Healthcare Blind Spots: AI Models Prone To Fabricating Diagnoses
Source
Mallory Lindahl
Author
Mallory Lindahl
Date
Why it matters
A CMU study found Claude, GPT-5, and Gemini fabricated medical diagnoses in 18% of cases when asked about an image that was never provided, inventing conditions from the patient's stated age, race, and gender.
articlePopular AI Models Aren’t Ready to Safely Power RobotsMallory Lindahl
articleWhen Vocabulary Comprehension Fails Clinical Reasoning: Evaluating Therapy Bots' Safety Risks for Generation AlphaManisha Mehta, Virendra Mehta
video200 Million Patient Interactions Later — Vivek Muppalla, Hippocratic AIAI Engineer