The Medicines and Healthcare products Regulatory Agency has picked Manchester University NHS Foundation Trust to host its newest regulatory sandbox, putting AI-enabled devices in front of real patients and clinicians before wider NHS adoption.

Manchester Sandbox Signals How the MHRA Wants AI Devices Tested Before Scaling

A regulator does not build a sandbox because it trusts the market to sort out the evidence on its own. It builds one because the existing route from prototype to NHS ward is too slow, too opaque, or too easily bogged by a glossy demo that never survives contact with a busy clinic. 

That is the problem the MHRA and Manchester University NHS Foundation Trust are trying to solve with the Manchester Sandbox, launched this week at MFT’s Oxford Road Campus and aimed squarely at AI-enabled medical devices, including tools built to flag patients at risk of complications from long-term conditions before those complications happen.

For a care technology sector that has spent years being told to demonstrate real-world value without ever being given a clear, sanctioned space to generate it, this matters. It also arrives at a moment when the government’s wider digital health agenda, including the NHS Federated Data Platform contract that new health secretary Yvette Cooper inherited, is under sustained scrutiny over exactly this question: how does the NHS know a technology works before it is embedded across the system.

A Sandbox Built on Two Years of AI Airlock Learning

The Manchester Sandbox does not appear out of nowhere. It extends the MHRA’s AI Airlock programme, which launched in 2024 to work through the regulatory puzzles specific to artificial intelligence as a medical device, from pre-determined change control plans to how a tool’s intended use can legitimately expand once it is live. The agency secured £3.6 million in DHSC funding earlier this year to run AI Airlock through to 2029, moving it off the annual funding cycle that had constrained longer studies. Manchester is where that groundwork gets tested against real patients rather than simulated case studies.

The distinction matters more than it might sound. Regulatory sandboxes, in the MHRA’s own framing, allow technologies to be deployed and assessed with genuine oversight, generating evidence on safety, effectiveness and workflow impact that a controlled trial or a developer’s own validation data cannot fully replicate. That is a different exercise from certifying a product works in principle, and it is the gap that has stalled plenty of otherwise promising AI tools on their way into routine care.

What the Manchester Sandbox Will Actually Test

MFT brings scale that few other trusts can match, sitting within one of the largest NHS organisations in the country and a wider Manchester innovation ecosystem spanning universities, research institutes and technology companies. That gives developers access to a large and diverse patient population, which the MHRA and MFT argue produces more reliable evidence than smaller, less representative pilot sites can offer.

James Frith, the Health Innovation Minister, said Manchester was “leading the way in getting the best new technologies into the hands of local patients faster,” while stressing that the process still carries strict oversight. Lawrence Tallon, chief executive of the MHRA, framed the ambition in similar terms, saying the challenge with AI and other emerging technologies is “understanding how they work in practice and where they can make the biggest difference.” Mark Cubbon, chief executive of MFT, put the trade-off more bluntly: patients “should never have to choose between access and safety.”

The Practical Stakes for Care Providers and Commissioners

For a domiciliary provider, an integrated care board, or a technology company building for community and social care rather than acute settings, the immediate relevance of a Manchester-based, hospital-anchored sandbox is not obvious. But the evidence standards it establishes will not stay confined to acute care. If the sandbox model produces a workable template for demonstrating real-world safety and effectiveness, commissioners across social care technology, remote monitoring and electronic care planning will be watching closely for whether a similar structure could apply to their own procurement decisions, where the evidence bar has often been set informally, inconsistently, or not at all.

Participation in the sandbox will not itself guarantee regulatory approval or NHS adoption, a caveat both organisations were careful to state. That distinction protects the programme’s credibility, but it also means providers should not read a sandbox place as an endorsement in itself.

An Uncertain Route From Sandbox to Scale

No participating companies or technologies have yet been named. The MHRA plans to invite expressions of interest from medical device developers, and MFT and the MHRA will hold a roundtable later this year with providers, innovators, researchers and patient representatives to shape the next phase. That timeline leaves plenty of open questions: which technologies get chosen, how findings get shared with trusts outside Manchester, and whether the evidence generated here genuinely shortens the path to wider adoption or simply adds another well-intentioned step to it.

The sector has heard promises of faster, evidence-led adoption before. What will separate the Manchester Sandbox from earlier initiatives is whether its findings actually change procurement behaviour beyond Greater Manchester, and whether the MHRA is willing to publish the evidence, including where technologies fail to clear the bar, rather than only the successes.