Contributor: Vidya Murthy, WEMBA’42
To learn more about Vidya, click here.

For years, healthcare treated cybersecurity like fire insurance: buy enough to satisfy an auditor, forget about it, move on. That's over. Three things are colliding right now that turn security into a precondition for doing business, not a line item: the FDA is embracing AI itself, attackers are weaponizing AI faster than most teams can patch for, and everyone in healthcare is racing to ship AI-powered clinical tools before the competition does.
Start with the regulator. Since 2023, the FDA can flat-out deny a device's market clearance over cybersecurity alone. No need to find a separate flaw first. Skip the documentation and you don't get a warning letter, you get a "refuse to accept" letter, and your launch date quietly becomes next quarter's problem.
It gets sharper. The DOJ has started using the False Claims Act (the fraud statute) against companies that certified cybersecurity controls they hadn't actually built, following Illumina's settlement over unpatched device vulnerabilities. Think of it like a SOX certification: exactly the kind of thing your general counsel and CFO want double-checked before anyone signs it.
Meanwhile, the FDA's using AI to move faster on its own end. Commissioner Makary rolled out Elsa, an internal AI tool that's reportedly turned three-day reviews into six-minute ones. An agency that reviews that fast expects the work in front of it to keep pace.
Attackers noticed the same AI wave and got there first. IBM found AI played a role in 16% of breaches last year, mostly phishing and deepfakes, and breaches involving ungoverned AI cost about $193K more on average. And the pace is only accelerating: healthcare logged 281 breaches in just the first half of 2026, second only to financial services, already on track to blow past all of 2025. Healthcare runs on urgent messages between clinicians, vendors, and payers, a perfect setup for exactly that kind of attack, and it's already the most expensive industry to get breached: 14 years running, at $7.42 million a pop.
And everyone's shipping AI anyway, for good reason: 75% of health systems use at least one AI tool now, up from 59% last year, and more than half who track ROI report 2x returns. Ambient scribes, remote monitoring, faster diagnoses: patients and clinicians want all of it. But every one of those tools is also a new door for the attackers above. Ship the experience without shipping the security underneath it, and you haven't built a better product. You've built a bigger target.
Bottom line: security isn't separable from the product anymore. A device that can't prove it's secure doesn't clear FDA review, or diligence on the next acquisition. A health system that can't catch a synthetic voice is one wire transfer from a bad quarter; a shaky compliance certification is one DOJ call from a worse one. The question for executives isn't "how much security satisfies the regulator" anymore. It's "how much do we need to actually do business, close deals, and keep the board out of a deposition." In healthcare, that number is never zero.
Contact Vidya at: [email protected]