The MHRA has secured £4.8 million to expand the AI Airlock programme. For anyone building medical device AI in the UK, this is significant news — and it warrants more than a passing glance.
Here's what the programme is, why the expansion matters, and what it means practically for digital health and medtech companies navigating the UK regulatory landscape.
AI Airlock is the UK's dedicated sandbox for medical device artificial intelligence. It's a joint programme run by the Medicines and Healthcare products Regulatory Agency (MHRA), the Department of Health and Social Care (DHSC), the NHS AI Team, and UK Approved Bodies.
The concept is straightforward: rather than leaving AI medical device developers to navigate regulatory requirements in isolation — often discovering compliance gaps late in the development cycle, at significant cost — AI Airlock creates a structured space where companies can work with regulators to understand and evidence compliance requirements before they go to market.
Participation isn't open to everyone. Companies apply, are selected based on their AI application and risk profile, and then work through a defined regulatory engagement process. The output is a clearer pathway to market authorisation, and — critically — a body of evidence that can support MHRA approval.
Since its inception, AI Airlock has served as a proving ground for some of the most technically complex medical device AI being developed in the UK. Diagnostic imaging tools, clinical decision support systems, predictive monitoring platforms — these are the kinds of applications the programme was designed for.
The funding expansion does two things.
First, it signals continued and growing government commitment to the UK's position as a leading environment for medical device AI development. The AI Airlock programme sits within a broader regulatory strategy — the MHRA's Software and AI as a Medical Device (SaMD) framework, the UK's AI Safety Institute, the government's life sciences strategy — and the additional investment reinforces that this is not a pilot programme. It's infrastructure.
Second, and more practically: more funding means more capacity. More companies will be able to enter the programme. More regulatory pathways will be available. More precedent will be established for how AI medical devices are assessed and approved in the UK.
For the medtech and digital health sector, this is net positive. The bottleneck in UK medical device AI has never been engineering capability — there is genuine innovation happening across the country. The bottleneck has been regulatory clarity. AI Airlock, expanded, begins to address that.
The timing also matters. The EU AI Act's August 2026 deadline for high-risk AI systems is creating significant pressure on developers with cross-market ambitions. The UK's AI Airlock programme offers a parallel pathway that — while distinct from the EU framework — can provide comparable regulatory rigour and market confidence for UK-focused or UK-first strategies.
If you're building AI that could be classified as a medical device, and the scope here is broader than many companies realise, the AI Airlock expansion has direct implications.
The regulatory bar is rising. The MHRA's approach to software and AI as medical devices has become progressively more rigorous. AI Airlock accelerates that trajectory. Companies that wait for regulatory requirements to become fully mandatory before taking compliance seriously will find themselves significantly behind peers who engaged early.
Regulatory engagement is a competitive advantage. Companies that have completed or are progressing through AI Airlock will have demonstrable evidence of regulatory engagement. In NHS procurement, where clinical decision support tools, diagnostic AI, and patient monitoring systems are all subject to procurement scrutiny, that evidence matters. Buyers are increasingly asking hard questions about how AI systems were developed, validated, and governed. AI Airlock provides a structured answer.
Independent assurance evidence is essential. This is where many companies underestimate the challenge. AI Airlock is a collaborative regulatory process, but it is not a light-touch one. The MHRA expects companies to bring evidence. Evidence that their AI performs as intended across relevant patient populations. Evidence that failure modes have been identified and mitigated. Evidence that clinical safety has been systematically assessed. Evidence that the development and validation process was governed by appropriate quality management systems.
Generating that evidence independently, without specialist support, is genuinely difficult. The intersection of AI development, clinical validation, and regulatory compliance requires expertise that sits across multiple disciplines. Companies that try to navigate this without support routinely encounter delays, rework, and in some cases fundamental challenges to their regulatory pathway.
The pathway is clearer, but it isn't automatic. Expansion of AI Airlock does not mean automatic entry, and it does not mean the regulatory questions become simpler. It means more companies will have structured access to regulatory engagement and that the bar for the evidence those companies bring will remain high.
Assurance, independent, expert-led evaluation of AI systems against defined technical and clinical standards, is the foundation of successful AI Airlock engagement.
The MHRA does not expect developers to mark their own homework. The whole premise of regulatory engagement is that independent evaluation provides confidence that a company's claims about its AI are substantiated. For diagnostic AI, that means clinical validation against relevant datasets and patient populations. For clinical decision support, it means assessment of how the AI performs under edge cases, across demographic groups, and in conditions that differ from training data. For monitoring and predictive tools, it means evidence that the system's outputs are clinically meaningful and appropriately governed.
Periculo works with digital health and medtech companies at exactly this point in their regulatory journey — building the assurance evidence that supports MHRA engagement, NHS procurement, and market access. We understand both the technical and clinical dimensions of AI assurance, and we've developed frameworks specifically for the UK healthcare regulatory environment.
The AI Airlock expansion means more companies will need this kind of support. The evidence bar isn't coming down, if anything, as the programme scales and precedent accumulates, it will rise.
If you're building medical device AI and haven't yet mapped your regulatory pathway, now is the time. Specifically:
1. Assess whether your product is in scope. The definition of AI as a medical device under UK regulations is broader than many developers expect. If your system processes clinical data and influences clinical decisions, it may be in scope regardless of how it's marketed.
2. Understand the AI Airlock eligibility criteria. The programme is not suitable for all AI medical device applications, but the expansion means more risk profiles and more development stages are being considered. Review the MHRA's guidance and assess whether your timeline and risk profile align with the programme.
3. Map your evidence gaps. What clinical validation data do you have? What safety testing have you conducted? What quality management processes govern your AI development? These are the questions the MHRA will ask — and being honest about the gaps now is far better than discovering them during regulatory engagement.
4. Get independent assurance support early. The companies that progress most smoothly through AI Airlock are those that arrive with evidence already developed. Building assurance evidence during regulatory engagement is slower, more expensive, and riskier than building it beforehand.
The UK regulatory environment for medical device AI is, at this moment, one of the more progressive and engaged in the world. The MHRA has invested significantly in understanding AI, the AI Airlock programme represents genuine regulatory innovation, and the combination of NHS infrastructure and UK AI talent creates real conditions for world-class digital health development.
The AI Airlock expansion is good news for the sector. But good regulatory infrastructure only translates into market access if companies engage with it properly — with appropriate expertise, appropriate evidence, and appropriate urgency.
The companies that will benefit most from the expanded AI Airlock are those that are preparing now.
Periculo specialises in AI assurance for digital health and medtech. We help companies build the technical and clinical evidence needed for MHRA engagement, NHS procurement, and market access — starting from wherever you are in your development journey.
If you're navigating AI Airlock, preparing for MHRA engagement, or simply trying to understand what regulatory compliance looks like for your AI system, we're happy to have a conversation.
Book a 15-minute call with Harrison →Periculo is a specialist security and AI assurance consultancy supporting digital health and medtech organisations across the UK. We help companies meet regulatory requirements, build robust governance frameworks, and demonstrate the evidence needed for NHS and MHRA engagement.