Subscribe
Share
Share
Embed
Johns Hopkins Bayview says worker uploaded 3,182 patients' data to AI tool
Johns Hopkins Bayview says worker uploaded 3,182 patients' data to AI tool
The voices on this episode are generated, and so is the script, from the day's post at Physician in the Loop. Every fact in the post links to its source.
Johns Hopkins Bayview Medical Center is notifying 3,182 patients that their information may have been exposed after an employee uploaded a spreadsheet of patient data to an unapproved outside AI platform, local outlets reported.
The hospital said the upload happened and was discovered on Sept. 21, that the data covered patients treated from May through September and that Social Security and financial account numbers were not included, according to WMAR-2 News. Hopkins said it asked the platform's operator to delete the data and is reinforcing its policy and training on unapproved tools, The Banner reported.
Separately, the Food and Drug Administration cleared two imaging software products, one that measures the aorta on chest CT and one that labels bone on automated musculoskeletal ultrasound, and brain-implant developer Coherence Neuro raised $45 million for investigational cancer devices.
In this episode
How technological innovation is changing medicine for patients and the people who practice on the front lines.
Voice 1: This is Physician in the Loop, the daily briefing on AI in medicine. It's Monday, October 26th, 2026.
Voice 1: Johns Hopkins Bayview Medical Center is mailing letters to 3,182 patients after a staff member uploaded their information to an outside AI platform the hospital hadn’t approved for that use.
Voice 2: What was actually in that upload?
Voice 1: A spreadsheet with names, admission and treatment dates, medical record and account numbers, treating department, medications and provider names. The hospital says Social Security numbers and financial account information were not included.
Voice 2: When did this happen, and when did they find out?
Voice 1: An internal review found the upload had occurred on September 21st, the same day it happened. The spreadsheet covered patients admitted or treated from May through September. Letters started going out October 22nd.
Voice 2: Did they manage to get the data back?
Voice 1: The employee deleted the prompt, the conversation and the document from the tool itself. Hopkins also asked the tool’s parent company to delete the data. The tool wasn’t named in reporting.
Voice 2: So there’s no confirmation the company actually did it, just that Hopkins asked.
Voice 1: That’s what’s on record. The hospital says it’s reinforcing its policy and training on unapproved tools, which suggests this wasn’t covered by existing guardrails before.
Voice 2: One employee and one spreadsheet, and now 3,182 letters.
Voice 1: The FDA had a quieter week, clearing imaging software instead of chasing a breach.
Voice 2: Coreline Soft announced October 23rd that FDA cleared AVIEW Aorta Metrics, software that automatically segments the aorta on chest CT images, classifies its anatomical zones and measures diameters in each zone.
Voice 1: Is this the company’s first clearance?
Voice 2: No, it’s their 13th, according to the release. The clearance itself was granted October 16th. The software also produces curved planar reformations, cross-sectional and 3D views, and analysis reports.
Voice 1: Any sense of how accurate the measurements are?
Voice 2: The release included no performance data at all, just the description of what the software does.
Voice 1: A second clearance landed the day before, for ultrasound rather than CT.
Voice 2: RIVANNA said October 22nd that FDA cleared BoneEnhance, a machine learning module for its Accuro XV system. It segments bone and surrounding tissue on automated volumetric scans and overlays color-coded labels on the B-mode image, while leaving the underlying image visible.
Voice 1: Did they actually test it against anything?
Voice 2: They tested it on 233 volumetric datasets, about 69,900 images from 173 subjects across five sites, against labels from three board-certified radiologists. Label accuracy ran from about 95% to 100% per volume. Those figures are the company’s own.
Voice 1: Who paid for the development work?
Voice 2: The Biomedical Advanced Research and Development Authority funded it, and RIVANNA says an injury-detection module is under investigation next.
Voice 1: Two imaging clearances in two days, neither with outcome data attached yet.
Voice 2: Different kind of story at Coherence Neuro, this one about money and brain implants.
Voice 1: The San Francisco neurotechnology company raised $45 million in a Series A, co-led by Founders Fund and the OpenAI Startup Fund. It’s developing implants that track electrical signals associated with tumors and use them to guide electrical stimulation.
Voice 2: Is any of this authorized for patients yet?
Voice 1: No. The devices are investigational and not authorized by FDA, and the work linking neural activity to tumor growth has so far been done in mouse models.
Voice 2: So where does the human side stand?
Voice 1: A first human study began in June, testing a cortical probe during brain tumor surgery at the Royal Melbourne Hospital in Australia. Cofounders Ben Woodington and Elise Jenkins wrote that the funding moves the company toward its first long-term human trial in glioblastoma.
Voice 2: A planned system, not a cleared one, with the money arriving well before the trial does.
Voice 1: Before we go, something to watch.
Voice 2: Conifer layoffs covering 1,037 people take effect November 2nd.
Voice 1: Is there any detail on which roles are affected?
Voice 2: The calendar entry doesn’t say. It just marks the date the layoffs take effect.
Voice 1: A thousand-plus jobs is a big number to land with no context attached.
Voice 2: It’s a date worth watching, even without the detail to explain it yet.
Voice 1: Sources and corrections are at physicianintheloop.org, where you can sign up for the free email briefing. This briefing was read by AI voices.