According to the government response published on GOV.UK on 6 October 2026, ministers have accepted all 44 recommendations made by the National Commission into the Regulation of AI in Healthcare, a commission led by NHS doctors whose report was published on 10 September 2026. The response is presented as a UK-wide plan for changing how AI-enabled medical devices are regulated, with the stated aim of speeding access to safe tools without weakening patient protection. Published alongside that response, the Medicines and Healthcare products Regulatory Agency has opened applications for AI Airlock Phase 3, its regulatory sandbox for AI-enabled medical devices. The MHRA says the next phase will test the same themes emphasised by the Commission, particularly post-market surveillance and lifecycle oversight.
The central policy shift is away from treating approval as a single event. The Commission concluded that AI-enabled medical devices should be assessed and monitored throughout their working life, and the government response accepts that model in full. In practical terms, that means closer attention to how products perform once they are being used in clinics, hospitals and community settings, not only when they first enter the market. This matters most for AI tools that can change over time. The published response says future regulation should be more proportionate and lifecycle-based, so that safety controls continue after deployment while effective products can reach patients sooner.
The earliest formal steps are now set out with clear dates. The MHRA says draft guidance will be issued by December 2026 on a new approach to managing changes to AI-enabled medical devices as those systems improve or are updated over time. In 2027, the regulator also plans to consult on how AI-enabled devices should be qualified and classified. Those two commitments are the first concrete rule-making actions in the response. For suppliers, commissioners and NHS adopters, they are the areas most likely to shape how AI products are regulated in day-to-day practice.
The response also backs the idea of staged authorisation pathways. According to the government document, that would allow promising AI tools to be used in the NHS earlier, under close supervision, while additional real-world evidence is collected. The policy aim is to avoid an all-or-nothing choice between full approval and no access at all. For patients, the safeguard is meant to be continued oversight rather than a lighter standard. For developers, it offers a route into supervised use, but with a clear expectation that evidence gathering continues after deployment.
AI Airlock Phase 3 is the first operational test of that approach. The MHRA says Phase 3 will focus specifically on post-market surveillance and lifecycle regulation, using real products and real regulatory questions to generate evidence for future guidance and policy. The agency has also secured three years of additional government funding for the programme. The application window opened on 6 October 2026. The MHRA has scheduled a webinar for prospective applicants on 22 October 2026 at 10:00, and says the first wave of innovators will be selected in November 2026.
Delivery is being placed under a formal governance structure. The published response says a newly established Programme Board will convene with work already under way across government, the MHRA, NHS partners and the devolved authorities on ten priority areas. A full implementation roadmap, with timelines and named responsibilities across all 44 recommendations, is due by Spring 2027. That sequencing is important because acceptance of every recommendation does not by itself change regulatory practice. The later roadmap, consultations and MHRA guidance will show how quickly the new model moves from policy statement to operational rules.
Patient safeguards are a separate strand of the response, not an afterthought. The government says it will work with health system partners to strengthen patient involvement in decision-making, improve access to redress when standards of care fall below expectations, and make sure patients are better informed when AI is used in their care. The response also says the gains from AI should be shared fairly across different communities, including groups that are often underserved by health services. That places transparency, accountability and fair access alongside speed and innovation as formal policy aims.
The institutional base for the reforms is broad. The Commission was established by the MHRA in September 2025 as an independent, non-statutory advisory body, linked in the notes to editors to commitments in the 10-Year Health Plan for England, the Life Sciences Sector Plan and the AI Opportunities Action Plan. Its recommendations followed a year-long evidence-gathering exercise involving more than 12,000 patients, clinicians and members of the public. Supportive statements published with the response came from the Department of Health and Social Care, the MHRA, the Regulatory Innovation Office, the Office for Life Sciences, ABHI and patient safety representatives. Taken together, those statements show that the programme is being framed as both a patient safety reform and a growth policy for health technology.
The UK-wide picture is slightly more complex because health policy is devolved. Welsh and Northern Irish ministers said on 6 October 2026 that they support the direction of reform and will continue working with the MHRA and health partners. The Scottish Government said it shared the Commission's ambition but would set out Scotland's own response in due course, while also publishing a separate AI policy framework for health and social care later in 2026. For NHS bodies, suppliers and patient groups, the immediate message is clear: the regulatory system is moving from a one-off approval mindset to ongoing assurance throughout a product's use. The next dates to watch are 22 October 2026 for the MHRA applicant webinar, November 2026 for the first Airlock selections, December 2026 for draft change-management guidance, and Spring 2027 for the full implementation roadmap.