Expert: Healthwatch England, Statutory NHS patient watchdog Healthwatch England, the statutory NHS patient champion, warns that ambient AI scribes "can put [patients] at risk by getting the names of drugs and illnesses wrong" - and that "these inaccuracies may persist in their records if the patient doesn't catch them". In each documented case it was the patient, not the clinician, who spotted the error. What the AI got wrong: 1. A dropped negation reversed a diagnosis. The AI scribe's summary of a consultation recorded the patient's MRI result as "demyelination" - the nerve damage that can lead to conditions such as multiple sclerosis - when the scan had actually found "null demyelination". The patient, herself an NHS health professional, had to query the record before the hospital corrected it: "This was eventually corrected but was a very traumatising experience to be given an incorrect diagnosis because of AI and then be told it's a typo." 2. A drug was substituted. An AI scribe confused the drug the GP had prescribed with a different medicine of a similar name; again it was the patient who identified the blunder. 3. An instruction disappeared. An AI-generated summary letter did not say that the hospital consultant had told the patient to seek a repeat prescription for their migraine from their GP - an omission that could have left them unable to get their medication. 4. The scribe invented content. Dr Shier Ziser Dawood, a London GP, recounted in the British Journal of General Practice that an AI scribe recorded her as telling a patient to "continue their Prozac" even though she had not prescribed or discussed that drug - a classic hallucination, in which the scribe referred to something that was never raised during the consultation. Why this matters: GPs and hospital doctors in England already use 27 different AI scribes, and the government's 10-year health plan for NHS England expects the technology to "liberate staff from their current burden of bureaucracy and administration". Yet the MHRA has decided not to classify AI scribes as medical devices, so there is no England-wide oversight to ensure they are safe and effective. Healthwatch wants patients told when a scribe is in use, clarity on how mistakes are reported and corrected, and consistent national standards for clinicians checking AI summaries. Dr Charlotte Blease, an expert in AI use in healthcare at Uppsala University, notes that GPs using ambient voice technology believe errors are more likely in consultations with more than one person, with patients with a complex medical history, and with patients whose first language is not English. The correction: an AI transcript is only as safe as the check that follows it. Every AI summary must be read and corrected by the clinician against what was actually said before it becomes part of the medical record - and patients need a clear route to challenge errors, because in these cases the AI's mistakes reached the record and were caught by the patient, not the doctor. Source: https://www.theguardian.com/society/2026/aug/31/doctors-ai-scribes-get-names-of-drugs-and-diagnoses-wrong-nhs-watchdog-warns