UK regulators warn of AI medical transcription errors endangering patient safety

British regulators have raised concerns over AI transcription tools used in NHS clinics, warning they may misidentify medical terms and endanger patient safety. A patient, also a medical professional, was alarmed when an AI-generated summary incorrectly noted she had 'demyelination,' a condition linked to multiple sclerosis. The error was later corrected, but the incident highlighted the risks of AI misinterpretation. Healthwatch England reported similar cases, including instances where AI confused medications and omitted critical information about treatment. The organization emphasized the need for clear procedures to report and correct AI errors, as inaccurate data could persist in medical records. The NHS aims to use AI to reduce administrative burdens, but experts warn that current tools require careful oversight and human verification to prevent errors. Rachel Power, CEO of the British Patients Association, stressed the importance of trust and transparency in AI systems. Meanwhile, the MHRA has not classified AI transcription tools as medical devices, raising concerns about a lack of regulatory oversight. Doctors like Shier Ziser Dawood have warned of 'AI hallucinations,' where tools fabricate information not present in consultations. Studies suggest AI errors are more likely in complex cases or non-English-speaking patients, yet many doctors still trust AI-generated records more than their own notes.