Johns Hopkins Bayview: patient data uploaded to an unapproved AI tool

Johns Hopkins Bayview Medical Center in Baltimore is writing to about 3,182 patients after a staff member uploaded a spreadsheet of patient information to a public AI tool the hospital had not approved. Nobody hacked anything here. Health data ended up with an outside AI company because one person was trying to get a task done faster, and that is a mistake many of us could make with our own documents.¹

Reporting checked October 28, 2026.

Too long; didn’t read?

• The spreadsheet covered patients admitted and treated between May and September 2026, with names, admission and treatment dates, medical record and account numbers, departments, medications and provider names.^{1, 2}
• No Social Security numbers or financial account details were included, the hospital says.¹
• Letters started going out on October 22 and offer 12 months of free credit monitoring.¹
• For everyone: think twice before pasting medical letters or other people’s data into an AI chatbot.

What happened

According to the hospital, on September 21, 2026, during an internal review, a staff member uploaded a spreadsheet of patient information to an outside AI platform that was not cleared for that use. The hospital found out the same day.¹

Johns Hopkins Bayview says the employee deleted the prompt, the conversation and the document from the AI tool’s history, and the hospital asked the company that runs the tool to delete the data. The hospital has not named the AI tool. It says it is tightening its policies and training, including limits on unapproved apps and AI tools and retraining staff on which tools are approved.^{1, 2}

The hospital began mailing letters to affected patients on October 22. They explain how to sign up for 12 months of free credit monitoring through Experian IdentityWorks.¹

What happens to what you type into a chatbot

Asking for deletion is the right step, but from the outside it’s hard to know whether a copy still exists somewhere. The UK’s National Cyber Security Centre explains that what you type into a public AI chatbot is visible to the company running it and is stored, and that the provider “or its partners/contractors” may be able to read it.³ Canada’s privacy commissioner adds that information you enter may be collected and stored, may be used to train the AI system further, and that platforms don’t always explain clearly where it goes.⁴

Many AI services let you turn off training on your chats or use a temporary chat mode, and workplace versions often come with stricter data rules. That is why hospitals and companies tell staff to use only approved tools: the same question can be safe in one tool and a data breach in another.

What affected patients can do

• Read the letter and decide on the credit monitoring. It is free and the letter explains how to enroll.¹ Since no Social Security or bank numbers were involved, the bigger risk is misuse of your medical and account details.
• Check your bills and insurance statements for treatments or prescriptions you don’t recognise. If something looks wrong, contact your health care provider first to see if it is a mistake.⁵
• Be wary of calls or emails that quote your hospital stay. A caller who knows your admission date or medication isn’t proof they’re from the hospital. Hang up and call back on the number on your letter or the hospital’s website.
• Ask questions. The hospital has set up a toll-free call center for affected patients; the number is in your letter.^{1, 2}

How to use AI chatbots without oversharing

• Leave out names and numbers. Canada’s privacy commissioner advises sharing as little personal information as possible, avoiding sensitive or identifiable details, and changing names and other identifiers.⁴ “Explain these blood test terms” works just as well without your name, date of birth or record number.
• Never paste other people’s data, whether it’s a family member’s medical letter or a spreadsheet from work, unless you have permission and an approved tool.
• Check the privacy settings of the AI service you use, and delete chat history you don’t need, if the tool allows it.⁴
• At work, use only the tools your employer has approved.

Sources

1. WMAR-2 News, October 2026: Johns Hopkins alerts 3,000+ patients after data sent to AI tool
2. CBS Baltimore, October 26, 2026: Johns Hopkins Bayview warns patients of data breach involving more than 3,100 people
3. National Cyber Security Centre (UK): ChatGPT and large language models: what’s the risk?
4. Office of the Privacy Commissioner of Canada: Your privacy and AI chatbots
5. HHS Office of Inspector General: Medical identity theft