X
Story Stream
recent articles

When digital infrastructure fails in finance or aviation, the backlash is immediate. 

Over the past week, a string of disruptions dubbed "Nightmare November" rattled commuters, travelers, and markets alike. Major banks stumbled, Microsoft Outlook went dark across corporate offices, and a Federal Aviation Administration telecommunications failure grounded roughly 20,000 flights. 

Almost on cue, policy makers and tech executives began warning of an "AI vulnpocalypse"—a world in which autonomous artificial intelligence discovers software bugs faster than engineers can patch them, threatening the networks that keep commerce moving.

Those disruptions, costly as they were, largely cleared within hours or a couple of days. 

Meanwhile, a far deadlier vulnerability is playing out across a sector that accounts for nearly a fifth of the U.S. economy—and receiving a fraction of the urgency. 

While Washington and Silicon Valley debate how to patch bank servers and air-traffic links at machine speed, hospitals and medical research centers are being plunged into multi-week blackouts that put patients’ lives at risk.

When an airline’s network crashes, planes stay safely on the tarmac until engineers reboot the system. A hospital cannot put a stroke, a trauma resuscitation, or a pediatric chemotherapy infusion on hold.

Yet modern healthcare runs on a brittle, hyper-interconnected web of electronic health records, automated pharmacies, laboratory analyzers, and bedside infusion pumps. 

Unlike a banking outage that resolves by lunchtime, a hospital ransomware attack leaves clinicians working in the dark for a median of 15 days. Doctors lose instant access to prior CT scans, allergy alerts, and real-time lab results. Nurses calculate high-risk medication drips by hand while couriers carry blood samples down stairwells. 

Research published in the American Economic Journal: Economic Policy found that during the first week of a hospital cyberattack, in-hospital mortality rises by more than a third, while neighboring emergency rooms suffer dangerous overcrowding from diverted ambulances.

The prevailing fixation on AI-discovered software flaws also misunderstands why healthcare is failing. 

According to 2026 threat data from Mandiant and Verizon, software vulnerabilities account for roughly a third of initial intrusions; most healthcare breaches still begin with stolen employee credentials (36%) or compromised third-party suppliers (roughly a quarter, and rising 60% year-over-year).  

Where AI changes the equation is velocity. Autonomous agents have compressed the window between a flaw’s discovery and its weaponization to "minus seven days"—meaning attackers exploit weaknesses before public patches even exist—while burying hospital IT teams under more than a thousand monthly enterprise software fixes.

Worse, the standard Silicon Valley prescription—"patch faster"—hits a structural wall in medicine. Roughly 60% of connected medical devices run on end-of-life operating systems for which manufacturers no longer issue patches, and 99% of U.S. hospitals operate clinical equipment carrying already-known security flaws. 

Once an AI-assisted intruder slips past an internet-facing firewall or a vendor portal, flat internal networks give it an open path to core hospital servers. Nor can hospitals simply push automated updates onto bedside gear without risking patient safety: a single defective security update in 2024 simultaneously crashed clinical workstations across a third of studied U.S. hospitals.

Just as troubling is the quiet sabotage of medical research. Clinical trials are the capital-intensive engine of biomedical innovation, bound together by strict regulatory audit trails, randomization portals, and continuous cold-chain monitoring.

 When cyberattacks hit academic cancer centers, device makers, or trial networks, experimental therapies stall. Even when cloud trial software remains intact, a compromised hospital network or vendor identity gateway locks researchers out. 

Narrow protocol windows for blood draws and tumor scans pass unmet; temperature-sensitive biologics risk spoilage; and months of clinical data face regulatory invalidation. In documented hospital attacks, oncology trial referrals and patient recruitment have collapsed for months. 

For patients with advanced cancers or rare genetic disorders whose last viable option is an experimental drug, a cyber blackout is not an administrative delay—it is a closed door.

Why does a single day of airport gridlock trigger congressional hearings while multi-week hospital shutdowns barely register beyond local headlines? Visibility is part of it: travel and banking outages hit millions of consumers at once, whereas medical emergencies unfold behind closed hospital doors across scattered regional systems.

The deeper failure, however, lies in regulatory accounting and financial incentives. Federal breach laws meticulously tally how many Social Security numbers are exposed in a hack, while ignoring clinical downtime, canceled surgeries, and diverted ambulances. 

Health systems, squeezed by thin operating margins, routinely defer costly infrastructure replacements, leaving 60% without basic data controls for AI tools. Washington’s response has been equally sluggish: updated federal hospital cybersecurity standards remain stalled until 2027, and bipartisan resilience legislation has languished in Congress.

Hardening financial networks and aviation infrastructure against AI-speed attacks is plain common sense. But a country that insures its wire transfers and flight schedules while leaving its emergency rooms and cancer research labs exposed has mispriced its risks.  

Because legacy medical hardware will remain unpatchable for years, real security will not come from an endless patching treadmill or post-breach credit monitoring. 

Health systems, insurers, and regulators must treat cyber downtime as a core patient-safety metrics. 

Retiring obsolete clinical devices, isolating bedside networks from corporate IT, locking down third-party vendor logins, and mandating air-gapped offline backups are as important as a correct diagnosis.  Preventing cyberattacks and outages are essential to delivering the best care .

Dr. Robert Goldberg is Vice President of Research Programs at the Center for Medicine in the Public Interest.

Robert Goldberg is the James F. Bender Clinical Professor of Finance at Adelphi University's Robert B. Willumstad School of Business. Before entering academia, he spent more than 20 years in investment banking, corporate finance and asset management.

Comment
Show comments Hide Comments