OT Vulnerability Management: Risk Decisions for Systems That Can’t Be Patched

What decision do you make about a vulnerability in a PLC or HMI that can’t be patched? ICS advisories, prioritization by reachability and exploitation information, a patch–mitigate–accept table and compensating controls.
Technical summary
Technology
CVE, CVSS, CISA KEV, SSVC, CSAF, ICS advisories
Topic
Vulnerability prioritization and patching decisions in OT
Level
Intermediate–advanced
Prerequisites
  • An OT asset inventory including firmware and software versions
  • A zone/cell-based network drawing
  • A change management process
In this article8 sections

In IT, vulnerability management is a “scan, prioritize, patch” cycle. In OT, this cycle gets stuck at the last step. A PLC firmware update requires a production shutdown, HMI software is supported only on the Windows version approved by the vendor, and for some devices a fixed version never comes out. In this guide we explain how the decision is made when patching isn’t possible or has to be postponed: information sources, prioritization, a decision table and compensating controls.

Why vulnerability management works differently in OT

The difference doesn’t come only from the difficulty of patching. How a vulnerability is found, its impact and its solution also change:

  • Detection: Authenticated vulnerability scanning on the OT network is often not appropriate. Matching is done by comparing the product and firmware information in the inventory with vendor advisories. If the firmware column in the inventory is empty, the vulnerability list is incomplete too; that’s why OT passive asset discovery is a prerequisite for the process.
  • Impact: The CVSS score measures technical impact, not impact on the process. The same vulnerability means low risk on a PLC on a test bench and high risk on a boiler control system.
  • Solution: A firmware update requires a maintenance shutdown, a commissioning test and sometimes OEM approval. For some products, the vendor publishes only a mitigation recommendation, or product support has ended.
  • Time: The patch may wait until the next planned shutdown; this period is measured in months. The time in between is managed with compensating controls.

The scanning, SLA and reporting cycle on the IT side is described in the vulnerability management program guide; the topic here is that program’s OT arm.

Sources of vulnerability information and vendor advisories

A single source isn’t enough for OT vulnerabilities. The main sources to follow:

  • Vendor PSIRT advisories: Security notices from vendors such as Siemens ProductCERT, Rockwell Automation, Schneider Electric and ABB. Affected versions and the vendor’s recommended mitigation steps are written here in the most detail.
  • CISA ICS Advisories: Publishes the notices of many vendors in a common format; gathers information such as remote exploitability and attack complexity on a single page.
  • CISA Known Exploited Vulnerabilities (KEV) catalog: Vulnerabilities known to be actively exploited. A strong prioritization input for OT products and for the Windows, VPN and remote access components used in OT.
  • Machine builder (OEM) and system integrator: The party that knows which versions the components inside the machine are on and how an update affects the machine warranty.

Tracking advisories by hand collapses as the number of products grows. Many vendors and CISA also publish advisories in the machine-readable CSAF format; automatic matching with the inventory can be built on this format. An advisory tracked by hand is an advisory that doesn’t get read.

ICS advisory evaluation checklist

  • Which devices in the inventory match the affected product and version range?
  • Through which interface is the vulnerability exploited: a network service, web interface, engineering protocol, local access?
  • Is authentication required, is user interaction required?
  • What’s the outcome: information disclosure, denial of service, code execution, program modification?
  • Is there public exploit code, and is it in the KEV catalog?
  • Does the vendor offer a patch, a configuration change, or only a mitigation recommendation?
  • Does the patch require a restart, a program reload or a hardware change?

Prioritization: reachability, exploitation and consequence

In OT, making the CVSS base score the only ranking criterion leads to the wrong work being done first. For example, a line is stopped for a 9.8-scored vulnerability reachable only from inside the cell, while a 7.5 on an internet-facing VPN device waits. A more useful model asks three questions together:

  1. Reachability: Who can reach the vulnerable interface? From the internet, from the IT network, only from the OT DMZ, only from inside the cell? How segmentation is actually implemented, not on paper, is decisive here. You don’t say “this PLC can’t be reached from outside” without reading the firewall rules; the common surprise is a forgotten “any” rule or a NAT opened for an OEM and never closed.
  2. Exploitability: Is there active exploitation (KEV) or public exploit code? Can exploitation be automated?
  3. Consequence: The asset’s role in the process and its criticality. Systems that carry a safety function, sit on a bottleneck line or have no backup come first.

CISA’s SSVC (Stakeholder-Specific Vulnerability Categorization) model turns this logic into a structured decision tree: exploitation status, automatability, technical impact and impact on mission/human safety are evaluated together, and the result comes out as Track, Track*, Attend or Act. For OT, the “mission and human safety” axis naturally overlaps with process consequences.

Decision table: patch, mitigate, accept

Status Recommended decision Scheduling
Active exploitation, interface reachable from IT or remotely Mitigate immediately; schedule the patch for the nearest window Mitigation urgent, patch at the first suitable shutdown
Public exploit exists, interface reachable only from the OT zone Mitigate and monitor; put the patch in a planned shutdown The next planned shutdown
No exploitation information, critical asset Put the patch in a planned shutdown; narrow access Routine maintenance schedule
No patch, or product support has ended Permanent compensating control + replacement plan Control immediately, replacement in the budget cycle
Access so limited it’s practically nil, low consequence A justified, time-limited risk acceptance Reassessment on a set date

The “mitigate” decision in the table isn’t a permanent solution replacing the patch; it’s an interim measure that keeps risk at a tolerable level until the patch. If the interim measure becomes permanent, the problem grows. A risk acceptance isn’t given indefinitely either: every acceptance has an owner and a reassessment date.

Compensating controls

If patching isn’t possible, the control is chosen according to how the vulnerability is exploited. The sentence “we did segmentation” isn’t enough; the control should concretely close the path to the relevant interface.

Exploitation path Compensating control
Network service (web interface, SNMP, FTP, etc.) If unused, disable the service; if used, allow only the required sources on the cell firewall
Engineering protocol Limit PLC access to approved engineering workstations only; enable the PLC protection level/password; monitor write operations
Remote access Remove direct access; an intermediary server, MFA and time-bound access
Windows-based HMI Application allowlisting, disabling unnecessary services, USB restrictions, limiting local administrator accounts
Denial of service Rate limiting, broadcast storm control, monitoring of critical connections

Every control must be proven to work. If a firewall rule was added, a connection is attempted from a segment not allowed to reach the vulnerable port, the block is seen and recorded. You can do this in a maintenance window even with a simple TCP connection attempt. A compensating control not verified by a penetration test or configuration audit stays on paper. Paper doesn’t stop an attacker.

Patch testing and vendor approval

Once the patching decision is made, implementation also proceeds in a controlled way:

  1. Compatibility check: HMI/SCADA vendors usually publish a compatibility list for Windows updates and antivirus products. An update not on the list may affect support coverage.
  2. OEM approval: Before updating a component inside a machine, the machine builder’s written approval is obtained; an unapproved change can affect the warranty and the builder’s responsibility for the machine’s safety functions.
  3. Testing: Applied on a test bench if possible; if not, on a backup line or a non-critical cell.
  4. Rollback plan: The current firmware, project file and system image are backed up; the rollback steps and how long they take are known in advance. A firmware update without a way back is a gamble.
  5. Commissioning: After the update, I/O, communication and safety functions are tested.

These steps are part of the change management process. For the engineering side of the change window, the OT change window guide on the industrial site gives details.

Recording decisions

OT vulnerability management being auditable within the enterprise cybersecurity program depends on the decision being kept together with its justification. For each vulnerability–asset match, the record has these fields:

  • CVE and advisory reference, affected assets and versions.
  • The reachability, exploitability and consequence assessment, and the person who assessed it.
  • The decision (patch, mitigate, accept), its justification and the approving authority.
  • The compensating controls applied and the verification evidence.
  • The planned patch date or reassessment date.

This record also shows which decisions need to be reopened when a new advisory comes out. When a vulnerability postponed for the same asset on the grounds of “no exploitation information” enters the KEV catalog, the decision should automatically fall into reassessment. If you try to remember this by hand, it gets forgotten. To set up the assessment in an OT context, the OT/ICS security service can be used, and for the scanning infrastructure on the IT side, the vulnerability analysis and scanning service. The consequence categories in prioritization should be kept on the same scale as the OT security risk assessment work; two different scales produce two different priority lists.

Frequently asked questions

Can vulnerability scanning be done on the OT network?

General-purpose active scanning can have unexpected effects on OT devices; it should be used with caution. The common method is to match the product and firmware information in the inventory with vendor advisories and, if needed, run a controlled scan in a maintenance window.

Is the CVSS score enough for prioritization in OT?

No. CVSS measures technical impact; it doesn’t take into account who can reach the vulnerable interface or the impact on the process. Reachability, active exploitation status and the process consequence should be evaluated together.

What should be done for a PLC with no patch?

Compensating controls are applied according to the vulnerability’s exploitation path: limiting access to approved sources, disabling unused services, enabling PLC protection levels and monitoring write operations. In the long term, a product replacement plan is made.

How long should a risk acceptance remain valid?

A risk acceptance isn’t indefinite. Each acceptance has an owner and a reassessment date; the decision is reopened early when the vulnerability enters the active exploitation list or the asset’s reachability changes.

Support

Contact us for support or a discovery session on this topic.

Contact form
[email protected]

Eretek Cybersecurity

Firewall, VPN, endpoint security and threat prevention. We reduce the attack surface and measure and report the risks.

Knowledge Center

Other articles