Truckee, Calif.-based Tahoe Forest Health System is notifying patients of a data breach after an employee uploaded identifiable patient health information to ChatGPT for data analysis.
Tahoe Forest said it became aware of the incident Sept. 18. The information uploaded to ChatGPT included patients’ last names, first initials, ages, sex, medical record numbers, chief complaints, diagnoses and emergency department acuity levels, according to a Nov. 16 statement from the health system.
Other information included emergency department and hospital discharge dispositions, arrival and discharge dates and times, lengths of stay, initial emergency department providers and, when applicable, the facilities accepting patients for inpatient care.
The health system did not disclose how many patients were affected, when the employee uploaded the information or whether the data has been removed from ChatGPT.
Tahoe Forest said it is notifying affected patients through MyChart and mailed letters.
In response to the incident, the health system has implemented restrictions on AI platforms within its organizational network and is providing employees with additional training on the appropriate and secure use of AI technologies.
Tahoe Forest is also working with ChatGPT representatives to determine whether the disclosed information can be removed from the platform. The health system said it is reviewing its AI use, data protection and information security policies and has taken corrective action in accordance with company policy.
The incident is being reported to the California Department of Public Health and HHS.
The disclosure follows similar incidents involving patient information uploaded to external AI platforms at other healthcare organizations, including Johns Hopkins Bayview Medical Center in Baltimore, Weill Cornell Medicine in New York City and Loma Linda (Calif.) University Health.