UK commission proposes 44 safeguards for AI used in healthcare

Regulation designed for adaptive systems

A UK commission has published 44 recommendations for regulating artificial intelligence used by the National Health Service and other healthcare providers. BBC News reported the recommendations on September 10 after the Medicines and Healthcare products Regulatory Agency released the commission's final blueprint. The central argument is that rules designed for static devices do not adequately govern software that can change as it encounters new data.

The proposals include continuous monitoring after an AI product enters use, the ability to withdraw approval when performance deteriorates, clearer penalties for developers that breach safety requirements and supervised pathways for testing new systems. Patients would also be entitled to know when AI is involved in their care and to obtain understandable information about the product. Those measures seek to preserve clinical accountability while allowing useful tools to reach hospitals and surgeries.

Public support came with conditions

The commission gathered evidence from more than 12,000 patients, clinicians, healthcare leaders, developers and members of the public over a year, according to the government release. Additional deliberative research found that accuracy was the public's leading priority, while human oversight, proportional regulation and protection against unequal outcomes were recurring conditions for acceptance. The recommendations therefore treat trust as an operational safety requirement rather than a communications exercise.

BBC reporting highlighted the practical challenge through AI note-taking tools already used in general practice. Such systems can reduce administrative work, but they may also produce inaccurate records or make patients less willing to disclose sensitive information. The commission's approach leaves clinicians responsible for reviewing outputs and proposes clearer information and opt-out expectations where appropriate. It does not recommend replacing professional judgment with automated decisions.

From advisory report to enforceable rules

The MHRA regulates medicines and medical devices, but the commission itself was independent and non-statutory. Its report is therefore a blueprint, not legislation already in force. Government and regulators must decide which proposals require primary legislation, revised medical-device rules or changes to procurement and clinical governance. That distinction is essential: publication establishes a policy direction, while legal obligations will depend on subsequent decisions.

The next tests will be whether ministers accept the full package, how continuous monitoring will work across changing models, and who bears responsibility when performance drifts after approval. Regulators will also need to coordinate internationally because many healthcare systems use the same developers and models. The immediate development is nevertheless concrete: a year-long official review has delivered a detailed regulatory framework just as AI moves from administrative support toward wider clinical use.