Book Appointment

ThreatOps in 2026: What Shifted This Year for Hospital Teams

Healthcare Careers

Every January, operations managers in healthcare administration draw up a list of what must be fixed by December. Security rarely gets a clean line on that list, yet by late 2026 the practice called ThreatOps has changed enough that it deserves one. This is a status report, written plainly: what moved, what did not, and what a small IT or administration team can do about it without a new budget. The scope is information systems only; nothing here touches clinical matters.

If the word is new to you, here is the short version. ThreatOps joins two older jobs, collecting intelligence about attackers and actively hunting for them, into one operation with an owner, a calendar and written outputs. The most thorough public description I know of is a long entry in the Ops Atlas catalogue, which is candid that the label itself is unofficial. No standards body defines it. The underlying work, however, is well established.

Dwell time went the wrong way

The headline number from the year comes from a widely read incident-response report published in March. It measured a worldwide median of two weeks for an intruder going unnoticed, three days longer than the prior year’s eleven. For the slower, quieter categories, such as espionage-style intrusions and schemes involving remote contract workers, the figure reached 122 days. The same report put vishing, meaning scam phone calls, in second place among first footholds, with about one case in nine.

A practical reading: more time undetected means more value in looking, not just waiting for alarms. A hospital office that relies entirely on automated alerts is betting that the attackers will trip them. Hunting is the hedge.

The attack framework moved, and old maps broke

MITRE shipped release 20 of its ATT&CK knowledge base in October. The Enterprise matrix now lists 15 tactics and 227 techniques, and two tactics were added, one called Stealth and one called Defense Impairment. Anyone who built a coverage chart on the earlier layout will find some cells no longer line up. The Atlas takes a sober view of that: tactics are not a sequence, and a colourful heat map is not a goal. The framework’s own guidance says not every technique is meant to become a detection.

If you keep such a chart, treat the version upgrade as a prompt to review the questions behind it, not a prompt to recolour boxes. Ask which behaviours matter for your environment, then check whether your data would even show them.

Maturity guidance got a practical update

A community maturity model for intelligence programmes issued its 1.3 update in January. The Atlas aligns it with other scales and proposes a five-rung ladder of its own: nothing at all, then consuming feeds in alert-only mode, then working from a written register of questions, then a hunting programme and finally intelligence-led decisions. To leave the second rung, it suggests holding at least 80 percent coverage of your questions for two quarters in a row.

Why does this matter to a small team? Because the ladder gives you permission to stay on a lower rung deliberately. Plenty of departments will never need the top step, and that is fine so long as you can say why.

Guidance from standards bodies is still thin

Two items are worth knowing. NIST’s cybersecurity framework, revised in 2024, includes outcomes about risk assessment and adverse-event analysis that map neatly to this work. And the large NIST control catalogue lists a control for hunting that is marked optional and absent from every baseline. Translated: auditors will not force you to hunt, so the motivation has to come from your own risk appetite.

On the sharing side, European rules allow voluntary exchange of cyber information, and the Atlas points to a clause in the NIS2 directive for that. It also notes that a set of United States sharing protections runs only until 11 December 2026, so anyone relying on them should check their current status.

What did not change

Several old truths stand. Feeds are for flagging, not for automatic blocking; FIRST’s staged guidance says early programmes should alert. More feeds do not mean more safety: the Atlas lists “more feeds, more protection” among its anti-patterns, alongside attribution theatre and treating hunting as overflow for the alert queue. And an indicator sweep that comes back empty does not prove you are unaffected. The Atlas uses the 2024 joint advisory on the Volt Typhoon activity to make the point, noting that conventional indicators were described as generally lacking while footholds had been kept for years.

A short plan for a small team

Here is what I would do before the year closes, in rough order.

  • Pick an owner, even a part-time one, and agree what they stop doing to make room.
  • Write five to eight questions, each tied to a decision someone can actually make.
  • Check that your log retention reaches back ninety days; without it, hunting has nothing to read.
  • Run one hunt of one behaviour and keep the note, whether it finds something or not.
  • Look at every feed you pay for and ask what it changed this year. Cancel quietly if the answer is nothing.

The retention point comes straight from the Atlas decision tree: with no ninety days of telemetry, fix the data first. Hunting on thin data produces false comfort.

Reading list from the same site

Administrators sometimes find the analogy helpful. Triage in a clinic is about deciding what needs a specialist and what does not, and the site’s piece on recognising when a back complaint calls for specialist hospital care shows that logic in a non-technical setting. Likewise, an overview of the usual reasons people develop back pain illustrates why ruling out common explanations first is a sound habit. Intelligence work follows the same pattern: check the likely causes before chasing exotic ones.

The bottom line

The year’s lesson is modest. Attackers are staying longer, the reference frameworks keep moving and the paperwork remains optional. None of that calls for a bigger tool. It calls for a named owner, a short list of questions, expiry dates on anything you act on and the humility to write down what you did not find. Start small, measure whether it changed any decision and keep what earns its place. This is general information about security practice for administrators, not advice for any particular organisation.