SamSam chose the door, the hour and the backups

The 2018 SamSam indictment and alert described ransomware run by people: entry through exposed remote access, attacks timed outside business hours, backups encrypted. Each choice maps to a different hospital control and owner, and a fourth exposure runs through shared vendors and labs.

Looking Back revisits past weeks one at a time, up to the most recent, and asks what it meant for the banks, health systems and public agencies we serve. The series was written in 2026, with the benefit of hindsight, and each article says which week it looks back on.

Two-minute summaryThe dek and the first sentence of each section, quoted from the article

The 2018 SamSam indictment and alert described ransomware run by people: entry through exposed remote access, attacks timed outside business hours, backups encrypted. Each choice maps to a different hospital control and owner, and a fourth exposure runs through shared vendors and labs.

  1. The week On Wednesday, November 28, 2018, the Justice Department announced that a federal grand jury in Newark, New Jersey had returned an indictment, unsealed that day, against two men in Iran.
  2. The door: remote access that looked legitimate The alert issued after the indictment was specific about entry.
  3. The hour: attacks timed for thin staffing The indictment alleged that the defendants maximized damage by launching attacks outside regular business hours, when a victim would find it harder to mitigate the attack.
  4. The backups: recovery designed for an attacker who looks for it The indictment’s most consequential detail for hospitals was that the defendants encrypted victims’ backups.
  5. The shared supplier: when the victim serves many hospitals Two of the six health care-related victims named in the indictment were not hospitals.
  6. Objections: old news, a solved problem, or the wrong focus One objection is that the SamSam pattern was already familiar to specialists by late 2018, so the indictment told hospitals nothing new.
  7. With hindsight Federal guidance after SamSam kept describing the same pattern.

The week

On Wednesday, November 28, 2018, the Justice Department announced that a federal grand jury in Newark, New Jersey had returned an indictment, unsealed that day, against two men in Iran. The six-count indictment charged them over a 34-month international computer hacking and extortion scheme built on ransomware known as SamSam.

According to the indictment, the defendants wrote the malware themselves. Beginning in December 2015, they would gain unauthorized access to victims’ computers through security vulnerabilities, then install and run SamSam to encrypt data. They demanded ransom in bitcoin in exchange for decryption keys. According to the indictment, they had collected over $6 million in ransom payments and caused over $30 million in losses. The first version of the malware dated from December 2015, with refined versions in June and October 2017.

The victims numbered more than 200, according to the indictment. They included the City of Atlanta, the City of Newark, the Port of San Diego, the Colorado Department of Transportation and the University of Calgary. Six were health care-related: Hollywood Presbyterian Medical Center in Los Angeles, Kansas Heart Hospital in Wichita, Laboratory Corporation of America Holdings (LabCorp), MedStar Health, Nebraska Orthopedic Hospital (now OrthoNebraska Hospital) and Allscripts Healthcare Solutions.

The indictment described a method as much as a crime. The defendants allegedly used online reconnaissance, such as scanning for network vulnerabilities, to select targets. They disguised their attacks to look like legitimate network activity and used Tor to communicate anonymously. They maximized damage by launching attacks outside regular business hours, when victims would find it harder to respond, and by encrypting victims’ backups. The charges were conspiracy to commit wire fraud and conspiracy to commit computer fraud. They also included two counts of intentional damage to a protected computer and two counts of transmitting a demand in relation to damaging one.

The same day, the Treasury Department’s Office of Foreign Assets Control (OFAC) designated two Iran-based individuals. It said they helped exchange SamSam bitcoin ransom payments into Iranian rial and deposited the rial in Iranian banks. For the first time, OFAC publicly attributed digital currency addresses to designated persons. It listed two addresses that it said had processed over 7,000 transactions since 2013 and interacted with over 40 exchangers, some of them in the United States. Treasury said OFAC compliance obligations were the same whether a transaction was in digital or traditional currency.

On December 3, 2018, the Department of Homeland Security’s National Cybersecurity and Communications Integration Center and the FBI issued an alert on SamSam. It said victims had reported in early 2016 that the actors used the JexBoss Exploit Kit against vulnerable JBoss applications, and that FBI analysis showed the actors using, since mid-2016, Remote Desktop Protocol (RDP) access gained by brute force or stolen credentials. Some credentials had been bought on darknet marketplaces. FBI analysis of victims’ access logs showed the actors could infect a network within hours of buying credentials. The alert explained the economics too. Network-wide infections were far more likely to bring large ransom payments than infections of single systems. Organizations that provide essential functions, it said, need to resume operations quickly and are more likely to pay larger ransoms.

The December 2018 alert drew a contrast that matters for hospitals. Many ransomware campaigns, it said, rely on a victim completing an action, such as opening an email or visiting a compromised website. The SamSam indictment and the alert described something different. People chose the victims, chose the way in, chose the moment and went after the backups. We read the scheme’s success as coming from those choices more than from the code.

That shifts how a health system should think about its defenses. Against an operator who chooses, controls have to be judged by whether they survive a determined person’s attention, as well as whether they stop automated spread. Each of the attacker’s choices described in the Justice Department’s account of the indictment and in the alert maps to a different part of a hospital’s defenses, with a different owner.

This article takes four of them. The first is the door: exposed remote access and stolen credentials, owned by infrastructure and identity teams. The second is the hour: attacks launched when staffing is thin, which tests security operations and the hospital’s ability to make decisions at night. The third is the backups: recovery systems that the attacker deliberately targets, owned by infrastructure and continuity planners. The fourth is the shared supplier: when the victim is an electronic health record vendor or a laboratory that many hospitals depend on, owned by vendor management and clinical operations. A final section weighs the main objections.

The four are distinct. The door is about prevention. The hour is about detection and decision. The backups are about recovery. The supplier is about dependence on someone else’s door, hour and backups. A hospital can be strong in one and weak in another, and each needs its own evidence that it holds.

The door: remote access that looked legitimate

The alert issued after the indictment was specific about entry. Since mid-2016, it said, the actors gained persistent access through RDP, using brute force or stolen credentials. Analysis of tools on victims’ networks showed that successful actors had bought stolen RDP credentials on darknet marketplaces. The alert added a point that defines the defensive problem: detecting RDP intrusions can be hard because the malware enters through an approved access point.

That last point is why the door belongs at the top of a hospital’s list. An intrusion through an approved remote access path looks, at first, like a user logging in. Firewalls permit it. Antivirus has nothing to scan. The attacker escalates privileges, as the alert and Treasury described, and only then deploys the payload.

Why hospitals keep remote doors open

Health systems have legitimate reasons for remote access that a security team cannot simply remove. Clinicians may read images and records from home. Vendors may maintain imaging systems, laboratory analyzers and building controls from their own offices. Billing and coding staff may work remotely. Each use creates a path from the internet into the network, and each path needs an owner who knows it exists.

The alert’s recommendations were blunt. Audit the network for systems using RDP and disable the service where it is not needed. Make sure cloud virtual machines with public addresses have no open RDP ports unless there is a valid business reason. Put any system with RDP open behind a firewall and require a virtual private network. Enforce strong passwords and account lockout, apply two-factor authentication where possible, and log RDP logins, keeping the logs for at least 90 days.

Credentials bought, not guessed

The alert’s finding that actors bought stolen RDP credentials changes what a strong password is worth. A long, complex password protects against guessing. It does not protect against theft, whether by malware on a user’s home computer, reuse of the same password on a breached site, or a vendor’s compromised system. Once the password is for sale, its complexity is irrelevant.

Three controls answer theft rather than guessing. Multi-factor authentication means a stolen password alone does not open the door. Disabling dormant accounts removes credentials nobody is watching, such as those of former staff or finished vendor projects. Monitoring sign-ins for unusual sources or times flags a valid password used by the wrong person. The alert also reported that some victims found unrelated suspicious activity on their networks during remediation, which the alert called a possible indicator that their credentials had been stolen, sold on the darknet and used for other illegal activity. A hospital that finds one stolen credential should assume others are in circulation and reset broadly.

Third parties at the door

One recommendation in the alert addresses a hospital problem directly: make sure third parties that require RDP access follow internal policies on remote access. A vendor that supports a device fleet may use one account across many customers, or keep credentials on its own systems where the hospital cannot see how they are protected. If those credentials are stolen, the attacker inherits a path into every customer that shares the arrangement.

Managing vendor access means three things. Each vendor connection should use an account specific to that vendor and that hospital, with strong authentication. Access should be enabled when needed and disabled afterward, or at least restricted to known sources. And the hospital should log vendor sessions so that it can tell a technician’s maintenance visit from an intruder using the technician’s password. Contracts should require all three.

What the door looked like by 2023

The guidance did not soften with time. CISA’s #StopRansomware Guide, updated in September 2023, says not to expose services such as RDP on the web. It says threat actors often gain initial access through exposed and poorly secured remote services and then move through the network using the native Windows RDP client. Its recommendations repeat the 2018 list: audit the network for systems using RDP, close unused RDP ports, enforce account lockouts, apply multifactor authentication and log RDP login attempts.

The consistency is itself informative. In 2023, the federal ransomware guide still told organizations to close the same door that the 2018 alert described. That suggests the problem is less a lack of knowledge than a lack of inventory and ownership. A hospital cannot close remote access paths that nobody has listed.

What an auditor would ask for

For a health system, the evidence for the door is concrete. It is a current list of every externally reachable remote access service, with its business owner and justification. It includes proof of strong authentication on each, account lockout settings, and logs of logins kept for a defined period. It covers vendor access, with the vendor named and the account’s scope stated. A regular external scan that compares what is reachable from the internet with what the list says should be reachable closes the loop.

How size changes the answer

A large health system can run a central remote access service and require every user and vendor to pass through it. A small rural hospital may depend on a single managed service provider who set up remote access years ago. For the small hospital, the most important step may be asking that provider for the list of every remote path into its network and the authentication on each. The question costs nothing, and the answer shows how much the hospital has delegated.

The hour: attacks timed for thin staffing

The indictment alleged that the defendants maximized damage by launching attacks outside regular business hours, when a victim would find it harder to mitigate the attack. The alert added that the actors could infect a network within hours of buying credentials. Together those statements describe a timeline a hospital’s defenses must match: from purchased credentials to infection within hours, launched when the victim is least ready to respond.

Hospitals operate around the clock, but their technology and security teams may not be staffed that way. Clinical units work through the night while the server team may be on call from home. An attacker who knows that works in the gap between the hospital’s clinical day and its technical day.

Detection that works at night

The first requirement is that someone sees the attack while it is happening. Signals of the SamSam pattern include a remote login at an unusual hour, a new administrative account, a sudden spread of remote execution across servers, and attempts to stop or delete backup jobs. Each is visible in logs if the logs are collected and someone, or something, is watching them.

Hospitals have three ways to watch at night. They can staff their own overnight shift, buy a managed detection service, or rely on automated rules that page an on-call engineer. Each has a weakness. Overnight staff are costly and hard to retain. A managed service may see the alert but lack authority to act. Automated paging depends on someone waking, understanding and responding within minutes. The right mix depends on size, but the test is the same: how long between the first malicious login and a human decision.

Authority to act before morning

Seeing the attack is half the problem. The other half is deciding what to do. Containment at night may mean disconnecting a server, disabling an account used by a vendor, or isolating a network segment that carries clinical systems. Each action can disrupt care. Where the person on call lacks authority to take such steps, they must first reach an executive, who may also be asleep.

Clinical operations offer a model: incident command, with predefined roles and authority to act in an emergency. Cyber incidents fit that model when the predefined decisions include technical containment. A hospital can agree in advance that the on-call security lead may isolate any server showing mass encryption, and that the administrator on call is told immediately rather than asked first. Writing those authorities down in daylight is what lets someone use them at night.

Downtime procedures as part of detection

Clinical downtime procedures also matter at night. When systems fail, nurses and physicians move to paper or standalone systems. If the shift to downtime procedures is fast and practiced, the technical team can contain aggressively, because the cost of taking systems offline falls. If downtime procedures are slow or untested, every containment decision becomes a choice between stopping the attack and keeping care running.

HIPAA’s Security Rule already requires an emergency mode operation plan: procedures to enable continuation of critical business processes for protection of electronic protected health information while operating in emergency mode. The rule frames this around protecting information. In a ransomware attack, the same plan determines whether care continues. A hospital that tests its downtime procedures overnight as well as during the day learns whether they work when the attacker prefers to strike.

An illustrative night

Consider a hypothetical community hospital on a weekend night. At 1 a.m., a remote login succeeds on a server used by an imaging vendor. Within an hour, a new administrator account appears, and remote commands start running on several file servers. At 3 a.m., backup jobs begin to fail. At 4 a.m., encryption starts.

Each step leaves a trace. Whether the hospital acts on any of them depends on choices made months earlier. If vendor logins are logged and alert on unusual hours, the first trace pages someone. If new administrator accounts trigger an alert, the second does. If failed backup jobs page an engineer rather than sending a morning email, the third does. If the engineer who receives the page can isolate servers without waking three executives, the hospital may contain the attack before encryption spreads. If none of those links holds, staff discover the attack when clinicians cannot open records at the start of the day shift.

The illustration follows the sequence the indictment and alert describe, and it shows that the hour is won or lost through a chain of small decisions.

Measures that show readiness

Useful measures for the hour are few and concrete. Time from a high-severity alert to acknowledgment by a person, by time of day. Time from acknowledgment to a containment decision. Coverage of critical servers by monitoring that can detect mass encryption or backup tampering. The date of the last overnight downtime drill and what it found. A hospital that tracks these knows whether its defenses keep the hours the attackers keep.

Logs the attacker cannot erase

An operator with administrator rights can delete logs as easily as backups. The 2018 alert asked organizations to log RDP logins and keep the logs for at least 90 days. Keeping them on the same servers the attacker controls defeats the purpose. Logs that matter for the hour should be sent, as they are written, to a system with separate credentials, so that the record of the night survives the night.

Those logs serve two purposes. During the attack, they let someone see it. After the attack, they let the hospital establish what happened, which drives decisions about breach notification, about which systems to trust, and about what to tell regulators and patients. A hospital without them may have to assume the worst on every one of those questions.

Differences by type of organization

The hour plays out differently across health care. An academic medical center may run a security operations center with overnight staff. An ambulatory network may close its clinics at night, which lowers the clinical cost of containment but may leave no one watching. A laboratory or a health IT vendor serves many customers continuously and may face the hardest trade-off, because its downtime becomes everyone’s downtime. That last case is the subject of the fourth section.

The backups: recovery designed for an attacker who looks for it

The indictment’s most consequential detail for hospitals was that the defendants encrypted victims’ backups. A backup designed for fires, floods and hardware failures assumes the threat does not know the backup exists. An attacker with administrator rights does know, and can find it, delete it or encrypt it before triggering the main attack.

CISA’s 2023 guide states the consequence plainly: many ransomware variants attempt to find and delete or encrypt accessible backups, making restoration impossible unless the ransom is paid. That turns backup architecture from an operational choice into a security control. A backup that the attacker can reach is part of the attack surface.

What HIPAA already required, and what it did not say

The HIPAA Security Rule requires covered entities and business associates to have a data backup plan, meaning procedures to create and maintain retrievable exact copies of electronic protected health information. It requires a disaster recovery plan to restore any loss of data. It treats testing and revision of contingency plans, and analysis of which applications and data are most critical, as addressable specifications.

Those requirements describe copies and restoration. They do not, in their current text, describe protecting the copies from an attacker who holds administrator rights. A hospital could meet the letter of the backup requirement with copies stored on systems reachable from the same compromised domain. SamSam showed what that arrangement is worth during an attack.

Separation, immutability and testing

The 2023 guide describes what backup designed for an attacker looks like. Maintain offline, encrypted backups of critical data. Regularly test their availability and integrity in a disaster recovery scenario. Maintain golden images of critical systems. Keep backups of infrastructure-as-code templates offline so cloud resources can be redeployed. Store source code or executables, and license agreements, with offline backups. Keep a hard copy of the incident response plan, as well as an offline version.

For a hospital, separation means the backup system does not trust the production directory, so stolen domain credentials cannot delete it. It means administrative access to backups requires separate credentials and strong authentication. It means at least one copy that cannot be altered for a defined period, whether through offline media or storage that enforces immutability. Each of those choices costs money and complicates daily operations, which is why they need an executive decision rather than a default.

Backups of what

A backup plan can cover copies of data and nothing else. A hospital recovering from ransomware needs more than data. It needs the directory that controls who can log in, because a compromised directory must be rebuilt or cleaned before anything else can be trusted. It needs configurations for network devices, firewalls and servers. It needs the installation files and license agreements for clinical applications; the 2023 guide recommends storing source code or executables, and license agreements, with offline backups.

A hospital that has protected copies of its databases but not of its directory or configurations may find that restoration stalls at the first step. Rebuilding a directory from memory, for an organization with thousands of users and service accounts, can take longer than restoring the data it governs. The inventory of what is backed up should therefore list identity systems and configurations alongside clinical data, each with its own restoration procedure and owner.

Restoration is the real test

A backup that exists but cannot be restored quickly is of limited use to a hospital. The clinical question is how long until the electronic health record, laboratory system, imaging archive and pharmacy system are back. Restoring from clean backups is slow when every server must first be checked for persistence. Without golden images and documented rebuild procedures, a team may spend days reconstructing configurations no one wrote down.

The 2025 HIPAA proposal addressed this directly. It would require regulated entities to have written procedures to restore critical relevant electronic information systems and data within 72 hours of a loss, and other systems according to a criticality analysis. It would also add a technical safeguard for data backup and recovery, moving the requirement from physical safeguards and applying it more broadly. As of the 2026 Unified Agenda, the final rule was listed as a long-term action, with final action projected for July 2027.

What a restoration test should prove

A meaningful restoration test answers several questions with evidence, not assertion.

  • Which systems are critical, in what order they must return, and who agreed to that order on the clinical side.
  • How long a full restoration of each critical system takes from offline or immutable copies, measured rather than estimated.
  • Whether restoration works when the production directory is assumed compromised and cannot be used.
  • Whether the team can rebuild servers from golden images and documented configurations without the people who built them.

Tests that only restore a few files from the latest backup do not answer those questions. Tests that rebuild a critical application in an isolated environment, on a schedule, do. The results belong in front of the executives who decide how much to spend on recovery.

The cost trade-off

Recovery designed for an attacker costs more than recovery designed for a flood. Immutable storage, separate credentials, isolated restore environments and regular large-scale tests all add expense and staff time. The case for spending it rests on what SamSam showed: when backups survive, a hospital can refuse to pay and recover on its own timetable. When they do not, its options narrow to rebuilding from scratch or paying, and after November 2018 a payment could also raise sanctions questions, since Treasury had designated two people who exchanged SamSam ransom proceeds.

The trade-off can be phased. A hospital can protect its most critical systems first, measured by the clinical order of restoration, and extend protection as budgets allow. What it should avoid is spreading modest protection evenly across everything, which leaves the systems that matter most no better protected than the rest.

The shared supplier: when the victim serves many hospitals

Two of the six health care-related victims named in the indictment were not hospitals. One was Laboratory Corporation of America Holdings (LabCorp). The other was Allscripts Healthcare Solutions. When an organization that serves many providers is hit, the effect reaches every provider that depends on it. For those customers, the attacker’s choice of door, hour and backups was made at someone else’s organization.

That dependence is the fourth exposure, and it differs in kind from the first three. A hospital can harden its own remote access, staff its own night shift and protect its own backups. It cannot do any of those things for its vendor. What it can do is know which vendors it depends on, understand how it would operate without them, and use its contracts to get the information and commitments it needs.

Mapping clinical dependence on vendors

The first step is a map. For each critical clinical process, such as ordering and resulting laboratory tests, prescribing, scheduling or billing, the hospital should know which external services it relies on. That includes hosted electronic health record services, reference laboratories, imaging reading services, and clearinghouses. The map should show which dependencies have a workable alternative and which do not.

The map changes downtime planning. A hospital’s downtime procedures may assume its own systems are down. When the outage is at a vendor, the hospital’s systems may be running but unable to exchange data. Laboratory orders may need to go by courier and phone. Results may arrive by fax and need manual entry. Those procedures need to be written for each critical vendor, as well as for internal outages.

Laboratories and the order-result loop

Laboratory dependence illustrates the supplier problem well. A reference laboratory receives orders from many hospitals and clinics and returns results to them. If the laboratory’s systems go down, orders may still be collected but results cannot flow back electronically. Clinicians then need another way to get results, and critical values need a reliable path to the right person.

The downtime plan for a laboratory outage should say how specimens will be sent and tracked, how results will be received and entered, and who will call clinicians with critical results. It should say how the hospital will reconcile records once the interface returns, so that results received on paper are not missed or entered twice. Where one vendor supports several critical processes, a single incident can trigger several of these procedures at once, and the plan should say who coordinates them. These are clinical safety questions. They belong with the clinical leaders who own laboratory services as much as with information technology.

What a contract can require

HIPAA already requires a covered entity to obtain satisfactory assurances from a business associate that it will safeguard protected health information. The 2025 proposal went further on one point that matters here. It would require a business associate to notify the covered entity, without unreasonable delay and no later than 24 hours, after activating its contingency plan. HHS explained that prompt notice could let the covered entity protect its own systems and activate its own plan.

Hospitals need not wait for a final rule to get that commitment. A contract can require prompt notice when a vendor activates incident response or contingency plans affecting the hospital’s services. It can require the vendor to describe its backup separation and restoration testing. It can set expectations for how the vendor will help the hospital reconnect safely after an incident, including what evidence the hospital will receive that the vendor’s environment is clean.

Reconnecting after a vendor’s incident

Reconnection is a decision with clinical and security sides. Clinically, the pressure is to restore the interface as soon as the vendor says it is ready. From a security view, the hospital needs some assurance that the vendor’s environment is no longer compromised and that the connection will not carry the attack across. A hospital that has agreed in advance what evidence it will accept, such as a summary from the vendor’s incident responders, can make that decision faster and with less argument.

Differences between hospitals and ambulatory networks

The supplier exposure is sharpest for organizations that outsource most of their technology. A physician group or ambulatory network that uses a hosted record system and a reference laboratory may have few systems of its own to protect, and therefore depends almost entirely on its vendors’ defenses. A large health system may host its own core systems but still depend on outside laboratories and service providers for parts of care. Each should size its vendor oversight to its dependence, not to its own internal footprint.

The supplier is also a door

The first and fourth exposures meet where vendors hold remote access into a hospital. A vendor whose own network is compromised may become a path into its customers. That is why the alert’s advice on third-party remote access belongs in vendor contracts as well as in technical configuration. The hospital’s map of vendors should show which ones hold access in, which ones hold the hospital’s data out, and which do both.

Objections: old news, a solved problem, or the wrong focus

One objection is that the SamSam pattern was already familiar to specialists by late 2018, so the indictment told hospitals nothing new. The alert itself traced RDP-based access back to mid-2016. Security teams that followed threat reporting would have recognized the method.

That is fair for specialists and less so for hospital leadership. An indictment and a joint alert made an official, detailed description of the method public, with hospital names attached, where executives, boards and auditors could read it. Its value was in making the pattern hard to dismiss as hypothetical. A health system that had not yet funded the controls in this article had a concrete, public reason to do so.

The objection that the indictment changed nothing

A second objection is that indicting people in Iran had little practical effect, since the defendants were unlikely to stand trial in the United States. On this reading, law enforcement action is theater and hospitals should ignore it.

Whatever became of the criminal case, the actions of that week had effects beyond the courtroom. Treasury’s designation and its first listing of digital currency addresses gave compliance teams identifiers to screen. The DHS and FBI alert gave defenders specific technical guidance. For a hospital, the useful part of the week was that guidance, which did not depend on any defendant’s arrest.

The objection that the focus should be on paying or not paying

A third objection is that the central hospital decision in ransomware is whether to pay, and that this article neglects it. That decision matters, and Treasury’s action that week, which designated two people who exchanged SamSam ransom proceeds, brought sanctions into it. But the decision is shaped by everything before it. A hospital with closed doors, a night watch, protected backups and mapped vendors is, in our view, less likely to face the decision at all, or faces it with real alternatives. One without them faces it on the attacker’s terms.

What we would tell a hospital CIO today

The four exposures give a simple structure for a board report. For the door, report the number of internet-reachable remote access paths, the share with multi-factor authentication, and the date of the last external scan. For the hour, report time from alert to human decision by time of day and the date of the last overnight drill. For the backups, report the measured restoration time of each critical system from protected copies. For the supplier, report the critical vendors with tested downtime procedures and contract notice terms.

Each figure is something a board can follow over time. Each also has a single owner who can explain a change. A hospital that can produce all four has turned the SamSam indictment’s description of an attacker into a description of its own readiness.

What would change the argument

The argument would be weaker if later ransomware had moved away from human operators toward purely automated spread, making the attacker’s choices irrelevant. The federal guidance that followed points the other way. CISA’s 2023 guide describes actors who exfiltrate data and threaten to release it, sometimes without encrypting anything. Those are choices too. A defense designed around an operator who picks the door, the hour and the recovery target remains the right design.

With hindsight

Federal guidance after SamSam kept describing the same pattern. CISA’s #StopRansomware Guide was first released jointly with the Multi-State Information Sharing and Analysis Center in September 2020 and updated in September 2023. It still told organizations not to expose RDP to the internet, to audit and log remote access and to require multifactor authentication. Its backup advice became more specific: offline, encrypted backups, regular restoration tests, golden images and offline copies of infrastructure templates. It also described double extortion, in which actors steal data and threaten to publish it, and noted that some actors now extort without encrypting at all.

Health regulation moved more slowly. On January 6, 2025, HHS’s Office for Civil Rights published a proposed rule to strengthen the HIPAA Security Rule. It would apply to covered entities and their business associates, which the proposal calls regulated entities. It cited the HHS 405(d) program’s Health Industry Cybersecurity Practices and HHS’s Cybersecurity Performance Goals among the guidance it had considered. It proposed a written technology asset inventory and network map, and procedures to restore critical systems and data within 72 hours. It also proposed a new technical safeguard for data backup and recovery, and a specific requirement to deploy multi-factor authentication. It also proposed that business associates notify covered entities within 24 hours of activating a contingency plan. Among the inputs it described were letters from the National Committee on Vital and Health Statistics recommending practices that included multi-factor authentication and offline backups. HHS said its proposals responded to changes in how health care is delivered, significant increases in breaches and cyberattacks, and common deficiencies OCR had observed in its investigations. It also observed that a small or rural provider might choose a cloud-based record system to reduce its own need to invest in data storage, backup systems and staff. That observation describes the same trade the supplier section of this article examines: less of the hospital’s own door, hour and backups, and more dependence on a vendor’s.

As of November 14, 2026, that rule had not been finalized. The 2026 Unified Agenda listed it among long-term actions, with final action projected for July 2027. That projection is the agency’s, and it may move. Until a final rule appears, the current Security Rule’s backup, disaster recovery and emergency mode requirements remain the binding baseline.

Treasury’s November 2018 action broke new ground for OFAC. OFAC publicly attributed digital currency addresses to designated persons for the first time and said compliance obligations were the same in digital and traditional currency. Treasury said the addresses should help the compliance and digital currency communities identify transactions and funds that must be blocked. Treasury also warned that persons who engage in transactions with the designated individuals could be subject to secondary sanctions.

What held up is the shape of the threat the indictment described: human operators who choose their entry, timing and recovery targets. What did not hold up was any hope that naming the attackers would change hospitals’ exposure. The controls that matter are still the ones the December 2018 alert listed, and the proposed HIPAA changes would make several of them explicit requirements if finalized.

Spotted an error? Tell us and we will check it and correct the article where needed.

  • ransomware
  • SamSam
  • remote access
  • backups
  • HIPAA Security Rule
  • third-party risk