The Medicines and Healthcare products Regulatory Agency published new guidance on 29 July 2026, developed with NHS England, to explain how existing medical device rules apply to ambient voice technology in health and care settings in Great Britain. For NHS bodies and suppliers, the immediate value is not a new approval route but a clearer statement of where the current legal boundary sits. According to the MHRA, the document is intended to remove uncertainty around a growing class of tools often described as AI scribes. These systems are being used to capture and summarise conversations between clinicians and patients, with the stated aim of reducing administrative work and improving clinical documentation.
The central point is straightforward. Where an ambient voice product is intended only to transcribe clinical conversations, summarise them, draft letters or suggest clinical codes for a clinician to check, the MHRA says it is not regulated as a medical device under the current framework. That does not remove governance duties. NHS England has issued separate guidance for England on the safe use of ambient scribing products, and organisations still need to decide what controls, information governance arrangements and working practices are required before deployment.
The position changes where software is intended to support diagnosis, treatment or prevention, or where it takes automated action without clinician review, such as placing orders. In those cases, the MHRA says the product falls within medical device regulation and must meet the relevant safety and performance requirements. For suppliers, that means product description and intended use matter a great deal. A tool presented as an administrative assistant may sit outside device rules, but the position can change once the software is designed or marketed to influence clinical decisions or act on them.
The guidance also makes clear what has not changed. Clinicians remain responsible for reviewing and verifying AI-generated transcripts, summaries and other outputs before those materials are used in patient care. Human sign-off is still the main safeguard on which these systems rest. For boards and executive teams, the message is equally direct. Deployment should be backed by clinical oversight, local governance, staff training and procurement processes, and organisations are expected to identify any change in functionality that could alter a product's regulatory status.
Lawrence Tallon, the MHRA's chief executive, presented the guidance as a step towards safer adoption by drawing a firmer line between lower-risk administrative functions and software that supports diagnosis or treatment. Rob Thompson, NHS England's chief digital, data and technology officer, said the two bodies had worked jointly to give NHS organisations a more consistent basis for assessment and rollout. The wider policy message is that national health bodies want quicker use of lower-risk tools across the service, while keeping formal controls where patient-facing clinical functions are involved. The MHRA also said the document reflects ongoing work by the National Commission into the Regulation of AI in Healthcare on the future framework for AI in health.
In practice, the guidance gives trusts, suppliers and clinicians a more usable test. If the tool helps with documentation and the clinician remains the decision-maker, it is generally outside medical device regulation under the current rules. If the tool is meant to guide diagnosis or treatment, or to act without review, regulation applies. For patients, the most important point is that this is a clarification of existing law rather than a weakening of safeguards. The administrative promise of ambient voice tools may support more clinician time with patients, but responsibility for safe use, accurate records and proper oversight remains firmly with healthcare professionals and NHS organisations.