AI Scribes Spread Rapidly Across UK Primary Care

AI scribes are rapidly becoming part of UK general practice, bringing documentation gains alongside unresolved questions over safety, liability and data governance. A 2026 survey published in npj Digital Medicine found substantial use of ambient voice technology among UK general practitioners. These tools listen to consultations, capture doctor and patient speech, and convert spoken interaction into structured medical documentation such as notes and letters. UK implementation has developed in a changing policy environment: NHS bodies encouraged use and issued guidance in April 2025, while NHS England sent a June 2025 communication asking GPs not to use AI scribes unless products met NHS standards for data management and patient safety. Liability remains with individual users or practices, and professional bodies have adopted a more cautious position.

Adoption Patterns Across General Practice

The survey included 598 UK GPs after two responses were excluded because of conflicting answers. Current use of AI scribes reached 40% in unweighted figures, with a weighted estimate of 43%. A further 23% had used AI scribes previously, while 37% had never used them. Among past users, most intended to use an AI scribe again or remained unsure, while only a small minority did not intend future use. Among never users, only 14% said they did not intend to use an AI scribe in future.

Current users applied AI scribes to an average of 60% of consultations, with use ranging from occasional deployment to every consultation. Past users had applied them to an average of 40% of consultations. More than half of current users had an AI scribe integrated into their computer system, compared with just over a quarter of past users.

GPs named 34 different AI scribes. Heidi Health and Accurx Scribe were the most common systems among both current and past users. Many never users were unsure which product they might use, although those who identified a likely option most often named Heidi Health or Accurx Scribe. Colleague recommendation was the most common reason for choosing a system, ahead of practice commissioning, independent choice after hearing about a product, social media advertising, health board or integrated care board commissioning, and vendor approaches.

Use Cases Extend Beyond Documentation

GPs mainly used AI scribes for documentation and consultation support. The technology helped capture patient interactions, record histories, summarise examination findings and convert unstructured dictation into organised clinical notes. Use was especially common for complex consultations, complete record keeping and reducing the time and cognitive load linked to typing or structuring notes.

AI scribes also supported communication with patients and secondary care. GPs used them to draft letters, generate or complete referral documents, send patient messages and reminders, and create clearer or more consistent communication across workflows. These uses reflected a wider administrative role beyond consultation note generation.

A smaller group used AI scribes for higher-level clinical decision-making beyond intended use. These GPs used the tools to interpret symptoms, refine diagnoses, identify possible differential diagnoses, cross-check management plans, select interventions and structure safety-netting. A few also used AI scribes for prescribing support, including repeat prescriptions or prescribing processes more generally.

Must Read:AI Decision Support Faces Data Access Test

Past users had stopped using AI scribes just over six months earlier on average. The most common reasons were lack of practice support and medicolegal or liability issues. Other reasons included loss of product availability, general concerns about use, high costs, limited perceived benefits, official communication asking users to cease use, negative media coverage, lack of integration, workplace change, uncertainty about approval, end of trial periods and personal preference.

Benefits, Risks and Uneven Uptake

Most GPs saw benefits for clinicians, patients, practices and the NHS, although current users endorsed benefits more often than past users or never users. Benefits linked to timeliness, including real-time documentation, were widely supported. Efficiency benefits, such as reducing administrative burden and time spent writing notes, also attracted broad agreement. Many GPs also endorsed safety-related benefits, including reduced feelings of burnout and fewer errors or omissions.

Perceived benefits were not universal. Only 40% of past users and never users agreed that AI scribes would reassure patients. Some additional benefits emerged, including support for GPs with specific needs such as neurodivergence or wrist injuries.

Safety and medicolegal concerns remained prominent. More than 60% agreed there were risks around inaccuracies, errors, misinterpretations and associated medicolegal threats. More than 50% agreed there were risks involving workflow disruption and lack of integration with electronic health records. Other concerns included limited functionality for some patient groups, disruption to shared decision-making and the possibility that administrative time savings could contribute to staff redundancy.

Adoption differed more by GP and practice characteristics than by local population factors. Men, GP trainers, GPs working more clinical sessions and GPs doing regular or occasional private work were more likely to use AI scribes. GPs with more experience also showed higher use in univariate analysis, while trainees and some other roles showed lower use than GP partners. Ethnicity and IT system were not associated with use in univariate analysis, although multivariable analysis found higher use among GPs working with SystmOne TTP compared with EMIS Web. Local authority characteristics showed little association with use.

AI scribes already have a substantial presence in UK primary care, with use shaped by peer recommendation, practice support, workload patterns and medicolegal confidence. GPs see clear potential for efficiency, faster documentation and reduced administrative burden, but unresolved concerns around safety, liability, patient consent, data protection and equitable benefit remain central. Current use also extends beyond documentation into communication and elements of clinical reasoning, increasing the need for clearer boundaries, stronger governance and coordinated implementation within general practice.

Source: npj Digital Medicine