MHRA sets out new UK medical AI regulation plan

If you hear the phrase AI medical device and immediately think it sounds distant or technical, it helps to bring it back to everyday care. These tools can include software that reads scans, flags risks, supports diagnosis or helps clinicians monitor patients. On Tuesday 20 October 2026, the Medicines and Healthcare products Regulatory Agency, or MHRA, used a speech at the AdvaMed conference to say the UK wants a new rulebook for that fast-moving world. In a government announcement, the regulator said Jessica Paulsen, its Deputy Director for AI and Software, would set out a more flexible approach to software and AI-enabled medical devices. The big message was simple: if medical technology is changing quickly, regulation cannot stay stuck in an older model built for products that rarely change once they reach hospitals and clinics.

That matters because AI tools do not behave like a traditional thermometer or a one-off piece of equipment. Software can be updated, retrained, patched and improved over time. Sometimes that is a strength. Sometimes it creates fresh risk. A tool that performs well in one setting may work less well somewhere else, or change after an update in ways that need checking. **What this means:** the MHRA is talking about lifecycle regulation rather than a single yes-or-no decision at the start. In practice, that means watching a device before launch, at approval and after it is already being used. For you as a reader, the key idea is that safety checks do not end on day one, especially when the product itself can keep changing.

The speech follows the MHRA 2030 Strategy, which the regulator says is meant to modernise UK oversight for the age of AI and personalised medicine. For medical technology, the plan includes an adaptive system for AI as a medical device, a new MHRA Centre for AI and Digital Health, better data on devices and stronger traceability so regulators can tell which tools are being used, where they are being used and how they are performing. The strategy also points to wider use of reliance and recognition routes with trusted overseas regulators, including the US Food and Drug Administration. That can sound dry, but the idea is fairly practical. If a trusted regulator has already reviewed a device, the UK may be able to use some of that work rather than starting from nothing every time. The promise is faster access. The test will be whether speed comes with clear evidence and proper safeguards, not weaker scrutiny dressed up as modernisation.

One of the clearest examples of this new approach is the AI Airlock, the MHRA’s regulatory sandbox for AI-enabled medical devices. A sandbox is not a free pass. It is a supervised test space where companies and regulators can work through problems early, using real evidence to understand how a product behaves before rules are applied more widely. The MHRA says this work is being backed by regional partnerships, including with NHS London and Manchester University NHS Foundation Trust. That matters because regulation is stronger when it is tested in real health settings rather than only in slide decks or lab conditions. If an AI tool is supposed to help doctors or patients, the evidence should come from places where doctors and patients actually are.

The government announcement also ties this work to a wider political move. In October 2026, the UK government accepted all 44 recommendations from the doctor-led National Commission into the Regulation of AI in Healthcare. According to the announcement, that backs a more proportionate and lifecycle-based approach to regulation. Proportionate is one of those policy words that can blur rather than clarify, so it is worth translating. It should mean low-risk tools do not face the exact same process as higher-risk ones, while high-stakes products still face tough checks. If an AI tool could affect diagnosis, treatment or safety, the burden of proof should stay high. That is the balance ministers and regulators now say they want to strike.

The international side of the story is just as important. The MHRA says its new Pioneer Programme with the FDA, announced in June, will put dedicated liaison roles inside both regulators to improve day-to-day cooperation, scientific exchange and work on new regulatory problems, including AI and medical devices. Both sides will still make their own decisions, which is an important detail. Cooperation is not the same as handing over control. **Why the FDA matters:** many medical technology firms want to work across both the UK and US, and separate rulebooks can slow that down. The MHRA says it wants fewer transatlantic barriers and more use of trusted international pathways, including for devices already approved by the FDA. It is also recruiting a Regional Deputy Director of US Partnerships in Washington, DC, to act as its dedicated FDA liaison. In plain terms, the agency wants a working relationship, not just a formal one.

Alongside the AI push, the MHRA says wider medical device reform is still moving ahead, including better post-market surveillance and proposed legal changes for a more responsive system. Post-market surveillance simply means keeping watch after a device is already in use: checking performance, spotting problems early and acting if safety concerns appear. For patients, clinicians and developers, the picture is becoming clearer even if the final rulebook is still being written. The MHRA wants the UK to be seen as a place where new medical technology can reach people sooner, but with stronger follow-up once it is in the real world. If that sounds like a careful compromise, that is because it is. The real question now is not whether AI belongs in healthcare, but whether regulators can keep it useful, fair and safe as it changes in front of them.