Network servers were the primary drivers of breach numbers in October 2026. Of the 34 reported breaches, 27 involved a network server, affecting 10,106,558 individuals out of a total of 10,116,362. This is a significant increase from the previous year's total of 1,419,860 individuals affected by network server breaches.
Two incidents account for most of that volume. iRhythm Technologies, Inc. reported a breach affecting 8,179,772 people, and Central Ohio Primary Care Physicians, Inc. reported 1,351,650.
So where should your next dollar go? Practitioners are divided. One camp argues for hardening the perimeter and servers themselves. The other suggests assuming compromise and minimizing what any single server can lose.
The Allocation Dilemma
You can't fully fund both strategies. IT security leads face a real allocation problem: invest in preventing server compromise (patching, vulnerability management, access controls, monitoring), or invest in limiting blast radius (segmentation, data minimization, retention limits, encryption architecture) so that a compromise exposes fewer records.
The October numbers sharpen the debate. Breach frequency rose modestly, from 22 to 27 network server incidents, but the impact rose dramatically. This suggests the problem isn't just that servers get compromised, but how much data sits on each one when they are.
The Case for Hardening Servers
Proponents of prevention start with the HIPAA Security Rule's structure. Risk analysis and risk management are Required Specifications under 45 CFR 164.308(a)(1). Technical safeguards, including access control, audit controls, integrity controls, and transmission security, all focus on the server layer. If you can't demonstrate disciplined configuration and monitoring there, you're facing a compliance problem before a breach even occurs.
Practical arguments include:
- Prevention avoids notification. Once an unauthorized party accesses a server holding unsecured PHI, the Breach Notification Rule is triggered. Segmentation reduces the number of affected individuals but doesn't stop the event.
- Server weaknesses are knowable. Unpatched systems, exposed management interfaces, weak or shared credentials, and missing multi-factor authentication are inventory problems. They respond to disciplined vulnerability scanning and patch cadence.
- Frequency matters. Twenty-seven network server incidents in a single month indicate many attempts are succeeding. Hardening addresses the attack vector that dominates the occurrence data.
- Recognized Security Practices count. Under the 2021 amendment to the HITECH Act, OCR must consider whether you had Recognized Security Practices in place for the prior 12 months when determining penalties and audit outcomes. Documented hardening aligned to NIST SP 800-66 or the HITRUST CSF earns real credit.
The hardening camp's bottom line: fix what attackers exploit and maintain a thorough paper trail.
The Case for Limiting Blast Radius
The other side sees a different lesson in the same data. Two breaches made up over 94 percent of the October network server impact. Whatever controls those organizations had, a single compromise exposed millions of records.
Their arguments:
- Perfect prevention doesn't exist. Attackers need one working path. Defenders need all of them closed, every day. Designing as though a server will eventually be compromised is realistic engineering.
- Impact scales with data concentration. A server holding years of records for an entire patient population is a single point of catastrophic failure. Retention schedules, archival of inactive records, and tiered storage reduce what's reachable.
- Segmentation limits lateral movement. Separating clinical systems, billing, and repositories means one foothold doesn't translate to every database.
- Encryption can change the legal outcome. PHI encrypted consistent with HHS guidance is not "unsecured PHI," which can take an incident outside breach notification. That only works if keys are managed separately and the attacker didn't obtain them along with the data.
The blast-radius camp's bottom line: the cost of a breach is driven by how much data was exposed, so engineer the exposure down.
Where Practitioners Actually Land
Most seasoned security leads don't pick a side cleanly. They sequence.
First, they treat hardening as the floor. Nobody serious argues against patching, MFA, logging, and tested backups. Those are baseline expectations under the Security Rule and the cheapest controls per unit of risk reduction.
Second, they use breach history to decide where to spend beyond the floor. The year-over-year pattern is useful here. Network server breaches affecting the most people were at a five-year high in October 2026, while email incidents fell to 4, down from 6 the year before and 15 two years earlier. That decline doesn't mean email security can be deprioritized; inbound filtering and authentication remain a common route to credentials that later unlock servers. But it does suggest that the attention email received has paid off, and that comparable discipline is overdue on server infrastructure.
Third, they let the risk analysis arbitrate. A small clinic with a single practice management server and a large device manufacturer with a massive centralized repository face different math. The larger the data concentration, the more the blast-radius arguments dominate.
Electronic medical record breaches also rose, to 64,728 people from 20,081 the prior year. Smaller in scale, but a reminder that the clinical systems themselves are targets, not only the infrastructure beneath them.
Our Take
Hardening wins as the first priority, and blast-radius reduction wins as the second. Reversing that order is a mistake. Teams that skip basic server hygiene to build elegant segmentation schemes are building walls around a front door that's propped open.
But stopping at hardening is the more common and more expensive error. The October data is a case study in concentration risk: frequency moved from 22 to 27, while individuals affected jumped from roughly 1.4 million to over 10 million. Prevention controls address the first number. Only data minimization, retention discipline, segmentation, and sound encryption address the second.
Practical steps for your team:
- Inventory where ePHI actually lives. Your risk analysis is only as good as your asset map. Identify every server holding Electronic Protected Health Information (ePHI) and how much each holds.
- Apply retention limits. If records no longer serve treatment, payment, operations, or a legal retention requirement, move them offline or destroy them per policy.
- Segment by function and sensitivity. Treat any repository holding a large share of your patient population as crown-jewel infrastructure.
- Verify encryption in practice. Confirm data at rest is encrypted, keys are stored separately, and administrative access is limited and logged.
- Document everything. Evidence of your practices over time is what OCR evaluates, and what Recognized Security Practices credit depends on.
The tradeoff is real. Minimization and segmentation cost operational convenience, and clinicians and analysts will push back on losing instant access to everything. That's a governance conversation worth having before an incident forces it.





