InnerWeb
Skip to main content

OT network incident at a small CHP plant: from a private APN to a joint command cell

In December 2025, according to a CERT Polska report, attackers entered the OT network of a small combined heat and power plant through the operator's private APN, switched a controller to STOP mode and shut down a steam turbine. It is the first known case of OT network entry by this route. Below is a generic use case based on those public findings: what failed, what a joint operations, security and IT command cell should look like, and how much time a single response path can save.

8 367
unique OT hosts visible from the internet in 2025
-115 min
restoration time difference in the illustrative estimate
Source: example calculation

The situation: a small CHP plant and remote access via APN

This generic use case concerns a small CHP plant in the district heating sector, running in cogeneration with a steam turbine and a water treatment line. Plant control was built from several PLCs. Remote access to the devices ran through the telecom operator's private APN, a mobile network the plant regarded as closed and trusted.

That assumption proved wrong. According to the follow-up report from CERT Polska, a single administrative device with default credentials and SSH enabled served as a bridge to three PLCs.

Background: the scale of the problem in Poland in 2025

In its 2025 Annual Report, CERT Polska recorded 8,367 unique OT hosts visible from the internet, 10,135 OT services and 22,045 OT notifications sent, 1.9 times more than a year earlier. The same report describes coordinated sabotage attacks in late December 2025 on at least 30 wind and photovoltaic farms, a manufacturing company and a large CHP plant. There was no impact on current energy production or heat supply, but there was an impact on IT systems and physical industrial devices.

How the incident unfolded

According to public accounts, the attackers were active for several days, between 18 and 29 December. The first steps were reconnaissance of the APN network and finding a device with default credentials. That device was then used to reach three controllers. On the morning of 29 December, at around 5:30, one of the controllers was switched to STOP mode, which halted the steam turbine and water treatment. The attackers deleted logs and reset network devices to make reconstruction of events harder.

  • Entry: a private APN treated as a trusted network
  • Bridge: an administrative device with default credentials and SSH enabled
  • Target: three PLCs in the control network
  • Impact: switch to STOP mode, steam turbine and water treatment stopped
  • Covering tracks: deleted logs, reset network devices

Impact on the public

The cogeneration outage was short, staff restored operation quickly and heat customers did not notice. That does not change the fact that it happened at a time of day and year when heat demand is highest and the night shift has the fewest decision-makers on site.

Time from stop to restart: illustrative estimate

Time from stop to restart: illustrative estimateBar chart comparing the stages of restoring operations in two variants: 280 minutes with separate chains and 165 minutes with a joint command cell.0255075100125minSeparate chains (Detection): 10 min10 minJoint command cell (Detection): 10 min10 minDetectionSeparate chains (Classification): 90 min90 minJoint command cell (Classification): 15 min15 minClassificationSeparate chains (Cutting access): 60 min60 minJoint command cell (Cutting access): 20 min20 minCutting accessSeparate chains (Verification and start): 120 min120 minJoint command cell (Verification and start): 120 min120 minVerification andstart
  • Separate chains
  • Joint command cell
Source: Illustrative estimate based on the assumptions in the Effect section

What failed in the response, not only in the safeguards

The technical errors are clear: default credentials, SSH enabled and trust in the APN network without additional segmentation. For incident response, the organizational questions matter more, and a plant in a similar situation must settle them within the first minutes. Is a turbine stop a failure, an operational event or a cyber incident? Who decides to cut remote access? Who makes sure no one is standing at the machine when it restarts? Who records the timeline before the logs are wiped?

In a plant without a shared response path, the answers to these questions are scattered. The operator thinks of a failure, security knows nothing about the event and IT finds out in the morning. As a result, classification as a cyber incident comes hours later, and evidence is destroyed by restarting devices.

The solution: a joint operations, security and IT command cell

A model response rests on a single command cell with one list of roles and one log. The shift manager activates the cell when a plant stop has no process explanation and an unauthorized or unexplained change has appeared in the control network.

The first 15 minutes

The shift manager classifies the event as a potential cyber incident and calls the OT administrator. The scribe opens a log entry with the time of detection. Security secures physical access to the control cabinet and establishes who was in the control room. The plant health and safety officer confirms that no one is working at the turbine.

Within 60 minutes

The CISO or OT administrator disconnects remote access through the APN, takes a copy of the configuration and locally available logs, and assesses which controllers were affected. The plant director decides on manual or emergency mode and on notifying heat customers. The 24-hour clock starts for an early warning to the CSIRT if the event meets the criteria of a significant incident.

Restart

Before a controller is switched to run mode, the team compares the program with a reference, checks who is present in the machine zone and only then gives approval. The list of people on site from access control makes it quick to confirm who is on the premises. During maintenance work, access control provides evidence of which external companies had access to the zone at the time of the event.

The effect: an illustrative restoration time estimate

The estimate below is illustrative and does not describe a real plant. It compares two variants of the same event: with separate health and safety, security and IT chains, and with a joint command cell. Assumptions, in minutes: detection by the operator 10; classification as a cyber incident 90 in the scattered variant and 15 in the joint one; cutting remote access 60 in the scattered variant and 20 in the joint one; verification of programs and zone safety 90 in both; restart 30 in both.

The total is 280 minutes in the scattered variant and 165 minutes in the joint one, a difference of 115 minutes. The moment of the CSIRT reporting decision depends chiefly on classification time, which drops from 90 to 15 minutes. The result is sensitive to the assumptions: if classification in the scattered variant takes one hour, the difference falls to 85 minutes.

Conclusions for the plant

First, treat a private APN as an untrusted network: client isolation, SSH and Telnet disabled on administrative devices, default passwords changed. Second, a register of external and subcontractor access to OT zones is part of incident response, not only of audit. Third, physical and logical access control to critical zones should give a single piece of evidence of who was in the zone and when. Fourth, a plant stop without a process explanation should trigger the incident path automatically, without waiting for morning.

Scope of InnerWeb solutions in this scenario

InnerWeb is a software producer, not a vendor of OT network security. It helps with the organizational and evidence side: event reports go to a team that takes them over and closes them, notifications go out by e-mail or SMS with a dispatch log, the list of people on site is built from gate scans, and the audit trail shows who changed a permit or an automation rule and when. Segmentation, APN isolation and network monitoring remain with the IT and OT teams and the operator.

Video

Frequently asked questions

Was this a significant incident within the meaning of the KSC Act?

The assessment belongs to the entity and the CSIRT and depends on impact and scale. In the generic case, a turbine stop in the energy sector justifies at least an early assessment for the 24-hour report under the Polish National Cybersecurity System Act (KSC).

Is a private APN secure?

Not by default. Treat it as an untrusted network, ask the operator about client isolation and do not expose administrative services on it without authentication.

Does a restart immediately restore security?

No. First the controller program is compared with a reference, remote access is cut and log copies are secured, and only then is operation resumed.

Sources

Let us build a joint incident path

We will show how reports, notifications and the list of people on site create audit evidence. Book a conversation or a demo.

Let's talk