TL;DR
The Medicines and Healthcare products Regulatory Agency (MHRA) has been awarded £3.6 million over three years by the Department of Health and Social Care to scale AI Airlock, the UK’s first regulatory sandbox for Artificial Intelligence as a Medical Device (AIaMD). The £1.2 million-a-year settlement covers 2026 to 2029 and moves the programme off yearly financial cycles for the first time since it launched in spring 2024.
What Airlock does
AI Airlock brings together the MHRA, DHSC, the NHS AI Team and Team AB (the UK’s Approved Body consortium) to test AI medical devices in controlled real-world conditions before full market authorisation. The pilot phase identified regulatory blind spots specific to AI: error-reduction techniques that ground model output in verified clinical information, explainability for clinician confidence, and post-market monitoring for drift and over-reliance.
Phase two, now underway, is examining diagnostic AI, pre-determined change control plans (PCCPs) — the mechanism allowing approved AI to update without re-certification — and scope-creep scenarios where a device expands its intended use. Technologies tested include large language models, voice tools and specialised diagnostics for cancer and rare diseases. Reporting on phase two is expected in summer 2026.
Why this matters
Multi-year funding sounds dry, but it is the single most consequential change the programme could have received. Sandbox programmes that run hand-to-mouth on yearly budgets cannot credibly invite long-running trials, and cannot attract the larger international AI medical device manufacturers that the Life Sciences Sector Plan is trying to draw to the UK. It also provides a concrete data point to pair with today’s Oxford heart-failure AI announcement — regulatory capacity and clinical evidence are building in parallel, which is the sequence UK medical AI has historically struggled to get right.
Dom Pimenta of TORTUS AI, one of the participating firms, said Airlock had the potential “to set a global benchmark for safe, effective and rapid deployment of clinical AI solutions”.
Looking forward
Phase three design and a new call for applications are due later this year. For UK medical AI developers, the practical signal is simple: the regulator is open for business and now has the budget to prove it.