Editor's Note: I'm writing this from my desk while ViVE 2026 unfolds in LA. The energy around ambient documentation and agentic AI is real. But so is the ransomware attack that just shut down the University of Mississippi Medical Center. That tension between innovation and vulnerability is the story of healthcare IT right now. If you're buying AI, you'd better be hardening your defenses at the same pace.

Australian AI developer Harrison.ai filed a petition last October asking the FDA to exempt certain AI-powered radiology tools from pre-market review, provided the agency has already cleared a similar product from the same company. The petition covers six product types and would shift the burden of evidence to post-market monitoring.

Here's what makes this explosive: the FDA must respond by mid-April. If it doesn't deny the petition, it goes into effect automatically. And STAT News reported on Feb. 23 that the FDA recently appointed a former senior executive from a Harrison.ai subsidiary to direct the office that informs agency policy on AI.

Why it matters: This could fundamentally change how AI medical devices reach patients. If approved, manufacturers could release new AI tools to radiologists without the 510(k) clearance process that currently serves as a safety gate. The conflict-of-interest question around the FDA appointment adds a governance dimension that CIOs and compliance teams need to watch closely.

Verified: STAT News (Feb. 23, 2026), FDA petition public recordRegulatory Watch

HHS Launches $500K EHIgnite Challenge (Feb. 2026)
HHS's Assistant Secretary for Technology Policy is offering $500,000 for health IT developers to turn raw EHR data into actionable insights for clinicians and patients. The challenge targets interactive patient tools, clinical domain filtering, and streamlined payer workflows. Webinar scheduled March 11.

Verified: Becker's Hospital Review (Feb. 24), HHS.gov

TEFCA Hits 500 Million Records Exchanged
HHS announced that TEFCA, the national interoperability network, has reached nearly 500 million health records exchanged. ASTP/ONC also released draft USCDI v7 on January 29, proposing 29 new data elements to strengthen nationwide interoperability. For health systems running AI on clinical data, better interoperability means richer training data and more reliable inference.

Verified: HHS.gov press release (Feb. 2026)

FDA Updates Guidance on Wellness Devices and Clinical Decision Support
In early 2026, the FDA issued updated guidance easing premarket requirements for low-risk wellness products (including many wearables) while clarifying where full device regulation still applies for clinical decision support software. Health IT leaders evaluating AI-powered CDS tools should review the new boundaries carefully.

Verified: Avio Medtech analysis (Feb. 20), FDA.gov guidanceImplementation Spotlight: Mount Sinai Scales Ambient AI to 2,000+ Clinicians

At ViVE 2026 in Los Angeles, Mount Sinai Health System shared that it plans to have more than 2,000 clinicians using AI-powered ambient documentation by mid-2026. The system reported promising early returns on clinician satisfaction and documentation time savings.

This is the largest announced ambient AI deployment at a single academic medical center. For CIOs evaluating ambient documentation, Mount Sinai's scale provides a meaningful benchmark for ROI modeling.

Single source: Chief Healthcare Executive (Feb. 24, 2026, ViVE coverage)

University of Mississippi Medical Center Hit by Ransomware (Feb. 19)
A ransomware attack forced UMMC to close clinics statewide and cancel most appointments. The attack took down their Epic EHR system. Level 1 trauma patients were served using manual procedures. The attackers made contact with the medical system after the breach.

This is the largest healthcare ransomware incident of 2026 so far. CNN, HIPAA Journal, and Mississippi Free Press all confirmed the scope. UMMC is one of Mississippi's largest health systems.

Verified: CNN (Feb. 20), HIPAA Journal (Feb. 20), Mississippi Free Press (Feb. 20)

Lazarus Group Deploys Medusa Ransomware Against U.S. Healthcare
The Hacker News reported (Feb. 24) that North Korea's Lazarus Group is using Medusa ransomware to target Middle East and U.S. healthcare organizations, with average demands around $260,000 and 366 claimed incidents. State-sponsored actors targeting healthcare infrastructure is an escalating pattern that security teams should factor into threat models.

Verified: The Hacker News (Feb. 24, 2026)Research Roundup

Generative AI Matches Expert Teams on Medical Data Analysis
Researchers at UCSF and Wayne State University found that generative AI could process large medical datasets far faster than traditional research teams, and in some cases produced stronger predictive models. In the study, even a junior pair (a master's student and a high school student) successfully built preterm birth prediction models with AI support. Published in Cell Reports Medicine (Feb. 17, 2026).

Key caveat: Only 4 of 8 AI chatbots tested produced usable code. The results highlight AI's potential as a research accelerator, not a replacement for domain expertise.

Verified: ScienceDaily (Feb. 21), Cell Reports Medicine (Feb. 17, 2026)

Stanford-Harvard ARISE Report: Clinical AI Shows Mixed Real-World Results
The ARISE network's State of Clinical AI (2026) report found mixed diagnostic results when AI tools move from controlled benchmarks to real clinical settings. This echoes what JAMA's October 2025 AI Summit Report flagged: we still lack settled benchmarks for what "good" looks like in healthcare AI evaluation.

Verified: TechPolicy.press (Feb. 24), ARISE network report (Jan. 2026)

FDA Clears Brain Mets AI for MRI (Feb. 21) - 510(k) clearance for AI software detecting brain metastases on MRI. Affects up to 40% of cancer patients. (Diagnostic Imaging)

NeuroQuant 5.0 Gets FDA Clearance (Feb. 23) - Updated AI neuroimaging software for enhanced MRI assessment of neurological conditions. (Diagnostic Imaging)

Median Technologies: eyonis LCS Cleared for Lung Cancer Screening (Feb. 9) - AI-powered SaMD for lung cancer screening received 510(k) clearance. U.S. commercialization underway. (Median Technologies)

Retia Medical: Argos Infinity Cleared (Feb. 17) - Cardiovascular intelligence platform for high-risk surgical and critical care environments. (AI Journal)Worth Your Click

This week crystallized a pattern: the regulatory framework for healthcare AI is being reshaped in real time, and not always by regulators. Harrison.ai's petition to bypass FDA pre-market review is the most aggressive deregulation proposal we've seen. Meanwhile, the UMMC ransomware attack is a stark reminder that every AI system we connect to clinical workflows is also an attack surface. The organizations that will thrive are the ones investing equally in innovation and resilience. If your AI budget doesn't have a matching line item for security, you're building on sand.

Should AI medical devices that are "substantially similar" to already-cleared products be allowed on the market without FDA review? Reply to this email with your take.