Doctors’ AI scribes get names of drugs and diagnoses wrong, NHS watchdog warns

by | Sep 14, 2026 | Health

Doctors’ AI scribes get names of drugs and diagnoses wrong, NHS watchdog warns

The NHS patient advocacy group Healthwatch England has raised concerns about artificial intelligence tools that transcribe doctor-patient conversations, citing multiple instances where the technology introduced medical errors into patient records.

In documented cases, AI scribes incorrectly transcribed diagnoses, including one instance where a patient was told she had demyelination when the correct notation should have been “null demyelination.” The technology also confused medication names with similar alternatives and omitted critical clinical instructions, such as guidance to obtain repeat prescriptions. Healthwatch noted that patients rather than healthcare professionals often detected these errors, raising questions about what mistakes might persist unnoticed in medical records.

The rapid deployment of these tools across the NHS is intended to reduce administrative burden on clinicians and support the service’s transition to digital operations. The government’s 10-year health plan emphasizes that AI scribes will “liberate staff from their current burden of bureaucracy.” Currently, 27 different AI scribe systems are in use by GPs and hospital doctors in England.

Healthwatch and patient advocacy organizations have called for clearer mechanisms to report and correct AI-generated errors. The Medicines and Healthcare products Regulatory Agency has determined that AI scribes do not require classification as medical devices, meaning there is no England-wide oversight system to ensure their safety and effectiveness. This regulatory gap has prompted warnings that healthcare providers could face legal liability for mistakes attributable to the technology.

Researchers studying AI implementation in clinical settings note that accuracy challenges may be more pronounced in complex scenarios, including consultations involving multiple participants, patients with complicated medical histories, and non-native English speakers. Despite these concerns, some evidence suggests healthcare workers believe AI-generated records can be more accurate than handwritten notes.

Article Attribution | Read More at Article Source

Article summary produced by Claude AI