
On the night of October 11, 2026, attackers pushed refrigerators in hundreds of French supermarkets to as high as 50°C. Les Mousquetaires said about 150 Intermarché stores were hit, with units held at 50°C for six hours, and that it threw the food away. Carrefour reported 80 affected stores. The Paris prosecutor described a large-scale attack with food losses and fires.
At the time of writing, investigators had not published how the attackers got in. Danfoss said the affected units "were running outdated firmware" and advised customers to migrate legacy systems. France's CERT-FR named Danfoss AK-SC and AK-SM, Copeland xweb and Dixell xweb controllers in its alert.
Read CERT-FR's recommendations as a gap list and they sort into three horizons:
One more line in the alert deserves your attention. CERT-FR tells operators to check their insurance contract terms before cutting the internet link to their refrigeration service provider. The fastest fix on the list depends on a contract clause someone signed years ago.
That is gap management in miniature. Here is how to run it across your own sites.
Your risk assessment produces findings. Sort each one by when it can be fixed, not only by how severe it is:
Give every gap one of three labels. High/Now means you act this month. Medium/Next means you schedule it, often into the next maintenance cycle, and apply a compensating control in the meantime. Low/Never means you accept it.
Dragos uses a similar Now, Next, Never triage for ICS vulnerabilities. In its 2026 Year in Review, it rated 3% of the vulnerabilities it analyzed in 2025 as Now, 71% as able to wait for compensating controls or the next maintenance cycle, and 27% as not warranting remediation. Your mix will differ, but the shape is typical: a few urgent items, a long middle and a real "never" pile. If you want a structured way to assign the labels, CISA's SSVC decision tree weighs exploitation status, technical impact, automatability, mission prevalence and public well-being impact rather than a severity score alone.
"Never" is a decision, not neglect. Write down the reason, who accepted it and when you will review it.
Annualized loss expectancy gives you a common unit for decisions: the cost of one loss event multiplied by how often you expect it per year (ALE = SLE x ARO).
A hypothetical: a cold storage loss at a distribution centre would cost $400,000 in stock, cleanup and downtime. You estimate one such event every 20 years, an annual rate of 0.05. The ALE is $20,000. A $5,000 annual cost to bring the controllers behind the firewall and onto supported firmware is easy to justify. A $250,000 replacement project, on this gap alone, is not.
Treat the frequency as an estimate, not a forecast. Reliable OT incident data is scarce. Use ALE to compare options against each other. Remember that consequence reduction is often the cheapest option on the list: Germany's BSI, responding to the French attacks, told operators to cut direct internet access to their controllers. An independent temperature sensor with its own alarm is a cheap way to limit the damage if that fails.
Handle the near-term, low-effort, low-cost items first. Then stop them coming back. Amend your service and maintenance contracts so the next technician:
The gaps you cannot fix with maintenance belong in the plans for equipment you are already going to replace. When a refrigeration rack, a BAS or a chiller plant's controls come up for renewal, put security requirements in the specification: supported firmware for the asset's service life, no direct internet exposure, and components assessed against ISA/IEC 62443-4-2. The trade-off is time. You will live with some gaps for years, so keep a compensating control on each one until the capital work arrives.
Track two numbers: how many High/Now items are open, and the age of the oldest. A gap register that only grows becomes the evidence file for the next incident report.
It is sorting the findings from your risk assessment by when they can be fixed, labelling each as High/Now, Medium/Next or Low/Never, putting a number on the options, and then building the fixes into contracts and capital programs.
Now means you act this month. Next means you schedule it, often into the next maintenance cycle, with a compensating control in the meantime. Never means you accept it, and you write down the reason, who accepted it and when you will review it.
Use annualized loss expectancy: the cost of one loss event multiplied by how often you expect it per year. Treat the frequency as an estimate, and use the result to compare options against each other.
Planning a new smart building, upgrading the systems in an existing one, or buying technology for a facility? Actimeta is an OT cybersecurity and technology advisory firm that helps owners get the design, the vendors and the security right from the start. Our building assessment service gives you clear remediation plans for the gaps you find. Our procurement advisory services help you write cybersecurity expectations into your RFPs and compare vendors before you buy. Our project management team verifies the work through commissioning. Book a free consultation to talk through your project.