The MHRA has published two reports on how artificial intelligence should be used in healthcare, after gathering views from patients, the public, healthcare professionals, industry and system partners. The reports are intended to inform future rules. The empty chair at the consultation has welcomed this development, provided nobody asks it to approve a chatbot before lunch.
Published on 11 June 2026, the reports bring together research and evidence gathered for the National Commission into the Regulation of AI in Healthcare. The MHRA also published findings from a Call for Evidence that attracted contributions from 760 people and institutions. This is the reassuringly substantial part of the announcement, where public participation is counted, recorded and placed somewhere safe, preferably in a folder that has been granted permission to open.
The broad message is that AI may offer real benefits in healthcare, but only if safety, effectiveness and accountability travel with it. People want useful innovation to move at a useful pace. They also want to know who is responsible when a system produces an answer that sounds confident, arrives instantly and is spectacularly unhelpful.
The reports identify demands that sound obvious until an institution attempts to make them happen. AI systems should be monitored after they are introduced. Their performance should be checked. Patients and professionals should understand how they are being used. Human oversight should remain more than a decorative phrase in a policy document, like a fire exit sign leading confidently into a cupboard.
This creates a modest difficulty for the technological future. It is keen to promise speed, convenience and improved safety, while regulation keeps asking slower questions. Does the system work? Who can challenge it? What happens when reality refuses to match the demonstration? Somewhere nearby, an administrator is being asked to turn those questions into a framework, then into guidance, then into a document explaining the guidance.
The MHRA says the evidence will help shape recommendations due from the AI Commission later this summer. Those recommendations are expected to support patients, encourage useful innovation and ensure new technologies deliver benefits in real healthcare settings. In other words, the machines may be allowed into the building, but they will still have to explain themselves at reception.
The empty chair has therefore endorsed a proportionate and adaptable rulebook. It has also requested that every future AI system arrive with monitoring arrangements, human oversight and one plainly written answer to the oldest question in public administration: who is going to sort this out when the screen starts smoking? The chair is available for the meeting, although it has sensibly declined to take minutes.
Source: GOV.UK.
