The Medicines and Healthcare Products Regulatory Agency (MHRA), working with NHS England, has set out how existing medical device law applies to ambient voice technology (AVT) used across health and care settings in Great Britain. The guidance, published on 29 July 2026, aims to draw a clearer regulatory line for an area where adoption has surged and legal questions have multiplied.
What the guidance changes — and what it does not
Ambient voice systems are increasingly deployed to record and summarise clinician–patient conversations, reduce administrative burden and support clinical documentation. The MHRA says those AVT products whose sole function is transcription, summarisation, drafting or suggesting codes for clinician review are not regulated as medical devices under the current framework. By contrast, products that support diagnosis, treatment or prevention, or that take automated actions without clinician verification, remain within the scope of medical device regulation and must meet relevant safety and performance requirements.
"AVT products intended solely for transcription, summarising clinical conversations, drafting letters, or suggesting clinical codes for a clinician to review are not regulated as medical devices under the current framework."
The guidance does not change the law itself; instead, it explains how the MHRA will apply existing regulation to a fast‑growing class of technology. It also aligns with the direction of travel signposted by the National Commission into the Regulation of AI in Healthcare, which is working alongside the MHRA and NHS England.
How responsibility and governance are allocated
Under the guidance, clinicians retain responsibility for reviewing and verifying AI‑generated transcripts, summaries and other outputs before those outputs are used in patient care. NHS boards and executive teams are asked to satisfy themselves that AVT deployments are supported by appropriate clinical oversight, local governance, staff training and procurement processes.
The MHRA highlights that any change to a product's functionality that might alter its regulatory status must be identified and reviewed. That means a supplier adding automated decision support or direct action capability to a system previously used only for transcription could shift the product into the regulated category.
- Not regulated as devices: AVT used solely for transcription, summarisation, drafting letters or suggesting clinical codes for clinician review.
- Regulated as devices: AVT intended to support diagnosis, treatment or prevention, or that performs automated actions without clinician review.
- Ongoing duties: Clinicians and NHS organisations remain responsible for verification, governance and monitoring of functional changes.
The distinction is consequential. If an AVT is a medical device, suppliers and NHS trust procurement teams must ensure compliance with safety and performance standards, technical documentation and, where required, conformity assessments. For tools outside that threshold, the regulatory bar is lower but the MHRA emphasises that clinical oversight and local governance are still vital.
Implications for suppliers, trusts and patients
For suppliers, the guidance offers much‑needed clarity on when to invest in the formal regulatory steps required for medical devices. For NHS organisations, it reinforces the need for robust local processes around deployment: procurement checks, staff training and mechanisms to detect changes in functionality that might change regulatory status.
For patients, the guidance underlines a continuing principle: human clinicians must check AI outputs before those outputs influence care. That duty is explicitly retained by the MHRA and NHS England in the new guidance.
Although the document does not amend statutory law, it is likely to shape procurement decisions and development roadmaps for vendors targeting the NHS market. The MHRA’s interpretation will also inform ongoing policy work by the National Commission into AI regulation in healthcare, signalling how regulators and health policymakers expect clinicians and organisations to manage the risks and benefits of voice AI.
