A phishing email at healthcare AI vendor Xsolis led to a breach exposing records for 1.4 million patients across seven health systems including Mayo Clinic, underscoring the risk of concentrating sensitive data with third-party AI platforms.
DeepHealth, a RadNet subsidiary, has received FDA 510(k) clearance for an AI system that detects breast ultrasound lesions and drafts structured reports, with validation data showing over 98% localization accuracy and a 37% cut in interpretation time.
Butterfly Network's FDA-cleared Gestational Age Tool uses AI to estimate how far along a pregnancy is from a simple ultrasound sweep in under two minutes, and is already deployed in Malawi and Uganda with Gates Foundation backing.
A new report from Disability Belongs and the Consumer Technology Association, released August 4, 2026, documents how people with disabilities are using AI tools right now across transportation, household management, healthcare, and employment.
Proscia has won FDA clearance for its Concentriq AP-Dx digital pathology platform to include a Predetermined Change Control Plan, letting it add support for new scanners without filing separate approvals each time.
Dartmouth's Therabot, tested in a 210-person randomized trial published March 2025 in NEJM AI, showed a 51% symptom reduction for depression and 31% for anxiety, offering the first clinical-trial evidence that a purpose-built AI chatbot can meaningfully treat mental illness.
After Samsung, Apple, PranaQ, and Happy Health all won FDA clearances for AI-powered sleep apnea detection on wearables in 2025, the FDA issued new January 2026 guidance clarifying which wellness claims wearables can make without full medical-device review.
Medtronic unveiled Touch Surgery Aide, a real-time AI computing platform for robotic operating rooms, featuring an FDA-cleared instrument tracking feature debuted at the Society of Robotic Surgery's 2026 meeting.
Researchers at Hong Kong University of Science and Technology built PRET, an AI pathology system that recognizes 18 cancer types from a handful of slides without needing retraining for each new hospital.
A multisite validation study found Duke Health's Sepsis Watch AI model, credited with a 27% drop in sepsis deaths at Duke, performed just as strongly after being deployed at a different community hospital system.