MHRA secures £3.6m to expand AI regulatory sandbox

The Medicines and Healthcare products Regulatory Agency (MHRA) has secured a funding uplift to expand its AI Airlock programme, a regulatory sandbox for AI as a medical device.

Following a successful second phase, the Department of Health and Social Care (DHSC) has allocated the programme £1.2 million per year for the next three years.

The funding will enable the programme to continue beyond the constraints of yearly budgets with longer-term testing models, with the aim of creating a more sustainable regulatory pathway for future AI medical technologies.

James Pound, executive director of innovation and compliance at the MHRA, said: “Securing this multi‑year funding boost marks a pivotal moment for AI Airlock and for the safe and responsible advancement of AI in healthcare.

“The programme has already shown how collaborative, real‑world testing can uncover regulatory challenges early and help innovators bring high‑quality, safe technologies to patients faster.

“This additional investment will allow us to scale up and ultimately strengthen our ability to ensure that AI‑powered medical devices can reach patients safely, efficiently, and with the confidence of robust regulatory oversight.”

AI Airlock is led by the MHRA in partnership with DHSC, NHS AI Team, and Team AB (the consortium of UK Approved Bodies).

The programme is a key enabler of wider AI regulatory reform and aligns with government priorities across the AI Opportunities Action Plan, the Regulatory Action Plan, and long-term strategic programmes including the 10 year health plan and the Life Sciences Sector Plan.

Dr Dom Pimenta, chief executive and co-founder of TORTUS, an AI medical scribe backed by the NHS Innovation Accelerator, said: “At TORTUS, we’ve found the cross-educational aspect of the programme particularly valuable – both in deepening our understanding of regulatory objectives and in sharing industry expertise at a time when AI is advancing at an unprecedented pace.

“The extension of this initiative is fantastic news, as it has the potential to set a global benchmark for safe, effective, and rapid deployment of clinical AI solutions.”

AI Airlock went through an early pilot phase in 2024, with a second round of projects opening in 2025.

The pilot highlighted several areas where AI medical devices raise regulatory challenges and showed that risk management must consider issues specific to AI, such as reducing errors by using techniques that ground model responses in verified clinical information.

It also found that improving how AI systems explain their recommendations is crucial for supporting clinician confidence.

The pilot underlined the need for ongoing monitoring once products reach the market, to identify changes in performance or over-reliance by users.

Phase two of the programme includes specific regulatory challenges for AI-powered diagnostic tools, pre-determined change control plans, and how AI devices may expand in scope or intended use.

The phase has explored technologies, including large language models, voice tools, and specialised diagnostics for cancer and rare diseases. Reporting for phase two is expected to be published in Summer 2026.

Insights from the AI Airlock are contributing to the National AI Commission’s work on the future regulation of AI in healthcare.