Deep dive: The Change Healthcare ransomware attack
Change Healthcare, 2024 · facts checked 2026-10-04
The facts
- UnitedHealth's chief executive told the Senate Finance Committee that on February 12, 2024, attackers logged into a Change Healthcare Citrix portal with compromised credentials; it had no multi-factor authentication. (UnitedHealth Group)
- The chief executive testified that once inside, the attackers moved laterally through Change's systems and stole data, then set off ransomware nine days after getting in. (UnitedHealth Group)
- According to that testimony, on February 21, 2024, an attacker that identified itself as ALPHV, also called BlackCat, launched ransomware that encrypted Change Healthcare's systems. (UnitedHealth Group)
- UnitedHealth told senators it cut off Change's data centres straight away, and that it had found no evidence the attack spread beyond Change. (UnitedHealth Group)
- In a February 27, 2024 advisory, the FBI, CISA and HHS said healthcare was the sector most often hit among ALPHV's leaked victims since mid-December 2023, likely after its administrator urged attacks on hospitals. (FBI, CISA and HHS)
- UnitedHealth's chief executive told senators that the decision to pay a ransom was theirs as chief executive, and among the hardest they had ever made. (UnitedHealth Group)
- UnitedHealth said that in January 2025 Change Healthcare put the number of people affected at roughly 190 million, with a final count still to be filed with HHS. (UnitedHealth Group)
- UnitedHealth later said the breached server was an older Change server not yet brought up to its MFA policy, and that it told teams to ensure MFA on every external-facing application. (UnitedHealth Group)
The panel
Jordan Pike AI host
UnitedHealth Group says its chief executive told the Senate Finance Committee that on February 12, 2024, attackers logged into a Change Healthcare Citrix portal with compromised credentials, and the portal had no multi-factor authentication. UnitedHealth Group says the chief executive testified that once inside, the attackers moved laterally through Change's systems and stole data, then set off ransomware nine days after getting in. According to that testimony, on February 21, 2024, an attacker that identified itself as ALPHV, also called BlackCat, launched ransomware that encrypted Change Healthcare's systems.
Jordan Pike AI host
UnitedHealth Group told senators it cut off Change's data centres straight away, and that it had found no evidence the attack spread beyond Change. In a February 27, 2024 advisory, the FBI, CISA and HHS said healthcare was the sector most often hit among ALPHV's leaked victims since mid-December 2023, likely after its administrator urged attacks on hospitals. UnitedHealth Group says its chief executive told senators the decision to pay a ransom was theirs as chief executive, and among the hardest they had ever made.
Jordan Pike AI host
UnitedHealth Group said that in January 2025 Change Healthcare put the number of people affected at roughly 190 million, with a final count still to be filed with HHS. UnitedHealth Group later said the breached server was an older Change server not yet brought up to its MFA policy, and that it told teams to ensure MFA on every external-facing application.
Jordan Pike AI host
If you run anything reachable from the internet, or you work in healthcare or supply it, this one is for you. We'll go in order: the way in, the days in between, the encryption, and what changed afterwards. Sam, start with the front door.
Sam Kowalski AI analyst
The front door was a login with valid credentials and nothing else asked of it. To a log, that looks like an ordinary employee signing in. Credentials get stolen in plenty of ways, so assume some of yours are already out there. The only thing standing between a stolen password and the inside is the second factor. If a remote-access portal doesn't require one, you've left it to luck.
Sam Kowalski AI analyst
Now the gap between getting in and the ransomware. That is the window defenders get. In that time the attackers moved around and took data, and the encryption was just the loud final step. Every earlier step touches something: authentication logs, internal connections, large outbound transfers. The pattern in nearly every breach is that the signals were there before the event and nobody was looking at them.
Nadia Haddad AI analyst (guest)
Two things from the policy side. First, the decision to pay a ransom lands on a person at the top, so a board should have talked it through long before the day it's needed. Second, the affected-people number is not final. Counting who is affected depends on knowing what was taken, which depends on logs and scoping. That is slow, and the headline figure can move.
Jordan Pike AI host
Sam, cutting off the data centres straight away is a containment call. What does a defender take from that?
Sam Kowalski AI analyst
Isolation is a decision you want made in advance, not at the moment of crisis. Know which links you can sever, who is allowed to sever them, and what breaks when you do. Fast isolation limits spread, but it takes services offline, so people need to have thought about that trade-off beforehand. And when a company says it found no evidence of spread, ask how long its logs go back. Without that, the statement doesn't tell you much.
Sam Kowalski AI analyst
The last lesson is the older server that hadn't been brought up to the MFA policy. A policy that says every external-facing application gets MFA is only as good as your inventory of what is external-facing. Older and inherited systems are where the policy and reality drift apart. And not all MFA is equal: push approvals are not phishing-resistant, so favour phishing-resistant methods on anything exposed.
Nadia Haddad AI analyst (guest)
For boards, put this in business terms: which systems, which customers, which revenue, which regulator's clock. That clock tends to start earlier than companies think, so work out your notification obligations with counsel before an incident, not during one. I'm not giving legal advice, just pointing out that nobody wants to read the fine print for the first time mid-crisis.
Jordan Pike AI host
So: Monday morning? Sam, you first.
Sam Kowalski AI analyst
List every external-facing application, including the old and inherited ones, and check that MFA is actually enforced on each one, not just written in a policy. Then confirm how long your authentication logs are kept, because that decides what you can ever say about an intrusion.
Nadia Haddad AI analyst (guest)
Take the board a one-page exposure summary: what an outage would stop, whose data sits behind it, and which notification duties would apply. Settle who decides on containment and on any ransom question before the day you need them to decide.
Jordan Pike AI host
To sum up: a stolen login, no second factor, days of quiet movement, then encryption. Close the front door everywhere, watch the days in between, and decide your hard calls in advance.





