Risks of AI medical transcription tools

The integration of artificial intelligence (AI) medical transcription tools into clinical settings has raised significant safety, clinical, and legal concerns.

Initial warnings from Healthwatch England highlighted specific patient safety risks within the NHS, where AI systems incorrectly recorded diagnoses and medication names. These errors, such as misidentifying medical conditions or confusing similar-sounding drugs, were often caught by patients rather than clinicians. Regulatory concerns were also noted regarding the MHRA’s decision not to classify these tools as medical devices, potentially limiting nationwide supervision.

Subsequent analysis expanded on these risks, noting that AI scribes may omit vital non-verbal information like facial expressions and emotional states, which are essential for accurate diagnosis. There are also concerns regarding patient privacy and the potential for patients to withhold sensitive information when they know they are being recorded. Legally, because medical charts serve as official documentation, AI ‘hallucinations’ or omissions create risks for malpractice claims and audits. Experts emphasize that AI outputs must be treated only as drafts requiring mandatory clinician review to mitigate liability.