Your doctor’s AI notetaker may be making things up, Ontario audit finds
Ars Technica AI · Kyle Orland · 2026-05-14
An Ontario auditor general report finds that all 20 government-approved AI medical scribe vendors produced inaccurate, incomplete, or hallucinated clinical notes in standardized tests, raising patient safety concerns.
Extraction
Topics: ai-healthcarehallucinationai-safetymedical-aigovernment-audit
Claims
- All 20 AI scribe vendors pre-qualified by the Ontario government showed accuracy or completeness failures in at least one of two simulated patient-doctor conversation tests.
- Nine vendors hallucinated patient information, 12 recorded information incorrectly, and 17 missed key details about mental health issues discussed.
- AI scribes fabricated nonexistent referrals for blood tests or therapy and incorrectly transcribed prescription medication names.
- The auditor general warned that AI scribe errors could lead to inadequate or harmful treatment plans with direct patient health consequences.
Key quotes
potentially result in inadequate or harmful treatment plans that may potentially impact patient health outcomes.
All 20 of those vendors showed some issue with accuracy or completeness in at least one of these simple tests, including nine that hallucinated patient information, 12 that recorded information incorrectly, and 17 that missed key details about discussed mental health issues.