The move has sparked concerns about whether enough research has been done to support AI scribe usage in health settings.
AI scribes are gaining momentum overseas, with the NHS England recently approving 19 new AI notetaking tools in a move aimed at freeing up doctors’ time for patient care.
The decision has raised concerns from the British Medical Association (BMA), which has questioned the pace of the rollout given that many health bodies still have questions on AI notetaking regulations.
The notetaking tools have been listed as ‘ambient voice scribes’ and have been approved to listen to consultations and transcribe in real time, generating clinical notes.
Both the chair of the BMA’s board of science David Strain and BMA GP Committee England policy lead for digital, IT, and data Mark Coley have expressed concerns.
“There are serious concerns about the speed at which this technology is being implemented given the Medicines and Healthcare Products Regulatory Agency (MHRA) is yet to conclude how AI needs to be regulated in healthcare,” they told The BMJ.
“That’s why it’s worrying to see the impulsive rollout of 19 medical devices that have not undergone evaluation, demonstrated accuracy, or evidenced that they can be beneficial to doctors and patients in a real life, busy clinical environment.
Related
“We would expect any other medical device used in the NHS to go through rigorous safety and efficacy assessments before coming to market and AI should not be treated any differently.”
Australia is still in the early stages when it comes to AI notetaking regulations with the TGA flagging back in July last year that it wants to conduct a review on the AI scribes.
Strong words were also directed to AI notetaker developers on current models not meeting basic accountability regulations.
“Developers of digital scribes claim they are not a medical device as their intended purpose is to summarise clinical practice notes,” the TGA said in its latest report on AI scribes.
“Users report digital scribes frequently propose diagnosis or treatment options for patients beyond the stated diagnosis or treatment a clinician has identified during consultations.
“This functionality indicates digital scribes meet the definition of a medical device and require pre-market approval in the Australian Register of Therapeutic Goods and are potentially being supplied in breach of the Act.”
The Australasian Institute of Digital Health (AIDH) has outlined its considerations for how AI scribes could be implemented back in June, focusing on how the tools should be used to “support, not replace, clinical judgment”.
Ensuring accountability of healthcare providers in using the scribes was another focus with timely reviews suggested as a requirement to ensure notetaking accuracy and alignment with relevant codes of conduct.
The NHSE first issued a tender in October 2025 to establish a registry of scribe products for healthcare, with the body then publishing this registry on 16 January.
Suppliers are required to comply with standards on clinical safety, technology and data protection as part of the registry, along with filling out a self-assessment questionnaire to be reviewed by NHS IT leads.
The decision to release the 19 named tools came after the NHSE cited a Great Ormond Street Hospital study that found AI scribes produced a 23.5 % in direct patient time during appointments.
However, the study also reported that time saved from using AI scribes was not being used to offer more appointments but instead reduce administrative burden and overtime working hours.
There are still reported gaps in the tools with the lead author of a 2025 BMJ paper University of Oxford senior digital health researcher Abi Eccles saying that the tools could still produce inaccurate content.
“There is no denying they have huge potential,” Mr Eccles and colleagues wrote.
“But there is a gap in the evidence to be explored.
“Somebody can say ‘I’m fine’ in very different ways, and it is important that clinicians’ interpretations of tacit signs or body language are not lost.
“Currently, we just don’t know and research is needed to look at its performance for those predicted to be most disadvantaged by it.”
