Healthcare AI rules are now open for public evidence, giving Britain the chance to explain to its future medical computers that a toaster is not a complicated patient and a kettle is not displaying symptoms.
The Medicines and Healthcare products Regulatory Agency launched a Call for Evidence on 18 December, inviting patients, clinicians, healthcare providers, technology companies and the general public to help shape the regulation of artificial intelligence in healthcare. A new National Commission will advise on the rules, bringing together clinicians, regulators, patient advocates and technology specialists.
This is the right cast. Before a machine joins a clinic, somebody should check whether it can distinguish a medical pattern from an appliance making an alarming noise. The consultation runs until 2 February 2026, giving everyone time to explain that turning it off and on again is not always an acceptable treatment plan.
At the Centre for Calm Machines, standards officer Gladys Algorithm welcomed the process. “We want safe, effective systems,” she said, in an entirely fictional contribution to this article. “We also want technology that admits uncertainty without booking a webinar about leadership.”
The evidence cited by the MHRA suggests that people are more comfortable with AI handling administration than patient care. The Health Foundation found that 54% of the public surveyed and 76% of NHS staff surveyed supported AI in patient care. Support was higher for administrative work, reaching 61% among the public and 81% among staff.
That distinction is not difficult to understand. Nobody wants a machine making a confident error about their treatment. Many people would happily let one stop them typing the same information into six different forms while a printer develops a personal grudge and the system congratulates itself for reducing friction.
Commission clerk Rory Cache said the consultation would gather views from every direction. “Then we will put them in a secure folder,” he said, also fictitiously. “The folder will ask for a second factor and decline to recognise its own password.”
The serious task is deciding who remains responsible when software behaves oddly, particularly as systems become more advanced and are used in new ways. The comic task is watching every institution promise human oversight while quietly giving a laptop a clipboard and calling it a multidisciplinary team.
Healthcare AI may improve support for patients and staff. It may also create a new professional role for somebody who stands beside the diagnostic screen and says, “That is a toaster, Kevin.” The role will require three qualifications, a safeguarding review and no authority to unplug anything.
Source: Medicines and Healthcare products Regulatory Agency.
