Johns Hopkins Bayview Employee Uploaded Patient Data to Unapproved AI Tool

# Johns Hopkins Bayview Employee Uploaded Patient Data to Unapproved AI Tool

https://gcdradar.vercel.app/incidents/johns-hopkins-bayview-employee-uploaded-patient-data-to-unap · AIR-2026-1783 · closed 2026-10-05

- Lane: Breaches & attacks
- Severity: 4 (Confirmed real-world loss): Patient data was exposed to an external AI service, constituting a real‑world PHI breach.
- Fix status: No fix needed
- First reported: 2026-09-21
- AI role: No AI role
- Products: public AI tool

## Summary

A Johns Hopkins Bayview Hospital employee uploaded a spreadsheet containing patient information to an unapproved public AI service, prompting notification to over three thousand patients. The hospital is seeking deletion of the data and reinforcing staff training.

## What to do

Notify affected patients, assess the AI service for data retention, and enforce strict data‑handling policies.

## Latest updates

- 2026-09-21: A Johns Hopkins Bayview Hospital Employee in Baltimore Reportedly Uploaded Patient Data to an Unapproved AI Tool (https://incidentdatabase.ai/cite/1792)

## Sources

- 2026-09-21 AI Incident Database: A Johns Hopkins Bayview Hospital Employee in Baltimore Reportedly Uploaded Patient Data to an Unapproved AI Tool https://incidentdatabase.ai/cite/1792

Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.