Louisiana’s ransomware outage was mostly its own decision
When ransomware hit Louisiana’s state servers in November 2019, officials said most of the disruption came from their own decision to shut systems down. Containment worked. The harder planning was for running services while dark and holding residents harmless afterward.
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
When ransomware hit Louisiana’s state servers in November 2019, officials said most of the disruption came from their own decision to shut systems down. Containment worked. The harder planning was for running services while dark and holding residents harmless afterward.
- The week Louisiana had already been through one cyber emergency in 2019.
- Going dark: the containment decision The decisive act that Monday was a decision to take services offline across state government, quickly, on incomplete information.
- Running dark: services while systems are down Once systems were off, the state still had residents who needed driver’s licenses, unemployment benefits and tax services.
- Holding the public harmless: the administrative side The governor’s November 22 declaration addressed a consequence that a technical response plan can easily miss.
- Coming back: restoration and proof Online services and email started to come back that Monday afternoon, hours after the attack.
- The case against going dark Three objections to a broad shutdown deserve an answer.
- With hindsight Several of the federal programs that now shape state and local cyber response did not exist in November 2019.
The week
Louisiana had already been through one cyber emergency in 2019. On July 24, the governor issued a statewide emergency declaration after malware hit several local government agencies, including school systems in Sabine, Morehouse and Ouachita parishes in the north of the state. The declaration made state resources available, including cybersecurity experts from the Louisiana National Guard, Louisiana State Police and the state Office of Technology Services (OTS). The Governor’s Office of Homeland Security and Emergency Preparedness (GOHSEP) activated its Crisis Action Team and Emergency Support Function 17, a newly created function for cybersecurity. The governor’s office said it was the first activation of that function. The state was coordinating with the FBI, state agencies and higher education partners. The governor had also established the Louisiana Cybersecurity Commission in 2017, an advisory partnership of public sector, industry, academic and law enforcement experts.
Early on the Monday morning of this week, according to the governor’s office, attackers struck the state government itself with an attempted ransomware attack on state servers. OTS responded by taking state websites and many online government services offline to stop the infection from spreading. The state activated ESF-17, the cybersecurity response team, again.
On Tuesday, November 19, the governor’s office reported that state websites and many online services were available again. Online services and email had started to come back on Monday afternoon, the day of the attack, though full restoration could take several days. Other services would remain limited as internet access was restored to agencies and workstations were repaired. The release said Office of Motor Vehicles offices would be closed until noon, and a later line said they would stay closed for the rest of the day. Unemployment insurance claims could be delayed until later in the day. People whose business could not be done online were told to call the agency directly.
The state was explicit about the cause of the disruption. Its release said the service interruption the public and employees saw was largely due to OTS’s aggressive response to prevent further infection of state servers, and not due to the attack itself. The Commissioner of Administration made the same point in the release. OTS said the ransomware was similar to the ransomware that had targeted local school districts and government entities across the country that summer. The state said it anticipated no data loss and had not paid a ransom, and that OTS was expanding its security presence through both systems and training. Louisiana State Police and several federal agencies were investigating.
The ESF-17 team, according to the governor’s office, brought together leaders from OTS, GOHSEP, State Police, the National Guard, state university systems and other cybersecurity experts.
On Friday, November 22, the governor declared a state of emergency. The declaration covered the Office of Motor Vehicles, the Department of Transportation and Development and the Department of Revenue. It let their heads take certain actions, including waiving fees and fines, to help members of the public affected by the service interruption. The governor said some people might have missed filing deadlines or incurred fees because of the outage, and that the declaration let the state correct this so the public would not be penalized unnecessarily. By then, the office said, ESF-17 had worked with agencies to restore access to most of their services.
A ransomware incident is easy to measure by what the attackers did. Louisiana’s own account of its November 2019 incident measured it differently. The state said the disruption its residents saw was largely the result of its own decision to take systems down, not of the ransomware. The statement read less like a confession than a description of a containment strategy that worked as intended, at a price paid in closed motor vehicle offices, delayed unemployment filings and missed deadlines.
Seen that way, the week was an exercise in operations more than in technology. Containment by shutdown is a choice to trade service availability for safety. Making that trade well depends on decisions taken long before an attack and on work that continues long after it. This article examines four of them, in the order an incident forces them. The first is going dark: who decides to shut systems down, on what criteria and how fast. The second is running dark: how services continue for the public and staff while systems are off. The third is holding the public harmless: the legal and administrative steps that keep residents from paying for the outage. The fourth is coming back: restoring service in the right order and proving the environment is clean. A fifth section weighs the case against shutting down at all.
The four parts have different owners. The first belongs to the security and technology leadership, with executive backing. The second belongs to agency program managers and service delivery staff. The third belongs to the governor’s office, agency heads and counsel. The fourth returns to technology, with program owners setting the priorities. An incident plan that covers only the first and fourth leaves the public-facing half of the response to improvisation.
Going dark: the containment decision
The decisive act that Monday was a decision to take services offline across state government, quickly, on incomplete information. That decision is easy to praise afterward and hard to make in the moment.
Who can pull the plug
Shutting down websites, online services and network access across many agencies affects programs that belong to different elected and appointed officials. If each agency head had to agree before their systems went dark, the delay would give ransomware time to spread. The state’s response depended on a central technology office being able to act across agencies at once.
That authority has to be settled in advance and in writing. It should say who may order a broad shutdown, for which systems, and under what conditions. It should name a deputy for when that person cannot be reached. And it should say how agency heads are told, because they will be the ones answering the public’s questions within the hour. The incident response plan should be approved at the top of the organization. CISA’s current ransomware guide recommends that the chief executive or equivalent review and approve the response and communications plans in writing.
Tiers of isolation
A shutdown need not be all or nothing. A sound plan describes tiers. The smallest is isolating a single host. Next comes isolating a network segment or an agency. Then cutting the network’s connection to the internet, and finally taking services down across the whole enterprise. Each tier has a different service cost and should need a different level of authority.
Tiers let responders act quickly at the low end without waiting for executives, while reserving the broad decisions for the people who will answer for them. They also make after-action review clearer, because the record shows which tier was invoked, when and why. An organization that has only one option, everything off, risks using it too often or hesitating too long.
What triggers the decision
Authority without criteria invites either paralysis or overreaction. A shutdown playbook should describe the signs that justify it. Examples include ransomware activity on more than one server, evidence that an attacker holds administrator credentials, or encryption spreading faster than responders can isolate individual hosts. The criteria do not have to be precise. They need to be agreed in advance, so that the person deciding is applying a rule rather than inventing one.
CISA’s #StopRansomware guide, revised in October 2023, describes the logic. Its first response step is to determine which systems were affected and isolate them immediately. If several systems or subnets appear affected, it advises taking the network offline at the switch level, because disconnecting individual systems during an incident may not be feasible. It also advises prioritizing the isolation of critical systems essential to daily operations.
The evidence trade-off
Shutting down has a cost beyond service availability. The same CISA guide says that devices should be powered down only if they cannot be disconnected from the network, because powering down destroys evidence stored in volatile memory. Network isolation keeps machines running for forensic work while cutting off the attacker’s spread. A playbook should prefer isolation at the network layer, reserve power-off for machines that cannot be isolated any other way, and tell staff which is which.
That distinction matters most for the people who carry out the decision. Help desk staff and field technicians told to shut everything down may pull the power. If the plan wants network isolation instead, those staff need to know it before the day, and they need a simple instruction they can follow under pressure.
Telling people without tipping off the attacker
The CISA guide also warns that attackers may watch an organization’s communications after a compromise. It recommends isolating systems in a coordinated way and using out-of-band methods such as phone calls, so the attackers do not learn they have been discovered and rush to deploy ransomware more widely. A state that coordinates its response over the same email system the attacker may be reading risks prompting exactly the spread it is trying to stop.
That Monday, email itself was among the services unavailable. Agencies need a contact tree that works without the enterprise email system and without the network: personal phones, a text alert service hosted elsewhere, or a call bridge. Building that list is dull work. It is also the first thing responders reach for in the first hour.
Rehearsing the decision itself
An exercise that tests only technical recovery leaves out the decision to shut down. A useful exercise puts the actual decision-makers in a room, or on a call, with an ambiguous picture and a clock. It measures how long they take to decide and what information they ask for. It also shows whether agency heads accept a decision made centrally, or call to argue.
The results should change the plan. If executives consistently ask for the same piece of information before deciding, the monitoring team should be able to produce it quickly. If decisions stall waiting for one person, the deputy arrangement needs work. Rehearsal turns authority on paper into authority that can be used at 4 a.m.
Law enforcement and federal partners
Louisiana State Police and several federal agencies investigated the November attack. Bringing investigators in quickly has operational consequences. They will want evidence preserved, which affects how machines are isolated and when they are rebuilt. They may also have information about the attackers’ methods from other victims, which can speed containment.
The contacts and the evidence handling rules should be agreed before an incident. A plan should say who calls which agency, what responders must preserve, and who decides when a system can be wiped and rebuilt. Without those rules, the restoration team and the investigators can work against each other in the first days, each with a legitimate goal.
Why the summer mattered
Louisiana had practiced this before November, though not by choice. The July emergency activated ESF-17 for the first time, to help local school systems. The November response activated it again, this time for the state’s own servers. An organization that has run its response structure once, for real, brings to the second incident a set of contacts, roles and habits that no tabletop exercise fully reproduces. The state’s own account credited its earlier investment in cybersecurity capability, including the creation of ESF-17, for the speed of the response.
For other governments, the lesson of the sequence is that response structures should be used, in exercises or in support of smaller incidents, before they are needed for the big one. A response team convened for the first time during a statewide incident is assembling while it responds.
Running dark: services while systems are down
Once systems were off, the state still had residents who needed driver’s licenses, unemployment benefits and tax services. The governor’s office told people whose business could not be done online to call the agency directly. A phone number is the minimum. A plan for running dark needs more structure than a phone number.
Knowing which services matter first
A shutdown does not affect all services equally. Some can wait a day with little harm, such as routine license renewals. Others carry deadlines with legal or financial consequences, such as benefit claims and tax filings. A few touch health and safety. The CISA guide advises prioritizing restoration based on a predefined list of critical assets, including systems critical for health and safety, revenue generation or other critical services, and the systems they depend on.
That list is equally useful for running dark. It tells program managers which services need a manual fallback ready to start within hours, which can simply pause, and which need public notice that deadlines will be adjusted. Building it requires program managers and technology staff to agree on dependencies, which may not be documented in one place.
Manual fallbacks
A manual fallback is a way of continuing a service without the system that normally supports it. For a benefits program it might be a paper or phone intake that records the claim date, so the claimant is not disadvantaged by the delay. For a motor vehicle office it might be a process for issuing temporary documents or recording transactions for later entry. For a revenue agency it might be a rule that filings received during the outage are treated as timely.
Fallbacks need forms, instructions, trained staff and a way to reconcile the manual records once systems return. Most of that can be prepared in advance and stored where it can be reached without the network. The CISA guide recommends keeping a hard copy and an offline version of the incident response plan itself, and the same logic applies to the forms and procedures that let services continue.
Reconciling the manual records
Every manual transaction taken while systems were down has to be entered later. That back-entry is a hidden second outage: staff who should be serving new customers are re-keying old ones. It is also a source of errors, because manual records are incomplete or hard to read. A plan should say who does the back-entry, in what order, how it is checked and how long it is expected to take. Simple forms designed for later entry, with the fields the system needs in the order it needs them, shorten this step considerably.
Deadline-bound programs
The release singled out unemployment insurance claims as possibly delayed. Programs like this run on schedules set in law or regulation, and a missed step can delay a payment that a household depends on. For such programs, the fallback must at least record that the resident tried to act on time. A timestamped call log, a paper intake form or a simple online form hosted elsewhere can preserve that record until the main system returns.
Program managers know which steps in their programs carry deadlines. Technology teams know which systems support each step. The fallback design needs both in the same conversation, which is why it belongs in continuity planning rather than in the technical incident plan alone.
Call centers under surge
Telling the public to call agencies directly moves the load to call centers at the moment they may have reduced capacity themselves. Phone systems that run over the data network may be affected. Agents may lack access to the records they need to answer questions. Call volumes will rise as residents find services unavailable.
A plan for running dark should include call center scripts for an outage, a way to add staff from other units, and a decision on what agents can and cannot do without system access. Even a scripted answer that confirms the outage, explains the relief available and records the caller’s details reduces repeat calls and complaints.
The workforce while workstations are repaired
The November 19 release noted that services would be limited as workstations were repaired. Every affected workstation has to be checked or rebuilt before it rejoins the network. During that time, affected employees may not be able to do their normal jobs.
Planning for that period is a workforce question. Which staff can work from unaffected devices, which can be reassigned to manual processes or public contact, and which must wait. Who tells them, and how, when email is down. How long can a field office stay closed before staff need different instructions. A plan that answers these questions keeps employees useful and gives the public more open doors.
The responders themselves
The governor’s office described the response team as working around the clock. Incidents of this kind put a small number of specialists under heavy load for days. Fatigue produces mistakes, and mistakes during restoration can reopen the door the team has just closed.
A workforce plan for incident response sets shift rotations from the first night, identifies who can relieve the core team, and settles overtime and compensatory time rules for public employees in advance. Mutual aid arrangements matter here too. Louisiana’s response structure drew on the National Guard, State Police and university systems, which widened the pool of skilled people available. Smaller governments can arrange similar support through state programs or regional partners before they need it.
Talking to the public
The state’s websites were themselves among the services taken down. The first public channel a government reaches for, its own website, may be unavailable at the moment it is most needed. Louisiana used press releases and statements by officials to explain what was available and what was not.
A plan should provide channels that do not depend on the affected network. Examples include a status page hosted on separate infrastructure, social media accounts with credentials stored offline, and pre-written messages for common situations such as closed offices or delayed claims. The messages should lead with what residents should do now, and only then explain what has gone wrong.
Dependent local governments
State systems may serve local governments as well. Parish offices, courts and local agencies may rely on state applications and networks for licensing, records or benefits. A state shutdown therefore affects services the state does not directly run. A complete plan includes a way to tell local partners what is down, what they should do, and when to expect restoration. It should also say which local systems connect to the state network and must be isolated with it.
Holding the public harmless: the administrative side
The governor’s November 22 declaration addressed a consequence that a technical response plan can easily miss. Residents who could not complete a transaction during the outage might have missed a deadline or incurred a late fee. Without action, the public would bear part of the cost of the state’s own containment decision.
Using emergency powers for an administrative problem
The declaration named three agencies: the Office of Motor Vehicles, the Department of Transportation and Development and the Department of Revenue. It allowed their heads to take certain actions to help affected members of the public. According to the governor’s office, the purpose was to correct missed deadlines and fees so that people were not penalized. The emergency declaration power was used here to change routine administrative consequences.
That use of emergency authority is worth planning for. Counsel can identify in advance which deadlines and fees can be waived under existing law, which need an emergency declaration and which cannot be changed without legislation. Draft declaration language for a cyber outage can sit alongside the drafts already kept for natural disasters, so that the step is quick when needed.
Finding the people who were affected
Waiving a penalty requires knowing who incurred it or would have. That is a data problem. An agency needs to identify transactions that were due during the outage window, applications that were started but not completed, and fees assessed because of the delay. If transaction logs or appointment records were themselves affected, the agency may need to rely on residents coming forward.
Two practical steps make this easier. The first is to define the outage window formally, with start and end times for each service, so that relief applies consistently. The second is to keep the logs that show attempted transactions in a place that survives an incident, so the agency can find affected residents rather than wait for them.
Limits on what a state can waive
Not every consequence of an outage is within a governor’s power to change. Some deadlines are set by federal law or by courts, and some fees belong to bodies outside the executive branch. A state declaration can address the deadlines and fees its own agencies control. Others may need a request to a federal agency, a court order or a decision by another body. Counsel should map these limits in advance, so that the public is told accurately what relief is available and from whom.
Telling residents relief exists
Relief that residents do not know about is of little use. Some may pay a late fee rather than ask for a waiver. Others may give up on a transaction. The administrative response should include a public notice that sets out which deadlines are extended, which fees are waived, for which period and how to claim relief if it is not applied automatically. The same notice should go to call center staff and field offices, so that the answers residents get are consistent.
Fairness across channels
Relief should also be consistent across the ways residents interact with government. Someone who tried to file online, someone who waited at a closed office and someone who called and could not get through all suffered the same delay. A relief policy that covers only online transactions, because those are easiest to identify, would treat those residents differently. The administrative response should be designed around the resident’s experience, not the system’s logs alone.
Who pays
Waived fees and extended deadlines have a fiscal cost that falls on the agencies whose revenue depends on them. Budget offices should know in advance that a cyber outage can reduce or delay revenue, just as a natural disaster can. Including that possibility in continuity planning avoids a later argument about whether relief was affordable.
Coming back: restoration and proof
Online services and email started to come back that Monday afternoon, hours after the attack. By the next day the state said websites and many online services were available again, and that full restoration could take several days. Restoration is where the containment decision is repaid, and where a second set of choices begins.
Restoration order
Not everything can come back at once, and not everything should. The order of restoration should follow the critical asset list from the running-dark section, adjusted for dependencies. Identity services, network infrastructure and security tooling have to return before the applications that depend on them. Public-facing services that carry deadlines should come next. The CISA guide also advises keeping track of systems not believed to be affected, so they can be given lower priority and the organization can return to business more efficiently.
A clear order helps program managers as much as technical staff. If the motor vehicle office knows its systems will return after the core network and identity services but before back-office reporting, it can plan staffing and public notices around that sequence.
Workstations at scale
The November 19 release tied the return of some services to the repair of workstations. Checking or rebuilding thousands of desktops and laptops is a logistics task as much as a technical one. It needs clean images ready to deploy and a way to track which machines are done. It needs technicians or remote tools that can reach field offices, and a rule for machines that cannot be checked quickly. Agencies with standard images and central device management can move much faster than those with many unique machines. The difference only shows during an incident, which is why standardization is worth defending in calmer times.
Proving the environment is clean
Bringing a system back too early risks reinfection. Each server and workstation needs to be checked or rebuilt, and the means of initial access identified and closed. The CISA guide advises examining existing detection and prevention tools and logs for evidence of other systems or malware involved in earlier stages of the attack. It notes that a ransomware event may be evidence of a previous, unresolved compromise.
This is the part of the response where the earlier evidence decision pays off or costs. If machines were isolated rather than powered off, investigators have more to work with. If they were wiped quickly to restore service, the organization may restore service without ever learning how the attackers got in.
Resetting credentials
When ransomware reaches many servers, responders should assume the attackers held administrator credentials. Restoration may therefore include resetting privileged accounts, service accounts used by applications and every user’s password. Each of those steps has an operational cost. Service accounts that are reset without updating the applications that use them break those applications. Mass password resets send large numbers of staff to the help desk on the same morning.
Planning for this means knowing where service account credentials are used, having a way to issue new passwords to staff who cannot reach email, and staffing the help desk for the surge. A state that has to repair workstations and reset credentials at the same time faces the busiest help desk days of the year.
Claims about data
The state said it anticipated no data loss and had not paid a ransom. Statements like these are important to residents, and they need to be supportable. An organization should be careful to distinguish data lost through encryption from data taken by the attackers, and to say what evidence supports each claim. The CISA guide now treats data extortion alongside encryption, and notes that attackers sometimes copy data and then threaten to release it. A statement that no data was lost does not by itself address whether any data was copied.
Restoring trust in public websites
When a government’s websites go dark during a cyber incident and then return, residents may reasonably wonder whether the restored sites are safe to use. A short notice on the restored sites, explaining what happened and confirming that services are available again, helps. So does restoring the most-used transactional services with care, because a second outage soon after the first would cost more trust than a slower return.
Hardening after the incident
OTS said it was expanding its security presence through systems and training. Done well, it targets the specific weaknesses the investigation found. The CISA guide lists preventive measures that apply broadly, such as maintaining offline, encrypted backups of critical data and testing them regularly in a disaster recovery scenario. Backups that the attacker cannot reach determine whether restoration is measured in days or weeks.
The after-action record
An after-action review should capture the decisions as well as the technical events. It should record when each tier of isolation was invoked, by whom and on what information. It should list which services were affected and for how long, which fallbacks worked and what relief was granted. That record serves three purposes. It improves the plan. It supports any later legal or audit review. And it gives the next set of decision-makers a precedent to work from.
Measuring the cost of the shutdown
Containment decisions are easier to defend when their costs are measured. Useful measures include hours of unavailability for each public service and the number of transactions delayed or moved to manual processes. Others are the value of fees waived, the deadlines extended, the time from detection to isolation and the time to restore each tier of service. Over several incidents and exercises, these figures show whether the organization is getting faster and whether its shutdowns are proportionate.
Differences by type of government
A state with a central technology office and an emergency support function can shut down and restore across agencies with one command structure. A county or city with a small IT team may lack both the authority and the staff to act across departments at once. A school district may depend on a vendor or a regional service center. For those organizations, the state’s July declaration offered a model: state resources, including the National Guard and State Police, made available to local governments under an emergency declaration. Local governments should know in advance how to request that help and what it can provide.
The case against going dark
Three objections to a broad shutdown deserve an answer.
The cure was worse than the disease
The first is that the state’s own action caused more disruption than the attack. By its own account, that is literally true for the service interruption residents saw. But the comparison is with an attack that was contained, not with one left to spread. The state’s November 22 release said its protective actions likely saved it from data loss and weeks of service outages. The day of closed offices was the price of making the attack small.
Shutdowns teach attackers to move faster
The second is that attackers who expect a shutdown will move faster, encrypting or stealing data before defenders can react. That is a real dynamic, and it is one reason the CISA guide stresses coordinated, out-of-band isolation. It argues for faster detection and better-prepared shutdown procedures, not for leaving systems running.
Shutting down your own network does not stop every attack
The third objection is the strongest and has grown stronger since 2019. Later incidents, including one in Louisiana described in the hindsight, exploited flaws in third-party software before its maker or its customers knew of them. Isolating the network once the flaw is known does nothing about data already taken. The hindsight section describes how this played out in Louisiana itself. The objection does not undermine the November 2019 decision, which addressed the threat in front of the state. It does limit how far that decision generalizes.
Central authority dilutes agency accountability
A further objection is that giving a central technology office the power to shut down every agency’s services weakens the accountability of agency heads for their own programs. An agency head answers to the public for a closed office, yet did not decide to close it. That tension is real. The answer is to agree the criteria and tiers with agency heads in advance, to inform them immediately when they are used, and to review each use with them afterward. Central authority works when it is exercised under rules the agencies helped write.
Where the argument is weakest
This article’s argument depends on the shutdown being well judged. A government that shuts down too readily, on weak signals, imposes real costs on residents for little benefit. Criteria agreed in advance, reviewed after each use, are the safeguard against that. The account the state gave publicly does not set out the criteria it used, so outsiders cannot judge how close the decision was. What can be said is that the state described its own actions candidly, and that candor is itself part of a sound response.
With hindsight
Several of the federal programs that now shape state and local cyber response did not exist in November 2019.
On September 16, 2022, the Department of Homeland Security announced the State and Local Cybersecurity Grant Program, which CISA describes as a first-of-its-kind cybersecurity grant program specifically for state, local, tribal and territorial governments. CISA says Congress established it and appropriated $1 billion to be distributed over four years. CISA says the program was designed to direct funding to the most vulnerable and least mature cyber entities, local governments, with states receiving funds through their State Administrative Agencies. DHS announced $91.7 million for the program’s fiscal year 2025 round. The funds were meant to help address risks to information systems owned or operated by, or on behalf of, those governments. The statute set the section’s requirements to terminate on September 30, 2025. The Consolidated Appropriations Act, 2026 (Pub. L. 119-75), signed on February 3, 2026, moved that date to September 30, 2026, and no later law we found moved it again, so the section’s requirements ended on September 30, 2026. CISA manages the program’s cybersecurity content, and the Federal Emergency Management Agency administers the grants.
The Cyber Incident Reporting for Critical Infrastructure Act of 2022, known as CIRCIA, requires CISA to write regulations requiring covered entities to report covered cyber incidents and ransomware payments. CISA published its proposed rule on April 4, 2024, and the comment period ran until July 3. It held four town hall meetings between June 15 and June 18, 2026. As checked on November 14, 2026, CISA’s CIRCIA page said that funding lapses had affected its ability to carry out the rulemaking and that it continued to work on the final rule. Reporting will not be required until the final rule takes effect.
CIRCIA also established the Joint Ransomware Task Force, through which CISA and its partners developed the #StopRansomware guide. Its October 2023 revision reads, in our view, like a checklist for incidents such as Louisiana’s. It advises isolating affected systems, taking the network offline at the switch level when several systems are hit, and powering devices down only when they cannot be disconnected. It also advises restoring by a predefined critical asset list and keeping hard copies of the response plan. The guide recommends against paying ransoms and warns that payments may carry sanctions risk.
Louisiana’s next large incident took a different form. In 2023 attackers exploited a zero-day vulnerability in MOVEit, third-party file transfer software, and had access to Office of Motor Vehicles files within one MOVEit server. The state’s published answers to residents’ questions said the OMV was running MOVEit version 14.0.4, the most recent patch available at the time, and that, according to the manufacturer, every version was exploitable by the same flaw. The manufacturer alerted customers on May 31, 2023, and announced a patch on June 2, which the state said it applied immediately. On the evening of June 14 the state finished reviewing the affected files and identified records of residents who had applied for or received driver’s licenses, identification cards and vehicle registrations. It said no other servers had been identified as compromised, and it pointed residents to steps to protect against identity theft.
What held up is the operating model. CISA’s guide now calls for an incident response plan and communications plan approved in writing by the chief executive or equivalent, and for immediate isolation of affected systems. What did not hold up is the assumption that the main threat was malware spreading across the network, which isolation could stop. In 2019 motor vehicle offices closed for a day after the shutdown. In 2023 the attackers reached Office of Motor Vehicles files through a flaw in third-party software on the OMV’s own MOVEit server, before the manufacturer had a fix, and no shutdown after the fact could have undone that access. Any government should now plan for both kinds of incident. The administrative lesson has aged best of all: when residents lose access to services, someone has to decide who bears the cost, and the answer is better written before the incident than after it.
Spotted an error? Tell us and we will check it and correct the article where needed.