Hospitals and regulators worldwide are grappling with the same challenge: how to safely implement operational AI. This is evident from two key developments this month, one in our local area and another across the Atlantic.
East Kent Hospitals University NHS Trust is seeking an ambient voice technology (AVT) or scribe solution from suppliers to reduce the time clinicians spend on documentation, improve note accuracy, and enhance medical coding. Published in December 2025, the trust's pre-procurement notice details five key objectives, from boosting clinical efficiency to improving coding accuracy. It invites vendors to submit details on their proposed delivery models and safety features, a move that supports a wider NHS initiative to establish supplier registries and standards for compliant AVT tools.
In December 2025, the U.S. Department of Health and Human Services (HHS) released an AI Strategy and a formal Compliance Plan, positioning artificial intelligence as a core element for enhancing internal operations, research, and public health. Five essential pillars build the Strategy: governance and risk management, infrastructure and platform design, workforce development, reproducible research, and care and public-health delivery modernisation. The accompanying Compliance Plan acts as an operational manual, translating guidance from the Office of Management and Budget (OMB) into practical steps for agencies, providing concrete instructions for establishing necessary oversight mechanisms, including audit trails and risk registers.
Ambient voice technology (AVT) offers more than convenience; it represents operational AI integrated directly into clinical workflows and patient records. This integration offers significant benefits, as pilot studies and market analyses indicate AVT can save clinicians substantial time (potentially thousands of hours per trust) and contribute to productivity goals. However, it also introduces concentrated risks. If we fail to ensure robust procurement, authentication, information governance, and cyber resilience, AVT will create vulnerabilities such as data-privacy issues, mis-transcriptions, biased summaries, and new attack surfaces.
The HHS approach, despite relying on a single legal framework, offers a valuable model. Its focus on governance, workforce readiness, and reproducible evaluation reflects the core elements emerging from recent NHS pilots. These shared priorities include: requiring staged, clinically-led validation; mandating vendors to supply auditable models and explainability; integrating cyber and disaster-recovery clauses into contracts; and assessing outcomes based on metrics like clinician time saved, documentation quality, and coding accuracy. Both regulatory bodies emphasise the necessity of accompanying AI adoption with staff training and transparent communication with patients regarding AI use.
Interest from East Kent and compliance efforts from HHS indicate that ambient scribing is transitioning from experimental trials to widespread implementation. Regulators worldwide recognise the need to govern AI, matching the genuine potential of these tools with the significant effort required to deploy them safely, reliably, and effectively across the healthcare system.
