UK Accepts All 44 AI Medical Device Regulation Recommendations: December Change Guidance and AI Airlock Cohort 3

UK accepts all 44 AI medical device regulation recommendations: December change guidance due and AI Airlock cohort 3 open

Published: 20 October 2026 | Category: Registration and market access

On 6 October 2026, the UK Government confirmed that it accepts all 44 recommendations made by the National Commission into the Regulation of AI in Healthcare. The Commission's report, published on 10 September 2026, concluded that the current regulatory approach—centred on one-off approval—cannot keep pace with AI technologies that continue to iterate, and that regulation should move to a lifecycle-based model.

On the same day, the Medicines and Healthcare products Regulatory Agency (MHRA) opened applications for cohort 3 of AI Airlock, its regulatory sandbox for AI medical devices. AI Airlock has completed a pilot and a second cohort and has received an additional three-year government funding allocation.

Neither the UK Government nor MHRA described the changes as deregulation. The official position is to enable safe and effective AI to reach clinical use faster while maintaining patient safety and public trust. The two statements should be read together.

Milestones already announced

• By December 2026: MHRA will publish draft guidance on how changes to AI medical devices should be managed when devices adapt and improve during use. For manufacturers, this means model upgrades, training-data updates and performance-threshold adjustments are likely to require a regulator-reviewable process rather than being treated solely as internal engineering changes.
• From 2027: MHRA will hold a public consultation on the qualification and classification of AI devices. This is the Commission's first recommendation and the starting point of the regulatory chain: whether software qualifies as a medical device, and which class it falls into, determines the pathway and the evidence required.
• By spring 2027: A full implementation roadmap will be published, setting out the timetable and responsible parties for each of the 44 recommendations, with progress reported periodically. The direction and some milestones are therefore known, but not all detailed rules.

What AI Airlock cohort 3 will test

The theme of cohort 3 is post-market surveillance and lifecycle regulation. According to the official description, real products and technologies will be used to bring developers, regulators and healthcare partners together to examine how AI devices are continuously monitored and managed after deployment. The evidence and experience generated will feed directly into MHRA's future guidance and policy.

Applications are open to eligible AI medical device developers. An application briefing webinar will be held on 22 October 2026 at 10:00 UK time, and the first selected companies are expected to be confirmed in November 2026.

MHRA also noted that the sandbox work is linked to regional partnerships, including with NHS London and Manchester University NHS Foundation Trust, to test and generate evidence in real clinical settings rather than only in theory.

International reliance and FDA engagement

On 19 October 2026, at AdvaMed 2026, MHRA further set out its device and AI direction under its 2030 strategy: establishing a modern, adaptable regulatory framework for AI medical devices; creating a new MHRA AI and Digital Health Centre; improving device data and traceability; and expanding reliance and recognition pathways with trusted regulators, including the US Food and Drug Administration (FDA).

This is supported by the MHRA-FDA Pioneer Programme announced in June 2026, which places liaison roles in both agencies for day-to-day collaboration, scientific exchange and coordination on emerging topics such as devices and AI. MHRA is recruiting a Regional Deputy Director for US relations in Washington to serve as a dedicated liaison officer at FDA. The official position remains that each agency continues to make independent regulatory decisions.

Important distinction: "expanded reliance pathways" is not the same as "mutual approval recognition." Reliance means a regulator gives greater weight to assessments already completed by its counterpart. Mutual recognition would mean that approval in one market directly equals approval in another. The public documents currently support the former, not the latter.

Issues not yet settled

The Commission's recommendations include exploring a "phased authorization" pathway that would allow promising AI tools to enter NHS use earlier under close supervision while real-world evidence continues to be collected. This remains an exploratory direction and is not yet a legal pathway on which manufacturers can rely.

Wider device reform, according to MHRA, includes strengthened post-market surveillance and legislative amendments introduced through the Government's health bill. The legislative timetable is not controlled by MHRA, which is one reason the roadmap places substantial work in 2027.

Implications for Chinese manufacturers

In the short term, the changes will most directly affect AI software products that are already on the UK market or are preparing to enter it: the evidentiary burden for change management will fall on manufacturers first. In the medium term, the outcome of the consultation on classification and qualification may change whether certain borderline software is captured as a medical device.

Caution: All the information above comes from public UK Government and MHRA publications. As of 20 October 2026, the December draft guidance had not been published. Any specific submission requirements inferred from these announcements are predictions and should not be used as a compliance basis.

Sources and scope

Sources:

• GOV.UK / MHRA — Government backs recommendations of NHS doctors-led AI commission (published 6 October 2026): https://www.gov.uk/government/news/government-backs-recommendations-of-nhs-doctors-led-ai-commission
• GOV.UK / MHRA — UK regulator brings new AI and MedTech agenda to AdvaMed (published 19 October 2026): https://www.gov.uk/government/news/uk-regulator-brings-new-ai-and-medtech-agenda-to-advamed

Verification scope and access date: This article is based solely on the public content of the above gov.uk pages as accessed on 20 October 2026. No secondary media reports were cited, and no assumptions have been made about the content of the unpublished draft guidance.

Applicability: This article discusses the regulatory arrangements for Great Britain and does not address specific differences in the rules applicable to Northern Ireland. The dates cited are official commitment milestones and may change. This article does not constitute legal advice. Whether a specific product qualifies as a medical device and which pathway applies must be assessed separately in light of its intended use, software functions and UK responsible-person arrangements.

AIMEILI regulatory interpretation

The UK is moving from a one-off approval model to lifecycle oversight for AI medical devices. The most significant near-term development is the draft change-management guidance expected by December 2026. Although its content is not yet public, the direction is clear: manufacturers should expect to document, validate and, where appropriate, notify changes to models, training data and performance thresholds under a structured process. Companies that already have UK-registered AI devices should review their change-control procedures now, because the first compliance burden will fall on them.

The 2027 consultation on qualification and classification is equally important for software products at the boundary of medical device regulation. The outcome may affect whether certain clinical decision-support, triage or workflow tools are regulated as devices, and if so, which class and conformity route applies. Manufacturers should monitor the consultation and consider whether their current UK regulatory position remains robust.

AI Airlock cohort 3 offers early regulatory engagement on post-market surveillance and lifecycle management. Participation is not an approval and does not replace registration, but it can provide valuable insight into MHRA expectations and help shape future guidance. The application webinar on 22 October 2026 and the expected November 2026 selection are immediate action points for eligible developers.

MHRA's international reliance agenda should not be read as mutual recognition. Expanded reliance with FDA and other trusted regulators may reduce duplication in some areas, but UK and US decisions remain independent. The MHRA-FDA Pioneer Programme and liaison roles are coordination mechanisms, not automatic market access. Companies should continue to plan separate UK and US evidence strategies.

The proposed phased authorization pathway for earlier NHS access remains exploratory. It could eventually create a new route for promising AI tools under close supervision, but until a legal basis is established, it should be treated as a policy signal rather than an available pathway.

Business impact

• Manufacturers with UK-marketed AI devices: Update change-control and post-market surveillance systems to capture model updates, retraining, performance drift and threshold changes. Prepare regulator-reviewable documentation and consider whether a predetermined change-control approach may be expected.
• Companies preparing UK entry: Reassess device qualification and classification, and build lifecycle evidence into the regulatory plan rather than treating approval as a one-time event.
• Distributors and UK Responsible Persons: Confirm that contractual arrangements allocate responsibility for change notifications, post-market performance data and incident reporting between the overseas manufacturer and the UK responsible person.
• CROs and regulatory consultancies: Prepare service offerings around AI lifecycle evidence, post-market performance monitoring, real-world evidence and change-management documentation.
• Strategic planning: Use the UK as an early regulatory testbed, but do not assume that UK decisions will be recognised automatically by FDA or other regulators. Maintain separate market-access strategies.
• Key dates to monitor: December 2026 draft AI device change guidance; 2027 consultation on qualification and classification; spring 2027 full implementation roadmap; ongoing AI Airlock cohort 3 milestones.