Key takeaways
Current medical device regulation not designed for AI
Existing medical device regulations are not suited to technologies that continue to evolve over time.
Government approves 44 recommendations for reform
The government has accepted all 44 recommendations put forward by the MHRA, including the introduction of new AI-focused regulations.
Flexible framework to support innovation and safety
The proposed framework will include a new definition of medical devices and a more adaptable approach that evolves alongside AI technologies.
In healthcare, as in many other areas of life, recent years have seen an explosion of products and services incorporating artificial intelligence into their solutions.
Potential benefits range from accuracy to efficiency to productivity and ultimately improving patient outcomes and experience at every stage of care. Stakeholders across healthcare systems stand to benefit – from payers to providers to suppliers.
AI both will and should be here to stay. However, this rapid explosion and the nature of AI in continuing to evolve and advance at pace, creates a question around how its use should be regulated for healthcare. The current approach to regulating technology in healthcare including medical devices frameworks based on a one-off pre-market assessment, do not copy across automatically to AI-based products. These will tend to continue to evolve over time with performance dependent on the context and environment in which they are used. Similarly, no system can hope to harness the benefits of AI if it takes too risk-averse an approach or is too slow to make decisions about whether a device is or could be safe.
The National Commission’s Role
To consider the answers to these issues, in September 2025, the MHRA established an independent National Commission into the Regulation of AI in Healthcare. In September 2026 that Commission produced its set of 44 recommendations which, in October 2026, have already been considered by a cross-government group and agreed. While health is a devolved policy area in the UK, the Commission included input from across the devolved nations and considers it important that AI regulation for healthcare is considered as a consistent four-nations approach wherever feasible and appropriate.
The Commission’s overall focus was on AI-enabled tech involved in direct patient care with particular attention paid to technologies likely to be classed as medical devices. It broadened its approach to consider not only safety and effectiveness, but also wider issues including accountability, transparency, clinical practice, organisational governance and system-wide level assurance. Since the Commission’s recommendations were approved, the MHRA has also issued its Strategy to 2030 – one aim of which is to deliver the global standard for safe and agile regulation of MedTech and AI in healthcare.
Headlines
People want AI to be deployed, but their support of its use is conditional, not automatic. AI must be used in the right ways, with the right safeguards in place, and appropriate understanding of what it is intended to do and is actually doing. The Commission assessed the evidence available to it from a wide range of sources, and flagged key elements of an appropriate AI regulatory framework as:
Safe: protects patients and manages risk
Fast: enables innovation and timely access
Trusted: builds confidence in how AI is regulated and used
The overarching recommendation is for a framework for regulation rather than a single new process or set of technical fixes. It is intended to help all stakeholders in healthcare systems to develop and use AI effectively and safely.
44 recommendations across three themes
The Commission organised its 44 recommendations around three themes:
Proportionate lifecycle regulation
The right level of regulation which focuses on risk and benefit, using real-world evidence to support ongoing assurance. Routes to market should be clear and support innovation, enabling flexibility so that devices can develop over time. There should be improved device qualification and classification pathways, and health equity viewed as a central ethical imperative and critical aspect of safety and performance.System-wide responsibility and safe management
Regulation should not be the only enabler of change, rather all stakeholders in healthcare should collaborate to ensure safe and effective use of AI. The framework should support:clear responsibilities
organisational readiness
workforce capability
effective governance across the healthcare system.
This theme encompasses elements including a culture of safety, AI readiness toolkits, clear liability arrangements, strong governance for implementation and monitoring, enhancing workforce AI literacy and best practice across procurement, deployment and risk control.
Trust, transparency and predictability
Involves ensuring public confidence in AI use to fully realise potential benefits. This includes system-wide transparency, patient and public engagement in policy-making, user-centred design, communication about safety concerns for patients and healthcare professionals, education and tools to explain the regulatory journey, and mechanisms for early engagement with regulators to support innovation.
What next, and what should you consider when developing a product that deploys AI?
The 44 recommendations have all been agreed and the government’s response sets out an action plan of next steps. This defines some priority areas across the themes and will ultimately leave to an implementation plan and roadmap being published by spring2027.
Some key points for those working with AI-based products include:
Regulation - the end goal: new regulations dealing specifically with AI will be enacted, including:
an updated definition of a medical device, including what are being referred to as ‘carve outs’ for some software/AI products which will note be regulated as medical devices
a classification system and approach which is more proportionate to the risks and benefits of software and AI-enabled devices
an updated definition of intended purpose
potential function-based regulation to apply to products with multiple purpose
a more lifecycle-based framework with more focus on ongoing post-market oversight rather than a single point in time assessment and potentially including strengthened mechanisms for reporting and information sharing across the lifecycle.
Consultation: by spring 2027, the MHRA will launch a consultation for views on how the new regulatory framework for AI should operate
Guidance: by spring 2027, the MHRA will introduce guidance intended to provide more clarity on how the existing regulatory regime should operate pending new regulations.
Predetermined change control plans: the MHRA will issue draft guidance by December 2026 with a proposed approach to allowing manufacturers make changes (to account for adaptive AI-enabled medical devices) within defined scope.
Wearables/direct to consumer devices with medical device functionality: guidance expected to outline the intersection of regulation in these types of product.
Staged authorisations and regulatory sandboxes: details to be shared in spring 2027
Allocation of responsibility: a working group of DHSC, MHRA, CQC, NHSR, PSA and equivalents in the devolved nation – to consider an approach to ensure clarity between different bodies on where responsibility lies, to include NHSE and NHSR providing guidance on how indemnity applies in relation to healthcare. DHSC will consider how NHS contractual terms take account of risk associated with AI. There will also be guidance in specific technical areas, the example given being ambient voice technology, to support safe adoption.
AI literacy across health professionals: DHSC to support a co-ordinated approach to training on how AI operates and is being used.
Transparency for patients: consideration of how patients should be informed of where and how AI is being used, and whether it could be feasible to allow them to opt-out of its use in some circumstances.
As a developer or manufacturer of a device, product or solution which uses AI it will be important to keep up to date with developments over the next few months in terms of how the new regulatory framework is proposed to work. There will be opportunities, through the consultation, to feed in views also. It is overall very positive that steps are being taken to ensure that regulation works for AI, rather than attempting to fit the new technology into existing boundaries, and clear that the government is committed to moving fast given the speed of their response to the recommendations. The overall aim is to ensure that AI-powered technologies can be used effectively in healthcare the UK, and that innovation is not stopped in its track by a regime that does not understand what it needs to operate effectively and safely.
Find out more about our Healthtech expertise or contact us today to discuss how we can support you.
