publishing a dedicated approach before the end of 2026. Until that draft appears, the prudent position is to maintain a documented change-management architecture — model versioning, retraining triggers, predetermined change control plans, validation and rollback provisions — that can be mapped onto the forthcoming guidance without redesign.

The phase 3 Airlock theme of post-market surveillance and total lifecycle management reinforces where regulatory attention is concentrated: deployment-context variability, signal detection under decentralised oversight, and timely communication of performance, safety, update and rollback information. These mirror the questions that other mature regulators are raising, which means evidence generated through the sandbox is likely to be reusable in other jurisdictions.

The MHRA 2030 Strategy and the MHRA-FDA Pioneer Programme add a reliance dimension. Reciprocal liaison posts and a Washington-based regional deputy director indicate a deliberate effort to align approaches with FDA, while both agencies retain independent decision-making. For dual-market developers, this supports building a single lifecycle evidence file that can serve both UK and US expectations, rather than parallel systems.

Finally, the qualification and classification consultation in 2027 should be tracked closely. How an AI device is qualified and classified determines the applicable conformity route and evidence burden; early engagement with the consultation is a low-cost way to influence and anticipate the outcome.

Business Impact

Immediate action — 31 October 2026: any organisation intending to participate in the current AI Airlock review round must submit by this date. Projects submitted later will generally be deferred to the next screening in spring 2027, which can delay access to MHRA feedback by six months or more. Applications should frame the proposal around one or more of the three phase 3 questions and set out a realistic 6- to 12-month test plan with defined evidence outputs.

December 2026 readiness: manufacturers of adaptive, retraining or frequently updated AI devices should prepare for the draft change-management guidance now. Practical steps include auditing current change-control procedures against the concepts of predetermined change control, version traceability, performance monitoring and rollback; identifying where current processes rely on undocumented assumptions; and assigning ownership for the gap closure. Companies that wait for the final guidance will be reacting to requirements rather than shaping them.

Post-market surveillance investment: the phase 3 focus confirms that PMS for AI devices will be assessed on real-world performance across heterogeneous deployment settings, not only on pre-market validation. Budget and system changes should reflect continuous performance monitoring, drift detection, complaint and signal handling under varying local oversight, and user-facing communication of updates. This is also likely to be the area where notified bodies and MHRA ask the sharpest questions.

Strategic positioning for 2027: the qualification and classification consultation and the spring 2027 implementation roadmap will define the medium-term operating environment. Regulatory affairs teams should plan to respond to the consultation, monitor the roadmap's allocation of responsibilities, and reassess UK launch timelines for AI-enabled products accordingly. The phased authorisation pathway remains exploratory and should be treated as a watch item, not a planning assumption.

Cross-market efficiency: with MHRA and FDA deepening cooperation while retaining independent decisions, developers active in both markets should align lifecycle documentation, change-control plans and PMS architectures to a common standard that satisfies both regulators, reducing duplication and shortening response times to future guidance in either jurisdiction.