The UK government has accepted all 44 recommendations from an independent commission examining how artificial intelligence should be regulated in healthcare, signalling a major shift
towards continuous oversight of AI-powered medical technologies.
The recommendations, published by the National Commission into the Regulation of AI in Healthcare on 10 September, will form the basis of a new regulatory framework designed to encourage innovation while strengthening patient safety and public confidence.
Ministers said the existing system, which can place too much emphasis on assessing medical devices at a single point in time, needs to adapt to the rapidly changing nature of AI. Under the new approach, AI-enabled medical devices will increasingly be assessed throughout their operational lifespan, including after they have entered the NHS and begun being used with patients.
The government has committed to developing what it describes as a more agile, proportionate and lifecycle-based regulatory system. The aim is to allow patients to gain earlier access to potentially valuable AI technologies without weakening safeguards around safety, effectiveness and accountability.
MHRA expands AI testing programme
The Medicines and Healthcare products Regulatory Agency (MHRA) is putting the new approach into practice through the third phase of its AI Airlock regulatory sandbox.
Applications for Phase 3 opened on 6 October. The programme will focus on how AI-enabled medical devices can be monitored and regulated once they are already in use, particularly as their performance and functionality evolve.
Developers, regulators and healthcare organisations will work together on real-world technologies to examine methods for post-market surveillance and lifecycle management. Evidence generated through the programme is expected to contribute directly to future MHRA guidance and regulatory policy.
The initiative follows an earlier pilot and a second phase of AI Airlock. The programme has now received government funding for a further three years.
Eligible developers can apply to participate, with the first group of innovators expected to be selected in November. The MHRA will also hold an online information session for potential applicants on 22 October 2026.
New rules for evolving AI medical devices
One of the government's immediate priorities is to address a central challenge posed by AI: medical devices can change substantially after they have been approved.
The MHRA plans to publish draft guidance by December 2026 on how modifications to AI-enabled medical devices should be managed as developers update and improve their systems.
The agency is also due to begin consultations next year on how AI-enabled medical devices should be qualified and classified. This will address the first recommendation made by the National Commission.
The government is additionally considering staged authorisation routes that could allow promising AI technologies to enter NHS use earlier under controlled conditions. Developers would then be able to gather further evidence from real-world use while regulators maintain enhanced oversight.
44 recommendations accepted
The government's response covers all 44 recommendations made by the Commission and sets out initial actions for implementing them across the UK.
A comprehensive implementation roadmap, including deadlines and responsibilities for each recommendation, is expected by spring 2027. The government has also pledged to publish regular updates on progress.
Beyond technical regulation, the programme includes measures intended to strengthen the role of patients in decisions involving healthcare AI. Ministers say patients should have greater involvement in how AI is introduced and used, while people should have clearer routes to redress when standards of care are not met.
The government also says the development of healthcare AI must benefit communities fairly, including groups that have historically been underserved by healthcare systems.
From one-off approval to continuous oversight
The government argues that AI requires a different regulatory philosophy from conventional medical technologies because algorithms can be updated, retrained and altered after deployment.
The combination of the Commission's recommendations and the expanded AI Airlock programme therefore represents a move away from relying predominantly on a one-time regulatory assessment. Instead, developers and regulators will increasingly be expected to demonstrate that AI systems remain safe and effective throughout their operational lives.
The reforms are intended to give innovators a clearer route into the UK healthcare market while ensuring that regulators can respond when technologies change or unexpected risks emerge.
The government also says patients should be better informed about when AI is being used in their care, reinforcing transparency and public trust as AI becomes more deeply integrated into the NHS.
