evidoria

← Back to browse

Good practice Imported

MHRA AI Airlock — the UK's First Regulatory Sandbox for AI Medical Devices, 40 Applicants, 4 Completed, Now Funded to 2029

United Kingdom · London · See the United Kingdom profile · See the London profile

Top 12% 80/100 · Ask Evidence Copilot about this practice

MHRA's AI Airlock pilot (to March 2025) took 5 of 40 applicants, 4 completing; its published report fed guidance, and DHSC committed £3.6m over three years for 2026–29.

Details

Promoter
Medicines and Healthcare products Regulatory Agency (MHRA), with DHSC and NHS England
Period
2024–2029
Keywords
regulatory sandbox, AI medical devices, health regulation

Description

The AI Airlock is the UK's first regulatory sandbox for Artificial Intelligence as a Medical Device (AIaMD), run by the MHRA in partnership with the NHS and the Department of Health and Social Care. The pilot was launched in May 2024 and ran to March 2025.
A public call drew 40 applications (31% generative AI, 23% machine learning); 5 were taken forward and 4 completed the full pilot. Each used a four-stage model (orientation, plan, test, review) across simulation, virtual and real-world testing environments, with regulatory gap analyses as outputs.
The MHRA published a 77-page pilot report in October 2025 with recommendations for AIaMD guidance, covering topics such as synthetic data, explainability and post-market monitoring. A second cohort of seven innovators was announced in October 2025, and in April 2026 the MHRA announced £3.6m over three years (£1.2m a year, 2026–2029).
Caveat: the pilot cohort was tiny and outputs are regulatory recommendations; no patient-outcome impact is yet evidenced, and testing in the real-world environment did not use AI outputs to make patient decisions.

Read the full analysis: https://assets.publishing.service.gov.uk/media/68ee1fb88427701993d5e02c/AI_Airlock_Sandbox_Programme_Report_Final.pdf

Implementation

Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Similar practices you may find useful