Large language models provide unsafe answers to patient-posed medical questions

Open in new window