Weill Cornell 放射研究把患者数据输入未授权 AI 工具,涉事研究已暂停并要求删除数据
Improper AI use leads to another data incident, more patient notifications
Weill Cornell Medicine 通知患者,其放射研究项目在 6 月将 NewYork-Presbyterian 的患者受保护健康信息输入 AI 工具并用机器学习处理,未经授权且违反机构政策。
Improper AI use leads to another data incident, more patient notifications
Weill Cornell Medicine says protected health information at NewYork-Presbyterian "was entered into artificial intelligence tools and processed using machine learning," without authorization and in violation of its policies.
By Mike Miliard,Executive Editor| November 10, 2026 | 11:00 AM

Photo: skynesher/Getty Images
A Weill Cornell Medicine radiology research project involving protected health information fed that patient data into an AI tool at NewYork-Presbyterian two months ago, and now patients are being informed that some of their PHI may have been breached.
WHY IT MATTERS
In anotice to patients this past week, Weill Cornell leaders said the privacy incident involved patient data and an AI model that was used in an "unauthorized capacity."
The New York City health system said the data used "was limited and there is no indication that any patient information has been misused."
WCM discovered the incident on Sept. 6 and immediately initiated an investigation.
It found that, during a radiology research study this past June, information from Weill Cornell Imaging used at "NewYork-Presbyterian or NewYork-Presbyterian Hospital was entered into artificial intelligence tools and processed using machine learning without authorization and in violation of WCM policies."
Upon discovery, "we suspended the research activity and requested that all involved data be deleted," officials said. "We have also provided additional education to reinforce compliance with WCM's requirements for the appropriate use of AI tools and the protection of patient information."
Despite the improper use of the AI tool, the health system told patients it had not so far found evidence that any unauthorized individuals had accessed their PHI.
That info included medical record numbers and radiology accession number (an internal tracking system for images), according to WCM.
The data also may have included dates and times of patients' radiology services, the place they received care and other "clinical information relating to patients' radiological examination(s), such as diagnoses, treatment, surgical and procedural histories and medication information."
The information did not include, however, identifying factors, such as patient names, addresses, telephone numbers, emails, Social Security numbers or patient-specific financial information, officials said.
THE LARGER TREND
This is the second data incident involving "shadow AI" at a major academic medical center in the past few weeks. In October, we reported on a Johns Hopkins breach, where a staff member at Baltimore's Bayview Medical Center reportedlyuploaded a spreadsheet to a public AI website that hadn't been approved for use – exposing data of more than 3,000 patients.
As bothgenerative and, increasingly, agentic AI tools become more popular and widely used in healthcare settings, experts have highlighted the need for IT and security teams to stay vigilant about which technologies are being put to use, and how.
The new risks these third-party tools pose for providersmust be met with more robust oversight efforts and staff education to help protect against the security and safety risks of unauthorized AI.
ON THE RECORD
"Weill Cornell deeply regrets any concern this incident may cause," said health system leaders in a statement. "We take the confidentiality and security of our patients' information very seriously, and are continuing to expand our extensive security measures including providing additional training."
More stories like this:
- Healthcare has a shadow AI problem – but closer collaboration can help
- AI ambitions outpace healthcare's infrastructure
- OpenAI agent breaches Australian Medicare portal
Mike Miliard is executive editor of Healthcare IT News.
Email:mike.miliard@himssmedia.com
Healthcare IT News is a HIMSS publication.
Follow us on LinkedIn:Healthcare IT News
VIDEO: The risks of shadow AI on clinician devices
Topic:
More Regional News
HIMSSCast: Improving nurse engagement with workforce tech
By Andrea Fox | November 13, 2026
Corewell Health's ambient AI cuts nursing documentation time by 52%
By Bill Siwicki | November 13, 2026
ePA testing project launched by ONC and CMS
By Mike Miliard | November 13, 2026
来源:AI Incident Database · healthcareitnews.com


