Ransomware Attacks: Incident Response Case Studies

Nov 10, 2026
Black and white portrait of a man with short hair and a beard, wearing a light-colored shirt with small patterns, looking slightly to the right against a background of vertical dark slats.
Black-and-white portrait of Dan Potter.

Most ransomware advice stops at prevention: patch faster, phish-test harder, lock down identity. That advice matters, but it skips the question that decides how bad an attack becomes: when ransomware does land, how well does your organization respond? The difference between a contained incident and a front-page crisis is rarely the malware. It is the decisions people make in the first hours, and whether they have made those decisions before.

Below are several of the most instructive ransomware incidents — from the attacks that defined the modern threat to the ones making headlines now. None is an Immersive customer story; they are public, well-documented events. For each, we look at what happened, the incident response decisions that shaped the outcome, and the lesson security teams can rehearse before it is their turn. We finish with what is changing fastest right now: AI in the hands of attackers. Throughout, Immersive subject-matter experts add the practitioner view.

A note on the older cases. WannaCry and NotPetya are from 2017 and Colonial Pipeline from 2021, but they are included deliberately: they are the most thoroughly documented incidents in existence, and the response lessons — containment, segmentation, recovery speed, decision-making under pressure — have not aged. The 2024–2025 cases that follow show those same lessons playing out at far greater scale, now accelerated by AI.

Why ransomware is an incident response problem, not just a prevention problem

Immersive's David Spencer, who works across technical product management, frames the shift bluntly: the threats themselves have not changed so much as the economics around them. Ransomware moved from nuisance to encryption, to today's double and triple extortion, where attackers also threaten to leak the data they stole.

Because you cannot assume prevention will hold, Spencer argues the strategic center of gravity has moved to recovery speed. “It's now more about the speed at which you recover. So identify attacks quickly, neutralize them quickly and move on, or get back to business as usual.”

That is the lens for the case studies that follow: not “how did they get in,” but “how did the response play out, and what would have made it faster?”

Case study 1 — Colonial Pipeline (2021): the containment call under pressure

What happened

On 7 May 2021, the DarkSide ransomware-as-a-service group breached Colonial Pipeline, operator of the largest refined-fuel pipeline in the United States. The entry point was a legacy VPN account with a reused password and no multi-factor authentication. Attackers moved laterally, exfiltrated roughly 100GB of data, and encrypted IT billing systems. Colonial proactively shut down its operational technology to keep the attackers away from fuel-delivery systems, paid a ransom of about US$4.4 million, and restored service by 12 May. The shutdown triggered panic buying and fuel shortages across the US East Coast; the FBI later recovered roughly $2.3 million of the payment.

The incident response decisions that mattered

Two decisions defined the outcome. First, containment: the choice to take OT offline to prevent the ransomware crossing from IT into fuel operations. Immersive's Kev Breen, who researches cyber threats, points to this as a textbook proactive separation — and a reminder that turning systems off is far quicker than turning them back on.
‍

“Colonial Pipeline suffered a ransomware attack that was close to their operational technology network, their ICS. And what they did is they took a proactive decision to separate those, to turn off their OT to avoid the attackers being able to jump on.” — Kev Breen, Immersive

‍
Second, the pay-or-don't-pay call. Breen's experience running crisis exercises is that this is the moment teams discover the wrong people are in the room.
‍

“The classic example with ransomware, to pay or not to pay … that decision usually lies with the legal team. Do you have your legal team in the room to be able to give you counsel on that decision?”— Kev Breen, Immersive

‍
There is also an OT angle that makes critical-infrastructure ransomware uniquely costly. Immersive's Sam Maesschalk, who specialises in OT/ICS, notes that downtime in these environments is brutally expensive — in manufacturing, on the order of $50,000 per minute — which is exactly why attackers increasingly target them.

The lesson to rehearse

Identity hygiene (MFA on every remote-access path) and IT/OT segmentation reduce the blast radius. But the decisive moments — whether to pull OT offline, whether and when to pay — are leadership and legal calls made under extreme time pressure. They should be rehearsed, not improvised. This is the core use case for a cyber crisis simulation that puts executives and legal counsel in the room together, and for OT-specific exercising for teams defending industrial environments.

Case study 2 — WannaCry and the NHS (2017): the cost of a known, unpatched vulnerability

What happened

On 12 May 2017, the WannaCry worm spread globally using EternalBlue, an SMB exploit leaked from the NSA by the Shadow Brokers. Microsoft had released a patch (MS17-010) two months earlier, but unpatched systems were hit hard. In England, around a third of NHS trusts were affected; roughly 19,000 appointments and operations were canceled, and some hospitals diverted patients. A security researcher's discovery of a kill switch slowed the spread, and the bulk of disruption was contained within about 48 hours. The UK government later estimated the cost to the NHS at around £92 million.

The incident response decisions that mattered

WannaCry was not a sophisticated, targeted intrusion — it was a known vulnerability meeting a slow, largely manual response. Staff locked out of machines reverted to pen and paper while teams worked through the weekend to contain the spread. The organizations that fared best were those that could detect, isolate, and restore quickly; the worst-hit were those discovering their patch and recovery gaps in real time.

This is where David Spencer's “speed of recovery” point becomes concrete: the time between detection and containment is where cost accumulates. A team that already knows the runbook — because it has practiced it — collapses that window.

The lesson to rehearse

Patch and vulnerability management are the prevention story, and Immersive's catalog includes dedicated vulnerability-management and patch-hardening labs. But the response story is muscle memory: responders need to have isolated infected hosts, triaged at speed, and run the recovery playbook before the real thing. Hands-on labs build that individual technical capability, while cyber drills pressure-test how teams coordinate when systems are down.

Case study 3 — NotPetya and Maersk (2017): when recovery is the whole game

What happened

In June 2017, NotPetya — a destructive wiper disguised as ransomware, attributed to the Russian military's Sandworm group — spread from a compromised Ukrainian accounting-software update and tore through global networks. Shipping giant A.P. Møller-Maersk was collateral damage: the malware propagated across its flat internal network and crippled operations in over 130 countries. Because NotPetya could not actually decrypt, paying was pointless. Maersk rebuilt from the ground up — roughly 4,000 servers, 45,000 PCs, and 2,500 applications in about ten days — famously saved by a single domain controller in Ghana that had been offline during a power cut. The incident cost an estimated US$250–300 million.

The incident response decisions that mattered

NotPetya is the clearest argument in modern security for two things: network segmentation and tested recovery. Maersk's flat network let the malware spread almost unchecked, and its recovery hinged on a lucky offline backup rather than a designed one. The immediate response — shutting everything down to halt propagation — was correct, but the scale of the rebuild showed how much resilience had been left to chance.

It also underlines a human truth from Immersive's Dan Potter, who works on cyber drills and resilience: teams that respond to incidents daily are not automatically ready for the once-in-a-decade event.
‍

“Take those teams that are operationally responding to incidents every day, use an immersive exercise to get them to respond to incidents that they don't face every day.” — Dan Potter, Immersive

‍
The lesson to rehearse

Segment networks so one infection cannot become an enterprise-wide outage, and test recovery as a deliberate capability rather than hoping a backup survived. A full-team cyber range exercise lets organizations rehearse a fast-spreading, business-wide incident end to end, and resilience scoring and reporting turns that practice into evidence of readiness over time.

Case study 4 — Change Healthcare (2024): the same mistakes, ten times the scale

What happened

On 21 February 2024, the BlackCat/ALPHV group deployed ransomware inside Change Healthcare, a UnitedHealth Group subsidiary that processes roughly half of all US medical claims. The attackers had entered around nine days earlier through a Citrix remote-access portal that was not protected by multi-factor authentication, then moved laterally and exfiltrated data on an estimated 100 million people — the largest healthcare data breach in US history. UnitedHealth paid about US$22 million. It did not get the data back: BlackCat ran an exit scam, and the affiliate took the stolen data to a second group, RansomHub, for another extortion attempt. The incident is expected to cost UnitedHealth in excess of $1.5 billion, and AMA surveys found around 80% of physician practices lost revenue while billing systems were down.

The incident response decisions that mattered

Two things stand out. First, the entry point was identical in spirit to Colonial Pipeline three years earlier: a remote-access service without MFA. Second, containment came late and blind — because responders did not initially know how the attackers got in, they severed connectivity to the data center to stop the malware reaching the wider UnitedHealth estate. That call worked, but the nine-day dwell time beforehand let the attackers steal data at scale.

Change Healthcare also became the definitive proof of a point Immersive's experts keep making: paying does not guarantee recovery. UnitedHealth paid and still lost the data twice over. The pay/don't-pay decision, again, was a leadership and legal judgment made under enormous pressure — and the cleaner answer is tested backups and a rehearsed recovery plan, not a ransom wallet.

The lesson to rehearse

Enforce MFA on every remote-access path, shrink dwell time with real detection capability, and make recovery a tested process rather than a hope. Hands-on labs build the detection and threat-hunting skills that catch an intruder during those nine days, and a cyber crisis simulation rehearses the pay decision with legal and leadership in the room before $22 million is on the line.

The new front line: AI and social engineering

The most important recent shift is not a new strain of malware — it is how attackers get in, and how fast. In September 2023, the Scattered Spider crew, working with BlackCat/ALPHV, broke into MGM Resorts and Caesars Entertainment not with an exploit but with a phone call: they socially engineered IT help desks into resetting credentials. The decision split is instructive. Caesars paid roughly $15 million; MGM refused, absorbed around ten days of disruption and an estimated $100 million hit, and rebuilt. Through 2025, the same playbook hit major UK and US retailers.

Now add AI. The balanced industry view — from threat researchers such as Halcyon — is that AI has not fundamentally changed ransomware tactics, but it has changed the economics: it lowers the barrier to entry and accelerates the attacker's workflow. Generative AI writes more convincing phishing and vishing, translates flawlessly, modifies code, and shortens the time between a vulnerability being disclosed and being exploited. Voice phishing surged several-fold across 2024–2025, the majority of phishing content is now AI-assisted, and deepfakes have moved from novelty to operational tool — as in the 2024 Arup case, where a deepfake video call cost the firm $25 million.

And Immersive's Kev Breen points to the flip side — the AI risk organizations create for themselves:
‍

“This is still a new and emerging technology, and we are still struggling to keep pace with it … one is shadow AI. In the same way that we've grown up with shadow IT … the same is true of Gen AI.” — Kev Breen, Immersive

‍
Here is the part that should shape how teams prepare. When Scattered Spider used deepfake video in mid-2025 to trick IT administrators into granting access and registering MFA, researchers found that something very low-tech defeated it: a simple, unscripted verification request — a human asking an unexpected question — broke the impersonation, especially because the people being impersonated were often out of the office. Detection tools cannot reliably spot a good deepfake; trained, skeptical people who follow a verification habit can. The right response to AI-accelerated social engineering is to shift teams from “spot the fake” to “verify the request,” and to rehearse that behavior until it is automatic.

That is squarely human-centric resilience. Workforce exercises and cyber drills build verification reflexes and decision-making across the whole organization — help desk, SOC, executives — not just a slide that says “beware deepfakes.” As Breen notes elsewhere, static playbooks go stale because attackers weaponize new techniques faster than documents can keep up; the durable answer is the ability to recognize and respond to a threat the team has never seen before.

Common threads: what separates the organizations that recover well

Across all four incidents — and the AI-driven social engineering now emerging — the same patterns decide outcomes:

  1. Containment is a decision, not a reflex. Colonial and Maersk both chose to take systems offline. Knowing when that trade-off is worth it — and who authorizes it — has to be agreed in advance.
  2. The right people must be in the room. The pay/don't-pay call belongs to legal and leadership, not the responder at the keyboard.
  3. Communications are a legal function under attack. As David Spencer points out, in a real ransomware event it is general counsel, not marketing, deciding what can be said externally.
  4. Recovery speed is the cost lever. The faster detection leads to containment, eradication, and restoration, the smaller the bill — in money, downtime, and trust.
  5. Readiness is rehearsed, not assumed. Every one of these lessons is something a team can practice before an incident, not learn during one.

How Immersive helps you prepare

Immersive exists to close the gap between knowing the lessons above and being able to act on them under pressure. On Immersive One, organizations prove, improve, and report cyber readiness across the whole workforce — not just the SOC. Mapped to the lessons in this article:

  • Cyber Crisis Simulation — puts executives, legal, and comms through the pay/don't-pay and containment decisions before a real incident forces them.
  • Cyber Drills — short, repeatable team exercises that build the coordination muscle memory WannaCry-style outages demand.
  • Workforce Exercises — builds the “verify the request” reflex across help desk and staff that defeats AI-driven vishing and deepfake social engineering.
  • Cyber Range Exercise — full-team rehearsal of a fast-spreading, business-wide incident, the NotPetya scenario.
  • Hands-On Labs — individual technical readiness across incident response, malware and ransomware analysis, digital forensics, and OT/ICS defense.
  • Resilience Score & Reporting — evidence of readiness over time for the board, auditors, and regulators.

‍

Immersive's customers use exactly this approach. Elastic moved beyond traditional tabletop exercises to measurable, multi-team cyber drills — benchmarking global teams and building real-world confidence rather than a once-a-year report.
‍

“Realistic simulation of current threats is the only way to test and improve response readiness, and to ensure that the impact of a real attack is minimized.” — Paul Jackson, Regional Managing Director, APAC Cyber Risk, Kroll

‍
Want to see how your teams would handle a ransomware incident before a real one tests them? Request a demo.

FAQ 

What is ransomware incident response?

Ransomware incident response is the coordinated process an organization follows to detect, contain, eradicate, and recover from a ransomware attack, then learn from it. It spans technical teams, executives, legal counsel, and communications, and works best when it is rehearsed before a real incident.

Should you pay a ransomware demand?

Whether to pay is rarely a purely technical call. As Immersive's Kev Breen notes, the decision usually sits with legal counsel and leadership, who weigh legal exposure, sanctions risk, and the uncertainty that paying restores data. Organizations that have rehearsed the decision with the right people in the room respond faster and more soundly than those deciding live.

What are the stages of ransomware incident response?

The widely used stages, aligned to NIST, are: preparation; detection and analysis; containment; eradication; recovery; and post-incident review. Speed through containment, eradication, and recovery is the single biggest driver of how much an attack ultimately costs.

What can security teams learn from Colonial Pipeline, WannaCry, and NotPetya?

Colonial Pipeline shows the value of decisive containment and identity controls; WannaCry shows how unpatched, known vulnerabilities turn into mass outages; and NotPetya shows why segmentation and tested recovery determine whether a business rebuilds in days or weeks. All three reward teams that have practised the response in advance.

How can organizations prepare their teams for a ransomware attack?

By exercising the whole response, not just the SOC: hands-on technical labs for responders, cyber drills and crisis simulations for executives and legal, and cyber range exercises for full teams — then measuring readiness over time. This is the basis of Immersive's Prove, Improve, Be Ready approach on Immersive One.

Is AI making ransomware worse?

AI has not fundamentally changed ransomware tactics, but it has changed the economics — lowering the barrier to entry and speeding up the attacker's workflow, from AI-written phishing and deepfake vishing to faster exploit development. The most effective counter is human: trained people who verify unexpected requests defeat even convincing deepfakes, so the priority is rehearsing verification behavior and response decisions across the whole workforce.

Published:
November 10, 2026